The Circumlocution Office is Alive and Well

The Circumlocution Office is Alive and Well

Duncan Steel, 2014 May 06.

I was intending to put up a post here simply so as to enable a new thread of discussion, due to the huge volume of comments and replies that have made my preceding post unwieldy. I had failed to put up a new post for while not so much for a lack of anything to say, but rather the lack of time to say and write it whilst travelling and attending to a range of other matters; my apologies for this.

So, I was going to put up a post with a title saying simply “Here’s a new post you can file comments under” but I wanted to convey the spirit of the investigation, and therefore came up with The Circumlocution Office. This wonder (and all-too-common feature) of government had its name invented many years ago by Charles Dickens, in his book Little Dorrit, although it was by no means a new phenomenon in the middle of the nineteenth century. Dickens introduces it with this paragraph:

The Circumlocution Office was (as everybody knows without being told) the most important Department under Government. No public business of any kind could possibly be done at any time without the acquiescence of the Circumlocution Office. Its finger was in the largest public pie, and in the smallest public tart. It was equally impossible to do the plainest right and to undo the plainest wrong without the express authority of the Circumlocution Office. If another Gunpowder Plot had been discovered half an hour before the lighting of the match, nobody would have been justified in saving the parliament until there had been half a score of boards, half a bushel of minutes, several sacks of official memoranda, and a family-vault full of ungrammatical correspondence, on the part of the Circumlocution Office.

In the century-and-a-half since Dickens introduced the world to its existence, in a formal sense, The Circumlocution Office has grow’d like Topsy so as to have sub-branches and sub-sub-branches in every department of every level of government.

On Monday I sent the following email message to the Joint Agency Coordination Centre ( in Canberra, Australia (a nation of which I am a citizen):

Dear Sir or Madam,

In the mass media today there have been various reports regarding a meeting of those investigating the loss of Malaysia Airlines flight 370, to occur on Wednesday in Canberra. For example, the following appeared in The Guardian (London):

Air Chief Marshal Angus Houston, said: “We’ve got to this stage of the process where it’s very sensible to go back and have a look at all of the data that’s been gathered, all of the analysis that’s been done, and make sure that there are no flaws in that.”

Might I ensure that you are aware of the analysis by a variety of people with knowledge of both the space/satellite sector and also the avionics and communications systems, all available on my website:

As an example, on April 2nd I posted an analysis that indicated that the Inmarsat modelling of the satellite-derived information appears to be incorrect in that a northern path for MH370 cannot be excluded:

I have already discussed and critically reviewed this analysis with a wide range of colleagues, including aeronautics/aviation staff at NASA-Ames Research Center, where I work part of the year.

Duncan Steel

The following is the complete (obviously pro forma) reply that I received:

The ATSB’s MH370 search group has been established since Australia’s involvement in the search for MH370. The group initially based at AMSA is now at the ATSB.

This group has been working closely with the MH370 joint investigation team of experts and in particular, the satellite communication subgroup, who will be visiting the ATSB this week to continue the review of information that will assist in progressing future underwater search planning. The work of the groups will be ongoing in the coming weeks as the underwater search planning progresses.

The satellite communication subgroup comprises experts from the ATSB, UK Air Accident Investigation Branch, the US National Transportation Safety Board, Inmarsat and their respective technical advisers.

Joint Agency Coordination Centre (JACC)

Apart from telling us that The Circumlocution Office is alive and well and has been transported to the Antipodes, the above reply is not entirely devoid of information: it mentions only “underwater search”, apparently confirming that the view remains that the MH370 flew south, regardless of any review of the satellite data. Perhaps such a review will indeed demonstrate beyond doubt that the aircraft flew south.

Almost precisely a century after Dickens named The Circumlocution Office, in 1955 Cyril Northcote Parkinson introduced his eponymous law. It has various forms, and corollaries, but the commonest is perhaps that “work expands so as to fill the time available for its completion.” Unfortunately its application in the sorry case of MH370 is that there is no known time limit for the search for the last resting place of MH370, and so myriad public servants [sic] will continue to be paid to do worthless, obscurantist pseudo-work until such time as the aircraft is found, and then they will find something else to do that is both time-filling and self-serving, so as to continue to waste the public’s money.


483 thoughts on “The Circumlocution Office is Alive and Well”

  1. Hi Bill,

    Thank you for your detailed analysis:

    You conclude “With all of the measurement error inherent in this reverse engineering, the ground speed looks to be a little over 500knots.”

    In my earlier comment (the first comment on this post):

    I stated “the last known speed of MH370 which I calculated at 500.36 knots” based on a similar analysis of the flight path from Perak Island to the end point of the Malaysian Military Radar.

    So we are in complete agreement.


    1. Richard,
      Yea, it sure is nice to see multiple approaches coming up with answers in such good agreement.

      But, I must confess, I selected my words in that post very carefully. “a little over 500knots” allows for agreement with the range of speeds that I have seen between 500 knots and 509 knots. I almost said ‘around ‘500 knots’, to pull in the 490knot numbers… But, that might have been too obvious. :)
      Given the precision that the available “data” provides, I view all of these numbers as close enough.

      1. Hi Bill,

        Returning to our discussion on precision and accuracy from the previous post, I would like to use the following analogy.

        If I am sailing a boat and want to know my position (without GPS) I can take a compass bearing from a lighthouse and know I am somewhere on a line with that bearing to the lighthouse. If I take second fix and get another bearing from a church spire then I can draw two lines and find where they intersect. This is still quite inaccurate however. But if I take a third fix from an aerial mast then with a third line my accuracy is increasing. I could carry on with a fourth and fifth fix if I wanted.

        What you have done in developing a consistent timeline for MH370 is similar. Assuming a constant air speed and constant winds and therefore constant ground speed over this short section of the flight from 17:59 UTC and 18:22 UTC with your 9 points along this timeline your accuracy will increase with the more points you have. With increasing accuracy, you can then allow yourself to be more precise.

        It is interesting in examining your 9 points that there was one point at 18:04:36 UTC that was not on the constant flight path where it appears that MH370 started to turn further to the north and then corrected to the original track. Unfortunately one point is not conclusive, it could have been an anomaly in the radar trace.


  2. Circumlocution Exhibit 558:

    “The raw data is not in the hands of Malaysia, Australia or MAS. What needs to be confirmed or released can only be done by Inmarsat themselves,” Hishammuddin told reporters at a media briefing this evening on the still missing Boeing 777-200ER aircraft.

    Depending on who you ask, nobody has the raw data, nobody owns the raw data, nobody may see the raw data and now apparently nobody is auditing the raw data.

  3. This will sound a bit crazy, but is it possible to plot the latest N/S route candidates on the western side of the ping rings? It seems to me that if there is a southern route that fits the BFO graph closely, there is also a second solution, symmetrical along the 64.5E meridian, running roughly along the east coast of Africa. Symmetry between north and south may not exist because the satellite is moving north or south, but what about symmetry east-west?

    While this isn’t a possible route for MH370, we haven’t received any hard confirmation on the ping identities, either.

    Could any of the pings have been mistaken? Would a western route look anything like a real flight route?

    Note that Johannesburg is very near the 00:11 ping ring. Is there a remote possibility that the pings were instead a flight from Dubai to JNB, for example, and unrelated?

    1. JS, my understanding would be that two mirrored flight paths options are roughly symmetrical about an imaginary great circle that passes through two points. One point is the Inmarsat satellite position and the other point is the “start” of the flight paths. This imaginary great circle for the MH370 scenario divides the ping rings into a “north” option and a “south” option. The aircraft’s speed limitations realistically limit it to a relatively steep “north and south” solution due to the distance from the flight path “start” to the innermost ping ring. In other words MH370 would have to fly very fast from the “start” point to a “western” portion of the inner ping ring.

      1. Hi Benaiahu,

        Js did not suggest that MH370 flew to Johannesburg. He suggested, that signals from a different plane, somwhere on the western side were mistaken to be from MH370. Some posters suggested that there is uncertainty about the ID or origin of the released ping and/or BFO data, hence, I presume, js’ question.


      2. Thanks MuOne, sorry JS, I misunderstood. I think my symmetry comments still apply in helping solve your question, just ignore the aircraft speed comments. A non MH370 “western” (N/S) symmetrical flight would be bisected by great circle running through the flight “start” on west side and Inmarsat satellite position. Hard to believe it’s possible they got the ID wrong, but nothing would really phase me now.

      3. phase you or faze you? (homophone corner) 8-)

        I am often out-of-phase: get up too late, get to bed too late.

  4. Hi Duncan,

    Please correct me, if I am wrong. It has been a long time since I did my PPL.

    If MH370’s track fits the ping rings you have published, then wind is irrelevant because wind has already been taken into account to determine the track of MH370 (as opposed to the heading).

    If MH370’s speed in relation to the satellite is one factor to determine the doppler shift, then wind is relevant because the ground speed of MH370 at a particular point in time is affected by the wind at that time.

    For example, if I flew with a heading of due north at a speed of 500 knots and there was a wind blowing from west to east at 50 knots and at 90 degrees to my heading, then my track would be 5.7 degrees to the east of due north and my ground speed 502.5 knots (as opposed to an air speed of 500 knots).

    My position will be along my track (heading plus effect that the wind has had up until now), which is relevant for the ping ring calculation, but my ground speed at that point will be affected by the wind, which is relevant to the BFO Doppler calculation.


    1. Richard,

      The wind is relevant in that , depending on the flight mode at the time, the wind will have an affect on the groundspeed, and the ground track as well. For southern tracks, near the outer ping rings, the winds can be 245 deg/ 50 knots. For northern tracks, winds north of the Bay of Bengal can be 255 deg/100 knots.

      There 3 possible flight modes, and assuming constant aircraft speed:
      (i) Constant [Great Circle] track – – – the groundspeed will vary and the aircraft [instantaneous] heading will vary
      (ii) Constant True heading [the Rhumb line] – – – the groundspeed will vary and the ground track will vary
      (iii)Constant magnetic heading – – – the groundspeed will vary and the ground track will vary with, wind and magnetic declination.

      Only the time for crossing each ping ring is known. The two unknowns are the groundspeed between the ping rings, and the ground track.

      1. Hi Brian,

        I agree with all you say.

        With respect to POSITION we need the sum of all wind effects (speeds and directions) up to that point on the track and the position on the ping ring has already absorbed those effects in calculating the FLIGHT MODEL.

        With respect to GROUND SPEED and TRACK at that POSITION we need the actual wind effect (speed and direction) at that moment as the BFO MODEL requires both the actual GROUND SPEED and TRACK in relation to the satellite motion to work out the satellite/aircraft component of the doppler effect at that POSITION.

        It must be possible with a detailed analysis of ping rings, burst offset frequencies, winds and magnetic variations to reverse engineer which mode the aircraft was in.


  5. Just talked with an old friend and retired 747 captain. He said that he and several of his B777-200 pals have been talking about MH370. They believe it is likely that MH370 suffered a fire caused by the O2 system, like Egypt Air in 2011. Here are some links on that fire.

    I asked if his 777 friends could explain how the aircraft could have continued to operate, and he said they were convinced that it could. They reason that if the O2 system that supplies O2 to the cockpit crew failed to deliver O2 to them, then when the fire burned through the fuselage, they would have had <30 seconds of consciousness. The fire would have caused all kinds of electrical loads to be shed, like in the Egypt Air case. The cockpit crew is comatose in seconds. The door is locked. Everybody in back may have had some O2 for a few minutes, but not long. What would the flight computer do? How long would the Crew have had to start a turn back? I will meet with these 777 pilots and get a better idea of the details they think support this scenario. they are also checking Honeywell sources for me. Meanwhile, I would like to hear from other 777 pilots. Is this a viable scenario?

    1. Thanks Mike.

      As a personal opinion – and readers in the UK will doubtless verify this – I could never cite the Daily Mail as a dependable source of information! In fact I even doubt the football scores published therein, unless I have another good source (or I had actually been at the match).

      1. As an outsider who has never read the Daily Mail before, I was suprised to find that it has generally had the best coverage of MH370, with the possible exception of the NY Times and the WS Journal. Go figure.

      2. This is the official final report of the EgyptAir cockpit fire:

        and here is a forum discussion about it:

        An interesting theory by a pilot on the connection of this case with MH370 which for some reason didn’t get public attention:

      3. This is the official final report of the EgyptAir cockpit fire:

        and here is a forum discussion about it:

        A very interesting theory by a pilot on the connection of this case with MH370 which didn’t get public attention:

    2. ” . . . according to an Airbus study in 2012, it takes just eight minutes for a fire to go out of control and the airplane has to be on the ground within 15 minutes, with the implication being that the airplane would be lost beyond this time. We know from the satellite pings that MH370 was airborne for up to seven hours after contact was lost. If any fire occurred strong enough to cause the loss of all radios and transponders, it certainly would have resulted in the loss of the airplane well before seven hours.” See

      1. Rodney,
        The critical detail to add to a fire scenario is the effectiveness of the cargo hold fire suppression systems. The fire would build up over a short period of time prior to being detected. fumes, smoke, flame do some amount of damage, then the fire suppression system either extinguishes the fire completely, or subdues it sufficiently to halt further significant damage. The system is designed to continue to suppress the fire, if the initial attempt doesn’t quite get it all out. But, if the fire continues generating heat longer than the suppressant keeps spraying into the hold, they the fire could resume. (I recall reading about the system being designed to continue to apply the fire suppressant for 30 minutes.) If the heat source remained that long (for example, overheated Li-Ion batteries continuing to smolder, but not aggressively burning ), the fire could accelerate again.

        The continued operation of the satellite subsystem is reasonable evidence that at least some power was unaffected. (I recall that the left power bus supplies the Satcom)

        If the location of such a fire damaged cables, the impact would be hard to predict. Loss of the cables from the main electronics bay (under the cockpit and forward of the forward cargo bay) to the transponder antennae and the VHF antennae and the HF antennae would be possible but unlikely. The location and size of the fire would have to be able to affect multiple sets of cable bundles heading to different locations on the top and bottom of the airplane to affect all communications.
        Damaging the data links between the main electronics bay and the satcom system would involve fewer distributed cables. (Still an unlikely event.)
        If a fire was in the electronics bay itself, then we go back to imagining scenarios that cause enough damage to take out the radios, while leaving the plane flying.

  6. In trying to reconcile how my projected path models differ from Yap FF’s excellent spreadsheet (Yap V8 with scaled D3) and other team member’s posted outputs it might be related to my mean satellite position for Inmarsat-3F1.
    A. My assumption is to use the mean position of the satellite only for the duration of the available data. In other words 0 ≤ (t) ≤ 461 minutes after March 7, 2014 UTC 16:30:00.
    B. My resultant mean satellite positions are:
    R_x, km =18146.14730 R_y, km =38065.55612 R_z, km =992.183767
    I was able to refine the mean positions once I used the polynomial curve fits, but in general my assumption and or mean satellite position math might be wrong. Lack of rest isn’t helping, any critical feedback would be appreciated.

      1. Great question. “Used” for example in (Yap V8 with scaled D3.xls) AD62:AL73 to generate “Estimated aircraft Doppler along line of sight, Hz”. Which feeds calculated D1 “Aircraft Doppler residual, Hz” then ultimately generates calculated BFO (D1+D2+D3, Hz).

  7. Hi Duncan,

    Mike Exner, in his comment: suggested it would be good if the BFO Model and the Path Model would align.

    I have taken Yap’s most recent Excel Spreadsheet describing the BFO Model (Yap V8 with scaled D3) and compared it with my most recent Path Model based on input from yourself, Eugene, Warren and Dr. Kuang, which I have discussed in various previous comments.

    I chose a speed of 500.36 knots for MH370 which I have shown in previous comments fits the initial flight path of MH370, ADS-B data, ping rings at 18:25:24, 18:26:54 and 18:27:48 as well as the Malaysian Military Radar trace. Yap chose your start point of 6.70N 95.30E, whereas I chose two slightly different start points for the Northern and Southern routes as I believe MH370 stated to turn earlier (as I have discussed in yet another comment) based on the input from yourself, Warren, Eugene and Dr. Kuang.

    The results are a good fit for both the northern and southern route as follows:

    NORTH Yap Richard
    UTC Lat. Long. Lat. Long.
    18:27:48 6.70 95.30 6.85 95.76
    19:40:30 15.33 90.47 15.56 90.57
    20:40:24 22.70 86.37 22.87 86.36
    21:40:42 29.61 81.21 29.76 81.04
    22:40:18 36.28 75.27 36.57 75.38
    00:10:48 44.51 62.67 43.81 61.89

    18:27:48 6.70 95.30 6.18 95.85
    19:40:30 -2.92 93.20 -3.60 93.32
    20:40:24 -11.12 91.79 -11.65 91.21
    21:40:42 -19.24 89.91 -19.74 88.95
    22:40:18 -27.41 88.20 -27.70 86.47
    00:10:48 -39.55 84.05 -39.69 81.90

    The northern route end point agrees +/- 1 degree of latitude and longitude and is near the Aral Sea on the border between Uzbekistan and Kazakhstan.

    The southern route end point agrees +/- 0.2 degree of latitude and +/- 2.2 degrees of longitude and is within the AMSA search area at the start of the Indian Ocean search on 18th March 2014.

    The relatively small discrepancies between Yap’s calculations and mine maybe as a result of the different starting points or other reasons which I will continue to investigate. I can make my spreadsheets available to anyone who wishes.


    1. Thanks Richard.
      Your submitted comment apparently was based on an assumption that various people investigating the BFO values are in agreement, and I know that not to be the case, and so I have edited out the clause that implied that to be true!

  8. Richard wrote:
    The next questions are:
    1. Can we agree on the position at 18:27:48 UTC for the Northern Flight Path of 6.8511N 95.7619E or for a Southern Flight Path of 6.1797N 95.8528E?
    2. Can we agree on the position at 00:10:48 UTC for the Northern Flight Path of 43.8100N 61.8900E or for a Southern Flight Path of 39.6853S 81.8992E?

    1. On your northern path, I would have it crossing the ping ring right where you do. Tailwinds at that point would have died down to 0.6 kts, so for a GS of 490.6, it would place it a couple of miles beyond the ping ring, which is not a bad error IMO.

    I have the turn taking place at NILAM, which is a major airway crossroad, and it would be at NILAM right when the 18:25-18:27 Doppler event was taking place.

    Probably, there was a 2nd turn somewhere, most likely POVUS. I don’t think the turn to pretty much due south happened at NILAM as that would have required crossing the norther tip of Indonesia. Looking at 9M-MRO purely behavioristically a la B.F. Skinner, she clearly demonstrated avoidance behavior of Indonesia by the turn up the Strait, so she would continue that behavioral pattern and gone out through the Great Channel rather than overflying Sumatra IMHO.

    After the 2nd turn, the run consisting of POVUS ISBIX MUTMI RUNUT is basically a straight rhumb line with a course of 189T.

    2. I have not worked out a waypoint path for northern scenarios, although I’m guessing there probably is a solution that works. Your ending point is quite close to mine about 48 nm to the SW of mine.

    I am curious how you arrived at that course? Does it take into account crosswinds?

    @ Tony Mach: I see on your blog that you now believe the waypoint way to be untenable because of the fuel situation. However, if Richard and I’s calculations are correct, she was flying at normal cruising speeds during the initial phase of the voyage. Cruising speeds are what they are because they are fuel efficient. She supposedly had 49.1 tonnes still onboard after the takeoff. Why wouldn’t this be enough?

    Harm wrote:
    As to your analysis of probable ground speed: your result seems to be very consistent also with the average ground speed that would be required to reach the 18:01:49 location in time starting from a bit beyond IGARI and acommodating a 2 minute, almost 180 degree turn from a bearing of 40 degrees to the Penang (or close-by) waypoint.

    The path from IGARI to the 18:02 position is fairly well constrained by time and the requirement to go around Penang. I calculated a turn based on a 25 degree bank angle (the highest notch on the bank indicator knob on the dashboard), with a turn radius of about 7 nm. With that I made it come out with a total path length of 338, that is for 500 kts. That’s probably about the most parsimonious solution that can be constructed IMO.

    Team members of the unofficial investigation should also note that I have uncovered a minor and a major discrepancy between the “Slide 2″ ping rings and the “Hussein Chart #1″.

    Slide 2 18:27 ping ring = 1897 nm
    Hussein Chart ping ring = 1898 nm

    The difference for the 19:40 ping ring is more drastic.

    Slide 2 19:40 ping ring = 1762 nm
    Hussein Chart ping ring = 1744 nm

    Probably the Hussein Chart should be considered authoritative, as it is the most recent and comes straight from the Ministry of Defense. I must confess that I like it a lot better: my 19:40 position between the ISBIX MUTMI leg was always the biggest error (19 nm). With the new ping ring the error is reduced to 1 nm–a nice, surprising confirmation of the model.

    1. Just to note that there are many comments with which I do not agree, but am putting them up anyway, such as the above.

      However, there are many submitted comments that are simply ignoring available evidence, and so they are not posted.

      1. Hey Duncan, lay it on me man! I gotta thick skin–if there’s a fatal flaw somewhere I want to know about it!

        Briefly, here is the theory behind the theory:

        I believe a careful examination of the initial phase of the flight up until about 18:29 UTC proves beyond a reasonable doubt that the post-IGARI flight was intentional. Yes, all sorts of contradictory logical possibilities spring to mind. However, in order to move forward, we need to form a working hypothesis. From my own experience oil drilling, when you’re paying $20,000/hour and you’re not exactly sure where the formation is going, you still have to do [i]something[/i]: so you pick a likely idea, try to avoid the worst case scenario, and drill ahead, because that’s the only option.

        Here we are in a similar situation. It’s a crisis because money and will power for the search effort is going to run out, probably sooner rather than later.

        So we got to go with what our gut interpretation of the evidence is: here it is that the initial phase of the flight was intentional. Therefore, in order to predict the location of the 9M-MRO, we must take this into account. We DO NOT need to worry about who–or what–took over the a/c. All we need to understand is the modus operandi that was demonstrated. That modus operandi was that it was navigitating via waypoints most likely with the LNAV navigational subsystem engaged, and that it was at standard cruising speed and altitude. The modus operandi also exhibits avoidance behavior of Indonesian air space.

        With that in hand, the simplest model of the future path will assume that this demonstrated behavioral pattern will continue: that he will navigate via waypoints at standard cruising speed and altitude, and that he will avoid Indonesian airspace. With this heuristic in hand, there is about only one or a few very waypoint paths that are consistent with the “ping ring” LOPs. Because the 19:40 and 20:40 ping rings are so close, you are pretty much forced into that ISBIX MUTMI RUNUT corridor.

        After RUNUT, what happens is anybody’s guess; but if he maintains the standard cruising speed and alititude until he runs out of fuel, then where he crosses the final ping ring is pretty tightly constrained, assuming he continues to run a more or less straight path.

      2. What you are suggesting there is broadly in line with my working hypothesis from quite early on: that some event left the aircraft flying by autopilot and moving from one waypoint to another and onwards in accord with decisions made by the software in line with defaults based on the array of waypoints in the relevant autopilot data file. This might well be wrong, but it seems to me to be as simple an explanation as one can arrive at in order to fit the Inmarsat-derived information.

        However, your suggestion that this must have involved a deliberate avoidance of Indonesian airspace so as not to be detected (if that is what you are actually suggesting), or indeed any other route choice based on avoiding radar or other detection methods, is flawed, I think. Here is why…

        If the aircraft had NOT taken a route which avoided (apparently) detection by radars etc in any airspace which is approached or crossed then we would have known for sure whether it had taken a northern route, or a southern route, in accord with the Inmarsat-derived information. And then we would not be discussing it. That is, our discussion and debate and mystification has a prior condition of the aircraft not having been detected after 18:22 UTC except by the satellite pings.

        In fact I was thinking about doing a post on this very matter, in which I would invoke what is known as the Anthropic Principle. For the time being let me use this analogy. Something often used by people in discussing probabilities is the story – I do not know whether it is true – that the first bomb dropped by the Allies on Berlin in WW2 killed the only elephant in Berlin Zoo. People invoke that as being a hugely unlikely event. And it was, a priori. But no-one talked about it, a priori. After the event occurred (if indeed it did) the a posteriori probability was one. That is, that first bomb did kill the only elephant in Berlin Zoo.

        The point is that no-one would talk about the elephant if it had not met its fate that way, whilst ignoring the lions, tigers, giraffes, apes, monkeys and who knows what else in the zoo. Similarly we are talking about (and worrying about) MH370 only because something happened to it which may have had an a priori probability that was very small, but nevertheless it did happen. Something like following a series of waypoints after an onboard calamity that left everyone incapacitated, the route automatically chosen leading it to avoid any detection perhaps by looping around Indonesia and then flying deep into the Indian Ocean before crashing into the sea far from land; or perhaps flying north and over eastern India and the Himalayas and western China and all the while escaping identification for reasons that we don’t know; or… Whatever the reason, arguing that it is an unlikely event is invalid, because it DID happen.

        A final note on probabilities. One might believe that an unbiased coin when tossed has a 50/50 chance precisely of landing on heads, and the same for tails. But I *know* this is not true. And here is why. One Friday evening about a dozen years ago I was at London’s Euston station awaiting a train to Manchester. Two Asian male students were playing around, spinning one-pound coins in the air. One dropped a coin onto the flat, smooth marble floor where it went kerplonk, kerplonk, and then ended up stationary balanced on its rim. They were amazed, and so was I. But that one highly-unlikely event means that there is NOT a precisely 50/50 chance of a tossed coin producing either heads or tails, no matter how many people spin how many coins for the rest of all time.

  9. Duncan,
    Terrific effort going here! I still believe we all can be of great service to the families of MH370 passengers as well as the authorities, if we also really focus on the first two hours of flight – in addition to the final five leading to the final location. I believe that if we could put together detailed time histories of location (including accurate estimates of all turn radii), altitude, vertical velocity, speed, heading, thrust settings, we could help determine if the cause was system failure with some piloting hopefully leading to an emergency landing, or untoward human intervention. There is considerable data (Inmarsat pings, ADS-B, primary and secondary radar, eyewitnesses, not-in-my-back-yard from Indonesia and the like), much of it untrustworthy without scientific sanity checking. Many of the contributors to this site have flight path test programs running for the last five hours and I would urge them to tackle this first portion of the flight. My background is from the engineering airline simulators at Boeing Seattle, but since I am retired from that position, I am using the high fidelity PMDG 777-200LR simulation at home to help check these possible flight paths in an attempt to diagnose the possible scenarios. You have assembled quite a team!


    1. Dave: Guarded Don is on it. He composing a very detailed time line combining ADS-B, Radar, BFO data, ATC transcripts, Official reports, etc.

      Don: How’s it coming?

      1. I’ll make time for it Thu morning, the ATC transcript is the last input & then edit checks.

        Dave: were you based at Customer Training & Flt Ops Support off 14th Ave S/Trenton St?

    2. Dave,
      Can you look at my hypothetical paths and comment on the feasibility of flying them. The simulator may enable you to fly a more realistic path.

  10. This is a cross post. I had not realized that Duncan had opened up another thread.

    “Some time ago I posted a proposed route computed using the spread sheet model of aqqa and input data based on Mike Exner’s work and others. I have continued to work with aqqa on improving the details in the model and have now done two paths to the North. They appear to be the only basic solutions in that direction (although the end point details are not completely understood). So as to be economical of space here, the end point coordinates are : 42.01 N and 71.56E for a path to the Northwest 30.10N and 94.00E for a path to the North.
    Graphs, tables etc may be found at

    Comments and suggestions would be appreciated.

    1. Sid & Duncan ~ The northwest path at 42.01 N and 71.56 E approximates Dr. Kuang’s Beshtash Valley crash site (30 km south of Talas, Kyrgyzstan). ~LGH~

      1. Thanks LGH: As per usual, you are a star.

        When is someone going to go and look in the Beshtash Valley? Are there any readers out there who could supplement Dr Kiang’s search for satellite imagery, in the hope that there might be some freely available (stranger things gave happened) with adequate resolution to show something suspicious?

      2. LGH, thank you for the comment and for looking at the data.

        I was aware of Dr. Kuang’s hypothesis and the satellite pictures he has posted and had made mention of it in an earlier post on this site. But, without a careful look at my results but others, and perhaps some refinement of the model, the end point that I found remains just suggestive.

        There is no reason not to consider the other Northern hemisphere flight path equally probable. One might suggest that each of the two flight paths could be considered representative of a different type of initiating event on board the aircraft.

  11. Dear Duncan,
    Some really great informed comments above as to who should be approached to ask for the original, unprocessed data on MH370. Would anyone be willing to craft a precisely worded 1 or 2 sentence request for information which provides less legal and semantic wiggle room than the currently popular “Inmarsat must release the raw data” ? Because this request has to be spot on, and focused on the right agencies to have any effect. For example, the MH370family committees would benefit greatly from a precisely worded (and targeted) model. I believe the MH370 families are the Malaysian authorities’ Achilles heels, and their precisely worded questions and pressure on the Malaysian investigators are the only route to squeezing out any further data or radar information.

    1. Hi roseny6,

      Good suggestion. Here, the Chinese families committee has just released an English version of their petition:

      The families still believe that it’s Inmarsat who should be approached for raw data. But as Inmarsat senior vice president Chris McLaughlin has told CNN that the company is limited by what they can share because Malaysia is at the center of the investigation so the data belongs to that country. “It’s a matter for the authorities to decide what they’re going to do with their data. It’s not something Inmarsat can release.”

      Precisely worded questions and presure on the Malaysian authorities are desperately needed.

      1. Below are the questions a few of us assembled off line today for background information for the CNN OutFront with Erin Burnett Show, scheduled for 7PM EDT today. Chris McLaughlin and Miles O’Brian were scheduled to talk about the Inmarsat data, but McLaughlin bailed about an hour ago. Said he missed a plane. Hmmm…

        Note that the “assumptions” we ask to be confirmed or corrected do not imply that any of us necessarily believe all these assumptions are true. Clearly, there is still disagreement on various facts. If these questions ever do get answered, then we will have a better understanding, regardless of whether everything is confirmed, or some things are corrected.

        Data and Information Desired from Investigation team

        Confirmation of, or correction to these assumptions:
        • MH370 used the Inmarsat Aero-H service via the global beam on I3-F1 and Perth LES for all “handshakes” and “pings”
        • There were at least 17 handshakes/pings recorded between 16:00 and 00:19 UTC
        • There were no AES responses to any attempts to make voice calls
        • All inbound communications were on frequencies near 1643.5 MHz up and 3614.5 down
        • Positive BFO values correspond to negative Doppler (i.e., moving apart)

        For every Handshake (Ping) time between 16:00 and 00:19 UTC, provide:
        • Time to full resolution available
        • Net Propagation Delay Times (AES Antenna to Spacecraft L band Antenna)
        • BFO values and description of observation methods, error sources and magnitude
        • LES or AES initiated?
        • Reason for, or source algorithm causing Handshake/ping
        • Data, if any, returned from transaction (including ACARS data where available)
        • Detailed explanation of D1, D2 and D3 definitions, errors and correct signs
        • Detailed explanation of AES Transmitter Doppler Offset algorithm
        o Under normal conditions (prior to 17:07 UTC)
        o If AIRINC 429 link to AIMS/IRS data was lost (no Nav data to SATCOM)

    2. Inmarsat could for sure be kindly asked to share the methodologies they had used for deriving their conclusions and/or some of the algorithms you guys tried to re-engineer.
      Still they could deny to disclose even that. But they wouldn’t be able doing so by referring to infringement of third party (Malaysian Airline) rights !
      They are the owners of their own applied methods and algorithms.
      Could that be of use to those of you experienced with the math work here ?
      I’d also sign this letter / email with the names and job titles (including email addresses) of those here, who are scientists or researchers, thus clearly indicating that it is not a request out of pure curiosity, but a request on expert level and a serious offer for assistance.
      Ignoring such a qualified request, I’d even say offer for collaboration, or rejecting the request would be very hard for Inmarsat’s public image and even contradict some of their PR (stating on the one hand they have consulted external auditors while on the other hand rejecting the request of experts with same or similar competence of their auditors).
      I’d send this email to Inmarsat’s CEO and the VP, Mr McLaughlin, but at the same time also to The Guardian, explaining to them the importance of this request.
      Anyway, I’m sure that at least Inmarsat’s technicians and technical consultants are following this blog. And that sooner or later these infos will leak anyway, as there are good people everywhere and such a request for getting these data is absolutely justified.
      Regarding the data Malaysian Airlines is owner of, I am not so confident they will be disclosed, nor do I have a good recommendation to whom ( within the Malaysian govt) such a request should be sent.

      1. Source :

        During a press conference in KL on Monday ? / Thuesday ? a question was asked to Hishammuddin Hussein ( MMoT ) by journalist Sumisha Naidu. Here is what she says on the MH370Families facebook account :
        ” I did ask about your request for satellite data, here’s what he said: “There’s been so many requests for so much information. The beginning it was transcripts of the conversation, then it was the cargo manifest,” said Mr Hishammuddin.

        “I’m trying to put forth the argument that if you do want the information, then you go through the panel of experts that we’ve established and if it doesn’t affect their investigations, by all means release it. But the point is this: any information released will not stop speculation.”

        So, if a letter or email is sent over to the “panel of experts” and to prevent right away a negative answer of the type “we can’t release as this would affect the investigation”, wouldn’t it be wise to start the email or letter by explaining why or how the release would not affect the investigation ?

      2. @airlandseaman

        “…Chris McLaughlin and Miles O’Brian were scheduled to talk about the Inmarsat data, but McLaughlin bailed about an hour ago. Said he missed a plane. Hmmm… ”
        I fully agree with the Hmmm …
        I guess he can’t miss the plane several times following several invitations from CNN and the overall pressure that is going up. I believe that, at least on this part, medias can somehow help to break parts of the circomlocution loop.

        And by the way, Mike, you did really fine on your 1st interview !

    3. Roseny6, the below is obviously a lot longer that what you have in mind, but it represents an attempt to translate Mike’s wish list below into legalese that maybe is a bit more precise than some of the Freedom of Information Act and other information requests I have seen:

      Please produce any and all non-identical documents and electronically stored information—including writings, models, drawings, graphs, charts, photographs, sound recordings, images, and other data or data compilations—stored in any medium from which information can be obtained either directly or, if necessary, after translation by you into a reasonably usable form (“document”) which comprise, refer to or relate to any or all pings, handshakes, partial handshakes, keep-alive messages, voice calls, attempted voice calls or other actual, attempted or partial contacts or communications between the aircraft operating as Malaysian Airlines Flight MH370 on March 7 and/or 8, 2014 (“Flight 370″) and any satellite, including without limitation, Inmarsat I3-F1 (“pings”).

      A document shall be deemed by you to “comprise, refer to or relate to” any pings if, without limitation, it discusses, shows or reflects:

      (a) whether Flight 370 used the Inmarsat Aero-H service via the global beam on I3-F1 and Perth LES for any or all pings; or
      (b) how many pings were recorded or occurred; or
      (c) any voice calls or attempted voice calls and any AES responses thereto; or
      (d) the up and down frequencies of any or all inbound pings: or
      (e) whether positive BFO values as reflected in the chart entitled “MH 370 measured data against predicted tracks” found at indicate negative or positive Doppler (i.e., whether the satellite and aircraft were moving farther apart or close together); or
      (f) the specific definitions, errors and/or correct signs associated with the terms “D1″, “D2″ and/or “D3″ in the illustration entitled “Doppler correction contributions” found at
      (g) any available time to full resolution of any pings; or
      (h) the propagation delay times (AES Antenna to Satellite L band Antenna) for any pings; or
      (i) BFO values, observation methods, error sources and magnitude for any pings; or
      (j) how, what or why any pings were initiated (e.g., LES or AES initiated); or
      (k) the source algorithm or other reason or mechanism associated with the initiation of any pings; or
      (l) any and all data or information, including without limitation ACARS data, included or referenced in any pings; or
      (m) the AES Transmitter Doppler Offset algorithm applied to or applicable to any pings, including without limitation (1) under normal conditions such as Flight 370 prior to 17:07 UTC on March 7, 2014 or (2) when the ping contains no navigation data, such as when the AIRINC 429 link to AIMS/IRS data for Flight 370 was lost.

      1. I’d phrase it a bit more succinctly:
        (1) Please make public the ping/handshake propagation time delays from which the aircraft-satellite elevation angles were derived; and
        (2) Please make public the original Burst Frequency Offset values along with a summary of the components that contribute to them.

      2. Amazing info everybody! So important to have a solid set of carefully crafted requests which prevent passing the buck due to semantic loopholes. Should a request for clarification of radar/comm data be added and addressed differently— this info is crucial too, if they would throw any more crumbs.
        Any thoughts on which agencies it should be addressed to and watertight wording? At present MAS is saying only the investigation can release data. Inmarsat is saying that MAS can. Does MAS has ‘ownership ‘ of the unprocessed Inmarsat data, and does it have to officially okay the investigation and Inmarsat to release any or all of the unprocessed and processed data?
        Perhaps the letter can take the form of 1.addressees 2.1 sentence explaining why or how the release would not affect the investigation, (and a nod to the fact that it is understood that no military secrets can be released) ? 3. Duncan’s succinct précis, with a link to the indepth tech details of the request as collated and written by Bruce ?
        Please correct me, and suggest tight wiggle-free wording of which agency/official should be approached. Perhaps Haxi, when it is finalized, you might be willing to help translate and communicate it the the Families, to augment their letter? That would be great!

      3. Thanks Lucy.

        It is just a shame – and an indictment of many involved – that a human tragedy like this, which also has implications for global travel and communications, should have descended into such a state of buck-passing, and acrimony, and obscuration of the truth, and displays of outright incompetence. As always, I tend towards the cock-up rather than conspiracy theories.

      4. Hi Bruce ~ I apologize, I misunderstood your reference to the aircraft 9M-MRO dispatched as Flight 370. Yes, the livery is “Malaysia” but I now see that you were referring to the airline “Malaysian Airline System Bhd (MAS)” so please disregard the first half of my previous comment of this airline versus aircraft distinction.

        ~ LGH~

      5. With LGH’s corrections I have submitted the above as a FOIA request to the NTSB.

        Fat chance.

      6. Bruce: I have every confidence you will be successful. And that the NTSB should also be in charge of all flying pigs (refer: Carroll, L.)

    4. Roseny6,

      Of course I’m more than willing to volunteer as a translator. And I’m sure there are people among the families who speak really good English.

      But for now, the biggest question is: who should be asked for the raw data. Please see my comment here:

      Inmarsat is saying that there is this “Convention on International Civil Aviation” which prevents the release of findings from an investigation without the consent from the state conducting the investigation.

  12. I would distill Duncan’s most recent replies and the hypotheticals regarding what level of data is possessed by whom and not being adequately shared to one simple statement that reaches back to the inception of this discussion:

    The data set as it has been presented does not conclusively indicate a flight trajectory to the southern Indian Ocean and the present location of search operations, and the analysis that produced such a conclusion yet remains in question and should be further tested by way of an open peer review process.


    Given that the sole source for the SOS transmission appears to be the China Times and it has not been corroborated officially or elsewhere in the press, it could very well be apocryphal. If the US Embassy in Bangkok had indeed disseminated this information to the China Times, it most likely would have disseminated it to other outlets; the Embassy would likewise most likewise have substantiated such information. The dissemination has not been corroborated in either case as far as I can discern, and thus I would suggest that there is a significant probability that the SOS transmission did not occur.


    1. Thanks Rand.

      As of now: no more comments will be put up here regarding a crash into the South China Sea, because there is no evidence to support the notion. Period.

      Similarly on the purported sonic ping detections in the Indian Ocean: no evidence available of their reality.

      Any other ideas that people want to promulgate must be consistent with the available evidence, which is very sparse. And that has to be positive evidence: a lack of detections in any area is NOT evidence.

  13. A useful tool.

    I have just come across this useful tool to assist with visualising and mapping the ping rings, or indeed anything else you might want to map. – – – and a variety of useful calculators too – – –
    This is a very versatile system for inputting almost anything you like and mapping the output on a whole variety of mapping applications. For example, it will map circles of a given radius about any point on the earth, and it does so – – -“a very high degree of precision (using the thoroughly nasty Vincenty Formula, which accounts for the flattened shape of the earth”.
    Which then got me thinking – – it ought to be possible to display the most recent ping rings, taking into account the satellite movement, and the flattened earth.

    I used Duncan’s latest ping ring times and radii – –, and extracted the satellite positions at the corresponding times from the chart here – – Then I constructed a very simple table in Excel to upload to gpsvisualizer. The output is available as a .kmz file which can be dragged into Google Earth directly.

    The result is a set of ping rings for all the 10 radii that Duncan lists. As a check on the accuracy, of course the 16:30 ping ring ought to go right through Kuala Lumpur international airport. Well, it doesn’t quite, but it is only about 15nM too far east. Similarly the ping ring at time 17:07 is displaced by about the same amount. Unfortunately I can’t determine where the offset error might be. On the other hand, if the ping ring at 00:11 was accurate to within the same 15nM that would be a major achievement.

  14. LGHamiltonUSA,

    The part about the US military base at Utapao having picked up an SOS call from MH370 at 2.43am had always been problematic for 2 reasons, firstly this military base would have been out of normal range of any emergency frequency radiowaves from MH370 if the plane was not that far from BITOD at that time and secondly the reported time does not fit the timeline for a forced ditching at the South China Sea.

    However, there could be an explanation. I had earlier in the TMF blog wondered whether the SOS call was actually picked by some other US listening post but attributed to the known base at Utapao simply to avoid disclosing the whereabouts or existence of this other listening post.

    The USS Pinckney was actually at the South China Sea at the time the plane went missing. Set forth below are relevant parts of the reports from the 7th Fleet on March 8th and 9th:

    [March 8, 9.13am] “……….USS Pinkcney is en route to the southern coast of Vietnam…..Pinckney was conducting training and maritime security operations in international waters of the South China Sea. The ship could be in the vicinity of the missing jet within 24 hours….. A P-3C Orion aircraft will also depart shortly from Kadena Air Base in Okinawa, Japan….”

    [March9, 9.32am] “……Pinckney was diverted from a training mission in the South China Sea to search for signs of the missing aircraft……..”

    The 7th Fleet is based in japan. Japan is one hour ahead of Malaysia in time. That would explain the reported time of the SOS call of 2.43am given by the US Embassy to China Times. In short, it would appear the SOS call was picked up by USS Pinckney at the South China Sea at 1.43am Malaysian time, and the ship reported the call in Japan time as it would, being part of the 7th Fleet.

    1. Alex,

      Any distress call on the emergency VHF frequency 121.5 would have been heard by all aircraft and ATC in the vicinity. This report you’re referring to is likely chaff.


      1. Thanks Scott.
        I know (I think) what you mean, but “chaff” may not be the best word to use here, given its meaning in the radio/radar domain… 8-)

      2. Scott, not necessarily so, for 2 reasons.

        Firstly, this SOS call was likely to have been garbled if the theory of the plane having been hit by positive lightning is correct. Lightning causes electromagnetization of radio equipment. Dont forget a pilot from another plane (MH88) made emergency radio contact with Mh370 at about 1.30am and reported hearing lots of static, interference and mumbling.

        Secondly, the plane disappeared from primary radar after BITOD meaning the plane had dropped substantially from 35,000 ft at IGARI to a level low enough after BITOD to lose primary radar coverage. At 1.43am the plane would have been just a few thousand feet above mean sea level (around 3000ft) and therefore the chances of a garbled SOS call being picked by other aircraft and ATC would have been slim. However, a ship in the vicinity is another matter altogether, especially a military ship with specialised equipment to pick up and decipher radio signals.

      3. A simple formula that includes the effect of the atmosphere gives the range of such a VHF distress transmission:

        Distance to radio horizon [miles] ~ sqrt(2 X height [feet])

        If the plane was at 1800 feet then the range would be about 60 miles, less than the distance to Hua Hin Airport, possibly the nearest listening post.
        Such a low altitude may be reasonable for a plane in distress asking to land at U-Tapao.

        By the way, here is the source to all this:

        The U.S. military said the driver had received a call SOS signal, said cabin facing disintegration
        2014-03-08 20:13:42

        Wang reported immediate WU Gui Feng
        The U.S. military said the debris found more oil received SOS signals
        20:25 on March 8, 2014

        and several machine translations:

      4. Alex, as Scott says no need for secret US listening posts or US Navy ships to pick up a distress call when there were lots of aircraft in the vicinity plus ATC listening to 121.5.

        In fact here is a posting on pprune by a pilot flying in the vicinity at the time listening to 121.5. Now I can’t vouch for his authenticity.

        “I was also flying in the area at the time of the disappearance, I heard the SGN controllers and the MH088 trying to contact MH370 on 121.5 and I certainly didn’t hear anything that resembled a coherent reply.

        There’s something in the SGN area which periodically causes 5-10 seconds bursts of buzzing static on all VHF frequencies. It’s well known to all of us based out here, I’m pretty familiar with what it sounds like, and that’s all I heard that night. Wish I had something more dramatic to report but I personally don’t believe any transmissions were made on 121.5 from MH370 at that time. ”

        And how far away was the USS Pinkcney? The report says that it could be in the vicinity within 24hrs. If we assume say a speed of 15knots for 24hrs = 360nm away at the time.

  15. Don,

    The evidence indicates the disabling event occurred at IGARI, not BITOD so the distance the plane could have covered from gliding would be calculated from IGARI. If the plane had suffered a catastrophic electrical failure from one of the things mentioned by Bill (fire, explosion, lightning), the pilots might not have been able to steer the plane as they wished. Also we do not know whether they managed to deploy the RAM turbine. Finally, there is the wind factor. The pilot who saw lightning said it was a NE monsoon night and the Kiwi on the oil rig described the wind as ‘NE – ENE averaging 15 – 20 knots”.

    1. Alex,
      With respect, I don’t see any evidence for your statement “the disabling event occurred at IGARI”. We don’t even know for sure that there WAS a “disabling event”.

      What we DO know is that the last ACARS communication from the aircraft was at 17:07 UTC, although we don’t know what prompted it or what its contents were.

      It seems to me, by analysis of publicly-available data from FlightRadar24, which gives Lat and Long at certain points, that MH370 actually passed IGARI at 17:17. That’s four minutes earlier than the time given in the preliminary report. By then, the aircraft had been flying at 35,000 feet for 10 minutes or so. At about 17:20 it seems to have changed heading from 25 to 40 (possibly consistent with a turn towards BITOD?) and was apparently on the same heading when the FR24 (ADS-B) data ended at 17:21.

      I know that FR24’s datasets aren’t entirely self-consistent and that they’re not entirely consistent with data from Flightaware, but none of the available ADS-B data suggests a disabling event at IGARI, as far as I can see.

    2. The facts released to date simply indicate that comms from the aircraft (voice, Mode-S transponder, ADS-B broadcasts, downlinked ACARS msgs) ceased in the vicinity of IGARI and BITOD. Your previous comment hypothesized a glide but now it’s loss of control, “the pilots might not have been able to steer the plane as they wished”.

  16. Yap, Henrik, Victorl,

    BFO for the 6 pings seems to be = [fixed offset of around 90] + [satellite velocity]. Both D3 and BFO were quite close to 90 when the plane was stationary at KL at 16.30 UTC ( D3= 83, BFO =88 ). I wonder if there is any significance in that.


    I had taken note of your previous suggestion. However, Inmarsat have not released the RT timings for the first 5 pings, so we do not know where the ping rings for the first 5 pings are situated. These RT timings will show whether the plane had crashed early on, or had flown on for another 6.5 hours. The measured BFO data for the 6 pings can be consistent with an early crash, since those values appear to reflect only the satellite velocity plus a fixed offset, as shown above.


    Fully agree with your analysis. If Inmarsat really want to, they can release the RT timings data even if the investigation team is against it. There is such a thing called a ‘leak’ and what sort of adverse consequences can there possibly be, if the data were to be leaked.

    1. Alex: The BFO de-composition is apparently far more complicated than you suggest there. What you are suggesting is phenomenology: seeing a pattern and trying to interpret it, which is a first step. However, the reality is often that a full explanation of an observed phenomenon is actually far more complicated. In this case, looking at only one small part of it, there are two contributions to the BFO that depend on the satellite’s motion: (a) Its velocity relative to the (stationary) Perth ground station; and (b) Its velocity relative to the aircraft (which is initially stationary at the gate at KL, but thereafter is moving). On top of that, stating things in quite simple terms, the shift in frequencies (and hence contribution to the BFO) might be the same in a fractional sense for both the uplink and the downlink in each of the links (a) and (b) above, but if the frequencies used for the uplink and downlink in each or either case are different (as they are) then the absolute frequency shifts (i.e. in Hz) which then contribute to the net BFO will differ.
      That verbal description may sound complicated to some, but the reality of the attempts at back-engineering the BFO graph and understanding what it means is far more complex: there has been a lot of debate on this off-line between various commenters you will have met here, and at times it has been quite heated; yet these are smart people doing their best to solve the problem.
      I am just saying: no, it may well not be as simple as “the satellite velocity plus a fixed offfset.”

      1. Duncan, understood. Regarding the formula for BFO, a layman would lean towards BFO= D1 + D2 only for the pings since these transmissions were initiated from the GES which presumably would know how to cancel out the C band doppler, meaning no D3. With no inputs of speed, position etc, the compromised SDU would not have been able to calculate D1 + D2 properly, so it is not suprising to see the BFOs a function of the satellite velocity. That still wont explain the apparent fixed offset, perhaps the SDU was geared to add an offset to get the end result to be within a certain frequency range in line with what has been suggested by @airlandseaman (if i have understood him properly). As always, when a layman puts forward a ‘mathematical comment’, there is a good chance it is a load of rubbish.

    2. Alex if your theory is correct that the plane crashed early on after passing IGARI into the South China sea somewhere on the 00:11UTC 40 degree ping arc with the SATCOM system floating for the next 6 hours or so then the distance measured for the last 5 pings for which the elevation angles (ping ring radii) have been released, see Duncan’s posting for the values, should simply differ by the distance that the satellite has moved between each of the pings.

      It’s really simple to check to see whether this disproves your theory.

      If you want people to take this theory seriously then show how the ping radii data for the last 5 pings matches your theory.

  17. I still did not succeed in reading and making sense of the tagging in the Malaysian Primary radar plot. I have however removed the tagging from the radar plot – nothing spectacular but the edited chart shows more clearly the point that I was trying to make with my previous posting.
    I hope this link will work to edited plot

      1. (re-using some material I have posted elsewhere. )

        The text next to each radar blip is a 24 hour timestamp
        This is the earliest and best resolution version that I have found:

        – The added text boxes on the chart highlight the time of the last blip (02:22H just off the left edge), and the time the plane was over Parek Island (over the tip of the right arrow at 02:20H. The first blip is farther to the right, a few minutes earlier than the 02:02H time shown.

        – I have uploaded two screen grabs so that you can see what I am reading.
        In this first one, I have zoomed into the timestamp on the left edge of the photo, and the timestamp on the right edge of the circle. For each one, the upper zoomed copy is overlayed with the text that I read. The lower zoomed copy is just scaled up larger than the original. (each is 8x enlarged)
        In the first one (left side), The blip itself is off the edge of the photo. The timestamp is mostly visible in green-ish yellow. The waypoint MEKAR is marked with a larger white ‘+’ sign and labeled in white. the diagonal white line is not labeled on the radar plot. But, it corresponds to the N571 flight track that runs through MEKAR. (The waypoints and flight paths can be referenced in
        The second enlarged area has two timestamps that I was able to figure out. One is readable (02:07:16 on the right side of the white circle), and the other is partially readable and I have filled in the other digits by estimating what the timestamp should be (02:40:36).

        – The second screen shows how I estimate the likely timestamp times. After reading just a few of the timestamps, I created a time scale that aligns with the blips. The scale started as just a linear timeline from 2:20 at Perak to 2:22 at the last point over the left edge (based on the text on the original photo.) Then, I started speculating on the unreadable digits on some more labels. (yes, this would be me guessing… ) As I filled in more numbers, I was able to adjust my timeline to be accurate to within a few seconds. After iterating back and forth for a while, I settled on a placement and scale for my timeline correlates quite well with the numbers I read and the numbers I guessed at.

        – The assertions about various altitude changes are not supported by what is seen on the radar photo. The claim that the plane went below 4,000 or 5,000 ft comes from the 53nm gap (~61 miles) (in the white circle) where no radar blips are seen. But, just calculating the time and distance, the plane did not slow down, so it must have stayed at a much higher altitude. At the Western end of the radar picture, the plane had to be at a high altitude for the Butterworth radar to track it out at 245 nautical miles (281 miles).
        A long range defense radar would be tuned to see as far out over the ocean as possible. It is not tuned to look straight up (The first, Eastern most blip is not seen until the plane was 57nm (~65 miles) West of the radar).

        – To calculate the plane’s speed across the radar plot, you need to use the actual elapsed time and distance. The visible plane blips cover around 187nm (~215 miles). The timestamps cover from some time before 02:00 until around 02:22. With all of the measurement error inherent in this reverse engineering, the ground speed looks to be a little over 500knots.
        {Note: the distances measured off of the radar picture are probably only accurate within 10%.


  18. Hi
    Not sure if this is helpful, or even relevant, but I have noticed discussions from time to time about who exactly has the satellite data and who analysed it. In the New Strait Times on the 21st March, the statement from Hishammuddin Hussein at a press briefing:

    “The investigations team received the complete raw Inmarsat satellite data, which included the six handshakes at approximately 3 pm on Wednesday, March 12.”

    “Upon receiving the raw data, the Malaysian authorities immediately discussed with the US team how this information might be used. The US team and the investigations team then sent the data to the US, where further processing was needed before it could be used.
    “Initial results were received on Thursday, March 13 at approximately 1.30 pm, but it was agreed by the US team and the investigations team that further refinement was needed, so the data was again sent back to the US.
    “The results were received at approximately 2.30 pm on Friday, March 14 and presented to the investigations team at a high-level meeting at 9 pm on Friday.
    The UK AAIB, who had also been processing this data independently, presented their results – which concurred with ours and those of the US team – at that meeting.”

    This at least shows who had access to the raw data for analysis. The next day it was reported that the USS Kidd was being deployed to the Southern Indian Ocean.

    1. “presented their results – which concurred with ours and those of the US team”
      Well. The Malaysians would not have been able to analyze the twice-refined Inmarsat data, so they must be basing their “results” on other data. What a load of crazy codswallop the Malaysian investigative team have released to the press.

    2. Hi JR,

      OR maybe it shows that Datuk Seri Hishammuddin Hussein does not know the difference between raw and derived data.


  19. Hello,
    Mathematically speaking the analysis provided on that website has demonstrated that the hypothesis of the south track, probably too quickly assessed by Inmarsat, is not unique and that the North track can still be an option. This is a great work, but the North track requires more additional hypothesis and complex scenarios.

    How a Boeing can avoid china and India radar military detection?
    It shall be clear that there is almost no way for a Boeing to avoid radar military detection. A boeing is not an flying saucer. The radar track will be detected, and crossing through China and India borders (even going through Bangladesh) will be seen. You may imagine that the aircraft was flying very low, but in that case, it would have been seen by some population on the way (low = noisy).
    Detected by a military radar, does not necessarily mean that military air controllers detected the flight immediately in real time, and awake their boss to launch their jets.
    If an aircraft follows civil routes, without being a threat for a specific area, military controllers may let him go, assuming that they have not received the civil flight plan (but they would expect to have an SSR-code at least). There is also the option for an aircraft without its SSR transponder, to fly near another aircraft to puzzle the radar (two close tracks considered by the controller has the same flight), but the other aircraft would have detected it using its own radar.
    Also, Air controllers may be sleeping in front of their screen, or some equipment may be under maintenance.

    BUT after the crash, all neighboring countries obviously analyzed in depth their radar archive, to find the aircraft, and would have found the radar track for the MH370. This is probably what happened to Malaysia, and the lack of radar data by India and China has convinced the authorities that the south track was the only option.
    So the north route implies that at least China is aware of the real track, but keeps that secret, and let other countries waste their time in the south? Why ?
    It could be somehow logical if the end of the route was in China (a type a spy novel), but less credible then the end point is far in Kazakstan. In that case TWO countries shall be in the same plot.

    For the final step there are two options :
    The MH370 crashed somewhere and, until now, no Satellite, no flight and nobody traveling in the area has seen the wreck (probably in fire). This is really impossible unless we have to go through a complex conspiracy scenario. How to move or hide a MH370 wreck without been detected by someone ? (an isolated area means huge transportation task which would be strange/unusual).

    The MH370 landed somewhere (for instance in Kazakstan), and has been hidden under some hangar. This is another conspiracy option.

    So even if Inmarsat company was a little bit too fast to claim for the south route, globally speaking this is more credible than the north one. But that right that the lack of wreckage is strange.

    1. Again, on the basis of my own experience and also various commentators here I have to differ from the view that an aircraft heading northwards (or indeed southwards) would inevitably be tracked by radar and the data stored. That might be the case in some (developed) parts of the world, but not everywhere.

      1. India-China border is not a piece of cake.
        If you just take the case of Ladakh area, crossed by some of your Northen routes solutions, the MH370 would have avoided that :

        “According to sources, India has deployed 70,000 troops including artillery regiments in Siachen and the Line of Actual Control in Ladakh. Also the Air Force deploys fighter jets in Srinagar and elsewhere in the valley. As compared to India, China deploys more than 200,000 troops along its border with India in Ladakh. That includes mechanized infantry units, the artillery and tank formations. China has reportedly built 14 major air bases in Tibet, 8 missile bases and 17 secret radar stations.”

    2. “So the north route implies that at least China is aware of the real track, but keeps that secret, and let other countries waste their time in the south? Why ?”

      Civilian planes have been shot down before, accidentally or otherwise. They have also experienced mid-air collisions, and that usually means somebody wasn’t flying where they should have been. Incompetence or equipment failures also lead to missed radar sightings.

      While I’m not suggesting any of these things happened, none can be ruled out at this time. If any had occurred, it wouldn’t be particularly shocking that details were kept secret. In fact, “Why?” is probably more appropriately asked as “Why not?”

      1. Again, people keep imagining that certain things imply other things. Here: “So the north route implies that at least China is aware…” It (a hypothetical north route) implies no such thing. Sherlock Holmes would be horrified by the (il)logical jump.

        One might suppose or assume that the aircraft taking some specified route would have been tracked by Country X, but one must remember that is indeed supposition and not evidence.

        On the other hand various commenters consistently ignore such evidence as is available: the ping rings, which negate any route that does not terminate somewhere near either the northern or the southern ping arcs at 00:11 UTC (with an apparent likelihood that the aircraft was still in the air at 00:19 UTC), along with a requirement that the aircraft was on each of the ping rings at 18:29, 19:40, 20:40, 21:40, and 22:40 UTC. That is, unless one decides to decline to accept the Inmarsat-3F1 derived ping rings (handshake radio flight time delays) as being valid evidence.

  20. To Feher Tamas

    That is simply the one more frustrating thing and the same goes for the australian Jorn/Jindalee OTH Radar. Even if the governement states they didn´t see the plane- and we assume it is a true statement- it simply does not exclude the posiibility, that the plane was still flying somewehere there.

    Keep up folks, you ar all doing an awesome job!


  21. Dear Sirs,
    The northernly MH370 path is simply not possible, unless we assume the Kremlin cooperated in a disappearance conspiracy! The russian “woodpecker” over-the-horizon, anti-ballistic missile, early warning radars are real monsters of steel with a few megawatts per antenna. Them, especially the Mishelevka site with its brand new phased radar arrays, would have seen MH370 fly over Asia, even if all of India, China, Pakistan or DPRK were asleep. That’s because all of those asian countries have nuke-tipped, guided ballistic missiles. Those can hit the big russian factories in the Urals, with just 10 minutes forewarn to spare, so the russian mil-space command must be 7x24x365 alert, if they are to live. The fuselage of a B-777 plane is even larger than the american “MX” heavy ICBM and surely would have been seen on russian “woodpecker” indicator displays!

    1. There is a huge difference between “would have seen” and “would have been able to see”. In addition, an OTHR system looking for ICBMs will tend to be ‘tuned’ to look for large Doppler shift targets, moving at far higher speed than an airliner. Again: absence of evidence is not evidence of absence.

    2. I second Duncan’s assumption. Ballistic missiles show trajectories fundamentally different from trajectories of airliners. Furthermore the processing of data received from high-power OTH radar is immensely complex. This together with the objectives OTH radar has to fulfill requires a high degree of automization and pattern recognition and match.
      In the case their OTH radar was technically able to catch MH370 (but discarded by the recognition algorithm, as no ballistic missile) and these “useless” overhead data recorded it still would be a huge effort digging into loads of raw data and checking for data, which could originate from MH370.
      Who should perform this analysis, even if technically possible ?
      It would not at all be the task of a ballistic missile defense squad dedicating resources to the search of a missing civil airliner of a third country.
      Moreover the military is always very cautious disclosing what they are capable of doing ( or even not capable of doing).
      The Australian military assisted in the search, because it was asked by the Australian government and the assumed crash site is located within the Australia’s maritime rescue area.

  22. Someone requested this exercise a while back, sorry got tied up with work. Duncan recommended using an ephemeris for the satellite positioning which really makes sense with only 12 positions and the difficulty, if not impossibility, of fitting curves to complicated orbital physics. Not to mention the unlikely chance that any one working these problems has additional Inmarsat pings… For input I used Duncan’s Inmarstat data as tabularized in Yaps’ spreadsheet V8. I used Yap’s V8 spreadsheet nomenclature for consistency. Hope this helps (maybe in the future if we get more data?) you are all amazing.
    Let me know if something looks goofy and I’ll run to ground, time permitting…
    Inmarsat 3-F1 Satellite positions (MS Excel friendly) equations for 0 ≤ (t) ≤ 461 minutes after March 7, 2014 UTC 16:30:00
    R_x, km =-0.0000000004571*(t)^4+0.0000005472218*(t)^3-0.0001686385053*(t)^2+0.1212962447607*(t)+18120.8581131172
    R² = 0.9999967323455
    R_y, km =-0.0000000012171*(t)^4+0.0000008132146*(t)^3-0.0001013375716*(t)^2-0.0650836643035*(t)+38078.8130575886
    R² = 0.9999949620252
    R_z, km = -0.0000000000000106*(t)^6+0.0000000000118582*(t)^5+0.0000000127586775*(t)^4-0.0000122246963201*(t)^3-0.0079298275861675*(t)^2+3.82762099977845*(t)+830.007896291936
    R² = 0.999999995895992
    R_dot, km/s =-5E-20*(t)^6+3.074E-17*(t)^5-8.7246E-16*(t)^4+1.927113494E-11*(t)^3-1.155759273138E-08*(t)^2-4.84038184841995E-06*(t)+0.00116793316624741
    R² = 0.999999958657692
    lat_dot, deg/s =-4E-20*(t)^6+5.122E-17*(t)^5-0.000000000000020064*(t)^4+4.38677004E-12*(t)^3-8.7318069799E-10*(t)^2-3.7894882565573E-07*(t)+0.0000870136290614873
    R² = 0.999999857383928
    lon_dot, deg/s = IF((t)>250,-0.000003,IF((t)>190,-0.000000016667*(t) + 0.000001166667,IF((t)>119,0.000000014085*(t)-0.000004676056,-0.000003)))
    Took liberties with this equation, keep your darts, I’m not much of a target!
    V_x, km/h =1.73444E-15*(t)^6-1.24993835E-12*(t)^5-5.4571011804E-10*(t)^4+4.3506141636507E-07*(t)^3-0.0000242992999037739*(t)^2-0.0108550931629452*(t)+6.88211832200568
    R² = 0.999996523352755
    V_y, km/h =-1.83347E-15*(t)^6+3.47161054E-12*(t)^5-1.83194645976E-09*(t)^4+2.139831152548E-08*(t)^3+0.000150090013431026*(t)^2-0.0163945715394789*(t)-3.62871429545248
    R² = 0.999999622633711
    V_z, km/h =4.4895E-16*(t)^6-4.56910718E-12*(t)^5+3.98838698179E-09*(t)^4+2.95566729580088E-06*(t)^3-0.00218816334449912*(t)^2-0.952301140696363*(t)+229.607679000095
    R² = 0.999999999588722
    If used in your Excel models, just substitute (replace all) your cell address time input for (t) in the equations. R2 is shown to indicate the equation outputs are not ultra precise, but perhaps fitting for an exercise or study. If someone wants these placed into a spreadsheet, let me know and I would be glad pull together and send via e-mail.

    1. What you have done is helpful. An alternate approach (and one I have been meaning to do myself) is to take Henrik’s equations for altitude, longitude, and latitude and transform them into into x, y, and z using the standard spherical transformation equations. Then differentiate these equations in time to get Vx, Vy, and Vz. This will avoid the need for any curve fits and produce the periodicity that is appropriate for all the parameters. I suspect that Henrik has already done this.

      1. Thanks Victor, I too tried to develop orbital motion equations but unfortunately the predicted values produced poorer results than the “poorman’s” polynomial curve fits for 0 ≤ (t) ≤ 461 minutes. For now I’ve settled on using the curve fitting polynomial equations for interpolation between pings on Duncan’s STK runs, I feel that works for me for the brief period of flight.

        BTW I find myself humbled by the team’s work here. Unlike the recent naysayer (obligate hibernator enjoying the spring thaw in mothers basement) I feel this team is absolutely brilliant and is not trying to prove out a single theory, but instead the team is scientifically exploring all possibilities given the light fare of publicly available raw data. This tireless, nebulous, team challenge each other in professional form while honing each other’s offerings to produce finer outcomes that get all parties involved closer to understanding this most challenging puzzle. “No man of sense who has been generally improved, and has improved himself, can be called quite uneducated as to anything.”

      2. Thanks Benaiahu.

        It would be feasible for me to output satellite positions and speeds at other times if there were a need for them, but I would need specified times at which they were required (in UTC). In principle this is easy; it’s just time-consuming.

  23. @ Duncan (I would appreciate a closer read),

    I’ve stayed away from analyzing the BFO data but started to look at it, yesterday. I looked at Michael Exner’s, Globus Max’s and some of the graphs showing LOS speeds. Having so many comments over this time, it’s hard to keep track of the overall progress. I have a few questions and I would appreciate it you or someone could respond.

    1. Was there any resolution regarding the differences between Michael Exner’s isolated a/c-to-satellite frequency offset which would result in ping rings with radii and the published ping ring? I think his data indicates an ever increasing distance from the satellite from 1829 to 0011. (I interpret a decreasing negative slope of the BFO as still opening. It also indicates an 0011 radius almost 200 nm shorter (2425nm) than Inmarsat’s published. (I used a different methodology using a spherical earth with a radius of 6378km +10 km elevation.)

    2. Globus Max created a chart with the delta velocities vs. the frequency shifts. I remember these shifts correlating with Inmarsat’s. Are the delta velocities used in the calibration oriented so they exclude all non-attributable delta velocities to the frequency offset? If not, this may be one source for the difference.

    I have some other questions, but will leave it at this.

  24. Yap, Henrik, Victorl

    First of all i thank Yap for making the effort to check out the suggestion that the plane could have crashed early on. Yap, can i ask u whether those BFO values u came up with were calculated on the basis that the SDU was operating ‘as normal’, or did u take into consideration that the SDU might not have been getting inputs for the plane’s speed, altitude and position etc some time after 1.07am as Victorl has theorized.

    1. Not sure what you mean by “as normal”. Since the plane’s speed is zero, D1 = 0. So the BFO is just D2 + D3.

    2. Alex have you taken the example point on the 00:11UTC ping arc in the South China sea that Yap used or any other point on that arc in the South China sea and calculated the distance from that point to the satellite sub-point for each of the pings to see how that distance compares for each ping with the ping ring distances that Duncan posted here –

      I suggested previously that you could use the program code that I provided publicly to do this sort of calculation or you could use the GPSVisualizer website –

      Duncan posted the satellite coordinates early on for each of the ping times.

      For your theory of the early crash in the South China sea the distances you calculate will have to be a fairly close match to the ping ring distances that Duncan published. I suggest checking that first before trying to match the BFO data.

  25. LGHamiltonUSA, once again u have come to the rescue. I think the extract from WSJ that u posted points to the eclipse on March 7 having affected the BFO reading for the 19.40 ping: ” The analysis was so advanced that it factored in the switch from solar to battery power, as well as activation of onboard heaters, to identify A SLIGHT CHANGE IN TRANSMISSION FREQUENCY DURING A SINGLE LINK UP BETWEEN THE SATELLITE AND THE PLANE” [emphasis added].

    That would explain why the BFO reading for this ping did not correlate as much to the satellite velocity as compared with the other 5 pings. I had been mulling over this discrepancy, so much so that a couple of weeks ago i asked Duncan whether the satellite velocity for this particular ping stated in one of his early posts had a typo error, to which Duncan replied that the figure was a screengrab from STK.

  26. Richard & Duncan,

    Cheryl #1 here.

    “You got the raw data?” “Nope, not me.” “They’ve got the raw data.” “We gave it to them.” “Nope, we never got the raw data from them.” “They own it.” WHAT??? It sounds like a good rendition of Who’s On First, What’s On Second, and I Don’t Know Who’s On Third. Agreed, farcical. Just trying to give you guys some well deserved comic relief.

    I have nothing to add here, just trying to follow along with all the great technical expertise, except that the circumference of the Circumlocuation Office seems to be widening as much as the search area and getting harder and harder to circumvent. Good luck, guys.

  27. Hi Duncan,

    Many thanks for your continuing efforts to maintain your blog and all commentators (and other lunatics) on a scientific path with the manner of a gentleman.

    BUT, you have certainly rattled Inmarsat’s cage.

    The raw data!!

    Inmarsat: “The raw data is owned by Malaysia.” Ah, but we have not given it to Malaysia.

    Malaysia: “We have never been given the raw data.” Ah, but we have passed it on to all the other international investigators (apart from the French investigators from BEA until Canberra).

    Is this a farce or a tragedy?

    I still think of MH370 as a tragedy, but Inmarsat, Malaysia and Australia are playing out a farce.

    And now, due to guilt feelings about getting the sums wrong (tragic), Inmarsat will in future give away for free the data that it is already collecting (farcical) ?!!??!! This makes the story of MH370 a tragic-farcical-farce-tragedy.

    And to cap it all, the independent audit by the original perpetrators gathering in Canberra appear to have concluded, that they were right all along.

    Yes Duncan, I am laughing, but only because crying would be too painful.


    1. Thanks Richard: your comments are always sane. Of course, we remember that we could all (or at least some of us) be wrong, and Inmarsat has been correct all along, and the aircraft is indeed at the bottom of the Indian Ocean. The aircraft might also be there, yet Inmarsat was wrong (in its deductions). Regardless, the Inmarsat information is the best (and perhaps only) information we have about where MH370 went after 18:29 UTC. We should remember that credit is due to Inmarsat for uncovering the best information about where MH370 may have gone, even if we wish that the company would make the original data available.

      Yes, in some tragic circumstances laughing is the only sane response, else one may follow the path to insanity.

      People ask why we bother over all this. There are three broad reasons, in terms of keeping on the tails of those conducting the ‘official’ investigations:
      (1) To ensure that the friends and family of those who were on board know that someone will keep pushing until MH370 is found;
      (2) To ensure that the flight recorder and/or other information sources that might tell us what happened to MH370 are uncovered, so that the likelihood of future events like this can be reduced; and
      (3) Due to the ‘geopolitical’ nature of this loss and its implications, to ensure that indeed the wreckage of MH370 is identified: without this there will continue to be strife between nations, and finger-pointing, and so on. If the route took MH370 to the Indian Ocean, so be it; if the route took it, instead, to Krygystan or western China or somewhere in the Himalayas, again so be it.

      1. Hello Duncan
        It looks that from France things are moving. The french court is opening a file for investigation because 3 french citizens were on board. Opening a file means experts will be appointed by the court to find reponsabilities. It will be BEA (Bureau d’Enquetes Aeronautiques). Because the Malaysian government did not deliver “certificate of death” (and they cannot) that means the passengers could be alive. In this case all the original data must be forwarded to the BEA including raw data of Inmarsat for investigations. If some body or institution or company refuses to forward to the french investigators, they will be sue under the terms of : “Non assistance a personnes en danger” ( Non assistance to endangered people) and “Entraves a la justice” (Obstacle to cooperate with the justice/court). The penalties could be a lot of millions dollars. So, maybe this action will make things moving in terms of raw data and Malaysian military radar reports !? Let see …

    2. I know this is splitting hairs, but I don’t think the statements about data are necessarily inconsistent.

      Malaysia would be correct in stating they weren’t given Inmarsat’s raw data if they’ve received any sort of converted data instead. This could be as insignificant as a hex-decimal conversion.

      Meanwhile, despite Malaysia possibly having no access to raw data, they still may technically own it. This would not be particularly unusual for a data hosting arrangement.

      Personally, I would interpret these statements as “Malaysia owns the raw data. Inmarsat gave Malaysia all the data they could actually use, but the data in actual raw form wasn’t requested. Malaysia did not get physical access to the data because it wasn’t necessary.”

      1. Hi JS,

        If Malaysia made the statement about MH370 ending up in the Indian Ocean with all lost without checking all the way back to the source data and only relying on derived data, then I think there statement is on very shaky ground. Inmarsat’s claim that there was a peer review by 4 other organisations is meaningless, if the review was only of derived data without the source data.

        But the question remains whether the Canberra Audit has access to the source data or only the derived data. In my view, the Audit is meaningless if it does not allow an independent investigation of the information derived from the source data.


      2. Duncan

        While I certainly agree that Inmarsat and the Malaysian authorities are using each other as PR shields at least to some extent and there is more than a modicum of circumlocuation going on, that should not obscure two obvious points.

        Inmarsat has given Malaysia more data than the Malaysians have released (unless you believe that Inmarsat is flat out lying) and Malaysia can unequivocally call for Inmarsat to release all the raw data from its aircraft.

        Perhaps Inmarsat has released as much as it contractually can or maybe not but Malaysian authorities can end this mobius strip of finger pointing. To me, at this point, the public pressure should be on Malaysia to take the stance that all the Inmarsat raw data should be available.

        (and that doesn’t even touch on the radar data that the Malaysians have withheld)


    3. I think that raw satellite data is owned by Inmarsat, and they have passed the raw data and the analysis to the UK AAIB, who has in turn passed the analysis to the Malaysian authorities. As this is an accident investigation (and a criminal investigation on top of that) Malaysian authorities have the final say what to publish and when. So, technically, Inmarsat is not allowed to publish the data they own (and they probably don’t want to publish it anyway), and the Malaysian authorities have chosen not to publish the data (a decision which they probably have discussed with the UK AAIB).

      1. Dale, probably it is like you assume or similar to that. But let me correct a thing I tried to explain generally in my previous posting: Inmarsat is NOT the owner of the MAS data ! They are the processor. Data owner or data processor make a big difference from legal perspective (and honestly said it is good it is that way).
        Inmarsat could definitely disclose, how they generally derive their BFO values, how they generally process all these data, what properties the receivers inside their satellite have etc. They are free doing that, as they are the owner of these procedures or algorithms.
        Only they won’t do that as details like these are often regarded as business secrets. But basically no third party could prevent them doing that.
        Local authorities, like criminal investigators or aviation accident examine may be empowered as per law claiming access to such data as long as they are needed for their investigations.
        But those authorities won’t have the knowledge to deal with those technical data on their own. Therefore I assume Inmarsat or some other experts would be asked to provide all available facts and expertise provided the laws enable these authorities to enforce such claims.
        Anyway the public ( or media as its representative) could demand answers to these questions, which would not infringe anyone’s personal privacy rights.
        Who those assessors are, by whom they have been appointed and what their exact objects of assession are, would be equally interesting in my opinion.

      2. Dale, I agree that Inmarsat owns at least the ping data. The AAIB in the letter linked by Mark One refused to disclose it in part because, “The data you have requested is owned by Inmarsat, a private sector UK organisation . . . . this Insarmat data has been provided in confidence to the AAIB.”

        I interpret AAIB’s comment to mean that Inmarsat has the power to publish the ping data or not as it pleases, but that the AAIB will not publicize accident investigation information it has been given on a condition of confidentiality (presumably imposed by Inmarsat itself).

        I also agree that the model Inmarsat used to locate the ping arcs and rule out the northern arc might belong to someone other than Inmarsat. Inmarsat is not an eleeymosnary institution. The model may well have been sold pursuant to a contract which vested ownership and disclosure rights in Malaysia. (Of course, if this were so, McLaughlin could simply have stated that the terms of our contract vest ownership of our work product in Malaysia as the ICAO Contracting State, but that would be contrary to his effort to obfuscate the situation, baffle his interviewer and focus on Inmarsat’s free tracking offer.)

        However I know of no basis for your conclusion that giving evidence to a safety or criminal investigation disables the owner of the evidence from continuing to exercise control over it. E.g., an eyewitness to a crime may report it to authorities without in any way losing the right to speak to the media. Accordingly, Inmarsat has issued press releases and given interviews releasing the information it has previously given to the investigators. Only now is it using the investigation as a shield to justify non-disclosure. Likewise, Malaysia released the Preliminary Report (which is marked “confidential”), notwithstanding that it included information previously given to and developed by the investigation.

        In McLaughlin’s most recent interview, he implies that Inmarsat is a “party” to the official investigation and thereby prohibited from releasing the data. I believe he takes this position to avoid disclosure under the umbrella of the position taken by the AAIB:

        Section 44 (1)(a) of the FOI Act provides an exemption from the duty to
        disclose information where disclosure is prohibited under any enactment.
        In this case EU Regulation 996/2010, Article 14. This requires that any
        material collected during an investigation “shall not be made available or
        used for purposes other than safety investigation”.

        I believe that Inmarsat is a sub-member of the investigation team, not an accredited representative, and is not subject to this provision with respect to information originally within its possession and control.

        “The International Investigation Team, which has already commenced its work, has accredited representatives from the US National Transport Safety Board (NTSB), the UK Air Accident Investigation Bureau (AAIB), China’s Aircraft Accident Investigation Department (AAID) of The Civil Aviation Administration of China (CAAC), France’s Bureau d’Enquêtes et d’Analyses (BEA), the Australia Transport Safety Bureau (ATSB) and representatives of Asean from Singapore and Indonesia.

        Included in the team are Boeing, who is under the US Accredited Representative and Rolls Royce and Inmarsat, who are under the UK Accredited Representative.”

        I do not pretend to be an expert in UK Freedom of Information Act law or in the applicable EU Regulation cited by AAIB. Even so, Regulation 996/2010, Article 14 at Par. 3 gives “the authority competent to decide on the disclosure of records according to national law may decide that the benefits of disclosure of the records referred to in Paragraph 1 and 2 [which include the type of information received from Inmarsat] for any other purposes permitted by law outweigh the adverse domestic and international impact that such action may have on that or any future safety investigation.”

        I conclude that while AAIB or other accredited representatives may have the legal right to decline to disclose the Inmarsat work, at least in the name of advancing the safety investigation they also have the legal right to disclose it. The Malaysians have insisted, “we have nothing to hide” and that “it is imperative for the government to form an independent team of investigators which is not only competent and transparent but also highly credible.”

        So no “adverse domestic or international impact” from releasing Inmarsat’s data or work product is apparent. As McLaughlin stated, Inmarsat has no potentially sensitive data such as military radar from other countries.

        In short, I see no legal impediment to Inmarsat releasing the ping or other data it owns, or to AAIB or Malaysia exercising their legal or possible contractual rights to release any Inmarsat data or information they have or to which they have access.

      3. “eleeymosnary: of or pertaining to alms, charity, or charitable donations; charitable.”
        You learn something every day.

  28. Talking of external reviews of the data, in this case the acoustic ping data.

    I would love to look at the data myself. However given the nature of my Defence work, my skill-set isn’t well publicised. So I mostly just appear to be another armchair expert!

    However I do know that ATSB has provided the raw TPL data to many external acoustic analysts. A couple of whom I know well and are much smarter than I. Unfortunately due to confidentiality issues they couldn’t tell me anything about what they observed, I am confident that the data is being properly reviewed.

    Whether those reviews become public information however…?

  29. Hi Duncan.
    Just listened to interview with Senior Vice President Chris McLaughlin of Inmarsat on CNN. Some ‘interesting’ replies to some of the questions.

    Some responses:

    1 The truth is…. it’s not our data?
    2 Data belongs to Malaysian authorities.
    3 We just shared a model (based on the data?) which has been ‘validated’ by other partners (including Boeing, AAIB,).
    How much was shared regarding how the model was derived?
    What did these other parties see?

    4 We don’t have a huge data pool – only 7 data points
    5 Hinted that the model was only a contribution to the investigation (and search area?) – other countries / parties added other sources of information including satellite tracking.
    6 We don’t want to hold back data – we don’t have it ?

    Not sure why they feel unable to release details of how they derived the model so that it can be prudently peer reviewed by others. How would this compromise other elements of the investigation. Are there data protection issues?

    Any comments?

    1. FWIW
      What I really would like to see is a TV interview between Mr McLaughlin ( and as many Inmarsat tech experts as he wants with him ) and people like Duncan, Mike, Victor, Henrik, … so that real technical questions can be asked and answered correctly. I have nothing against TV reporters but this subject is now so and too technical to be treated like this as it is too easy to somehow elude their questions with vague answers.
      But I guess I’m dreaming !

    2. Neil

      I watched the interview with McLaughlin closely and several times. He muddles (in good PR spokesman fashion) the answer to the question of whether they have released all the raw data in order to say that they have given it all to the Malaysian authorities rather than saying whether they have released all the data.

      It seems there must be some kind of contractual obligation that Inmarsat has to their clients that transmit data not to release it without their permission.

      My conclusion after watching him dodge the direct questions is that the release of the raw data is solely in the Malaysian authorities hands.


      1. In a comment to a different post, Arthur Sorkin provided a link to the UK AAIB’s response to a request for information under the UK Freedom of Information Act:

        The response stated that the information being sought was given by Inmarsat to the AAIB, who passed it on to the Malaysians, etc etc. (I paraphrase, not wishing to be accused of any circumlocution here….)

        The AAIB then invoked EU Regulation 996/2010, Article 14 by way of an excuse for not complying, suggesting that under the terms of that Regulation they might be breaking UK and European law by acceeding to the request. And of course that might indeed be so.

        Nonetheless, I can’t help recalling Mr Bumble’s famous remarks about the law, which Dickens so memorably captured in “Oliver Twist” (chapter 51, I think).

    3. Neil, Duncan:

      Thanks so much for your post Neil.

      If it’s not obvious to people by now, there are a multitude of problem’s with Inmarsat’s narrative.

      “1 The truth is…. it’s not our data?”

      Or Inmarsat punts when it’s convenient. Otherwise, how could the UK’s AAIB be in a position to refuse to provide (Inmarsat) ping data to France’s BEA – in response to a request to do so made by MALAYSIA? Recall Duncan’s post:

      “My information is (as noted in other comments from me) that the UK AAIB has refused to make the ping data to others. I do not mean ‘the public’.

      “2 Data belongs to Malaysian authorities.”

      Well, based on the above and what follows, that’s appears not to be the case:

      “Malaysian officials admitted that Malaysia has no access to Inmarsat’s raw data and algorithm in terms of how Inmarsat gained the southern corridor conclusion”

      “Not sure why they feel unable to release details of how they derived the model so that it can be prudently peer reviewed by others. How would this compromise other elements of the investigation. Are there data protection issues?”

      Now you’re getting to the heart of the matter Neil. Re-upping in case you missed:

      1. Hi Nihonmama and Duncan

        There might be some misunderstanding but I’m not sure. I think it’s worth pointing out.

        When my weibo contact told me what she was told during the Beijing briefing on April 29, she referred to the presenters as “Malaysian officials”.

        BUT, there’s a big difference between “Malaysian government officials” and “Malaysia Airlines officials”. As we see, those Malaysian officials presenting slides to the families were likely from Malaysian aviation department and Malaysia Airlines. It’s clear that Malaysia Airlines didn’t have Inmarsat’s raw data. At this stage however, we are not sure whether the Malaysian government had access to it or not.

        According to CNN, a senior Inmarsat official has suggested that the Malaysian government has their raw handshakes data, and it’s up to the Malaysian authorities to decide what’s to release.

        Earlier, Malaysian Defense Minister and acting Transport Minister Hishammuddin Hussein made this comment regarding families’ demanding for Inmarsat’s raw data: Those information might rouse more speculation, which would have negative effect. He said it was better for the International Panel of Experts to decide which information should be made public.

        I would assume the International Panel of Experts is the International Investigation Team put together by the Malaysian government.

    4. It is really not my intention to defend Inmarsat, but there are certain contractual clauses in any contract regarding data privacy.
      To crucial terms are: “data owner” and “data processor”.
      In the business relationship we are referring to, Malaysian Airlines (MAS) is data owner and Inmarsat is data processor.
      Certainly in the “object of agreement” part it is governed which data Inmarsat has to provide to MAS in which format, frequency etc.
      This is probably data MAS needs for running its business / has subscribed to / is paying for – and Inmarsat is also officially offering to their clients.
      Inmarsat has no contractual relationship with the public or with UK authorities. Hence from legal point of view these data are no public data and even if there were some freedom of information act in the UK, it would not apply to Inmarsat (which is a private enterprise and the contract is with MAS and not the British government).
      Coming back to the contract between Inmarsat and MAS and their roles as data processor and data owner:
      Let’s assume MAS had subscribed to only a set of ACARS data A’, while Inmarsat – for whatever reason – is always collecting the full set of ACARS data A and Inmarsat further needs a set of technical metadata M for computing those.
      MAS would under the terms of the contract be entitled to receive A’. Inmarsat need not give to MAS the full data set A and they would not give to them M, also because no third party without deep technical knowledge would be able to take use of such data.
      Nevertheless, in the general definition of the above terms, Inmarsat would only be the “data processor” of A, A’ and M, not its owner.
      So legally Inmarsat is not allowed to publish any of its customer-related data without the consent of MAS !
      As I would regard MH370 as a case of broad public interest one can claim all kind of data available, including technical metadata, secondary computed data etc. shall be made available or shall even be published ( or at least made available to independent researchers) in order to leverage kind of croudsourcing capabilities.
      But this wouldn’t be Inmarsat’s decision, but rather MAS’s decision !
      So we should ask MAS what is their take on that !
      Apart from that bilateral contractual relationship I would assume there may be yet an other party, legally eligible to get insight into some or all of the data, like an authorized aviation investigation body (no idea, whether this is the case and who it might be) or Boeing, in the case Inmarsat has such an agreement with them and this would be a mandatory part of any Inmarsat standard agreement.
      Of course there are also the big intelligence agencies, who have their ways of getting hold of any data without any contract ;)
      Inmarsat will for sure have some intelligence ties, some ground stations are said to be operated in cooperation with intelligence agencies.
      Only thing is that intelligence agencies, in my opinion, do not care about MH370s whereabouts, as it is not their primary business assisting in SARs and thus risking their work gets exposed to the public.
      Somebody in this forum would like to see a public discussion between Inmarsat’s VP McLaughlin and Duncan. Mr McLaughlin is no scientist, not even a technician by education. Most of Duncan’s questions would not be understood.
      However Inmarsat could be asked, referring to the public nature of this case and some “moral obligations” to disclose how they computed some of their technical parameters, like BFO values, satellite capabilities (e.g. transceiver frequency stability) or working methodology in order to allow for validation of computations done elsewhere, like by people in this forum.
      This won’t violate the interests of third parties of theirs, but it may get them into trouble, in case money has been spent on searching in areas, solely based on their (wrong) computations.
      But Inmarsat’s opportunity could be to share this risk by involving third party researchers, like some experts here in this forum and hence get less blamed for mistakes, which may remain undetected.

      1. Gerry,
        Thank you for these clarifications.
        Regarding your comment ” … Somebody in this forum would like to see a public discussion between Inmarsat’s VP McLaughlin and Duncan. Mr McLaughlin is no scientist, not even a technician by education. Most of Duncan’s questions would not be understood … ”
        I’m the one … 8-)
        What you’re saying is fair and perfectly understood. This is why I specifically said ” … ( and as many Inmarsat tech experts as he wants with him ) … “. With 2 non-scientist guys on a TV show, we were likely not going to learn anything new.
        Actually, it’s not the public discussion or the TV interview (to keep the words I used-but I agree it’s the same) which is important. What’s important is a meeting between InMarsat experts and independent experts so they can talk the same language and get some clarifications. But I guess I didn’t convey correctly my thoughts. That meeting doesn’t have to be public and as a layman I don’t have to have all the details. I presume a NDA could probably be signed as well between the parties.

      2. Gerry, you are absolutely correct in your description of the responsibilities and roles of the parties in this scenario.
        However, MAS’ contract for service is with SITA: SITA contracts Inmarsat to provide satcom datalinks along side its own VDL services and those it accesses from other partners.
        Contractually, that’s a very important distinction as you highlight.
        Technically, it’s a very important distinction also. As Bill H discussed some days ago, the architecture of data comms is discussed in layers.
        One of the layers concerns the datalink: the datalink that Inmarsat provides to SITA and is used by MAS aircraft is only part of a chain of links beginning at the aircraft, through the Inmarsat GES, over a terrestrial datalink (provided by any number of global telcos) back to SITA, and on….
        For anyone to suggest that Inmarsat’s datalink management exposes lat/long position data of the aircraft, ships, etc, for which they are carrying data is simply wrong.
        Do note, Inmarsat may offer end user services that are location oriented but that is not what is involved in the MH370 scenario.

      3. Haxi, Duncan

        Hi Haxi: Thanks for your post.

        While Malaysian Airlines (MAS) and the Malaysian government are technically distinct entities, from a legal (and financial) perspective, they are conjoined.

        The majority stake in MAS is held by Penerbangan Malaysia Berhad’s (PMB), which in turn is 100% owned by Khazanah Nasional Berhad, the investment and holding arm of the Malaysian government. So in essence, MAS is a portfolio company of the Malaysian government.

        For purposes of this investigation, MAS, as the owner of the disappeared aircraft, is an interested party — as is Boeing (the aircraft manufacturer) and Rolls Royce (engines manufacturer).

        So, if MAS is a party in interest in this investigation (it is), but it does not ‘have’ Inmarsat’s raw data, does that mean that they are not in possession of the data (but have seen it) — or that they are not in possession of the data, AND haven’t seen it?

        Whichever answer you choose, it’s a problem.

        Moreover, IF MAS, a party in interest to this investigation, does not ‘have’ Inmarsat’s raw data, how could senior MAS executives participate, with straight faces, in a press conference wherein the Malaysian PM informed the world that MH370 ‘ended in southern Indian Ocean’?

        Because, as the “senior Inmarsat official has suggested…the Malaysian government has their raw handshakes data…”

        Which essentially means that MAS and the Malaysian government are one in the same.

  30. Duncan,
    You are leading here an amazing “open source” analysis project. Comparing it to the official one is like comparing the wonderful open source Firefox to the horrible proprietary Explorer. It works so well because people are best motivated by an internal drive to autonomy, mastery and finding purpose (and being appreciated by their peers). Official projects limps because mere financial compensation works only for simple jobs and may have paradoxical results if too high.
    You approved some comments that were a ridiculous personal attack on you and nothing more. Some neighbor you asked to stop throwing garbage via their window?
    There is no need to approve such comments, they are completely off topic, not even an interesting conspiracy theory.

    1. Thanks Ron. I filter most of them out: worse than simple noise. But some trolls are so weird one has to laugh, it’s the only sane response to their delusional behaviour.

  31. Some things that I noted on the Malaysian military radar plot:
    – The radar blips are tagged with 5 or 6 letter and/or number words – of the tag to the 4th last blip before the circle (in which there are no blips) I read the 2 last positions as “TA”
    not all the blips are marked in the same way – it appears that for instance the tag to the last blip before (e.g. on) the circle ends with “H”, like the tag to the 3rd but last blip.
    – The offset of the 4th last blip (roughly 5 nm) suggests that not all the blips relate to one and the same object (as the different tagging would support)
    – Even if discarding some of the clearly offset blips, the track (at least up to the circle) clearly is somewhat twisting/swaying
    – The sampling rate on the plot in any case is less than 10 seconds (calculated with an a/c speed of 500 kts), the silence between the last 2 blips before the circle is 90 seconds, the silence within the circle is at least something like 5.6 minutes.
    – The first blip on the other side of the circle is West of VAMPI, at the most exactly on VAMPI (this shows if one knows that the blips are tagged)
    – Whereas on the first part of the track no other a/c flew at around 02:02H, this is different for the second part of the track – almost exactly 4 minutes after the assumed passage of MH370 at VAMPI, UAE343 arrived there and flew the same route to MEKAR and beyond (data from While refraining from thinking about “shadowing”, maybe the blips of UAE343 pollute the radar plot presented.
    – The very last blip on the plot is slightly south of the line connecting MEKAR and NILAM – the next blip is not on the screen anymore but part of the tag is – taking into account the distance from the right-hand-side of the tag to the blip to which it belongs, places this next blip on the continuation of the line connecting VAMPI and MEKAR (this is actually how UAE343 continued, e.g. UAE343 did not make the slight starboard turn in the direction of NILAM).

    Maybe the above is some food for thought (and maybe somebody is better than I am in making the tags readable). Or maybe somebody knows more about the systematics of such tagging.

    About the gap in the blips just before VAMPI:
    was there ever anything said about that during the presentation of the chart ???
    Apart from the a/c for some reason becoming invisible to the Butterworth radar, I could think of:
    Interruption in the functioning of the radar
    Interruption in the recording of the data
    Nevertheless the decreasing density of blips before the circle would seem to support that indeed for some reason the a/c became increasingly invisible to the Butterworth radar.

    1. Harm, you did an amazing work!

      Some questions and comments:
      * The blip tags indicate secondary radar capability or some software enhancement to primary radar?
      * The blip tags being different means the track is a combination of tracks from two radars or there was more than one aircraft tracked?
      * Did you use a de-blurring filter on the image?
      * I think I read something on the blip gap caused by some topographical barrier.
      * Someone on connected these blips and got a highly wobbly trajectory that would require crazy bank angles.
      His interpretation was that the blip locations are not accurate, more than their diameter would imply.

      1. Hi Ron,
        1) I did read that primary military radar can be enhanced with secondary radar. In fact also ATC appears to have mixed primary and secondary radar capability. I have not yet found tracking images, e.g. derived from a time series of radar plots. Maybe in such tracking images the usual (more extensive) secondary radar labels are replaced by simpler tags. Certain seems to be that the a/c shown on the first part of the Malaysian image was not associated with any secondary radar label and therefore some other tag must have been added – I really do not have a clue as to how the tags are created
        2) I believe that there are at least 2 objects but i guess that we see only 1 a/c in the first part of the track – in the second part of the track I am not so sure
        3) I used sharpening tools but that did not provide much improvement. I saw more by playing with respective RGB color brightness and Gamma
        4) I remember that as well – somebody talked about a hill on Penang Island being in between of the a/c and Butterworth AB. Somebody else observed that actually the radar of the Butterworth AB is on the Penang island hill. The latter makes sense to me.
        5) It crossed my mind as well that the swaying (or wobbliness) might have been caused by inaccuracy. What in my mind speaks against that is the fact that there are also pretty smooth, slightly curved path portions. In the case of inaccuracy I would expect to see more randomness …. I did not see any previous discussion on this subject. I will try to find the thread on

      2. Harm, here is some data on the Butterworth radar:

        1.1.4 In the Kuala Lumpur and Kota Kinabalu FIRs, radar services are provided using the following civil/military ATC Radars:

        k) A 50 NM Terminal Approach Radar with co-mounted 250 NM conventional SSR located one each at Butterworth, Labuan and Kuantan Air Forces Bases;


      3. PS
        to 5) Also the Thai military stated that the a/c that they saw was following a twisting path. That rather supports a possibility that the a/c consistently did not follow a straight track, for which I see 2 possible causes:
        1) a technical problem
        2) a pilot problem (either a partially incapacitated pilot or a non-pilot flying)

      4. Press reports on the Thai radar data from news agencies (when possible):

        Associated Press
        Mar. 18, 2014

        “Montol said that at 1:28 a.m., Thai military radar “was able to detect a signal, which was not a normal
        signal, of a plane flying in the direction opposite from the MH370 plane,” back toward Kuala Lumpur. The
        plane later turned right, toward Butterworth, a Malaysian city along the Strait of Malacca. The radar signal
        was infrequent and did not include any data such as the flight number.
        He said he didn’t know exactly when Thai radar last detected the plane. Malaysian officials have said Flight
        370 was last detected by their own military radar at 2:14 a.m.”

        See also:

        The AFP agency was cited in a Vietnamese paper in English:

        Thai radar saw ‘unknown aircraft’ after MH370 vanished

        An “unknown aircraft was detected at 12:28 (local time, 1:28 am Malaysian time), six minutes after MH370
        vanished” in the East Sea moving southwest back towards its origin in the Malaysian capital Kuala Lumpur
        and the Strait of Malacca, he told AFP.
        That timing corresponds with the last transmission from the aircraft’s transponder at 1:21 am Malaysian
        time, which relayed information about the plane’s altitude and location.
        Although the signal was sporadic, the aircraft was later again picked up by Thai radar swinging north and
        disappearing over the Andaman Sea, Monthon added.
        “It’s not confirmed that the aircraft is MH370,” he said, without giving the exact times of the later sightings.

      5. Hi Ron,

        it is a bit late to reply to this post but better late than never ;-)

        The radar of Butterworth AB is indeed on top of the hill (Western Hill) on the island of Penang. In fact the location is very strategic as it overlooks (in an unobstructed way) the Malacca Straits from a height of 833 meters. Meaning that it can spot objects at sea level (at straight LOS) at a distance of 55 nm or an a/c flying as low as 3000 ft at a distance of 120 nm. In order for the radar to detect an a/c at its maximum indicated range of 250 nm the a/c should fly at least at FL300 (again assuming straight LOS).

        I had overlooked that actually in a different thread here on Bill had already covered most of what I said. He assumed that the tags are time stamps, e.g. a software enhancement to the primary radar images. The thought is logical but despite that it did not really help me in reading the tags – the resolution of the picture is too low – twice the present resolution would be a clear improvement. If Bill is reading this: it would be interesting to know how he succeeded in reading some of the tags.

        Some posts below this post I have posted a link to the radar plot from which the tags are eliminated. From that it is clear that the track does not give any support to a (slight) starboard turn at MEKAR in the direction of NILAM. The track rather supports the a/c continuing on the same heading as between Vampi and Mekar, just like UAE343 did.

    2. Hello Mr. Harm: I took another close look at the radar today. There are actually two independent sources: the screenshot from the March 21st Beijing conference, and the chart Mr. Hussein put up on his fb page. I overlaid them both on my model and readjusted them as best I could. Actually, the screenshot has better resolution than the fb chart, and was able to get the waypoints to line up pretty much exactly.

      Now, the fb chart has two points listed: “Air Defense Radar Point, 18:01:49″, and Updated Last Air Defense Radar Point, 18:22:00. The latter is within about 1 nm of where the final screenshot point is, but the former is 8.5 nm away. Thus, we have 4 points, and 4 possible combinations to estimate ground speeds. Time in this case would be 0.3364 hours.

      But we also have the 18:27 ping ring: it is close enough in space and time to provide a 2nd constraint. Depending if there is a turn toward SANOB or a continuation on course to IGOGU, one can come up with 8 combinations of distances and times.
      time dist speed route
      0.336 166.2 494.1 R2-R1
      0.336 167.0 496.4 U-R1
      0.336 175.5 521.7 U-A
      0.336 174.6 519.0 R2-A
      0.421 203.0 482.4 S-R1
      0.421 211.5 502.6 S-A
      0.421 204.5 485.9 I-R1
      0.421 213.0 506.1 I-A

      It’s hard to choose any particular combo for a principled reason. However, by pooling the data, hopefully errors will come out in the wash. I get an average ground speed of 501 knots, which is reasonable. According to wind data provided by Brian Anderson here, there should have been a tailwind component of around 10 kts, so that’s a TAS of 491 that is pretty darn close to the standard cruising speed listed in the Wikipedia article.

      1. Hi Warren,

        Fascinating analysis! Well done!

        You state “I get an average ground speed of 501 knots”.

        In the very first comment on this post (please see above), I stated “the last known speed of MH370 which I calculated at 500.36 knots”.

        I am pleased to see that we are in agreement on our analysis of the speed in the initial part of the flight at the time leading up to 18:27:48 UTC.

        The next questions are:
        1. Can we agree on the position at 18:27:48 UTC for the Northern Flight Path of 6.8511N 95.7619E or for a Southern Flight Path of 6.1797N 95.8528E?
        2. Can we agree on the position at 00:10:48 UTC for the Northern Flight Path of 43.8100N 61.8900E or for a Southern Flight Path of 39.6853S 81.8992E?

        I would appreciate very much hearing your analysis for the latter part of the flight.


      2. Hello Warren,
        Mr Hussein has many many pics on FB but I did not see the chart there. I guess that you refer to the updated radar plot chart with also ping ring Information incorporated thereon. Indeed for my purposes a lesser quality.
        As to your analysis of probable ground speed: your result seems to be very consistent also with the average ground speed that would be required to reach the 18:01:49 location in time starting from a bit beyond IGARI and acommodating a 2 minute, almost 180 degree turn from a bearing of 40 degrees to the Penang (or close-by) waypoint.

        I guess that the tags I was referring to are flight identification codes (FLT ID). These have up to 7 (e.g. can be less) alpha-numerical characters. I guess that there is a code for unidentified a/c.

        Kind regards,
        Harm – which, for the records, is my first name ;-)

      3. Warren, when you say “the chart Mr. Hussein put up on his fb page” you mean this one?

        Studying the radar track may tell us if it’s compatible with the ping rings. I read the plane had to go supersonic to get on time from the track’s end to the old ping ring. With the new ping ring it had to fly back. All these data points have errors whose size is unclear but they may still be useful to support or disprove the identification of MH370 with the aircraft that made an “air turn back” to Malaysia. The Malaysian International Investigation Team apparently couldn’t resolve this issue as evident from the evasive wording in the Preliminary Report .

      4. Yes, that is the chart.
        Your later comments, Ron, appear to be way off the mark. The ping rings/arcs available are from 18:29 UTC onwards, unless you are wanting to investigate the ping rings through to 17:07 (i.e. times for which actual positions of the aircraft are available).

  32. May 10, 2014 : Australian ship returns to MH370 search

    During the 1st mission in the search area, the Ocean Shield track was always visible on MarineTraffic
    (the track is not visible anymore with the free version)

    For this new mission, when Ocean Shield left Perth, the track was visible. But suddenly, it’s not visible anymore !
    Why ? Going to a new search area ? Why is this hidden to the public now ? Is this following instructions from someone ?

    On a side not, the JACC is not even updating their page.
    Yes, the Circumlocution office is really alive !

    1. @Louis:
      MarineTraffic on 6 April indeed decided to open up Sat-AIS, which is a premium subscription service, to the public “due to public demand in the course of the search for MH370″ as they said.
      Free AIS, which uses VHF and only works for line of sight, similarly as ADS-B for airplanes. But unlike ADS-B, which is broadcast from an airborne plane at high altitude and hence can be tracked up to 450 km, terrestrial AIS is transmitted by ships, traveling at sea level. Hence coverage is limited to 50-max.100km, i.e. coastal waters.
      Now, MarineTraffic switched off their free Sat-AIS service for the Indian Ocean area again a couple of days ago.
      This is why you could see Ocean Shield leaving the harbor, navigating out on the ocean and suddenly “disappearing”.
      That’s happened when Ocean Shield left the coverage of the VHF-AIS receiver(s) located somewhere on the Australian coast near Perth.

      for the Indian Ocean During the 1st mission in the search area, the Ocean Shield track was always visible on MarineTraffic
      (the track is not visible anymore with the free version)

      So, no mysterious instructions commanded by someone, simply a commercial decision by MarineTraffic.

      1. @Gerry
        Thanks Gerry for this explanation.
        I was just wondering why without implying anything. I just regret this commercial decision as seeing the actual track could eventually help people on this blog in seeing / identifying that a new search area has been decided or to see that Ocean Shield was returning to the same area. A commercial decision, in this context, is to me something that is very hard to accept and even to understand.

  33. Regarding eclipse:

    As noted in the WSJ report posted by LGH, a couple of things happen when geosynchronous spacecraft experience a solar eclipse. During most of the year, the spacecraft is in full sun 24 hours a day and the spacecraft is powered directly from the solar panels. The panels power all payloads and keep the batteries charged. But for a period lasting ~45 days centered on spring and fall equinox, the solar panels do not receive sunlight all day. The eclipse lasts up to 70 minutes on the day of equinox. This varies a bit depending on the inclination. On March 7, 2014, the eclipse period would have been ~25 minutes, but it would have peaked at ~19:42, which as luck would have it, coincided with the 19:40 handshake. Thus, the spacecraft would have been powered by batteries at that time, and some auxiliary heaters would have been turned on to keep temperature sensitive parts of the spacecraft close to normal temperature. This is an interesting fact, but the BFO data does not suggest any significant effect on the transponder local oscillator offset, nor would I expect one. The LO is certainly well insulated, precisely to minimize this effect. I’ll be interested to see how close my back of the envelope calculations match Duncan’s more exact STK analysis.

  34. Inmarsat Press Release – 0700 BST, Monday 12 May 2014 : Inmarsat to provide free global airline tracking service

    “… Because of the universal nature of existing Inmarsat aviation services, our proposals can be implemented right away on all ocean-going commercial aircraft using equipment that is already installed. Furthermore, our leading aviation safety partners are fully supportive of expanded use of the ADS-C Service through the Inmarsat network. This offer responsibly, quickly and at little or no cost to the industry, addresses in part the problem brought to light by the recent tragic events around MH370.”

    … right away on all ocean-going commercial aircraft using equipment that is already installed …
    MH370 had the equipment. There was likely a test period before this press release. Was MH370 tracked with that new service during this last flight ? Hence the reluctance to release raw data until now and before this press release ?

    1. Quite likely. It has been suggested before that a good possibility for Inmarsat not wanting to release the raw data was not due to conspiracy or malignancy, but rather just a protection of propriety technology or non-disclosed services that are not appropriate for public release. Time will tell…

      1. Thanks Luke.

        Yes, agreed. Throughout I have (I hope) said the same thing: that I see no sign of ill-thought on the part of anyone (well, apart from the trolls), including the governments involved, just incompetence and/or the desire to protect their own jobs and power structures.

        And of course I freely admit to a lack of competence on my own part in certain matters, for example the analysis of the BFO values. There is no such thing as a universal expert. Anyone who has a command of even one part in a billion of all human knowledge would be astonishingly smart.

    2. ADS-C is part of the story for FANS (future air navigation system). The challenge with evolution in the airline industry is that its global : all airlines and all air traffic services moving towards the same goal – it’s not a lean, agile team. Together with aircraft that have an economic service life of 15 to 25 years, change is slow. That’s the way it is: change happens region by region, driven by a multitude of priorities.
      I believe ADS-B implementation is moving faster: its cheaper to implement on the ground and VHF range covers the areas where surveillance is most needed: departure and arrival. But ADS-C has its benefits over oceanic and remote regions.
      Link to a SITA doc that illustrates and discusses ADS-C in context with its supplementary technologies: (warning: acronym hell)
      ADS-C doesn’t stand or fall on Inmarsat’s offer of free transit for the data packets: data transit is a fraction of the cost for ADS-C. ADS-C requires a ground infrastructure in the ATS systems, it’s planned to be implemented alongside CPDLC (controller – pilot data link comms) and AFN.
      Do consider that all those technologies simply automate the existing manual procedures that were expected to be carried out as 9M-MRO continued across the Gulf of Thailand crossing the FIR boundary into Vietnam’s air traffic control area (but its systems ceased to operate) – I have to pose the question: how is FANS including ADS-C, going to avoid a repeat of an incident where an aircraft’s communications “go off the air”?
      Let’s just leave Chris McLaughlin to do his job, i.e., telling ‘a good story’ for Inmarsat.

    3. It is a good thing it is free so not every country and company will have to send up their own satellite…

      Not even crowdsourcing would manage that!

      And then there’s binding agreements.

  35. First, the Malaysian Insider (quoting David Gallo) and now, the WSJ:

    “Searchers preparing to resume the underwater hunt for Malaysia Airlines Flight 370 increasingly suspect that some of the electronic signals detected last month didn’t come from the jetliner’s black-box flight recorders”

    1. Not just some of the signals, all of the signals. They did not come from the black box pingers. Too many things that don’t add up. The frequency being that far off, while at the same time the signal strength goes up? Not possible.

    2. Hi Nihonmama,

      Well, we’ve already discussed how the good doctor has no expertise to be giving us his highly valued opinion on the expected performance of Dukane’s pingers …

      Anyway, thanks for the WSJ link. Because this is behind a paywall, I’ll just give a quick summary here :

      The captain of the Ocean Shield (who presumably has good connections with the Australian investigators – my comment) says that increasingly, only the first two acoustic detections (April 5) are being considered relevant to the search for MH370. The April 8 detections were only 27kHz, which is considered too low (I don’t know if that’s the manufacturer’s view, this is not mentioned). Also, the Bluefin 21 has not been able to scan one third of the previous area, because some is too deep and because of up-to-70m peaks and crevasses toward the north-west. New equipment will be sought, to be decided Wednesday.

      As you are probably aware, Ocean Shield is about to resume sonar scanning, this time in the first acoustic detection area (previously, the second detection area was scanned).


      1. Thanks Nihonmama and Joe: Now let’s all calm down a little. Please, no personal comments here, either in terms of people being ‘experts’ or ‘fools’ or ‘unqualified’ or anything in between.

        Once arguments get personal, logic and reasoning go out the window, and that’s what we don’t want. Please remember the quote I gave about a true scholar being one who is delighted when an error in found in his/her reasoning: the implication being that with the error identified, the mass (including he/she) can move closer to the correct answer. So, let’s leave our egos at the door as we walk into the friendly debate.

      2. In response to Joe, but for others who may not be aware: paywalled articles (WSJ and other) can almost always be accessed for free (one-time) by plugging the title of the article into Google.

      3. The pings were actually at 33khz which is right in the range of every other transmitter and transducer used in the ocean. There is a paper that was written on the Air France pingers that sat in the ocean for 2 years and when they did their initial testing on the beacon it was at 34khz, but it was also only putting out a fraction of the signal strength it should have been putting out. There was some kind of a current drain that was causing major issues with the unit. It is my belief that the reduced frequency was based on the current drain keeping the piezo from operating correctly rather than the oscillator changing frequency.

        My contention from day one has been proven by the examination of the AF beacon, that the first thing to go on one of those units would be signal strength and there is no way that you are going to get frequency drift of that magnitude without reducing the signal strength to a level that would be undetectable from more than a foot away.

      4. Hi mung,

        You should not make glib statements about “other transmitter and transducer used in the ocean” in comparison to the Ocean Shield acoustic detections. When you look closely at these, you find differences in frequencies, coding. etc. If you have a specific model in mind, then a comparison can be done.

        You are correct in that the AF447 pinger output level (after recovery) was extremely low and its current consumption was somewhat high.

        However, you should not jump to the conclusion that this relates to the low 34kHz tone it produced after recovery. In a further test after 18 days at 45C (moisture removal, would also reduce some mechanical stresses), the pinger then produced 36.2kHz, almost within specifications, yet the output level had not changed (if anything, it was even lower) and the current consumption was almost unchanged.

        For reference :


      5. Differences in frequencies? Uh, no. 33khz is 33khz it doesn’t matter what the source is. The timing of the pulse bursts may differ, but we haven’t really heard what they claimed to have picked up. Even the media is now saying that what they heard could have been another transmitter.

        Now, lets talk about oscillators. Short of a serious malfunction, there is nothing that will cause an oscillator to drift that far in frequency. The only possible cause of a frequency shift in that circuit would be something impeding the piezo’s movement. I would have loved to have seen them on the AF investigation take the board and connect that to a new piezo.

        There are way too many things that don’t add up with the beacons and the pings they heard. I have 0 confidence that they heard the beacons from the flight data reporters.

    3. I just posted the following on Rodney Thomson’s blog — — A logical inference of the May 12th Wall Street Journal article is that the Joint Agency Coordination Centre has been sitting for some time on information that invalidated what Angus Houston told the press on Wednesday, April 9th: “On Monday I advised that the Towed Pinger Locator deployed by the Ocean Shield had detected signals consistent with those emitted by aircraft black boxes on two separate occasions. I can now tell you that Ocean Shield has been able to re-acquire the signals on two more occasion—late yesterday afternoon and late last night, Perth time.” See Note the phrase “re-acquire the signals.” While Mr. Houston may have thought that was true on April 9th, surely the JACC learned within a few days that the signals had not been “re-acquired” because the frequency was much too low for the source to be the same. Just another reason to be skeptical of the JACC.

  36. Duncan,

    I sure am sad that you placed yourself out there where the whole world can see that not only do you not understand any analysis or physics to tat matter, You are so vain, that you just wanted to be ranked on Google and in the end, the Google search will top the charts as you being a complete wacko.

    Nothing you have done in you report even confirms to = True. In fact, its all false and fudged numbers with an opinion unsupported by any real computation.

    Furthermore, There is NOT even a suggestion of a Northern Arc, or any facts to your theories and fudged calculations as well as stolen work as it is plagiarism and incorrect as well down right inconsiderate of the families.

    All the language and history shows, not only are you unable to understand any of the Data, you do not understand how to calculate the required steps if it was handed to you on a silver platter.

    Although Inmarsat initially issued a report showing a Northern and Southern Corridor, No such Corridor ever existed and has only placed conspiracy theories and whack jobs to compile numbers and fudge results.

    Moreover, all the people who assisted you in good faith not only were hoodwinked by your falsehoods and wild scenario, they now are much less knowledgeable than they were on their own findings, as they did all the work and was mislead by your pipe dream.

    So Mr. Ping Ring, keep chasing your circles as thats all you have done was lead everyone in a circle of ignorance and scammed them in your desire to be goggled.

    Have fun with your misleading and unfortunate misguidance/



    M E Marciniak @felinenut

    1. Hi Narciniac!
      I’m psychologist and you sure are a great kid. Duncan, feel free to moderate this comment out of the northern or sourhern hemissphere, but let’s all appreciate how much the disappearance of MH370 has gripped the collective imagination.
      All the best to our latest guest blogger and everybody working a little harder.
      Yours sincerely,
      Anders I

    2. Wow, I love trolls like you. Since you have no ability to actually prove the theories on this site wrong, you just call names and shout conspiracy theorist at anyone who questions the official story. Why don’t you come back here and provide some actual facts that prove that you are right and Duncan is wrong? I would love to see you try.

    3. Thanks for posting-
      I needed a chuckle…
      This is sure to humorously amuse all the followers of these posts!

    4. Yes Duncan, I have been meaning to talk to you about your desire to be goggled. It really has to stop!

  37. Sorry Duncan, i should have cited source references. Re KL ATC tracking the plane up to 1:30am on their primary radar, the head of Malaysia DCA said on March 8th Mh370 was last seen on ATC radar at 1:30am, he repeated this statement in the days that followed. For eg from the Star which is one of Malaysia’s main newspapers, on March 8th:

    ” The last signal position of Mh370 on the Department of Civil Aviation’s radar was at 1:30am Saturday. Director-general Datuk Azharuddin Abdul Rahman said this was MH370’s last position on the radar before the signal disappeared at 1:30am. ‘The signal suddenly disappeared’ Azharuddin told reporters at the Operation Coordinating Centre at the Sama Sama Hotel here”.

    From the WSJonline on March 13th: ” Azharuddin Abdul Rahman, director general of the Department of Civil Aviation told a news briefing that air traffic control lost contact with Flight 370 on its secondary radar system at 1.21am Saturday before losing contact on the primary radar at 1:30am. At that point the plane was cruising at 35,000 feet above the South China Sea, he said. As is standard international practice, Malaysian controllers use two radar systems, a primary and a secondary, to monitor their airspace.”

    Regarding the tracking by Saigon ATC on their primary radar of MH370 up to BITOD, Saigon ATC had said 3 times in the recorded calls between the 2 ATCs on that night listed in the schedule to the Preliminary Report that they last saw the plane at BITOD. At 1.38.19 – “… radar target was last seen at BITOD.” At 1.41.21 – “……… Ho Chi Minh advised that the observed radar blip disappeared at waypoint BITOD”. At 2.18.53 – “……… HCM confirmed earlier information that radar contact was lost after BITOD…….”.

    Curiously, Malaysia, on their part, omitted mentioning in the Preliminary Report that their ATC had tracked the plane until 1:30am. The main text of the report merely says ” At 1.21.13 the radar label for MH370 disappeared from the radar screen at …..KLATCC”. If the entire signal/blip had disappeared from their radar at 1.21am, it would make sense to say MH370 disappeared from their radar screen at 1.21am. The fact that only the radar label disappeared must mean the blip remained, as the DCA had repeatedly said in the days following the plane’s disappearance. It also begs the question as to why Malaysia deliberately declined to say in the Preliminary Report that they had tracked the plane until 1:30am on their ATC primary radar. This ATC primary radar recording has not been produced and has reportedly been ‘sealed’.

    1. The Vietnamese and Malaysian ATC narratives do not match.

      airlandseaman: “I have been taking on faith the reports that MH370 turned at IGARI and followed the path relentlessly reported on news channels. IF MH370 was in fact tracked to BITOD before dropping off the radar, it raises serious questions about the entire story and assumed path after loss of vioce contact. Maybe there is a good explanation for the decision to ignore these reports, but the May1 report makes no effort to reconcle the IGARI turn assuption and these BITOD reports.

      Alex Siew: “Regarding the tracking by Saigon ATC on their primary radar of MH370 up to BITOD, Saigon ATC had said 3 times in the recorded calls between the 2 ATCs on that night listed in the schedule to the Preliminary Report that they last saw the plane at BITOD. At 1.38.19 – “… radar target was last seen at BITOD.” At 1.41.21 – “……… Ho Chi Minh advised that the observed radar blip disappeared at waypoint BITOD”. At 2.18.53 – “……… HCM confirmed earlier information that radar contact was lost after BITOD…”

      From an investigative and evidentiary perspective, this discrepancy is not only significant, it is fertile ground for unpacking and further inquiry, particularly given that the Malaysians have already SEALED some information, including radar data. To say that this is irregular would be beyond an understatement. So I’ll repeat myself: how does sealing (of anything) possibly occur in the middle of an investigation, when you don’t yet have the plane, bodies or the black boxes?

  38. Hi Duncan,

    Sorry about the essentially duplicate postings above, about the alternative satellite LO frequency. The “Comment awaiting moderation” preview wasn’t working for me last night, so I didn’t think the first posting ‘took’. Feel free to delete one of them.


  39. @ Johan Ohlsson,
    If I’ve got it right (I’m no technical wizard and *certainly* no expert on radar), KL-ATCC would have been using the plane’s ADS-B transponder data just as FR24 did, so that the radar blip on their screen would have been accompanied by an identification label showing which plane it was. I’m not sure how that works in combination with primary radar.

    Someone please correct me if I’ve got this wrong?

    In the Preliminary Report, KL-ATCC spoke of the “radar label” disappearing at 17:21.13. I believe this “label” would have been generated from the ADS-B transponder data. (Mind you, KL-ATCC also said they’d observed the plane at IGARI at 17:21.04, which I reckon is too late – so who knows?)

    The “Actions Taken” appendix to the preliminary report claimed that Ho Chi Minh established radar contact with the plane over IGARI but that their radar “blip” vanished when the plane got to BITOD. It’s not clear whether by “blip” they meant “label” or whether they were referring to a primary radar “hit”. (Can anyone clarify that?)

    But if either party kept a record of whatever transponder data they captured, I somehow doubt they’ll be sharing it with you and me, any time soon…

    1. Thanks Mark One,

      Sometimes you have to shoot even without a target…

    2. Re: Mark One ( )

      The ‘transponder’ (Mode-S radar returns) and the ADS-B are two different mechanisms to report a/c position & identity.
      The transponder responds to secondary radar surveillance scans with its Mode-S reply, civil air traffic services use this technology to track commercial aircraft.
      ADS-B is a regular broadcast, from the a/c, of the same information.
      ADS-B is becoming more prominent in air traffic management. Australia now mandates a/c must use ADS-B in their airspace when flying above FL300 to maintain continuous surveillance. Other regions will follow.

      ADS-B, as a broadcast, is openly available to anyone with a capable receiver, hence the feeds to FlightWare and FlightRadar24 from ‘enthusiasts’ around the world.


      1. Thanks, Don. I was hoping for some clarification, which you’ve helpfully provided.

        Despite having read up feverishly on radar technicalities over the past few days, I’m only now beginning to understand the details. If I now comprehend it correctly, the Mode-S transponder returns information to the secondary radar (SSR) interrogations and also passes information through to ADS-B, which then broadcasts it to “enthusiasts”.

        That would imply that the reason FR24 received no more data was because the Mode-S transponder was no long communicating with the ADS-B system, hence no more ADS-B output. And it would explain why, simultaneously, the radar label vanished from the ATC screen in Kuala Lumpur.

      2. The transponder units can provide the Mode-S return and the ADS-B broadcast, different antennas but same ‘box’.

        The enthusiast reception is just a side effect of the broadcast. As I said, air traffic services are adopting ADS-B to varying timetables around the world.

      3. Mark One,
        Just as a note:
        As primary radar has a limited range, the label disappearing from screens at KUL ATC could, hypothetically, be a question of distance, and not the
        malfunction. The plane was leaving Malaysian airspace.

        I think it is established with the report that their atc did not notice in 17 minutes that the “plane” (label?) “was gone from the screens.” (Quoting Guardian). The atc responsible for Mh370 might have considered him done with it as they said “good night”, and focussed on another plane instead. Noone noticed it coming back, apparently. All this might be in the report, but I don’t remember.

        “Plane disappeared from screen” is thus not an exact measure of transponder failure, as I think you too hinted at. The report might be more exact.

  40. Henrik, Victorl, Yap,

    Regarding Henrik’s statement that ‘the LOS aircraft movement is not visible in the BFO graph because the aircraft transmission correction algorithm is working perfectly”. Would this be inconsistent with Victorl’s theory that there was a disabling event after 1.07am with the SDU not getting inputs on the aircraft’s position, speed etc from thereon?

    Victor’s theory seems to be backed up by considerable evidence. The SSR and ADS-B signals were lost at 1.21 am suggesting electrical failure of some sort. The plane continued to be tracked by both KL ATC and Saigon ATC on primary radar up to 1.30am all the way to BITOD before it disappeared from primary radar. BITOD is 37nm from IGARI meaning the plane had traveled at an average speed of 250 knots from IGARI to BITOD. The loss of primary radar coverage after BITOD indicates the plane was declining in altitude from IGARI to BITOD. 250 knots happens to be the gliding speed for a 777 ie its speed if there is no engine power. The loss of speed from 470 knots to 250 knots and the drop in altitude would suggest the plane had lost power at IGARI and was gliding from that time onwards. At the reported glide ratio of 18:1 for a 777, it would mean the plane can stay up in the air for another 24 or 25 minutes and glide for more than 100 nautical miles. This would be consistent with the alleged SOS call from the plane picked up at 1.43am by the US military base at Utapao as reported by China Times on March 8th citing the US Embassy at Beijing, with the pilots on that call saying the cabin was disintegrating and they needed to make an emergency landing.

    As regards the nature of the disabling event, there is some evidence that the plane was possibly struck by positive lightning at 1.21am. Could it be the BFO chart for the 6 pings merely reflects the movement and velocity of the satellite from a plane which had already crashed early on at the South China Sea (say at the intersection of the plane’s last known flightpath-BITOD 40 degree and the portion of the arc crossing the South China Sea)? An early crash would mean the rear upper fuselage where the SDU was located, floated above water after the crash and the SDU had some sort of battery backup to respond to the satellite’s interrogation, to emit those faint and irregularly timed pings. It would also mean the arcs for the first 5 pings, which have not been released, would overlap with the arc from the final ping, after discounting the movement of the satellite.

      1. If 9M-MRO had glided 100nm from BITOD it would’ve reached the Ca Mau peninsula & not even got wet, even IGARI to dry land on Ca Mau is only 120nm. It’s real flat there too.
        A certain sense of deja vu to this line of debate.

  41. Good question LGH. The answer is that crystal oscillators generally get better (lower drift rate) with aging. As they age, they sluff off molecules causing a very small decrease in mass, thus increase in frequency. This trend tends to decrease with time as all those pesky loose molecules depart. The LES Pilot tracking system measures the satellite transonder LO offset and compensates for it on the ground, before the downconverted IF signals go to the demodulators. The BFO value is sampled after this corretion, thus transponder LO offset does not appear in the Doppler calculations. As the LO reference oscillator ages, and the drift slows down, the job of the Pilot system gets easier.

    The accuracy of the BFO measurements is not materially affected by the aging. Diurnal changes are very predictable, and easily modelled out, but shorter term temperature changes driven by eclipse events take a bit more special handling to tune out. This is, no doubt, what was going on with I3-F1 on March 7 because the satellite would have been going into an out of eclipse that close to equanox. (Earth caused eclipses can continue for up to 45 days around equinox with the satellite being in the shadow for up to 70 minutes each day.)

    It would be nice if someone skilled in such things could tell us the exact time span I3-F4 was in eclipse on the 7th.

    1. Amazing that you mentioned eclipses because I had been meaning to mention that after someone had written somewhere that temperature would be slowly-varying, and an eclipses is contrary to that.

      This is something I can do easily in STK. Wilco.

      Other readers might also like to note that this is why your home ‘satellite TV’ pictures can sometimes (i.e. some times of day) become more ‘snowy’ than usual around the equinoxes: the satellite you are using (and therefore the direction in which your antenna/dish is pointed) can align with the direction of the Sun, and the Sun is a strong radio emitter.

      1. Hello Mike ~

        Chapter 3 has a discussion about the correlation of anomalies and SEUs with the eclipse seasons:

        Data Management of Geostationary Communications Satellite Telemetry and Correlation to Space Weather

        “To understand and mitigate the effects of space weather on the performance of geostationary communications satellites, we analyze sixteen years of archived telemetry data from Inmarsat, the UK-based telecommunications company, and compare on-orbit anomalies with space weather observations.”


  42. Hello Team ~

    Any temperature issue may or may not have a significant impact on the Doppler shift and BFO values, however considering that I-3F1 is in orbit beyond its expected economic life I can only question whether vital components such as it’s OCXO are operating at optimal performance levels. If performance has degraded I then question what adjustments Inmarsat has needed to apply to the raw data to have derived the predicted path analyses:

    VictorI wrote: “If the SATCOM was operating as intended with a low value of residual D1, why are position/speed/heading data missing from the ping packets?”

    Henrik Rydberg wrote: “All crystal oscillators exhibit a small temperature dependence that can lead to substantial errors when looking at the hertz scale.”

    Investigators incorporated arcane new calculations reflecting CHANGES IN THE OPERATING TEMPERATURES of an Inmarsat PLC satellite as well as the communications equipment aboard the Boeing 777 when the two systems exchanged so-called digital “handshakes.”


    1. LGHamiltonUSA,

      Many thanks for this supporting reference. Regarding why Inmarsat is interested in the temperature dependence, my thinking is that it affects the baseline contribution to the value at the last ping ring, and hence greatly affects the actual angle along the ping arc where the aircraft is to be found. In our latest development, we are ‘cheating’ by using Inmarsat’s model data, present in the graph, to extract that baseline.


      1. Hello Henrik ~ I’m really glad you found the information useful. The satellite operating temperature reference in the article always struck me as very unusual. ~LGH~

      1. From the WSJ link provided by Bill we glean a significant nugget of information:

        “The last major refinement of Inmarsat’s data came in early April, when the team analyzed the cooling of the satellite, which orbits around 22,000 miles above the equator, when eclipsed by the Earth. The analysis was so advanced that it factored in the switch from solar to battery power, as well as activation of onboard heaters, to identify a slight change in transmission frequency during a single link-up between the satellite and the plane.”


  43. I have been looking a bit closer at the path between the radar sightings and the 19:41 ping ring. In order to match the BFO, it seems the flight must have started to head north before it made a very tight turn to the SW. Indeed, it looks like the plane ‘skirted’ the peninsula.

    The path in the link below roughly matches the ping rings and the BFO data, reproducing the strong peak and the right level after the turn. A constant speed of 490 knots is assumed, similar to Warren’s waypoint paths. More refinement will follow, but I thought I would share this now anyways.


    1. Henrik,
      Did your path intentionally ignore the “Military Radar” photo that appears to show the plane passed over VAMPI and then MEKAR, with ground speed > 500kts, exiting the screen around 02:22:12 MT (18:22:12UT). If you omit the radar, then the plane’s speed and timeline for when the plane passes this area could be different.

      If there was a turn to the right (clockwise) in progress during the 18:26, 18:27, 18:28 time period (to match the BFO), that turn may have continued (clockwise) until it was heading the correct direction to intersect with the 19:41UT ping ring. …Or, the turn might have finished with the plane heading in some other direction, with another turn later bringing the plane around to head towards an intersect with the 19:41UT ping ring. Does the data and path modelling indicate one of these is more likely than the other?


      1. Hi Bill,

        I was not paying particular attention to the MEKAR point, but the speed and timings of the legs amount to roughly the same positioning, so there is no contradiction. In fact, the ‘bad’ points can be removed with much the same qualitative behavior, see updated link below. The main point is the turn north before it goes west. Basically, it seems very reasonable to have assumed a northern heading on the plane, leading to the initial search towards the adaman island.

        Regarding possible continuations, I have been looking especially for such possibilities. I do not see any other solution around this time of the flight than north-west-south. As to why, it is anyones guess.


  44. Voice370, representing families of the passengers and crew, plea for release of the raw Inmarsat data.

    “5. We implore the Malaysian Government to share and release the raw Inmarsat satellite
    engine ping data for 9MMRO (every ping from Friday, March 7 12:00 until the final signal was received globally so that it can be subject to broader analysis by relevant experts.”
    Read more at:

    Also Duncan, Tim, and Michael have been mentioned in The Atlantic, which features and cites Duncan’s graphs.
    “So it should be straightforward to make sure that the math is right. That’s just what a group of analysts outside the investigation has been attempting to verify. The major players have been Michael Exner, founder of the American Mobile Satellite Corporation; Duncan Steel, a physicist and visiting scientist at NASA’s Ames Research Center; and satellite technology consultant Tim Farrar.”

  45. It seems that some commenters have misunderstood or taken out of context some of my comments about the BFO analysis. I wish to clarify. In particular, the table I posted here:

    …has been interpreted by some to mean that I believe the TX offset WAS zero after 17:07. However, that is not true. I posted that table to illustrate that, in the unique case for offset=0 after 17:07, the result would be “pure Aircraft LOS Doppler”, thus LOS speed. That would be a fortuitous failure mode if it happened. But I did not claim, or intend to imply that that is what actually happened. I have said from the start that offset = 0 is ONE possibility. But I also suggested other possible post 17:07 offset values, including the possibility that it continued “stuck on” the last known good offset, or possibly defaulted to some “flywheel algorithm”. Since then, the work of others, notably Henrik, Yap and Victor, seems to suggest that the “flywheel algorithm” is more likely than a zero offset, or other fixed constant. I tend to agree.

    Rockwell’s patent (and model SATCOM 906) use the AES received Doppler to calculate the transmit Doppler offset. But this technique has a problem. The received carrier is not continuous under some conditions (Class 2 operation). It can have gaps up to 5 seconds. To meet the offset limit specification (100 or 185 Hz) during 5 second gaps, Rockwell uses a “flywheel algorithm“. See patent at Section 13 here:

    They describe two methods, one that uses a fixed offset value, and one that uses a fixed value plus a rate of change, in both cases based on the offset and offset rate of change prior to the loss of received carrier. Rockwell claims that more sophisticated models are within the scope of the patent, suggesting the possible use of sinusoidal turn dynamics discussed in the patent.

    While Rockwell needed to incorporate such a flywheel default algorithm to qualify the method for class 2 operation, best practice would seem to dictate incorporation of a similar default algorithm even if known 5 second outages were not a problem, as in the Honeywell method. Honeywell engineers would likely incorporate a similar algorithm to insure compliance if IRS/429 data was lost briefly. The default algorithm in the event of an extended 429/IES data outage is less certain, but it is reasonable to assume that the short term default might be of the form Offset(t)=offset(t=0) +OffsetRate*t. In essence, this is what Henrik, Yap and Victor are finding empirically. IOW:

    BFO=C band Doppler + L band LOS Doppler + Offset(t=0) + OffsetRate*t (if IES/429 data not available)

    Of course, left running indefinitely, such an algorithm would exhibit something analogous to PID Loop “integrator wind up” limits if OffsetRate was not zero. So there might be some limit put on how far the algorithm would be allowed to coast. What it would do after a minute or two is harder to guess.

  46. This discussion is at the limits of my mathematical abilities but I do know that smart people need to keep pushing this case or it will be dropped. I do not for a second believe that this plane disappeared – in the sense that no one , anywhere, has an idea where it is.

  47. Victorl, Yap, Henrik,

    I had posed a question in Duncan’s preceding post, unfortunately just before it closed for comments, so i think u guys must have missed it.

    The BFO chart for the 6 pings seems to be tracking the satellite velocity for that period. Why is that? Is it because the SDU was not getting input of the aircraft’s position, speed etc from AIMS and thus the aircraft’s side of the equation was taken out from the Doppler calculation?

    From 2.29am the velocity of the satellite was decreasing as it headed northwards reaching the apex at 3.36am. We dont have a ping at exactly 3.36 am but the BFO for 3.40am was lower than for 2.29am so that would be consistent with the decrease in the velocity of the satellite for that period. From 3.40am (or 3.36am) the velocity of the satellite kept increasing and on the chart the BFO values kept increasing as well. At 2.29am and 4.40am the velocity was roughly the same at 49 n 47 knots respectively and the BFO values for those times are also the same around 140 hz ,with the reading for 4.40am slightly lower than 2.29am, just like their respective speeds. In fact, if one deducts a fixed offset of 90 from the BFO values, one gets roughly the satellite velocity in knots as the balance, for 5 of the 6 pings (eg at 2.29am, 140-90= 50, velocity was 49 knots: at 8.11am, 250-90= 160, velocity was 159knots at that time). The only exception is the ping at 3.40am where the velocity was 3 knots, but BFO(110) – 90= 20.

    There may be a reason why the number for 3.40 is a bit off. The satellite had just reached its northern apex a few minutes ago at 3.36am. In an earlier post, in answer to my query, Duncan said the satellite on that night reached 20 knots at around 4.03am. It may be that after 1.07am (subsequent to the disabling event) the SDU was using default satellite data (where the point of the northern apex is reached at an earlier time), accounting for the above trend BFO.

    1. Alex,

      “The BFO chart for the 6 pings seems to be tracking the satellite velocity for that period. Why is that?”

      Yes, this is precisely what seems to be happening, and the key observation that the current (meaning the last couple of days worth of posts in this thread) model is based on.

      It is really very simple: the line-of-sight aircraft movement is not visible in the BFO graph, because the aircraft transmission correction algorithm is working perfectly. However, we do still see the effect of the satellite movement, because the algorithm assumes the satellite to be at a fixed position.

      As a result, give or take other contributions to the graph that do not depend on the flight path or satellite, the BFO at the later pings is in essence a plot of the speed of the satellite times the latitude of the aircraft. This is why we can say for certain that the plane went south.


  48. Finding a linear function to give BFO from aircraft movement, assuming a transmit correction based on a fixed satellite is nice, but a little worrying because of the amount of parameter selection and adjusting involved to fit a small number of points.

    To put it into perspective, can I give an imaginary conversation at Inmarsat HQ between A (a boss) and B (a worker bee):

    A: “thanks for the map with 4 likely simple paths based on ping rings, B, I reckon it went south, but can we rule out the 2 northern paths some way?”
    B: “hmm – I know, we can get the BFO from Perth logs, and at each path point, from the assumed aircraft position and velocity, we can calculate what the aircraft transmitter would have removed from the Doppler by applying its frequency offset, which I know is based on a fixed satellite assumption – I’ll work it out and report back.”
    A: “thanks for the BFO graph, but it looks meaningless, its way out!”
    B: “well, we can ignore the first few points, as we are not completely sure where the plane was and it may have been turning, but notice that the later parts of the 4 predicted BFO lines are pretty linear, and if I just shifted them vertically they would be close to the linear part of the measured BFO points, and if I rescaled them it would be a very good fit.”
    A: “yeah, but even I know that is cheating, of course straight lines can be matched if you use 2 adjustments!”
    B: “OK, I have a great idea – I will get as many BFO measurements from the “straight and level cruise” part of as many known flights around that time as I can, and I will divide them randomly into a training set and a test set, then I will get the best linear fit on the actual known flight position and velocity against Perth BFO logs using only the training data, and then see if those parameters work on the held out test data, and I will report back.”
    A: “good, make sure the collected data is from flights using the same transmitter as MH370.”
    B: “amazing, it worked on the test set, looks like we have built a good simple BFO predictor. The only thing that bothers me is that there is absolutely no physical meaning to the offset and drift rate parameters I just fitted.”
    A: “don’t worry about that – how does it work on the last parts of our 4 hypothetical MH370 paths?”
    B: “a fairly good fit to all, but a slightly better fit to the south paths”
    A: “fantastic, the search team will be happy, give us some of your charts, and I’ll have our graphic artists and PR team put out a public release – this makes a nice story”
    B: “I could work on a few more refined paths, say by taking wind effects into account.”
    A: “No, don’t bother, that effect will be trivial, and time is of the essence if that black box is to be located, and anyway we have the search area reasonably pinned down now, great work B, you can take a break.”

  49. MuOne, Richard

    One more:” Immarsat swears that it is absolutely certain they have the right device ID for the terminal. If it turns out they are wrong there are going to be some very very embarrassed people in the UK”

    About a week after the above post, someone on pprune asked a question on the spoofing issue and a very specific (and illuminating) response was provided – by a poster trained in the satellite business. And then, pprune admin deleted it – #8789. I am still kicking myself for forgetting to screencap it.

    1. Nihonmama, maybe the deleted #8789 response still lives in Google cache? Try prefixing the URL with “cache:”…

      I think this blog readers can devise better conspiracy theories than those filling the Internet, they just prefer to first exhaust the non-conspirational ones. To illustrate this point, at the risk of being deleted, here is another conspiracy theory:

      The Malaysians are not sure that MH370 turned back after contact was lost. They say the plane that came back “behaved like a commercial airline” and was “deemed” non-hostile. Now assume for a second it was something like a Boeing E-3 Sentry or a Boeing E-767. With its new turbofan engines the old E-3 Sentry can fly for about 11 hours so in principle it could:
      * Lift the MH370 SATCOM ID and co-pilot phone’s ID during a fly by
      * Imitate MH370 to the Inmarsat satellite (using similar antenna and equipment)
      * Connect briefly with the Penang cell tower while flying near it
      * Produce the famous satellite pings while flying the official route
      * Drop into the ocean non-standard 33.331 KHz acoustic pingers (where do you buy these?)
      * Land safely in Australia or some other place

      This conspirational scenario would require MH370 to quickly crash or land so there wouldn’t be two sets of SATCOM signals with the same ID.

      Now to fight fire with fire here is another conspiracy theory: All conspiracy theorists are unwittingly serving the Martians as Reverse Search Engine Optimization (RSEO) agents deflecting and discrediting facts and ideas that the aliens don’t like. Nice?

      1. You forgot the bit about returning JFK to the White House after his half-century imprisonment under Area 51 with Marilyn Monroe.

        Never forget that many people watching the X-Files and the like think they are viewing documentaries.

        Now here’s what must be a real conspiracy. In 1967 the Australian Prime Minister, Harold Holt, went missing whilst swimming alone in the sea at a beach near Portsea, Victoria. He was never seen again. The proof of the conspiracy is that the perpetrators later named a swimming pool for him in Melbourne.

      2. Ha ha.

        You know, Ron, the one thing conspiracy theorists will never accept is that regardless of if the plane really went down in the jungles of Cambodja, the black box will still turn up in the South Indian Ocean…

      3. Hey Ron:

        Someone (with a background in mechanical and aeronautical engineering) responded in a LinkedIn group to a post of Ari Schulman’s Atlantic piece. And he quoted Hercule Poirot: “when reasoning doesn’t match facts, something is wrong with facts”. ;)

        I will try again to see if can surface that deleted pprune post re satellite spoofing. Thanks!

    2. Satellite spoofing is an interesting idea, partly because Iran claims to have spoofed GPS to hack a drone in flight to land in Iran. However, I really don’t think it possible to spoof a satellite to a specific aircraft without being close to the aircraft or having a directional antenna tracking it. This sort of spoofing would probably only be done by a state for military purposes.

      If any state made MH370 disappear, it can be assumed there has been effort to hinder recovery.

      1. Satellite spoofing is an interesting idea, partly because Iran claims to have spoofed GPS to hack a drone in flight to land in Iran. However, I really don’t think it possible to spoof a satellite to a specific aircraft without being close to the aircraft or having a directional antenna tracking it. This sort of spoofing would probably only be done by a state for military purposes.

        Inmarsat is satcom, not navigation. Satcom would not be a primary spoofing target for taking an airplane, it would be secondary. An easy way to detect GPS spoofing in an aircraft would be GPS deviation from inertial navigation or magnetic compass headings.

        If any state made MH370 disappear, it can be assumed there has been effort to hinder recovery.

  50. Duncan: Thank you for the last. I applaud your democracy-in-action approach to this blog.

  51. MuOne, Richard

    Mu: your scenario —

    “A (Wildly speculative and bordering on conspiracy theoretics) scenario, could be “land the plane – collect the system – mount to some other aircraft- send off other aircraft including system down into the southern ocean, where it is never to be found” —

    may not be as wild as you believe. See this which includes a 2013 TR (MIT) article on how the AIS system (ship tracking) was spoofed.

  52. Ron,

    The ping and Doppler data for each timestamp come from the same transaction between the airplane and the ground station.
    All of the satellite ‘data’ Inmarsat has is data gathered at the ground station based on the ground-satellite-plane communications. No data is retained on the satellite, and any data that may be retained on the plane is not accessible right now.
    From each transaction, Inmarsat has been able to extract data (which is not published) to calculate the results that have been published-
    – The “ping” time measurement data (used to determine the original ping ring(s), then the elevation angles, which lead us back to new ping rings)
    – the received frequency in some form (used to plot out the BFO, which lead to South=’Yes’, North=’No’ — and hence, all of this discussion :-)

    Some elaboration-
    The satellite subsystem on the plane can only communicate with one satellite at a time. It would only be switched to another satellite if it was heading out of the coverage area of the current satellite. The ping rings indicate that the flight remained well within the coverage area of the Inmarsat 3F1 satellite (elevation angles 40-55 degrees). From the ground network perspective, the connection must stay with a single satellite so that the correct ground to satellite link is used.
    To switch to a different satellite, the current connection must be ended (airplane log off), then the antenna wpould be aimed to the new satellite, and a then new connection is established (airplane log in).

    1. Dear Bill and Ron, isn’t this manual switching of satellite connections you describe exactly what this pilot is saying MAS flights do on routes to China? ( I have been beating my head against a wall though to get it verified, because I think if this is true, this switch off point is when things went nuts )

      “It is Mas procedure to switch ACARS, VHF, and High Frequency selection off but this is only for flights to China as the service provider for Mas does not cover China. Some if not all pilots switch them all off for a while and then later switch SATCOMM back on to force the system into SATCOMM mode.”
      Read more: MISSING MH370: ACARS cannot be disabled – Latest – New Straits Times

      I think he is referring to the satellite coverage when he says service provider , as SITA handles MAS in China . And perth Inmarsat station receives traffic from both satellites I think?

      1. Just to remind people that even if the same ground station is used to receive data from multiple satellites, generally that will be through separate antennas (dishes). You can’t have high gain *and* wide beams.

      2. “To switch to a different satellite, the current connection must be ended (airplane log off), then the antenna wpould be aimed to the new satellite, and a then new connection is established (airplane log in)” —-
        But this wouldn’t this fit the reported ‘ switch comms off, wait few minutes, switch back on to connect to new satellite’ procedure (above) that MAS does on flights to China? What would it mean for the reported data if it did actually connect to the China coverage satellite? And wouldn’t Perth then have 2 sets of data from 2 different connections?
        I also wonder if the satcomm started to malfunction at this switch over point, maybe that is why SITA has been unduly reticent about sharing raw data, as it would be rather embarrassing.

      3. roseny6,
        see Rons success at discussing this above. The key word is “voice”. There will be more to dig up there.
        They are not switching satellite but (satellite) service agreement.

      4. Again roseny6,
        as I have understood it, the reporter is not referring to the log-out and log-in when switching between satellites, but the shutting down of VHF to open up for Satellite comm., or, as Ron also writes, switch to the voice mode of the VHF3 radio, if I got that one right. If there are any more special particulars to this case they will no doubt be sorted out by seven horsemen.

      5. SITA is MH’s service provider, the crew has the option to select SATCOM as only datalink. Satcom datalink service provided to SITA by Inmarsat is global, period, no national exclusions. That even the ground msgs at were seen over satcom, when VHF was in range, evidences that data over satcom rather than auto or vhf was the preference here.

    2. Bill (and Duncan and others),
      There are to you knowledge no other vendor or player that is keeping similar (and processable) values in connection to the flight transponder, as long as it was operable? I realize this is just about time, and that it may not make any difference.

      I have noted the flightpath-plotting with flightradar-values, but you can’t get BFO curves out of them, can you?

      1. Johan,

        If you’re referring to the ADS-B flight transponder, the page at states that FR24’s data is collected from a large number of receivers, often privately owned. The data is then compiled (I presume, “on the fly”) and converted into “live” maps that can be viewed on its website.

        The “B” of ADS-B stands for “broadcast”. Anyone can pick up the transmissions of any transponder, although you need to have the necessary receiver and must be within range.

        FYI, the “live flight” of MH370 can still be seen at As the URL suggests, this presentation begins about eight minutes after takeoff.

        FR24 also has an archived (“pinned”) version of the flight but for unknown reasons this version seems to be based on slightly different data:

      2. Thanks Mark One,
        My smartphone is hardly not the optimal instrument, for one who is not immediately familiar with the names and the vocabulary. I meant ADS-B (if the flight transponder works through that). My thought was that KUL ATC, other primary radar, and perhaps even flight scanner enterprises might have not only the “tracking” but also the “signals”, just like Inmarsat, which could be turned into BFO values to compare with the ones from Inmarsat. But I guess it doesn’t work like that, for one, and that they aren’t saving them, if it is even possible.

  53. I doubt you would want to post this as it is very long, but I thought you might find it useful/interesting as an evolving reference. A guy called CopperNickus put it together on reddit and regularly updates it .UTC MYT Source Note
    16:00 00:00 ACARS Init [2] (corrected from radio)
    16:07 00:07 ACARS Preflight [2] (corrected from radio)
    16:12 00:12 ACARS Engine start [2] (corrected from radio)
    16:25:53 00:25:53 Radio Contacted ATC [5]
    16:26:53 00:26:53 Radio Transferred to Ground Control [5]
    16:27:45 00:27:45 Radio Pushback and start approved [5]
    16:28 00:28 ACARS Pushing back [2]
    16:30 00:30 ACARS Engine power up [2] [7]
    16:32:13 00:32:13 Radio Request taxi approval [5]
    16:32:42 00:32:42 Radio Taxi approved [5]
    16:35:53 00:35:53 Radio Contacted Lumpur Tower ATC [5]
    00:36:30 Radio Runway A10 32R holding at runway [5]
    00:38:43 Radio Runway A10 32R approved for runway [5]
    00:40:38 Radio Runway A10 32R approved for takeoff [5]
    16:41 00:41 ACARS Takeoff [2]
    16:42:10 00:42:10 Radio Directed to Altitude 18000ft (FL180), chg course to IGARI [5]
    16:43 00:43 ACARS Ascending, Course Change to IGARI [2] [7]
    16:46:51 00:46:51 Radio Contacted Lumpur Area Radar ATC, Directed to FL250 [5]
    16:50:06 00:50:06 Radio Directed to FL350 [5]
    16:56 00:56 ACARS ? Ascending / Reached FL350 ? [2] [7]
    17:01:14 00:01:14 Radio Confirmed FL350 [5]
    17:07 01:07 ACARS ? Reached Cruising Altitude ? [2] [7]
    17:08:00 00:08:00 Radio Confirmed FL350 [5]
    17:19:24 01:19:24 Radio Handoff to Ho Chi Minh ATC (HCM-ATC) [5]
    17:19:29 01:19:29 Radio Last radio contact
    17:21:00 01:21:00 Report Arrives IGARI [8] (est)
    Transp Last position tracked by FlightRadar24 ADS-B (transponder) FL350 471kts 040[10]
    … … Transponder disabled [9]
    … … Brief HCM-ATC Tracks MH370 from IGARI to/near BITOD (36nm NE) on radar, no verbal [4]
    17:28:00 01:28:00 Report Thais track bogey westbound from IGARI area [9] (UFO if you prefer)
    17:30 01:30 Report JAL750 radio contact with MH370, garbled transmission [11]
    … … Report MH088 radio contact with MH370, garbled transmission [12]
    17:37:00 01:37:00 Report Missed ACARS check-in
    17:38:19 01:38:19 Brief HCM-ATC contacts KL-ATCC MY …[4]
    17:57:02 01:57:02 Brief … Continued discussion [4]
    18:02:00 02:02:00 Radar Radar contact near Palau Perak, west of MY
    279R 89nm from Butterworth AB, N5°41.99′ E98°55.34′ [6]
    Consistent with turnaround between IGARI and BITOD at 450kts [14]
    18:03:46 02:03:46 Brief MAS OPS reports plane responding from Cambodian airspace [1][4]
    This entry conflicts with 18:15/2:15 as to when MAS OPS was notified [4]
    18:07:00 02:07:00 Brief KL-ATCC asks HCM to verify Cambodian airspace (Phnom Penh FIR)
    Attempts to locate MH370 via ATC continue for hours (Hainan, Singapore, Hong Kong, Beijing) see [4]
    18:13:22 02:13:22 SkyV Arrival VAMPI 450kts [13] [14]
    18:15 02:15 Sat KL-ATCC Watch Supervisor contacts MAS OPS [1][4] conflicts with 18:03:46/02:03:46 [4]
    MAS OPS sends ACARS messages, may have triggered the next 3 sat/doppler events
    18:20 02:20:36 SkyV Arrival GIVAL 450kts [13] [14]
    18:22:00 02:22:00 Radar Radar contact beyond MEKAR waypoint 295R 200nm from Butterworth AB (WMKB) [6]
    Conflicts with actual location of MEKAR at ~285R 240nm from WMKB [6]
    Conflicts with [14] Matches: N6°51.76′ E97°20.95′
    18:26 02:26 Sat Doppler indicates turn (implies sat traffic, not a ping) [7]
    18:27 02:27 Sat Doppler indicates western heading (implies sat traffic, not a ping) [7]
    18:28 02:28 Sat Doppler indicates turn (implies sat traffic, not a ping) [7]
    18:35:54 02:35:54 Brief MAS OPS indicates plane at N14.90000 E109.1500 east of Vietnam. (later retracted) [4]
    18:39 02:39 Brief MAS/Marfan sat phone call resets timer [2] [3] (but no doppler data ?)
    18:53:51 02:53:51 Brief Flight MH386 asked to attempt contact with MH370 [4]
    19:41 03:41 Sat Keep alive Ping [7]
    20:41 04:41 Sat Keep alive Ping [7]
    21:30 05:30 Brief Kuala Lumpur Air Search and Rescue Centre activated [4]
    21:41 05:41 Sat Keep alive Ping [7]
    22:41 06:41 Sat Keep alive Ping [7]
    23:13 07:13 Sat MAS/Marfan sat phone call resets timer [2] [3]
    00:11 08:11 Sat Keep alive Ping [7]
    00:19 08:19 Plane Partial comm initiated by plane [2]

    1 Briefing: Phone calls were not mentioned in this briefing, but OPS messages were.

    2 Blog post on briefing:

    3 Blog post on briefing:

    4 MY summary:

    5 Radio transcript:

    6 Radar plot:

    7 Doppler release:

    8 News Report:

    9 News Report:

    10 Flightradar24 FB post:

    11 JAL750 report:

    12 MH088 report:

    13 Report:

    14 Skyvector:,94.41943361097128&chart=301&zoom=10&plan=G.2.749508673586428,101.72145082334617:G.2.92233136294601,101.6102142510735:F.WS.IGARI:F.VV.BITOD:V.WM.VKB:V.WM.VPG:G.5.695833,98.922000:F.WM.VAMPI:F.WM.GIVAL:F.VO.IGREX

    Ping Timings Reddit Thread

    Source Column indicates where the data originated

    ACARS: ACARS data from family briefing, some overlap with Doppler [7]

    Brief: Malaysian family or news briefing

    Plane: means unknown system on the plane

    Radio: From ATC radio recording/transcript [5]

    Report: News report

    Sat: Satellite data from doppler and elevation charts

    Calculated and known locations:

    IGARI 17:21 01:21 450kts, takeoff, course change, est.

    Radar 18:22 02:22 450kts, turn somewhere between IGARI and BITOD

    VAMPI 18:33:20 02:33:20 Skyvector est.

    GIVAL 18:40:40 02:40:40 Skyvector est.

    1. @ roseny6

      Thanks for this exhaustive timeline summary.

      I would caution that different sets of conflicting data from FlightRadar24 exist (see posts elsewhere in this blog). It’s not clear why these datasets disagree, or which, if any, is truly reliable.

      Also, it seems to me that the FlightRadar24 data (assuming sufficient reliability) show MH370 passing IGARI at 17:17 UTC, rather than the 17:21 you source from Reuters. The change of heading from 25 to 40 has been interpreted by others as a turn towards BITOD. If the turn is “real”, it seems to begin at 17:20 (cf three FlightRadar24 data point headings: 28, 40, 40). A minute or so later, data reception by FR24 ceased.

      This is obviously one of the crucial moments in the flight. To summarise (and approximating timings for simplicity), we see:

      17:08 Plane reports: “maintaining FL350″ (repetition of TX of 17:01)
      17:17 Plane passes IGARI
      17:19 ATC instructs: “contact Ho Chi Minh”. Plane responds: “Goodnight”
      17:20 Plane begins turn (?) towards BITOD (?)
      17:21 FR24 receives final data from plane

      I understand the distance from IGARI to BITOD to be 37nm. At the speed given by FR24 (average 473kts) and assuming IGARI was passed at 17:17 UTC, I think the plane would have been about 6nm short of BITOD when its transponder fell silent.

      Would more technical brains than mine kindly review these musings and give feedback?

  54. I emphasize that the implication of Henrik’s and Yapp’s work is that the BFO data really can be used to distinguish between North and South routes, and the South route matches the measured data, given the assumptions that went into the analysis such as constant speed and the algorithm used by the SATCOM to apply the frequency correction term.

    1. Let’s suppose that Henrik and Yap have accurately replicated the model that Inmarsat used to predict the north and south paths. The implication is that despite the SATCOM’s performing as intended, the BFO grows to a large value of 269 Hz . If you look at 00:11 UTC, for instance, here are the contributions (based on Yap’s spreadsheet): residual D1 = -11 Hz, D2 = 52 Hz, D3 = 101 Hz, (drift + offset) = 127 Hz, BFO = -11 + 52 + 101 + 127 = 269 Hz. (The measured value was 252 Hz.) Even taking out the C-band shift leaves you with 269 – 127 = 142 Hz of shift, of which 101 Hz is due to (drift + offset). For a SATCOM that is properly functioning, I wonder whether frequency shifts like this are likely? Or does the data imply that the SATCOM was not functioning properly, possibly due to bad or absent inertial data, and the actual residual D1 was much higher and the (drift + offset) much lower? If the SATCOM was operating as intended with a low value of residual D1, why are position/speed/heading data missing from the ping packets?

      1. Victor,

        Glad to hear that you have been able to reproduce the data from the TX offset model. I believe you may have slightly different linear offsets, but I does not matter much at this stage. Did you optimize with both ping rings and BFOs, and do the paths differ from what you have seen before?

        The doppler shift from the L-band is no more than 40Hz, so that is not a big value at all. The only question is whether a drift of 10 Hz per hour is reasonble, whatever the cause. Bear in mind that the frequency budget applies to the channels within the band, but the whole band can certainly drift without causing any issues – and it certainly does at serveral points along the chain through the converters and down to earth. Given the amount of headache that has gone into the rationale for this or that limit, without really getting us anywhere, I will ignore that aspect for now. It is what it is.

        Regarding the pings, it has been stated here before (sorry for not finding the reference) that those consist of the link-level protocol (look up L1/L2 in the OSI model). Hence, there are no traditional packets being sent at all. This is all consistent with the systems that usually send packets being taken off line.


    2. Henrik,
      You are to be commended in that you have reverse-engineered Inmarsat’s model by applying your D1 correction and linear offset. This is a potentially a big advance. However, I am not sure that the Inmarsat has properly modeled the behavior of the SATCOM. Perhaps I do not have enough of the humility you have cited. Perhaps it is because I am projecting my own humility, and the willingness to accept that I often err, onto the actions of others. A number of us are trying to dig into the actual performance of the Honeywell MCS4200 SATCOM unit to understand better how it applies the Doppler correction. It very well may be that Inmarsat is correct in their assumptions. However, until we have physical evidence that the plane went down in the search zone, a number of us will question all assumptions. The drift/offset values (180 Hz offset, -0.115 Hz/min drift) that are necessary for your model to collapse the data seem very large in light of the 100-150 Hz Doppler shift budgets that are available. I am not faulting your analysis, which appears to be sound and clever. Rather, I am questioning Inmarsat’s analysis.

      1. Thanks Victor.
        Yes, I am of the opinion that a plea to humility is much the same as a plea to authority: invalid.
        The essential thing required to get the truth is that everything should be questioned, tested, verified; and such questioning is never concluded, even for the law of conservation of momentum.
        We should try to conduct this questioning, testing, verifying as ladies and gentlemen, with proper respect for the ideas of others. If for no other reason we should conduct ourselves this way because then it is easier for the others (or indeed ourselves) to put our hands up and say: “OK, I was wrong. Now let’s move forward.”
        If, in such debates, someone’s ego gets bruised then perhaps that ego was too large in the first place.

    3. If the drift/offset is occurring in the SATCOM, then it will shift the frequency of the data on its assigned channels relative to the channel/frequencies assigned to other aircraft. I don’t think the whole band would shift, as it does due to C-band (downlink) Doppler.

      1. I was thinking of the drift in the satellite, which is out in space with old technology, suffering from huge temperature variations due to solar eclipse. The aircraft TX offset would also matter if the error was in the ppm range, but people have for the time being convinced me that it is accurate to the ppb level.

        May I also refer to the latest report regarding the temperature experiments, relayed by LGHamiltonUSA.


  55. I have gone through the work of Henrik and Yapp by following the inner workings of Yapp’s spreadsheet. Here is my interpretation of what they did:

    BFO = (D1 + Dcorr) + D2 + D3 + + Co + C1*t

    Where Co + Ct applies a linear correction to the data and Dcorr is the correction to the L-band Doppler shift calculated by assuming a stationary satellite at its mean position. The contribution of satellite velocity D2 is assumed to be uncorrected, as is the downlink Doppler, D3. The drift term C1 is 0.115 Hz/min and the offset is Co is 180 Hz.

    The variation in BFO for the North and South paths at 19:40 UTC is mostly due to the differences in (D1+Dcorr) for the North and South paths because D2 is small. At later times, the difference between the North and South paths is mostly due to the differences in D2 as the correction Dcorr becomes more accurate as the plane moves away from the satellite.

    I like this model. It is better than anything I came up with. Is 0.115 Hz/min drift a reasonable value?

    Yapp: I would stay away from trying to correlate the term Co + C1*t with D3. I think the concept of frequency drift correlated with time is better.


    1. Victor, Yapp, Henrik,

      I posted some thoughts on the D1 contribution to doppler and/or AES side doppler correction above, which in short discussed the “heading-speed” vector vs “COG-VMG” vector (i.e. actual velocity).

      What are the implied errors in velocity for a 0.115Hz/min drift? Are these in the order of the prevailing winds (or rather their projections onto the LOS vector)? If so, the drift term could be related to the plane moving relative to a large scale weather system, where the local wind speeds steadily increase/decrease from centre to outside or vice versa.

      If, as proposed, the applied D1 correction is based on heading and speed, rather than the actual (wind affected) velocity vector (COG and VMG), the introduced doppler difference would appear to drift in time, as with increasing time, the plane would find itself in regions with ever increasing/decreasing wind speeds and therefore increasing/decreasing correction errors.

      Maybe C0 + C1*t can/should be related to (D1 + Dcorr), i.e. the AES side offset calculations. It could represent a first approximation of the “wind effects” potentially present in the BFO data.


      1. Hi Will,

        If the TX correction depended on anything but the ground velocity, it would turn up as a factor on the radial doppler correction. As such, it would be directly visible in the BFO, because the radial doppler is so much larger than the other terms. Even a 5% error in that estimate would distort the BFO graph substantially (I checked). It would basically be sad if it was the case!

        Since so many other systems depend on accurate positioning, there seems to be little reason to assume anything but a true position-based velocity, but never say never. Systems including gyros and accelerometers are in abundance these days (anyone has a cellphone?), so if the AFC includes such a system, it will function properly even if the position feed is absent for long periods of time (read minutes or so).

        As an anecdote, inertial systems can be remarkable accurate these days. A friend told me about one particular case, where such a device was mounted on a fighter jet which was subsequently sent out to do all sorts of maneuvers. When it came back a couple of hours laters, the integrated position from the intertial system was only off by a couple of kilometers.


      2. Hi Henrik,

        Thanks for contemplating my post in such detail. I take that as a bit of a confidence builder that my thoughts contribute a tiny bit and are not just noise. Unfortunately, I don’t have the time or means to build my own model and test my theories or post results of my own, so I rely on others to test my thoughts, if they find them plausible or worthwhile. So, thanks for that.

        There is a possibility to sanity-check, i.e. verify/falsify the theory with the early pings, where there is a large discrepancy between measured and expected data during the initial start and climb phase of the flight. I assume the initial speeds are relative low, hence wind effects comparatively large. Later at cruising speed, wind effects are comparatively smaller. Accurate wind data for the start time and region (airport) should be available.

        I could imagine that inertial systems could be used/considered in the software as a mere back-up system, since (I assume) e.g. the GPS system is orders of magnitudes more accurate and its data contribution shouldn’t be “polluted” by the less accurate inertial data. However, a periodic processing of all available data may still be done to re-synchronize everything and rebase the individual contributing sensing systems.

        In this context I note the odd dip in the fourth BFO data point, bringing the measured data very accurately back to the predicted graph point. This is then followed by a large discrepancy again in the fifth. That dip “appears” to interrupt a relatively stable rate of error/diversion between measured vs expected data points 1,2,3,5. I understand that these points are rather sparse and one should not read too much into the interconnecting graph lines, but I do have that niggling feeling that I can’t shake yet.


  56. I believe that post-17:07 L band Doppler data that are believed to be free of AES TX corrections may not be.

    I submitted a comment on April 28th that describes a conceptual approach to determining potential MH370 tracks and speeds between successive points on satellite ping rings. It was summited because it shared some similarities with the approach outlined in Duncan Steel’s April 27, 2014 blog, “A graphical technique for investigating the MH370 flight path”. The value of this approach is its potential of providing MH370 track/speed solutions for successive ping pairs — if complete Inmarsat data were available.

    Once I saw the Doppler data presentation in I decided to expand my simplistic model to a global form to see how it worked. The results were disappointing and generally not worthy of reporting, except for the overall conclusion above, which is the apparent cause.

    The model’s track/speed solution for each ping pair is INCONSISTENT with its average speed computed from the track segment length and elapsed time. In general, the average speed was greater than solution speed. The solution speed is assumed to be constant over a great circle track and is derived from sub-satellite and aircraft positions, bearings, ping rings, elevations, and radial speed components. Each radial speed is the result of transforming the Doppler shift to LOS speed and then to radial speed. The average speed is simply the
    track length divided by elapsed time. If the data are consistent and the math is correct, these separately developed speed measures should be comparable.

    Factors other than Doppler shift are better understood and appear to have less leveraged impact on model results.

    The model is contained in an Excel workbook with VBA macros and is available for review and manipulation at: . I offer it primarily for others to verify my observation or to point out my error(s). Down the line, it might serve as a cross check for other modeling.

  57. Scott Alexander
    Great analysis. I do remember seeing fuel consumption charts somewhere–either this blog; or PPRUNE; or They were good graphs with consumption at various altitudes. I’ll try to remember where I saw them, if no one else chimes in

    1. Thanks for the props Barney. I’ll look there.

      It would also be extremely helpful if someone knows the confirmed fuel weight that MH370 had onboard. I’ve seen some numbers in the past elsewhere, but their reliability is unknown. One would think that this rather basic piece of info would have been released in the Malaysian preliminary report, but I didn’t see it in there. Perhaps it was and I missed it.


    1. Hopefully the media will be talking more to Mike and/or the other stars of this blog again, e.g., when the results of the audit and the new search plan are revealed. Here go some suggested media bullet points:

      1. The northern arc search corridor was ruled out by mistake.

      2. The satellite and radar information released so far shows no reason to favor a southern arc search over a northern arc search.

      3. The northern arc search apparently was ruled out due to a simple error in subtracting the aircraft-to-satellite delay component instead of adding it when calculating the total Doppler effect for the northern route.

      4. The satellite and radar data need to be released publicly so that independent scientists subject to peer review with no stake in justifying past errors can help refine the search area.

      5. It is not a matter of searching in the wrong ocean; it is a matter of searching in the most likely locations in both the northern and southern hemispheres.

      1. Bruce,

        I agree with point 4 and 5 on your list, but points 1-3 are completely false in my mind. Please check the developments in this thread before digging deeper into the sand.


  58. Just wanted to let you know (and mainly Warren Platts) that I tried to reconstruct a flight path using waypoints and a constant cruise.

    I first tried a guestimate of a “simple” waypoint flight directed to the south pole (SPOLE). The available ping rings clearly rule out such a flight.

    Next I tried Warren Platts “RUNUT” and followed by 189 heading scenario. The crosswind plus magnetic deviation clearly rule out that course using a 189 mag track mode. The crosswind rules out a 189 great circle scenario.

    I think a cruise towards RUNUT and then a great circle track of about 191 (give or take, leads somewhere off the west coast of South America) could be brought into accordance with ping rings (as per Malaysian “Google Earth” graphics released) and winds. However this is only a guesstimate and ignores doppler (I could not reconcile the available BFO data with any flight path, and I do not trust available doppler/LOS reconstructions).

    I have written a little more here:

    I will hopefully add some of my data from Google Earth (waypoints, image wind/mag overlays, measured distances) to the blog post when I come around to it.

    1. Tony,

      what would the maximum ground speed of this scenario amount to?


      1. Just as a possibility and out of a sense of completness:
        I remember reading somewhere on (sorry, can´t find now, as the thread is 60 pages long meanwhile), that due to incapatitation of pilot/aircraft, someone might have tried to enter the original target (Peking) into the navigation system, and a mistakenly could have set the correct number but (i.e. in smoke diminished conscious) choosed south instead of north.
        The overall direction would then fit the ping rings, but i do not posess the math skill to calculate the exact path.

        With deep thanks to all involved so far


  59. Duncan,

    Please add this to my last comment.

    The scale factor I used is equal to (Mean of Uplink Frequency and Downlink Frequency) / (Downlink Frequency)

    (1643.5 MHz + 3614.5 MHz) / 2 / 3614.5 MHz = 0.73

    I don’t know why this works, but it could be related to how the doppler is affected at the satellite from RX at uplink frequency to TX at downlink frequency.

    1. Yap: Transponders do not work that way. It is very simple.

      Fout = Fin + LO

      …where Fin is the receive frequency, Fout is the transmit frequency and LO is the Local Oscillator frequency. We do not know the exact channel frequencies 9M-MRO used on MH370, but we do know them within ~1%.

      D7 range 1626.5 to 1660.5 MHz
      D8 Range 3600-3629 MHz (assuming global beam backhaul; spot beams 1-6 are slightly different)
      Inbound Uplink Frequency (C band) 1643.5 MHz wavelength 0.182411 m
      Inbound Down Link Frequency (L band) 3614.5 MHz wavelength 0.082942 m
      Transponder Local Oscillator 1971 MHz

      1. Sorry, L and C band notes are reversed. should be:

        Inbound Uplink Frequency (L band) 1643.5 MHz wavelength 0.182411 m
        Inbound Down Link Frequency (C band) 3614.5 MHz wavelength 0.082942 m

      2. Mike,

        Thanks for the clarification. Now I have a question. If the LES sends up a pilot signal to the satellite and it is immediately relayed back to the LES, what would be the total doppler?

        Uplink doppler: v * f_uplink / c
        Downlink doppler: v * f_downlink / c

        Total doppler: v * (f_uplink + f_downlink) / c

        So if one is using a pilot signal to extract the doppler, the result would be related to the mean of f_uplink and f_downlink, no?

      3. Hi Mike,

        Or … LO = 5.2GHz (5258 Mhz, assuming the same carrier frequency numbers)


  60. The manual may be ‘replete’ with the phrase ‘primary power’. It’s also totally devoid of the phrase ‘secondary power’.

  61. Hi Duncan,

    I am inspired by and have great appreciation and respect for your selfless efforts (along with that of the many other contributors) attempting to scientifically and mathematically define the possible flight paths of MH370, with little more than the few breadcrumbs of information made public and confirmed to date.

    I am writing to offer some observations and thoughts from more of a qualitative aviator perspective (what I’ll term as a 50,000 foot level).

    I am hopeful that some portion of this input may incite some of the over-the-horizon brain trust contributors in your blog to perhaps consider additional analysis that may help further constrain (or may unfortunately further widen) the proper search area using a slightly different angle. It may also (through the mental gymnastic process alone) help confirm or invalidate some of the assumptions made by others in conducting their own analysis.

    I fully understand that your main purpose initially has been to confirm or deny the Inmarsat analysis and the resulting conclusion (southern flight path only) that has been drawn primarily from the satellite handshake data.

    Having said that, for the purposes of this input, I am assuming (and truly hoping) that there is some other source of independent data (e.g., radar, acoustic, seismic, satellite, etc that has not been publicly disclosed for unknown reasons) that has led the investigation to have what appears publicly at least, to be blind confidence in the southern flight path.

    This may or may not be a good assumption as it would seem that after a two month search yielding not one shred of physical evidence that a crash anywhere even occurred, one would think if such independent data exists, the Malaysians surely would have somehow found a way to release it for the sake of the families, at a minimum (without compromising other countries’ security, sources, or capabilities).

    Perhaps this other data will be released sometime soon if it does in fact exist. If it does not exist, I would join the many others that believe the investigation will be left with almost nothing to go on (except the very suspect 4 acoustical pings thought to be detected from the FDR/CVR on the final day of their expected battery life at a degraded frequency well outside the design spec frequency, within one day of deploying the TPL at a less than optimal depth for detection, immediately after a major shift of the primary search zone ~700 NM to the NE, days after receiving the most promising leads from numerous countries’ satellite information depicting suspected surface debris, and detected in a dispersion pattern well outside the design spec range of the pinger devices themselves). Wow, just writing the above rambling in parens made it almost difficult to believe, let alone that it could somehow inspire anyone to have great confidence.

    In any event, assume that MH370 did take a southern route.

    My BLUF (bottom line up front) is this:

    1. The possible final arc locations may be further constrained by examining the effect on fuel consumption due to both the assumed cruise altitude during the completely unknown portion of the flight combined with the postulated additional climb/descent events that may have occurred after the initial climb to cruise altitude.

    2. The search area may extend a potentially large distance along the flight path route (and possibly even a considerable distance in a different direction off the flight path) depending on how the final partial handshake is actually interpreted.

    The unknown cruise altitude over the last 5 1/2 hours of the flight, combined with the unknown number of climbs and descents postulated between IGARI/BITOD through the last military radar contact (which could result in wide variations in fuel consumption) remains a very significant unknown no matter how accurate the flight path can reasonably be determined by the Inmarsat data (even though this is already partially accounted for in the different speed assessments constrained by the handshake arcs).

    Additionally, depending on the interpretation of the final partial communication with the satellite, that will also add to the uncertain size of that area (in potentially various unknown directions off the flight path and from various unknown altitudes). The most likely interpretations I have read/thought about being:

    – First engine fuel starvation
    – Second engine fuel starvation
    – Blanking of Inmarsat antenna due to plane turning/falling off in one direction or the other and then entering a nose low spiral at fuel exhaustion
    – Deployment of Ram Air Turbine (RAT) at fuel exhaustion
    – Autopilot disengagement
    – Some other unknown anomalous or failure event
    – Impact

    All but the impact scenario above would likely introduce significant differences in distance traveled dependent on airspeed, altitude, and the plane’s most likely behavior at onset of those events (wings level glide, stall to nose low spiral, or some possible combination of both).

    What I’m trying to suggest is this: Assume that we know the flight path/s (whatever it is/they are), and then evaluate whether the assumed cruise altitude plus the effect of climb/descent events may rule out some of the possible terminal locations.

    Then, also determine how large any search area would need to extend to cover the numerous possible ranges the plane may have achieved based on the various possibilities of what the final partial handshake could actually mean.

    This may be a bit of stating the obvious but I’m only stating it because all endeavors to date appear very narrowly focused on determining the flight path to the final arc where the plane may have ended up, which ultimately is partially constrained by the fuel consumption due to speed, but may be further constrained if these additional climb/descent events actually did occur.

    I would propose the following altitude scenarios, from which someone with access to the 777 fuel consumption charts could determine a reasonably accurate total range for each scenario. It would also be good to factor in prevailing winds effect on fuel consumption/groundspeed. The first portion of the flight from KL including engine start, push back, taxi, takeoff, initial climb to 35K’, cruise to IGARI/BITOD is known with certainty and would be the same throughout each scenario, then:

    1. Cruise altitude of 35K’ from IGARI/BITOD to terminal location

    2. Cruise altitude of 35K’ from IGARI/BITOD with an idle descent to 5K’ until over the Straights of Mallacca near the end of the path where it was reaquired by Malaysian military radar, then a climb back to 35K’ to terminal location

    3. Cruise altitude of 35K’ from IGARI/BITOD with an idle descent to 15K’ until over the Straits of Mallaca near the end of the path where it was reaquired by Malaysian military radar, then a climb back to 35K’ to terminal location

    4. Cruise altitude of 35K’ from IGARI/BITOD with an idle descent to 5K’ until over the Straits of Mallaca near the end of the path where it was reaquired by Malaysian military radar, then remain at 5K’ to terminal location (may be a throwaway due to fuel constraints)

    5. Cruise altitude of 35K’ from IGARI/BITOD with an idle descent to 15K’ until over the Straits of Mallaca near the end of the path where it was reaquired by Malaysian military radar, then remain at 15K’ to terminal location (may or may not be a throwaway)

    These scenarios are loosely based on what Malaysia has put out in the media and are suspect based on the source alone, combined with the inherent inaccuracy in determining altitude solely from primary radar. There may be better scenarios to examine and this approach would certainly be helped considerably if the Malaysians would release the raw military radar data or the altitude fluctuations that have been conclusively determined, if any have. I excluded the one postulated where it may have climbed initially from 35K’ to 45K’ then later descended primarily because early on in the flight with a relatively heavy fuel load remaining, it may be unlikely the aircraft could actually attain its designed service ceiling height, let alone exceed it.

    I believe it is estimated that MH370 had enough fuel onboard for a normal flight of ~ 7 1/2 – 8 hours, normal flight being defined as start, taxi, takeoff, climb to cruise altitude, enroute descent to destination, approach to landing, plus normal reserve for weather/alternate destination. My thought is that examining the effect that these different scenarios have on overall fuel consumption/range may further constrain the flight paths possible.

    This may or may not be a productive suggestion and perhaps somebody has already assessed it, but I throw it out there in an effort to help make this effort just slightly less than “trying to complete a crossword puzzle without even knowing which squares are black”, the absolute most classic Duncan quote to date.

    If somebody could provide a vector to who may already have access to the 777 fuel charts, I’d be happy to help. Apologize for the lengthy post.


  62. Did the plane go north or south?

    The conclusion depends on the interpretation of the BFO graph. There are only two options here, so it is possible to follow the consequences of the two options in details.

    1. The BFO graph shows expected-actual.

    This looks reasonably at first, in particular since that is what the Inmarsat report states, albeit in a text written for laymen. It also looks reasonable, because it means the BFO graph contains no contribution from the aircraft at the time when the aircraft is on the ground.

    However, there are a number of consequences that are hard to epxlain.

    First, it implies that the contribution from the moving aircraft increases steadily over the whole flight, which is not consistent with a working “TX correction”, i.e., the notion that in theory, a moving aircraft should not visible in the graph at all. As a consequence, something must have happened to/in the aircraft to make the TX correction malfunction, minding that the satellite dish was still functional.

    Second, the resulting calculations show a path that goes north. Curiously, the shape of the BFO from that solution looks very much like the southern path in the Inmarsat BFO, but upside down. Still, the conclusion must be that Inmarsat mixed up north and south, even though they did not even need to make calculations to arrive at their conclusion.

    2. The BFO graph shows actual-expected.

    This looks unreasonable at first, because it introduces an offset in the graph that is not mentioned anywhere.

    However, BFO stands for burst *frequency* offset, so perhaps we could imagine other contributions to the graph, such as oscillator error.

    First, it implies that the contribution from the moving aircraft is steadily small through the flight. Not zero, but small. This is to be expected if we are to see any difference between a northern and southern path. A TX correction that does not cater for a moving satelitte fits the bill. The overall aircraft contribution becomes small, and it is mathematically possible to extract the northern-southern difference.

    Second, the resulting calculations reveals a distinct difference between all north-moving and south-moving paths. The eye graph. This bolsters the idea that it is possible to make comparison to other flights, and see the similarities. Moverover, and perhaps most importantly, the predicted ‘best match’ path goes south, to the same area as predicted by Inmarsat.


    1. Henrik,

      I recently realized that there is another wrinkle in determining the sign of the BFO. Some downconverters invert the frequency spectrum – such is the case if the local oscillator frequency is above the incoming C band signal. In that case, if the aircraft is approaching the satellite giving a signal that is higher in frequency, once you go through the receiver chain, the output signal will be lower in frequency. Argh!

      1. sk999,

        Ah, yes, excellent point. So if this is the case in the BFO graph, we are actually looking at a signal that is inverted in sign compared to what would otherwise be expected from the definition. Hence, it is possible that the Inmarsat report is fully consistent, and really shows what we would normally define as ‘actual – expected’.

        It does not have to be this way, of course, but the mere possibility should really make us disregard the sign statement and instead look at the path labels in the graph. Those labels literally have a billion-dollar worth of trust attached to them (and I mean that in a good sense, not in a huge-coverup-conspiracy sense).


    2. Henrik,

      Thanks for your analysis of consequenses of the two different interpretations of the BFO graphs.

      As a non-native english speaker I was initially confused about what the expressions actual-expected vs expected-actual meant. I fell into the trap of reading them as linguistic constracts. It later clicked, they are mathematical expressions:
      1. BFO.graph = BFO.actual – BFO.expected
      2. BFO.graph = BFO.expected – BFO.actual

      Just thought I share my little mis-reading experience, in case it helps others to avoid suffering the same ;-)


      1. Yes: and US typesetting often uses endashes rather than emdashes, and usually-as here in this sentence indeed around these few words-uses hyphens to encapsulate side comments without spaces before and after the hyphens/dashes. Thus this is not only a problem for non-native English speakers. If I were not going to use spaces around those hyphens/dashes then I would like to use long, long dashes; but prefer some space. We spend a lot of time worrying about fonts, and not enough time about the gaps between…

  63. Following up on my last comment, I have tried a scaled D3 instead of using a linear time varying BFO offset like Henrik. The resulting fit also look very good, if not better.

    So BFO = D1 + D2 + scale * D3 + constant offset

    Scale = 0.75
    Constant offset = 155 Hz

    Calculated South Track 450 Knots with scaled D3:

    Calculated North Track 450 Knots with scaled D3:

    Calculated South Track 350 Knots with scaled D3:

    Here’s the spreadsheet:

    1. Yap: Might I suggest that you look at what has been done by Mike (airlandseaman) and VictorI to see whether you can identify any errors?

      1. I think it’s just down to different interpretations of the BFO. I have tried VictorI’s April 30th interpretation before and my results agreed with his for his northern track.

        VictorI’s April 30th analysis:

        BFO Interpretation #1 :
        i) Different BFO correction values for times before 18:00.
        ii) After 18:00, no correction is used. D1+D2 is the actual aircraft-satellite doppler.
        iii) Sign of the BFO needs to be inverted.

        Although the results match Inmarsat’s *measured* BFO values for this northern route, I have not been able to find any southern route that matches. It doesn’t mean this interpretation is wrong. It’s just that it’s not possible to reproduce Inmarsat’s North and South Track predictions based on this interpretation.

        The other interpretation assumes that D1+D2 is a residual doppler offset after a correction has been done by the AES. The correction assumes that the satellite is stationary at its mean position. This is the interpretation that I (and Henrik and RichardC) have used.

        BFO Interpretation #2 :
        1) D1 + D2 = residual Doppler after a correction by the AES that assumes a stationary satellite at its mean position.
        2) A linear time-varying offset (Henrik), or a fixed offset with a weighted D3 (me).
        3) No sign inversion

        The results match well with Inmarsat’s Measured and Predicted BFOs. In particular, it shows the difference between northern and southern routes. This does not necessarily mean that Inmarsat’s analysis is correct. We still do not know many details about their data. We have simply shown that based on certain simple assumptions, we are able to reproduce the main features of Inmarsat’s results as shown on page 2 of their Annex.

      2. Thanks Yap.
        This just to note that I gave ‘true geostationary satellite’ (altitude 35,000 km, zero inclination, zero eccentricity) to aircraft at near-KL positions to Mike on Friday and he is using those to check out the idea that Inmarsat indeed estimated BFOs using that default satellite (i.e. not the actual satellite).

    2. Yap,

      Thanks for these numbers. Since it seems a bit harder to explain why D3 would enter with anything but a factor of one, I would not read too much into it. It is much more likely that the flight path in question is not straight, which may affect the BFO fit much more.


      1. Henrik,

        I was playing with the idea that the downlink frequency is 2.7 GHz (S-band ?), because that will explain it. But I don’t think that’s the case.

    3. Yap,

      Several people have now done the calculation the way you have and arrived at similar conclusions. I checked your numbers against my program (not a spreadsheet, so totally independent) and they come out pretty close, and the small differences seem to be due mainly to differences in the assumed path of the aircraft.

      The question is why is there a fraction of the satellite-Perth Doppler (what people call D3) and/or a ramp and/or an offset remaining in the BFO. The Inmarsat Annex slides do not tell the entire story. One of the ground stations (possibly Perth itself) sends up a pilot signal that is relayed by 3-F1 and is received by Perth in order to measure the frequency errors introduced by the satellite itself (which apparently can be quite substantial). In one mode of operation, the ground station uses the pilot signal to remove the frequency offset from all the downlink C band channels in an AFC-controlled downconverter before they are decoded. Since the pilot signal also includes any contribution from D3, it also means that D3 is removed from the aircraft signal before its frequency is ever measured. This note from MITEQ shows how its receivers work. Whether the Perth electronics work this way is unknown. I have been trying to think of different ways to introduce a residual ramp of the right amplitude and sign but haven’t come up with anything convincing.

      1. sk999,

        You gave me an idea. Have you considered the possibility that if one is using such a pilot signal to extract the D3 doppler, the result would be related to the mean of f_uplink and f_downlink, which in this case happens to be 0.73 of the downlink frequency?

    4. Yapp and Henrik:
      Thank you for your contributions. Your work is very interesting. I like that you were able to use a single formula for BFO to handle North, South, and before and after 18:00 UTC, which I was not able to do. I’ll dive in and try to understand it a bit more.

      1. For those observers/readers who are following this thread but are understandably mystified by many of the comments regarding the technicalities of the BFO graph interpretation: VictorI, Mike (airlandseaman), Henrik, Yap and others are all trying to work out what the BFO information can deliver in terms of the speed(s) of MH370 during the flight, but without the benefit of knowing what Inmarsat knows about what specifically contributes to the BFO values. This has proven very complex. They have arrived at (at least) partial understandings, but their analyses are by no means identical, nor do they necessarily arrive at the same results. Therefore they are comparing methods, assumptions and outcomes. This is excellent scientific philosophy, and (one hopes) the way to get to the correct answer, in the end.

  64. Henrik,

    I was able to get the same results as you for the BFO fit using. The fit is remarkably good at all time points, with no sign inversion necessary.

    Calculated South Track 450 Knots:

    Calculated North Track 450 Knots:

    Calculated South Track 350 Knots:

    Here’s the updated spreadsheet if anyone is interested. Let me know if you spot any mistake.

    1. Yap,

      I am pleased to hear it!

      To be honest, I cannot say that I am surprised, since I was not the first to arrive at these figures. RichardC10 made the calculcations more than five weeks ago (I erroneously stated two weeks in an earlier post, sorry about that).

      Now, regarding your setup, I am sure you have noticed that the fit is not perfect. I am currently pondering why, and think the reason is that the plane flew closer to the satellite before turning south, and then at a higher speed. Is this something you would agree with, or otherwise has the means to falsify? The details may not matter much to the direction of the end position, but it may well affect both the understanding of what happened and the fuel starvation at the end, hence the actual position in the east-west direction.

      Looking forward to hearing from you again. :-)

  65. “It is Mas procedure to switch ACARS, VHF, and High Frequency selection off but this is only for flights to China as the service provider for Mas does not cover China. Some if not all pilots switch them all off for a while and then later switch SATCOMM back on to force the system into SATCOMM mode.” (Perhaps quoted pilot means that they have to connect to a different SITA satellite for China bound flights when leaving Malaysian airspace?) MISSING MH370: ACARS cannot be disabled – Latest – New Straits Times
    Is it possible that the outgoing FMS data-link to the satcomm failed when the pilots attempted to reboot the comms systems? Ingoing was fine, but outgoing failed?. The coincidence of the comm systems (possibly) being routinely turned off and rebooted, in the same time period that mh370 went dark, seems remarkable, and if this statement is correct, there is something here to dig into. The pilot’s remark is so casual, it rings true to me. Maybe someone here knows some MAS pilots to ask?

    1. I saw that one too, roseny6. It sounds a bit odd. I have seen no one else mention that.

      Anyone who can verify or pass a judgement on that?

      The fuse went.

    2. No VHF on the ground in China (still?), and parts of the South China Sea?, so the system is forced into satcom mode (when not doing it automatically).
      The pilots might need to reboot or “kick-start” this through turning off and turning on again. Either it didn’t start up again, or they forgot to reboot it? The latter might not be the case since there seems to be vhf where they were.

      There is by the way an overlap between Inmarsat’s satellite coverage there, starting right on the Malaysian peninsula. But the “Indian” one covers all of China.

      Do they have to do anything to show up as China Southern Airlines Flight 748 (CZ748/CSN748) on Chinese flight-tracking by any chance? I guess not.

      1. Explanation (in part) for non-specialist readers: VHF (wavelengths around 10 metres) can be used to ‘broadcast’ from the aircraft to the ground direct, without needing directional beams. However, VHF will not pass through the ionosphere much of the time, and this (plus many other reasons) means that we must use UHF (wavelengths like 10 cm) for aircraft-satellite communications (or indeed your satellite TV links). This means that in general the satcom is NOT ‘broadcast’ but rather needs directional antennas, although the beams *might* be wide: consider the fact that from geostationary orbit a beam that is 14 degrees wide (+/- 7 degrees from the nadir direction) will cover the whole of the Earth visible below). Back on VHF: because it does not go beyond the horizon (for transmitters near the ground or on aircraft) the area that is covered is only about 1,000 km wide and so the same frequency bands in the VHF can be used for different regions around the globe without interference.

      2. Thanks Duncan,
        This is outside my competence, but there is seemingly a critical moment there. I assume someone has looked into this already. I don’t remember just now what has been said about how this works in all respects.

        Better abled than me might look at the swissair accident, I believe it has been mentioned here before:

        The interruption of vhf coverage is there described as accomplished through a “saturation of the system” or by switching to voice mode on the vhf3 radio. We have perhaps been through this. This “saturation” is perhaps half a euphemism for turning it off.

        And then mh370 lost all the above plus the transponder, or perhaps more than that, and turned around. Or detected smoke and turned around. Or something else.

        Something was not working properly.

    3. I tried to understand the statement: “It is Mas procedure to switch ACARS, VHF, and High Frequency selection off but this is only for flights to China as the service provider for Mas does not cover China.”

      The famous pings were communicated via satellite 3F1 global beam so MAS planes probably have high gain antennas. They could probably communicate with the satellite over China even if they didn’t have such antennas as 3F1 have a regional beam (no. 6) directly over this country. The problem doesn’t seem technical and it’s not clear why this “service provider” didn’t cover China.

      It seems that China, like India, doesn’t allow in-flight satellite connectivity for foreign airlines in its airspace because they can’t control/monitor the contents . Inmarsat are negotiating with the Chinese and seem to have reached some solution:

      Will Horton
      Chinese airspace becoming friendlier to carriers with inflight connectivity
      July 11, 2013

      The blog says “Inmarsat’s Classic Aero services are already approved” but it’s not clear if the approval became effective yet.

      It wouldn’t be surprising if Vietnam have a similar policy so MAS pilots turn off SATCOM and probably also VHF/HF before they enter it.

      Maybe this blog readers can tell us more about China and Vietnam policies?

      1. Ron,
        I cannot add anything about country policies…
        The inflight connectivity being discussed in the article is related to passenger internet access. SwiftBroadBand is a high(er) speed data link that Inmarsat offers so that the airlines can offer WiFi access to the passengers. (It can also be used for in flight entertainments systems and other things.) Providing internet services to customers within the country would be draw the attention of some governments.
        MH370 was only using the low speed data link service for communications to the plane and crew. This is the Aero service, mentioned in paragraph 4: “Inmarsat’s Classic Aero services are already approved.”


      2. Bill, thanks.

        I understand the Horton/apex blog talks about internet connection for the passengers/entertainment system and saw the sentence “Inmarsat’s Classic Aero services are already approved”. I’m trying to push this just a little bit further in search of an explanation for this mysterious article:

        MISSING MH370: ACARS cannot be disabled
        New Straits Times
        19 March 2014

        My questions are:

        * Was the Chinese approval of Classic Aero services effective on March 8
        * Does Vietnam have similar policies as China
        * Do satellite communication restrictions extend to VHF/HF

        These questions sound crazy to a western ear and maybe they are silly in China/Vietnam but surprises happen.
        I guess a pilot forum would be more suitable for such questions, e.g. PPRUNE or

      3. Margaret, thanks a lot!
        The man is a treasure but some interpreting is required. I think our BFO experts will be interested in his comments, for example:

        A note about links to the page opens with comments closed, the target comment is top line, click on the subject (not user) to open it.

        * In the first comment Mandala499 says that Malaysia Airlines (MAS) subscribes to SITA network and because of some Chinese legal problem SITA doesn’t have a roaming agreement with ARINC. Since ARINC is supposed to provide ACARS in China MAS is flying there voice only. If true this is a Malaysian safety breach they tried to shift on the pilots.
        * In the second comment he talks about the New Straits Times article. He brings coverage maps of ARINC and SITA VHF networks and it’s clear ARINC have better coverage in China.

        May 8 2014

        ACARS is part of the Aeronautical Telecommunications Network, and it comes under special telecommunications and also basic telecommunications law of different countries (it’s not FIR specific, but country specific). ACARS through SATCOM is covered by the ATN provisions of a country’s law, and also by the laws covering SATCOM. VHF ACARS is provided by ARINC and SITA, except for Japan which has it’s own VHF ATN.
        BUT, regardless of which of the 2 you subscribe to, you do have co-roaming agreements between the two, and the two covers the Japanese ATN too, to provide a seamless service, with only minor difference in transmission costs if you’re roaming.

        Now, China is covered by ARINC for ACARS Datalink, and as far as I know, ARINC is the designated VHF ATN Datalink provider. But, China being China, roaming agreements are not as simple as over other countries, this is due to data routing requirements under Chinese law, and this is an issue SITA have always avoided.
        MAS/MH (not MAH as some write here), is with SITA and probably don’t have the roaming approval for China.

        Under SATCOM, for Inmarsat 3, the same situation apply in terms of the data routing for China.

        Mar 25 2014

        Direct links to the coverage maps:

      4. Ron,
        The pinging in question would go to the Indian Ocean satellite irrespective of all else — from what it seems. If they did not log-off it (see far below). It is Acars, pilots’ and other communication that went via VHF that is switched to satellite, if I am not wrong. This may normally not be a critical move at all, I don’t know for sure. And the flight transponder is or should be independent of this, wasn’t it so?

        What it means is, 1) that the “switching off” of Acars is even less necessary of ill intent (it could also result from trying to establish source of smoke in cabin) and 2) that this turning-off and on, on the face of it looks more closely connected to the blackout and turnaround than has been suggested. But it doesn’t explain the transponder, so, reasonably, they are impaired by a common external destruction. Or, it is a question of how the cables have been laid together in this particular machine.

        But should not switching from vhf to satellite be expected to be standard procedure for any plane in the world, especially with Acars? The pilot recounted in the article suggests it is extra-customary. So either MAS has its own patented lever and cocktail of system reboots for China here,
        or the journalist is simply clueless, which is probably the more likely
        (and me too, obviously). Or are everyone else going with satellite directly these days (ads-d?)? Is there risk of any “integrational” mishaps here, with transponder, ads-d, vhf, acars, and entertainment systems? I know several have touched upon this.

        This would lean towards cable fire, I guess. In the avionics bay or wherever all these cables gather.

        I saw several here with that technical expertise. Any comments?

      5. There’s confusion above between aeronautical services & passenger services over satcom.
        The aero (MH operational & voice) services will not be affected by an individual nation’s policy on ‘internet’ access for passengers which airlines may carry via Swift64/SwiftBB, the aero services are agreed by the ICAO. China is a participant in ICAO activities, esp in this region as its people are driving the growth in air traffic movements.
        Given that all the range and doppler data being discussed here derives from comm events over the satellite datalink, including those seen events while the aircraft was on the ground, its safe to conclude that satcom was the default data selection (rather than auto vhf/satcom datalink selection) for the whole flight.

      6. Further to my post above,

        The other important distinction to make about communications is voice vs data. The context for everything discussed here is data: VHF datalink services have the acronym VDL.

        SITAs VDL coverage over MH370’s route and that whole region is poor, the VHF voice coverage is ‘OK’, hence we’re only concerned with satcom datalink events over Inmarsat’s network.

        SITA global VHF voice coverage (Sep 2012):
        SITA global VDL data coverage (Sep 2012):

      7. Don,
        Ok, I’ll let go of the bone.

        There is an urge to jump to conclusions, i guess.

    1. Victor,

      What is your definition of D1 and D2, respectively? I am asking because it varies substantially amount contributors, and it is good to at least get such uncertainties out of the way.


    2. Hi Victor,

      Well done! A lot of hard work, but you got there in the end to understand where Inmarsat went wrong!

      Coming back to your initial focus of finding possible flight paths that fit the information you have gathered, are you still of the view that the best fit is Qamdo Airport or are there other good fit end points that that you can share?


    3. Wow again, Victor, and thanks for the reformat. It’s crystal clear now.

      I have a thought. What is the likelihood that the calculations Inmarsat did were performed in Excel? It appears that at least the graph is an Excel graph. Could the entire set of calculations have been done in a single Excel?

      If so, then if the north path was caused by a sign error, could the rest of the discrepancies have been caused by single-row errors in Excel equations? For example, if a formula was revised but not dragged to every row? Could all remaining discrepancies at multiple ping times be explained by the same changed formula?

      Accountants look for transposition errors by checking to see if a discrepancy is divisible by 9.

      Spell-checkers identify correct words by finding the word with the lowest “edit-distance” from the typo-ed word, telling me I meant to type “To” instead of “Yo” because the switch from T to Y represents the fewest numbers of bad keystrokes needed to mess up a word. (As opposed to “Up” which requires 2 mistyped keys.) Although I really did mean “Yo” but that’s another story.

      So my sort-of rhetorical question here is: How many mistakes are needed to make your most accurate spreadsheet yield Inmarsat’s graph at every point?

      1. Interesting thought! Change a constant in a cell, such as speed on leg or something and by dragging to other cells excel incrementing by one each row… There might be the unexplained linear trend.

        For the uninitiated, Excel is not a good tool for scientific or engineering modelling.

  66. Bill, been following all your comments with great interest. Re the issue whether the Satcom components have redundancy or can be termed ‘non-critical’, here is a passage from a manual on the Iridium satellite system:

    “2.3.4 Reliability/Integrity…………..As all the avionics, the AES will be designed so that [mean time between failures] is as long as possible whereas the [mean time to repair] is as short as possible. These two requirements will apply to essential airborne units such as the satellite data unit, communication management unit, beam steering unit and the antenna sub system. This may be achieved by main/hot standby configuration of the critical units stated above, as well as automated change over mechanism within each unit.”

    Also the Honeywell’s manual is replete with the phrase ‘primary power’, suggesting there may be secondary backup power somewhere.

    Finally, the Honeywell MCS 4200/7200 system provides for the option of a dual Satcom configuration using 2 SDUs and 2 antenna systems, in a master vs slave setting or otherwise. In this regard, i note MH370 had both a low gain antenna on top (shark fin shape) and a high gain side mounted Ball Airlink antenna according to Don. The manual says each SDU is “capable of being the master or the slave” but “[I]f the slave is only equipped with an LGA, it cannot function as a true slave, but only as a standby backup system for low-rate packet-mode data services, ready to take over as master in case the original master fails”. It is unclear whether a LGA+HGA system would necessarily signify a dual SDU system, elsewhere in the manual it is stated ” The HGA transmits L-band RF signals from the HPA to a satellite, and receives L-band signals from a satellite for the SDU and the HSU. When installed, the LGA supplies backup communications (packet-data only) for the HGA by providing low-rate, Classic Aero (only) packet-data communication services”.

    According to the manual on 777 from Alteon, the “left main ac bus” sends 115v ac to the various Satcom components. Assuming MH370 had suffered electrical failure knocking out the avionics in the front bay, do u think the Satcom avionics at the back could still have remained powered by this left main ac bus?

    1. Alex,
      My comment was pointing out the irony – that the less critical system remained operational, while other systems went offline (or were manually taken offline.) Specifically, the transponder, ADS-B and ACARS.

      There is no reason to believe that redundant SDUs were installed.
      (btw- since the airline only subscribes to the Aero service, the limitations would not matter much.)
      For the satellite link to be operating, the plane must be providing power. There is a lot of power used to maintain the sat link. if there were power faults, the priority would be to keep flight critical systems powered. So, I would expect the satellite power to be shut off as needed to keep other equipment operating.

      If the avionics bay was ‘knocked out’, it could not have been caused by loss of power.
      Many pieces of gear have provisions for either AC power or DC power, as well as provisions for redundant power. How they are utilized is up to the airplane manufacturer and the build options chosen by the customer.

      1. As I have understood it the plane needs to be in the air to have “secondary power” and needs to have fuel to have primary. Anything else would need more or less to be solved like the FDRs. If the plane has (crash)landed there is no reasonable need for a battery driven system or satellite connection in each plane. Either ask the flight tower for help or analyse the FDR.

        The idea that planes might disappear completely is not accounted for. In more than 9 out of ten times they don’t.

  67. Richard:

    Couldn’t reply to this earlier of yours:

    So doing so here. You said:

    “a time of 7 hours 37 minutes from take off to the last ping at 00:19 UTC and a fuel endurance of less at 7 hours 31 minutes…is too tight as a fuel planning basis for a hijack. Of course, there are many factors that affect fuel consumption that have been discussed at length in previous posts, but I would not rely on these factors if planning a hijack.”

    Very reasonable assumption.

    BUT – are we absolutely sure, that from take-off to last ping, MH370 was airborne the entire time?

    “Satellite pings that weren’t turned off showed Flight 370 operated for almost seven hours after last making contact, Najib has said. That may have taken the plane more than 3,000 miles from where it was last tracked and pushed it to the limits of its fuel load, if it was airborne the whole period. ”

    1. Hi Nihonmama,

      Even if MH370 was crossing the ping rings at right angles and taking the shortest distance between the ping rings, then it was moving all the time and with an increasing speed toward the final ping ring at around 00:19 UTC, which implies MH370 was airborne.


      1. Hi Richard:

        Airborne implied indeed. Merely pointing out that Malaysia’s PM Najib went on the record (albeit subtlely) with that possibility to the Bloomberg reporters – and the editor let it stand.

      2. Richard, Nihonmama,

        One scenario marrying both your assertions/speculations could be something like the following:

        To conform to the ping ring data, inferred speeds etc, it is not required that MH370 as a whole is airborne, but merely the transponder and associated systems.

        A (Wildly speculative and bordering on conspiracy theoretics) scenario, could be “land the plane – collect the system – mount to some other aircraft- send off other aircraft including system down into the southern ocean, where it is never to be found”

        From memory of ping ring images, the closely spaced “possible turn” ping rings do seem to run through/near/along the nicobar and andaman islands. Could these indicate aligning MH370 with a landing strip there? I guess, we would see more AES initiated pings during an actual landing maneuvre and, in the absence of further pings, such a scenario should still be discountable.

      3. Richard,
        “with an increasing speed toward the final ping ring”
        An inference one can draw from that is that the aircraft was on a magnetic [True might also work] track at a heading something less than 180 thus having the winds [approx 270deg/50-60 knots] assisting the groundspeed.

  68. If the TX correction is based on a fixed satellite position, we can work out what the basline in the BFO graph is.

    Just as the difference between the northern and southern paths reveals information about the position along the ping arc, the average of the two paths reveals something about the way to get there.

    Assuming the TX correction is perfect in the radial direction, the average of the northern and southern path will provide the baseline, since (D1 + D2) will then be zero (give or take a very small error, ~1 Hz, from the combined north-south corrections due to the satellite not being at the equator).

    So I did this for Inmarsat’s predicted paths, then subtracting the known C-band doppler component. The remainder contains a bump close to the point when the two paths diverge, which is explained by the different paths aligning themselves. Subtracting this contribution, I arrive at a linear fit for the baseline:

    D_baseline = 175 + (124 – 175) * t / t_max [Hz].

    The source of this basline is not yet know, but it must be emphasized that the only assumption needed to arrive at it is that the AC TX correction does not depend on the velocity of the satellite. There are several patents and methods described that works this way, thanks to Mike and Ron for posting them here. Whether such an algorithm was used on the particular aircraft in question remains to be seen.

    As a speculation, I would wager that the baseline is due to the temperature dependence of the local oscillator at the satelite. The flight takes places at +- 4 hours around midnight from where the satellite is positioned. During a good deal of the mid portion of the flight, the sateliite is in the shadows. The temperature would be dropping fast for several hours after the flight started.

    All crystal oscillators exhibit a small temperature dependence that can lead to substantial errors when looking at the hertz scale. The frequency will drop when moving away from the optimal working temperature . The effect is likely to be several kHz, and probably is adjusted before reaching the BFO measuring point, as Mike points out. However, extrapolating from suggestions on the TMF blog that this particular satellite system was not built for such a large eccentric orbit, there may well still be several hundred Hz visible at the point of the BFO measurement. I therefore find it very hard to disprove that the AC TX correction on this aircraft was independent of the satellite velocity.


    1. Just to note that the eccentricity of the satellite orbit is not large, but the inclination is (for a ‘geostationary’ satellite).

    2. Henrik, Inmarsat uses an oven controlled crystal oscillator on the airplane. There would need to be a thermostatic failure for the frequency of the crystal oscillator to be effected. GPS satellites use atomic clocks synchronized to atomic clocks on earth. It looks like Inmarsat is synchronized with GPS. Time drifting with temperature would be a serious design flaw I doubt exists on Inmarsat satellites.

      1. Karl,

        accurate time is nice, but it is not the same as temperature dependence on the satellite LO oscillator. The bent pipe is a dumb device, in that respect. The 20TEC.pdf document has a pretty good picture of the errors involved.


    3. Henrik:

      Your statement that : “All crystal oscillators exhibit a small temperature dependence that can lead to *substantial errors*…’is not really accurate in this context. Yes, the transponder LO surely does have some temperature dependence. But it is small to begin with, and highly predicatble. The temperature change over one orbit period is small. The orbit period is 24 hours, so the changes are very slow, and the dependence is routinely and easily measured on the ground with reference oscillators better than 10^-12. This is SOP. They do this with a pilot carrier. So I seriously doubt that the transponder LO, after correction, is contributing more than a few Hz error to the BFO observations. The fact that the residual at 16:30 is only 1-2 Hz proves this.

      1. airlandseaman,

        This is still assuming something about the point of measurement. Either the effect of the temperature dependency is large, and the residual after correction may still be visible, or the effect is small, in which case it may well be included at the point of measurement. The point is that we do not know.

        Regarding the offset being a couple of hertz, you are referring to the residual at 16:30, which depends on the sign interpretation, and is therefore a circular argument.


  69. Please allow me to clarify my last post regarding models for the BFO data. My goal was to replicate the model that Inmarsat used in their North and South predictions to better understand the underlying assumptions. I do not necessarily subscribe to the theory that the SATCOM worked in the way suggested by the model. The model assumes that after 18:00, the SATCOM had aircraft data supplied via the data link and an offset correction to the frequency was applied to (partially) compensate for the D1 Doppler shift due to aircraft motion. This implies knowledge of the position of the satellite, the aircraft’s speed, position, and heading. This is an item under investigation.

    1. Glad to read this clarification Victor, I had written a note about concluding that Inmarsat incorrectly calculated its predictions for the north trajectory… based on our own models… In essence, there are so many models we could fit the the measured BFO ! (in the note, an autoregressive/a time varying model of AES compensation) more or less perfectly… but in the end, nothing guarantee us they are correctly describing how the AES compensation scheme actually works !

  70. I finally got models to fit the BFO data presented by Inmarsat. It required that the contribution of satellite motion to the Doppler shift be incorrectly inverted for the Northern path. The graph comparing the measured data, the Inmarsat’s predictions, and the models can be found here:

    Based on this work, it is even more likely that Inmarsat incorrectly computed the prediction for the North path.


    1. Hi Victor,

      It looks like your predicted path coincides with the south path, do I read the graph correctly? What does the correction mean here?


      1. The predicted paths are data points from the Inmarsat graph. I tried to match their data with my model, in attempt to reverse engineer their assumptions. For the South path, I could get my model to match their prediction by appropriately choosing the Doppler shift correction term, which assumes that the SATCOM is trying to minimize the overall frequency shift by applying a correction that compensates for the Doppler shift. For the North path, I could get my model to match their predictions only by intentionally introducing a math error on the sign of the D2 (satellite velocity) term. When this term is corrected, I could find a North path that matches the measured data. The implication is that the difference that Inmarsat sees between the North and South paths is due to their math error. When paths are appropriately chosen and the calculation is performed correctly, one could not distinguish between and North and South path because either can match the measured data.

        I will post another image that might be easier to understand. After working on this graph for hours, what is obvious to me is not obvious to others.


    2. Wow. I think I get this, but can you confirm?

      The “Predicted” are from Inmarsat’s graph.

      The “Model” are your attempt to recreate the Inmarsat graph, sign error included?

      So if the sign error wasn’t present, your South model would still be accurate, but the North would change?

      Is it possible to show both the correct and incorrect North models?

      What do you make of the 21:40 discrepancy?

      What a PR nightmare, if it turns out that the path was chosen based on a flipped sign – an error true peer review would have caught 6 weeks ago.

      1. Your interpretation is correct. I do not know the reason for the discrepancy at 21:40 UTC. Perhaps my model for the North path is a bit different than Inmarsat’s.

      2. JS,

        As relayed by several people in this thread, Inmarsat used direct comparison between the actual BFO measurement and other actual BFO measurements, for other similar flight paths. No calculations involved. In addition, they have a model that fits the measurements. If there are any sign errors involved in the analysis here, they are surely made by us, not Inmarsat. Let’s be reasonable and humble.


  71. Duncan:

    You’re welcome re the last. And here’s more.

    Just had a VERY interesting chat with a well-placed old friend/colleague from my consulting days in Asia (she’s still in that part of the world). Re what information is in those pings and why Inmarsat would be withholding that data.

    Some of you may suspect this, others may know it, but I share for the benefit of all:

    Whether MAS subscribes to a location tracking service or not, the aircraft satcom terminal sends its position WITHIN THE REPLY PING HEADER. Airlines don’t get billed for this, because it doesn’t count as payload data.

    In essence, the location data related to the assets of many clients is being tracked through the Inmarsat satcom without their consent.

    1. That assumes that the SATCOM knows where it is. There is a theory that the data link to the SATCOM was broken and therefore data related to the speed and position of the plane were not known to the SATCOM. That would explain why Inmarsat does not know the location at the time of each ping.

      1. Correct Victor. Very aware and the possibility that the datalink to SATCOM was broken cannot be ruled out. Simply passing on information that I was told that may be relevant.

      2. It also assumes that the unspecified source is familiar with the AMS(R)S protocol spec. The spec has been discussed at tmfassociates and the link interface data units (LIDUs) concerned with the ‘pings’ (the log-in verifications) do not carry such information.

    2. The one premium service that MAS subscribe to essentially does also work as a locator — the fms Wind Uplink service. .It analyses the route ahead from the current position to the next selected waypoint and advises pilot/autopilot on altitude, course, speed. @ 600kg fuel benefits on a normal long haul. I thought it might have explained one of the triple pings, as SITA Wind Uplink auto-queries the FMS when a new course change occurs, but I guess not.

      1. Roseny6


        Why do you ‘guess not’?

        I discussed this sometime ago over at TMF’s blog. The service, according to SITA, “provides key data to the aircraft flight management system for in-flight re-analysis of weather conditions on the route ahead.”

        That data would be uplinked to the a/c FMS as ACARS msgs, it’s derived from the World Area Forecast System which distributes updates every six hours, i.e. one at 18:00 (allow batch processing time/workflow within SITA’s data centres and 25 min delay before reaching the a/c is feasible). We know the satcom datalink was functioning but no msgs getting though to endpoints beyond it ( MH OCC did attempt to send msgs to the a/c that do correlate to satcom events on the BFO plot ) so repeat attempts to send the WX update msg is likely.

  72. Richard & Duncan

    I agree that there are innumerable alternate scenarios. My point was to note that what has been repeated ad nauseam in my variations; that the plane took a northwest and then southerly route to “avoid radar detection” is hooey. If it went south after 18:22, either something happened around 18:22 +/- or, as Duncan noted, it took the route it did for some other reason than trying to “avoid radar detection” or it went north.

    The base line “story” repeated in the media and no doubt hinted at in background by the authorities inferring an intent to fly around Indonesian territory that resulted in final location should be dismissed.


    1. Hi John,

      I agree. They did not make a very good job of avoiding Malaysian or Thai radar and I am sure they did not fly around Indonesian radar either.


    2. Hello John,

      I also agree that the plane did not avoid flying through Indonesian radar space. However, one nagging detail here is that radars at both Banda Aceh radar and at Lhokseumawe should have seen the plane (at least at the 18:22 point located by military radar where it was at 35,000′ altitude). And as far as I know, we simply don’t know why this is so. So I don’t preclude the possibility that the pilot knew something here that we clearly do not, i.e., that he knew a priori why it would be unnecessary to avoid these sites.

      As for the military radar, the pilot could have purposely popped his head up momentarily so as to be last seen heading north, before dropping from view and making a turn southward. The plane made a similar turn at IGARI after dropping below the Kuala Lumpur radar horizon. That seems like a basic maneuver to misdirect. So the south track could have been intentional IMO.

      1. Charley,

        Judging from the BFO data, the flight may well have continued NW for a while longer, before heading W, then S. The signature of the turn is that of a direction more straight towards the satellite, and we cannot tell from the pings how long that lasted, if it started before the first ping in the series of three. There might be a narrow turn involved, but it is not the only explanation.


  73. Richard:

    A logical inconsistency has been bothering me. And support for the 2 event hypothesis or the north route. And maybe an addition to the pro or con.

    It has been widely speculated in the media and hypothesized that the northwesterly route after crossing the Malay peninsula was to avoid Thai radar and/or to follow established commercial airline routes. However, since the route, assuming it went south, has now likely been shown to have flown over/near Sabang/Banda Aceh that rationale should be disregarded for the south route. I note that there is a commercial airline route that extends the route after the Malay peninsula to MEDAN and beyond. If the person in control of the plane knew at the time of initial diversion that the intended destination was in the south Indian Ocean, there seems to be no logical reason to have turned northwesterly for hundreds of miles after Penang (and Alain and others can weigh in on the MEDAN route) and then flown over Thai radar and lands anyway.

    Conversely, if the person in control of the plane was trying to avoid Thai radar and/or to follow established commercial airline routes AND planning on a subsequent northerly route, then the route Malay straits route makes sense.

    It seems logical to assume that either 1) the plane went north or 2) something happened around the 18:22 time frame (the 2 occurrence hypotheses) to alter the previously intended route.


    1. Thanks John.
      However I would note that there are other possibilities e.g. the autopilot software (plus waypoint data file) was in total control from 18:22 or possibly earlier and then the aircraft flew a route determined exclusively by the autopilot.
      And I would imagine that others could propose other ideas; that is NOT an invitation, though!

    2. Hi John,

      A good example of a hijack followed by a malfunction theory.

      I originally thought of Iran as a possible hijack destination (because of the 2 passengers from Iran with false passports) but I discounted it because of the ping ring information. But the diversion from IGARI to the Straits of Malacca and then heading North West via NILAM would be the flight path for Iran. Following the standard flight path N571 (as I have indicated in previous comments as a possible case for MH370 at around 18:27:48), eventually leads to the Gulf of Oman on the border between Muscat and Tehran ATC, just south of Iran. Crossing India consistently on a standard flight path would make you less suspicious to radar.

      Then (as you say) a malfunction occurred soon after 18:27:48 which had the effect to disable pilot control and to realign the flight path with an end point which fits both ping ring and doppler information and thereby thwarting the hijack but dooming the plane.

      The problem of adding radar detection (Malaysian, Thai, Vietnamese) or non-detection (India, China, Australia) or non-statement (Indonesia, Uzbekistan, Kazakhstan) or radar avoidance to the PRO and CONTRA is that we do not know the difference between a detection/non-detection/avoidance or an unwillingness to make a public statement about radar capability (or lack of capability).


    3. Hi John,

      Currently analysing, but it seems clear from the BFO data that the aircraft did not fly over indensian territory, but at least ‘skirted’ the peninsula, as so populary expressed some time ago. I am not sure of direction yet, but straight from, say, NILAM and south does not fit the BFO very well.


  74. “Dr David Gallo, a senior scientist at Woods Hole and co-director of the successful search for the wreckage of AF447, wrote in an email to me: ‘I don’t know any underwater acoustic people that think the pings have anything to do with the plane.'”

    Via, who posted the above comment in response to Ben Sandilands’ (@planetalking) post on The Atlantic article. planetalking:

    1. Hi Nihonmama,

      And just what would an oceanographer know about electronic circuit design and the particular performance of the Dukane pinger at a depth of almost 5km???

      Said pinger uses an LC oscillator (NOT a crystal) to generate its output tone, with a ferrite slug as adjustment of the inductor (and the capacitor is probably the piezo transducer itself). That’s why it has a lousy +/-1kHz tolerance for its tone output, and that’s just under typical conditions.


      1. Hi Joe:

        Very well reported: the team of scientists at Woods Hole, using a Bayesian statistical model, were instrumental in finding the AF447 black boxes – after numerous failed attempts and infighting amongst Air France, Airbus and the air crash investigators. David Gallo was the co-leader of the Woods Hole team.

        I am neither a statistician nor an oceanographer. But if you are of the opinion that Gallo does not have the bona fides or is otherwise uniquely unqualified to comment on the underwater acoustic search for M370, I’d suggest you pose your question to him.

      2. Thanks Nihonmama: I was going to say that the spirit of this website excludes any appeal to authority (or statements against anyone on the basis of specifics of education and/or experience and so on). Anyone is qualified to comment on anything, so long as their comments are science- and evidence-based.

      3. Joe, have a look at this video clip of a ULB pinger being dismantled.
        The 37.5 Khz [carrier] frequency is generated by a little piezo electric device, and a small handful of electronics, to ensure a stable frequency, and a stable pulse rate. It is certainly not an LC oscillator. It would be possible to build a simple RC oscillator with a handful of parts costing about $10 that would meet the stability specification, but using the peizo electric device, similar to that which you would find in many simple battery driven watches should ensure even greater stability. The power consumption of the electronics is tiny [think how long your watch runs on a little battery], and how accurate it is over a much wider temperature range than the ULB will encounter. The electronics are potted in a silicone compound for g-protection. The pot-core inductor, with the little ferrite tuning slug, is used to create a larger energy pulse to drive the piezo-electric ring inside the casing. Very likely it is also “tuned” with the ferrite slug to maximise the efficiency of the energy conversion process. Any variation in L will not affect the base frequency. It is this part that of the circuit that consumes the battery power. The pressure at depth is managed by the solid cylindrical tube that the parts are mounted in.
        A frequency “drift” of more than 10% is just not plausible.

  75. Ron,
    Using 8 minutes for your banked turn calculation seems too long. At least one read of the three timestamps as picked off the graphs to the second are:
    18:25:24, 18:26:54, 18:27:48 (of course, these may be wrong…)
    And, The third data point appears to indicate the plane had completed the turn, and was already on its final heading out across the ping rings.
    So, the section of turn seen in the pings is somewhere between 1 minute 30 seconds (00:01:30) and 2 minutes 24 seconds (00:02:24). Doubling those times would lead to a turn 3-5 times quicker that 8 minutes.

    1. Thanks Bill.

      Yes, “the section of turn seen in the pings…” – But the complete turn would be a little longer than that. I’d go for 3/4 minutes. If one picked 4 minutes then adjusting from Ron’s 8 minutes can be done in the head. Not even a need for the back of an envelope. By the way, the demise of snail-mail and thus the reduction in the used-envelope supply is greatly hampering scientific research.

    2. Bill, I think you are referring to my comment:

      If the width of the 3 pulse interval is: 2:24 minutes and all previous assumptions remain then:

      omega = pi / (2 X 144) 1/s = 0.0109
      v = (470 X 1852) / 3600 m/s = 241.789
      g = 9.8 m/s^2 = 9.8

      tan(theta) = (omega X v) / g = 0.2691
      theta = 15.063 degrees

      This is still a mild turn.
      If the satellite’s elevation was about 50 degs the antenna can’t lose line of sight with it no matter what.

      If the 3 pulse interval is 90 seconds we get:

      tan(theta) = 0.4306
      theta = 23.297 degrees

      Still a mild turn and no SATCOM disruption.

  76. Harm,

    I am unsure whether elements of Richard’s synopsis of can be considered illogical. Sure, the validity of his statements are open to being tested, but then this he has acknowledged by providing several. Also, his comment has several generalized assumptions, among them: what is presently known regarding flight trajectory has been informed by Malaysian primary radar; the Malaysians have confirmed that the aircraft traveled back over Malaysian in a southwesterly direction before exiting Malaysian airspace over the Straights of Malacca.

    Richard has thus provided a concise expression of what a ‘2 event’ flight path trajectory would look like without venturing into more speculative frames. It is what amounts to a strict analysis of the ‘behaviour’ of the aircraft (human pilot + autopilot). Finally, that Richard is highly informed regarding aircraft autopilot systems is something that cannot be disregarded. In my mind at least, this elevates the superiority of his position as a behavioural analyst regarding the early phases of the flight.

    In short, I believe Richard was generalising in the interest of being concise. So, in terms of the behaviour of the aircraft, you can either have my long hypothetical ramblings or his concise, well informed summaries!


    Regarding witnesses to the flight, I am thinking more along the lines of human elements of the processes of the Royal Malaysian Air Force or the Royal Malaysian Police. You know, radar operators, signal corps officers, squadron leaders, etc. With the right mix of pressure being applied by the International Investigation Team and perhaps an investigative journalist or two, perhaps the RMAF and RMP will share more about what they know regarding that segment of the flight while it was in Malaysian airspace. These are actually the two big fat elephas maximus in the room.

    As for possible airstrips available for a landing in Malaysia, they are numerous, as Warren has so indicated. The fact that Langkawi was not utilised along with other airports (e.g., Penang) is one indication (in my eyes) that a ‘pure’ malfunction at IGARI is less likely. Yet, if the aircraft was indeed diverted at IGARI it would most probably have had a destination. What was that destination? Simplicity provides the most likely (although not necessarily correct) answer(s).


    I don’t know if the 2 event hypothesis has gained all that much in popularity; has it? I and a few others have been stipulating it for weeks that it is the only way to resolve an intentional diversion with a terminal flight trajectory, and people have generally only viewed me as a speculative kook. As for what the flight recorders will reveal, don’t lose heart just yet, we yet have a possibility of new information now held on the part of the RMAF or the RMP being made public and open to scrutiny.


    Again, I believe Richard was generalising and in the interest of remaining concise only stipulated one PRO and one CONTRA argument for each element. You are correct that a hijacking could have entailed an intended diversion to a distant, central asian destination. What I believe Richard is trying to indicate, however, is that barring this scenario, an intentional diversion is difficult to reconcile with a terminal flight trajectory, as it only produces “the aircraft was hijacked with the intent of engaging a terminal flight trajectory to the point of fuel exhaustion,” which can be argued is less than likely given the relative probabilities of other hypothetical constructs.

    1. Dear Rand, dear Richard,

      indeed I misunderstood the emphasis. I was mixing events and facts. Indeed the turn at Igari was an event and the pings at Nilam and the later pings around 00:10:48 and 00:19:00 may each be indicative of a further event.

      1. Hi Harm,

        No worries! Please continue to question everything!

        We refine precious metals with fire.


  77. The listing of events by Richard to me appears to be not completely logical:
    1. MH370 turned back around 17:20:38 near to waypoint IGARI, or just thereafter, closer to BITOD.
    2. 3 pings around 18:25:24, 18:26:54 and 18:27:48 within a time frame of around 2 minutes 24 seconds near to waypoint NILAM indicating an aircraft problem requiring correction or recovery.
    3. 2 pings around 00:10:48 and 00:19:00 which are around the same time fuel exhaustion is predicted”

    There are no pings at around 18:26 that, on their own, prove the a/c to have been near NILAM. This conclusion can only be arrived at if at the same time accepting the radar contacts presented by the Malaysian Air Force as being linked to MH370. Even event 1, e.g. turning before reaching BITOD, can only be accepted as a fact if concluding that otherwise the a/c would not have had enough time to be at the right spots in the Malacca Straits at the right time.
    We should acknowledge that there are events/facts that are, or seem to be, pretty firm (distance to sat on the basis of time delays, a/c positional information from secondary radar or ADS-B) and events/facts that are more disputable (eye-witnesses, unconfirmed sources of information, alleged blackbox pings, etc).
    This is by no means a criticism of the conclusions reached by Richard – I too think that the a/c turned shortly after IGARI and flew at just above 500 kts to be near Nilam at the time of the 18:26 pings. I merely object to the black-and-white discarding of some facts/events and the selection of a few as the only relevant ones. According to me, there is a spectrum of events, some very plausible, some less plausible, on their own. One should however not discard the possibility that less plausible events, seen in combination, may render an event that at first sight seemed to be very pertinent, less likely. In the present case for instance an official confirmation of a presently dubious event like the co-pilot cell phone having connnected at around 01:50 to a Penang comm tower would suddenly render without doubt that the a/c did turn west and that the radar contacts in the Malacca Straits do correspond to MH370.
    Some evidence may be on the back-shelve for the time being, but as long as it is there, one should not completely disregard it.

    1. Hi Harm,

      I apologise that I appear to be illogical. My summary was not clear.

      You state “There are no pings at around 18:26 that, on their own, prove the a/c to have been near NILAM. This conclusion can only be arrived at if at the same time accepting the radar contacts presented by the Malaysian Air Force as being linked to MH370.”

      What you state is not true.

      My summary was based on this previous posting (please see below), where I had already discounted the Malaysian Military Radar (which I personally accept, but others have questioned).

      In this posting I showed that even if you ignore the Malaysian Military Radar and just take the point IGARI and work out the nearest point on the ping ring at 18:27:48 to IGARI, the minimum speed of MH370 was 417 knots and the nearest point is not far from NILAM.


      1. This is still unverified, but looking at the BFO behaviour, it may well be that the speed was higher than 300 knots also for the remainder of the flight. Rather than slowing down and following a great circle towards the end, it somehow seems to first have gone further south, maintaining its previous cruising speed. The end position is likely the same, but it may affect the reasoning about what actually happened.

    2. Referring to Harm’s guess that the tags on the Malaysian radar track points might be “flight identification codes”, this does not appear to be compatible with Malaysian authorities stating that they are not sure that these radar tracks were produced by MH370. Is it possible instead that these tags indicate the time associated with each track point, with the time given as hh:ssH, which is the same format as the time shown in the two boxes in the radar figures? This format would have six letters ending in H as Harm has observed for two of the track points, and would indicate information available to the radar operator. As for the track point ending in TA, perhaps this indicates a point coming from a radar other than the Butterworth radar, such as a Thai radar, for which the time is not known as well. AThai radar has been quoted as spotting MH370 in the Malacca Straits. See

      1. Hi Ron,

        a posting that I had overlooked at the time was one made by Bill. He actually took up the same issue and came to the conclusion that the tags might be time stamps. I see the logic in that although, even if assuming that the tags were time stamps, it did not much help me in reading the tags. For me the tags are too short to be of the format 0202 and indeed 02:02H would better fit the length. But I see the H only in 2 of the tags and this is therefore by no means conclusive evidence. If the pic would have been in an only 2 times better resolution, then I am sure that I would be able to read the tags …..
        The tag possibly ending with TA pertains to the blip shortly before the circle with no blips and is clearly off-track. I do not think that that blip can reasonably be associated with MH370, whether from Malaysian or from Thai radar. I would however tend to exclude that some of the blips originate from other than Malaysian radar. Or for that matter from any radar other than the Butterworth radar.

        As long as I am not able to read the tags, I see at least (only) the following 2 hypotheses:
        1) The track after the circle without blips, e.g. starting at VAMPI, could possibly be related to another a/c, like uae343 that, according to ADS-B info, followed the same track about 4 minutes later
        2) If looking at the blips only, e.g. ignoring the tags, the radar plots do not at all support a turn towards NILAM.

  78. Regarding Inmarsat’s reluctance to release the calculations and data, it could simply be because they are working to file a patent on it first. I wonder if this blog can be considered an example of prior art? ;-)

    Some people may have reflected on the fact that although the calculations by Victor and Mike in essense use a BFO of inverse sign to my own calculations, we still end up with the same solutions. The reason is simply that the radial doppler component, which obviously enters the two methods with different sign, has no impact on the final position along last ping ring. The only difference is that small detail of whether the flight went north or south.


    1. Inmarsat are wisely keeping a low profile. The MH370 affair brought them public relations and company shares jumped a bit but in the long run clients will wonder if they are being tracked like MH370 was. If everyone will think BFO is problematic it could only help.

      Interestingly the president of global government business for Inmarsat, warns us not to expect privacy using satellite communication:

      1. Satellite companies use the same solutions as cable companies to maximize bandwidth usage – perhaps even more so, since bandwidth is so precious on those links. I happen to be developing the inner workings of those systems. As a side effect, tracking is most certainly possible. However, there are plenty of ways to defend against tracking if one really wants to. Most people just don’t care, and rightfully so.

      2. Henrik thanks.
        It was just a guess they want to keep a low profile. Sorry if it sounded otherwise.

  79. A small bit of data that may evaded your keen eyes.

    Flight Radar 24 save and display only a sample of the data points they get from
    ADS-B receivers. On March 9 admin Mikael Robertson was asked in the forum to
    search if they had more data points on MH370. He quickly answered:

    We normally save data once per minute on cruising altitude. Now we have checked
    all databases and logs and have managed to reproduce some extra data that was
    not available first, so we have more or less 2 points per minute for the last minutes.
    I don’t think we will be able to get more data than this.

      1. Yes, I noticed and tried to find an explanation.
        8535 decimal is 2157 hex. Maybe this helps?
        Maybe we should ask Mike.

        I didn’t check but it seems FR24 have fuller coverage of MH370 than FlightAware.

      2. Regarding the strange squawk code in Mike’s data table. I looked again at the FR24 forum. Someone asked Mike:

        “The data shows a Squawk Code out of limits! Octal numbers are 0000-7777. Has someone tampered with the code column Mike?”

        and he answered:

        “I’m just pulling the data from DB. Not sure about the structure and format. Could be some other format. For example the flight ID 2d81a27 on is the same as Flight ID 47716903 in the picture.”

        Since we don’t know in which format the squawk code is kept in the FR24 DB and how Mike pulled the data it’s difficult to understand what happened. I’ll think about it. Hopefully the other data is OK, format mismatches usually create noticeable errors.

        By the way, Mike’s table doesn’t include data points with zero altitude seen at the end of MH370 replay. His answer:

        “There are some 0 yes, but I don’t want to speculate where they come from. It could just be bad signal.”

      3. Ron ~ Thanks for your further efforts. I’ll file this discrepancy in my 9M-MRQ Doppelgänger Conspiracy File. ~LGH~

    1. Mikael Robertsson: admin at FR24? He’s a co-founder at FR24! I’ve been asking FR24 for a few such dumps for 4 adjacent flights passing west of Penang, they haven’t yet obliged. I’ll check the forum, that must be where the ‘organ grinders’ hang out.

    2. Thanks for the link, Ron.

      Oddly, the “extra data” is at odds with the collated FlightRadar24 data published at Pprune on 22 March by the user “LFRT “:

      In LFRT’s collation, we see the expected squawk 2157 – where a squawk code is provided, since the data were assembled from various sources, as explained in the post.

      There are several discrepancies between the datasets, though, especially as the data from the live flight tracker page differ from those from the “pinned” (archive) page – according to LFRT. I’ve not seen an explanation for this.

      Notwithstanding, and judging from the coordinates given, it looks as if MH370 passed IGARI at approx 17:17 UTC, flying at 473 kts on a heading of 25. The last datapoint in the list is at 17:21, a couple of seconds after MH370 said “Goodnight” and vanished into the informational void.

  80. Hi Duncan,

    I believe both you and your key contributors such as Mike, Victor, GlobusMax, Eugene, etc. have the joint expertise to offer the MH370 families to analyse the raw data from Inmarsat and quite quickly show the MH370 flight path taken, if the Governments of Malaysia, China and Australia respond positively to the open letter sent today by 350 of the MH370 family members.

    Sarah also signed this letter and I know that you were in touch with her directly.

    I will gladly help where I can. I spent 5 years in developing auto pilots (unfortunately not the MH370 autopilot) and automatic landing systems. I spent a further 5 years on the NASA Space Shuttle where I built a simulator of the Shuttle and on the ESA Spacelab where I was the hardware software co-ordinator for the integration and test phase.


  81. Following AndRand’s comment:

    Some measure of the compatibility of the Malaysian radar track with the new ping rings
    can be found if the distance from the track’s end to the 18:29 UTC ping ring is calculated.
    Dividing the distance by 18:29 – 18:22 = 7 minutes the average velocity can be computed.

    The distance to the ping ring is the minimum distance covered in these 7 minutes so the
    calculated velocity would be the minimum ground speed. Assuming no significant winds it
    should be smaller than the B777 max speed in the relevant height.

    In the other direction if the velocity is assumed the corresponding points of transit can be
    found on the ping ring and used in computing northern and southern tracks.

    Note that incompatibility of the radar track with the ping rings wouldn’t be surprising since
    the Malaysians themselves are very careful not say clearly that MH370 came back.
    For example the Preliminary Report only says:

    “A playback of a recording from military primary radar revealed that an aircraft with a possibility of MH370
    had made an air-turn back onto a Westerly heading crossing Peninsular Malaysia. The search area was then
    extended to the Straits of Malacca.”

    Such careful wording appears also in the prime-minister’s CNN interview:

    Note that the Thai Air Force was also careful:

    The story on the co-pilot’s phone connecting with a Penang cell tower came later and
    raised suspicions on technical grounds:

  82. To Duncan et al,

    Kudos to all of you who have painstakingly gone through all the data and are trying to put a ‘jigsaw’ puzzle together.
    Am sure the families appreciate the efforts made here.
    Is there anyone ‘on the ground’ that have made an attempt to look at the suggested places, I.e the Bestash Valley etc ?

    To Dale.C, you’re spot on about the line of sight and the temp loss of SAT connectivity via the Beam Steering Unit have to ‘realign’ to pick up again the next spot/wide beam as the aircraft turns/banks . Hopefully with that new info more logic can be applied and good data be obtained.

    Be Well,

    1. Duncan:

      This is a duplicate post. Never sure where to reply.


      The AirLink steered phased array antenna does have shadow ed AZ angles, especially straigt forward and straight back. But no transmission is required to re-estabish the tracking. That is done with the AIMS/IRS data when available, like this (MCS-4200 manual):

      The BSU or ACU is used in the HGA system to convert tracking and pointing
      coordinates (that is, aircraft–relative azimuth and elevation) from the SDU
      into signals needed to select the antenna array elements in combinations
      that point the antenna beam in the desired direction towards the satellite.

      … and after the loss of that nav data, it would probably default to a AZ/EL scan looking for the P Channel (a pilot channel), adjusting AZ/EL beam steering to maximize signal strength. This is all done using a receive signal only.

  83. Duncan ~ Are you posting comments and replies chronologically or at random? And thanks for the new search bar. I agree it has its limitations & still testing it. ~LGH~

    1. Posting acceptable comments and replies as I meet them, which I guess means chronologically. Note that the times associated with comments are the times when they were submitted, and not when I approved them for posting. And also note that just because they appear here does not imply that I agree with the arguments put forward in each case! Just facilitating exchanges of views and information.

      1. Yes, I’ve just noticed after my most recent comment that EST is shown, previously NZ time. I quite enjoyed looking into the future…..

      2. I. too, think preservation of this blog and its comments is in the historical interest. For that reason, I hope Duncan does NOT carry through with his earlier “threat” ( ;-) ) to delete comments pertaining to pings, acoustics, and etc. As a record of various considerations and in order to maintain consistency of participants’ thinking (“No, I SWEAR I read that somewhere!”) , I believe they should stay. Of course, the final decision is Duncan’s, but I hope he errs on the side of preservation.

  84. Duncan,

    If this has already been shared, please disregard, but I thought this article posted on the The Atlantic’s website would be good reading for those visiting your blog. Based on the article, it’s not clear if those cited (including you) were directly contacted or not, but the article seems to present things consistent with what’s been shared on your site, at the very least at a more general level.

    Again, keep up the good work.


    1. Anyone reading this website should now go read that article in The Atlantic. It is clear, well-written, and essentially sets out the reasons for the various efforts from many people on this website.

      1. That was my thought too. Glad to see everyone’s great work here starting to get some broader attention.

      2. Duncan: very glad to see that the earnest and important work being done here has finally gotten media attention. Tweeted this am.

      3. Wonderful article, particularly for non-technical readers, marred only minutely by the misspelling of “linchpin” as “lynchpin” in the fourth paragraph. However, I’m guessing that was a Freudian slip, pertaining to the first syllable and certain folks involved in the investigation. ;-)

      4. Thank you Duncan and Nathan. helped me to a better understanding of the science as I do not consider myself a qualified or technical person.

        However, Duncan, a couple of questions:

        Your arc of the satellite moving in a north-south ellipse, the distances shown being plus and minus about 1.5 degrees of latitude, are these projected down to sea level or at the satellites altitude?

        Ground speeds are calculated from the satellite ping rings for the aircraft at rest in KL. What then is the calculated “ground speed” of the Perth ground station? Should this also be taken into account?

        ps: no conspiracy theories this week.

      5. Q1: It’s the same thing! When the satellite is above latitude 1.5 degrees it’s at declination 1.5 degrees.

        Q2: Your initial sentence makes no sense in terms of what is done. When the aircraft is at rest the LOS speed satellite-aircraft is due entirely to the satellite’s motion, obviously enough. Similarly the satellite-ground station LOS speed is due exclusively to the satellite’s motion. And that *is* included in the BFO.

  85. Hi Duncan,

    After 26 posts and 2,021 comments, I thought I would try and summarise as follows:

    1. MH370 turned back around 17:20:38 near to waypoint IGARI, or just thereafter, closer to BITOD.
    2. 3 pings around 18:25:24, 18:26:54 and 18:27:48 within a time frame of around 2 minutes 24 seconds near to waypoint NILAM indicating an aircraft problem requiring correction or recovery.
    3. 2 pings around 00:10:48 and 00:19:00 which are around the same time fuel exhaustion is predicted.

    1. No attempt to make an emergency landing after turning back from near IGARI as MH370 maintained an average speed of at least 478 knots to the end point of the Malaysian Military Radar. MH370 did not slow to 250 knots and descend to 5,000 feet to try an emergency landing approach.
    2. No transponder data or communication from MH370.
    3. No crash site or debris found.

    THEORIES (I have left out any conspiracy theories and only summarised theories mentioned on this web site)
    1. Malfunction (Electrical Fire, Cargo Fire, Lightning Strike, System Failure, Electrical Fault, Mechanical Fault, Decompression, etc.)
    2. Incapacitation (Pilot, Crew, Passengers, Decompression, Hypoxia, etc.)
    3. Hijack
    4. Suicide

    1. Malfunction – PRO – Diversion – CONTRA – No Emergency Landing, Flight continued at least 7 hours 37 minutes.
    2. Incapacitation – PRO – No Emergency Landing, Flight continued to Fuel Exhaustion – CONTRA Flight followed waypoints for at least 1 hour after IGARI (which could all have been pre-programmed).
    3. Hijack – PRO – Diversion, No Emergency Landing – CONTRA – Flight continued to near Fuel Exhaustion.
    4. Suicide – PRO – No Emergency Landing, Flight continued to Fuel Exhaustion – CONTRA Diversion, Flight continued at least 7 hours 37 minutes.

    Each theory has pro and contra, therefore a combination of two or more problems has been proposed which may align to two or more events listed above. e.g. 2 Malfunctions, Hijack and Malfunction, Malfunction and Incapacitation, etc.

    1. Northern Route appears to have been intentional Pilot input (e.g. Qamdo Airport as proposed by Victor).
    2. Southern Route appears to have been without intentional Pilot input (e.g. 300km NW of the recent search area as proposed by Henrik).


    1. > 3 pings around 18:25:24, 18:26:54 and 18:27:48 within a time frame of around 2 minutes 24 seconds near to waypoint NILAM indicating an aircraft problem requiring correction or recovery.

      As I understand it, the pings can be initiated by a sharp turn itself (the fuselage of the plane blocks the line-of-sight between the plane’s antenna and the satellite), and the antenna has to re-orient itself to the satellite. No aircraft problem needed, and this is the most likely explanation for the 3 consecutive pings.

      1. Thanks Dale: I had not seen that suggestion before. Very useful input. It is a self-consistent explanation for the three pings being closely spaced AND having large, quickly-changing BFOs indicating a sharp turn with much of it being in a direction away from the satellite. (Kick myself and think: why didn’t I think of that?)

      2. Dale:

        The AirLink steered phased array antenna does have shadow ed AZ angles, especially straigt forward and straight back. But no transmission is required to re-estabish the tracking. That is done with the AIMS/IRS data when available, like this (MCS-4200 manual):

        The BSU or ACU is used in the HGA system to convert tracking and pointing
        coordinates (that is, aircraft–relative azimuth and elevation) from the SDU
        into signals needed to select the antenna array elements in combinations
        that point the antenna beam in the desired direction towards the satellite.

        … and after the loss of that nav data, it would probably default to a AZ/EL scan looking for the P Channel (a pilot channel), adjusting AZ/EL beam steering to maximize signal strength. This is all done using a receive signal only.

      3. Hi Dale,

        The reason I believe that the 3 pings reveal an aircraft problem is not because of the time interval or the fact that the aircraft was turning (which it was) but rather comes from my analysis of the track at that time matching the ping ring data.

        Please see:

        I show 2 tracks for a northern and for a southern route. The northern route shows a turn followed by 2 further turns to recover the aircraft following a designated flight path N571. The southern route shows 2 turns followed by a correction having over compensated.


      4. airlandseman: I don’t know the specifics, but I would be surprised if the pings are generated due to an aircraft problem occurring during the turn and not generated due to the turn itself. The antenna re-orient might require the SATCOM modem to log-on again to the satellite after losing line-of-sight or whatever the specific reason.

        Richard: the plane’s orientation (pitch / roll / yaw) during the turn re-orients the plane’s velocity vector in relation to the satellite. Yes, pitch and roll
        and true heading are supposed to be taken into account in the pre-compensation, but we don’t know specifics and that might affect re-constructing the turns.

      5. I tried to estimate the severity of the “Possible turn” appearing in the BFO graph just before 18:30 UTC. The basic formula can be found in:

        Substituting: v = omega X r

        we get:

        tan(tetha) = (omega X v) / g

        where tetha is the bank angle.

        Assuming a 180 degrees turn over 8 minutes (twice the width shown in the graph to make it symmetrical) and the cruise speed of MH370 at IGARI:

        omega = pi / (8 X 60) 1/s = 6.545 X 10^-3
        v = (470 X 1852) / 3600 m/s = 241.789
        g = 9.8 m/s^2 = 9.8

        tan(tetha) = 0.1615
        tetha = 9.173 degrees

        This is a mild turn even for a B777. Boeing bank angle protection doesn’t kick in at twice this value. No chance it harmed structural integrity.

      6. Thanks for that, Ron, and the Doppler patents.
        I think you intended to type ‘theta’ rather than ‘tetha’ but the symbols matter not, so long as they are consistent!

      7. Hi all,

        This is perhaps the most intriguing sequence of the flight right now.

        Turning and pinging? Would an attempt at overriding the autopilot or other set value for course and navigation, without disconnecting it in the proper way, generate pings? Or is this strictly about the communication as such, between plane and satellite, and perhaps also engines and structure, would you say?

        I there any consensus about this yet? It does not seem like that here.

        Or could it be the sign of something else deteriorating at that time?

    2. Thanks Richard: That is exceptionally helpful as a checklist or set of guiding notes (but not ‘rules’) that everyone might keep in mind.

      As of yet I am not convinced over the final argument there (north means intentional, south means automated flight) for various reasons. For example, a northern route following automated waypoint choices might be feasible (and VictorI has noted how his ‘best fit’ route can also be closely fitted by a series of standard waypoints); and a southern route could have been steered deliberately (although I doubt it).

    3. Fantastically concise Richard.
      Personally, I share the popular opinion that the truth lies in a some kind of combination of proposed theories.
      I very much hope- as does everyone here- that the wreckage is one day found, however I am pessimistic about what our ability will be to accurately confirm these theories to a state of certainty even once the FDR and CVR are evaluated. The malfunction theory would be discovered easily enough, but any element or presence of any of the other theories that involves human action, intent, or mental state- would continue to most likely remain speculative, even after all crash data is gathered.

    4. “3. Hijack – PRO – Diversion, No Emergency Landing – CONTRA – Flight continued to near Fuel Exhaustion.”
      I don’t think that the CONTRA here applies if you assume that the hijack plan was to land the plane on an unlighted, makeshift runway or airfield. Under that scenario, the plane could not have landed until at least dawn (which at the time of the last satellite ping was just breaking in a good portion of western China, and had not yet broken in Kazakhstan and Kyrgyzstan), and it would also have been desirable to reduce the weight of the plane as much as possible to land on such a runway or airfield.

      1. Hi Rodney,

        If I wanted to hijack an aircraft to a particular location, I would choose an aircraft with a planned destination that required at least the same amount of fuel that I required for my hijack destination, not less.

        I accept that, for example, Qamdo Airport following Victor’s route could be reached before fuel exhaustion.

        However, in general, a time of 7 hours 37 minutes from take off to the last ping at 00:19 UTC and a fuel endurance of less at 7 hours 31 minutes (please see the comment from LGHamilton at the start of the previous post from Duncan) is too tight as a fuel planning basis for a hijack. Of course, there are many factors that affect fuel consumption that have been discussed at length in previous posts, but I would not rely on these factors if planning a hijack.


      2. “If I wanted to hijack an aircraft to a particular location, I would choose an aircraft with a planned destination that required at least the same amount of fuel that I required for my hijack destination, not less.”

        All things being equal that makes sense, Richard, but consider that all variables would not have been under the hijackers’ control. For example, let’s suppose that the hijacking was carefully planned by a sophisticated group that had: (1) determined a suitable landing location in central/western China and (2) detected a hole in Chinese radar that existed only in the wee hours of the morning at a particular border location. Ideally, they might have preferred that the aircraft have more fuel, but they did not have another option that would have allowed them to slip past Chinese radar. So, they may have done a precise calculation that MH370 would have just enough fuel to enable them to carry out the hijacking almost entirely under the cover of darkness, while still landing safely as dawn was breaking. Far-fetched? Perhaps, but what hypothesis isn’t?

      3. Thanks Rodney, but another language warning in that many will read that and think that a hijacking did take place; and of course we do not know that.

        “Far-fetched? Perhaps, but what hypothesis isn’t?” – The answer to that is the simplest one. Pilots and all on board incapacitated by some factor that we do not know, and the aircraft flew on with the autopilot in control.

        Ockham is a small village in Surrey, southwest of London. There are many lovely pubs in the surrounding countryside where one may sit and ruminate on the ways of the world, or you can take a walk around the Royal Horticultural Society’s splendid gardens at Wisley whilst thinking about what might have happened to MH370.

      4. “Far-fetched? Perhaps, but what hypothesis isn’t?” – The answer to that is the simplest one. Pilots and all on board incapacitated by some factor that we do not know, and the aircraft flew on with the autopilot in control.”

        But on the northerly route, Duncan, how could radar detection have been evaded by autopilot? Just dumb luck?

      5. I simply don`t understand why people have a problem with the sparsity of radar detections no matter what path the aircraft took. That is, I do not believe that much luck would be needed, as such, based on my own experience of radar (which is limited) but also my knowledge of how radar works and its limitations (which is greater). I am not saying I am definitely right – no-one should do that – but I am saying that I would not have anticipated a priori any radar identification of a civil airliner travelling over any of the putative regions or routes, just as I would not have anticipated any acoustic evidence of a route over the Indian Ocean and subsequent crash, nor would I have anticipated any rapid identification of sonic pings from a locator beacon on the bottom of that ocean.

    5. airlandseaman:

      Classic aero protocol permits the AES to relogin in the event of a loss or degradation of P channel (defined as failure to read the synchronization word for 10 consecutive seconds).

      I found an article about a test of a low-gain SATCOM system (though dealing with a 747, not a 777) explicitly saying that “satellite elevation angles below the aircraft horizon during banking maneuvers were shown to have a significantly deleterious effect on SATCOM communications”. Granted, it would take extremely sharp banking to put the satellite below the aircraft horizon in our case, since the turn occurred when the satellite was 50 deg. above the horizon, but it’s a possibility.

    6. Richard,
      I like the simplification of your summary.
      …But, I want to immediately complicate it with additional conditions. Mostly #1 Malfunction.

      1a) What if the original malfunction included loss of Throttle Control?
      – No opportunity to descend, thus, no turn towards any runway.

      -1b) What if, following the realization that they could not land, they programmed the FMS to fly a holding pattern out over the Straights, while they worked on a Plan-B. But at this speed, the coordinates entered for the holding pattern were not possible, so the Nav system abandoned the turns, and went shooting off, holding the last heading from the holding pattern.

      1c) What if, following the realization that they could not land, they were flying to dump fuel and turn back to land… And, the plane broke more… Loosing all control, and continuing off on the last heading they were on.


      1. Thanks Bill.
        Input from experienced pilots here have indicated that in a major emergency no attempt would be made to dump fuel. An immediate landing would be attempted as being the least of the many evils.

      2. Hi Bill,

        A good example of the 2 Malfunction theory (first loss of throttle control followed by “the plane broke more”).

        That is what a simple structure is there for, to add complication. I had actually produced the simple structure for my benefit because after 26 posts and meanwhile over 2,092 comments, many of which referenced other material, I could easily lose the thread(s). I have even considered (if Duncan would allow) downloading all posts, comments and references into an Access Database allowing SQL searches including fuzzy searches and complex selection criteria.

        As an avionic expert, you will know that some avionic systems are built without any redundancy as they are not mission critical, others are duplicated, yet others are triplicated or even quadrupled if they are absolutely safety and mission critical. The shuttle even had 5 main computers, 4 from one company and a 5th built by a separate company, who were not even allowed to talk to the other company.

        The throttle must be a safety and mission critical system. It would take a lot to make it fail altogether both as an automatic as well as a manual device.


      3. “I have even considered (if Duncan would allow) downloading all posts, comments and references” – Richard, anything that might assist the search for MH370 is OK with me.

        If you want to do that (especially since the ‘Search’ facility on this website is rudimentary) please do so and let us all know where the resultant database is available.

        Apologies that I cannot do this myself: this is the first time I have run a webpage (and it was initially set up by my 21yo son) and I am not at all skilled in this or possessing the finesse required. On HTML and WordPress I’ve got the brain of a five year old and I bet he was glad to get rid of it (apologies to Groucho Marx).

  86. I wished you had received a more satisfactory reply to your message to the JACC and they would look into your analysis in detail. One of key factors previously which swayed me into thinking that a Southern route was more likely were the multiple pings which were said to be from the black box. I was always a little confused though as to why the pings were spread out like that instead of being clustered closer.

    And lately it seems that even the pings aren’t really reliable:

    Once again, thanks for all the efforts put in on deciphering what little we have and continuing to push for the available data to be looked at again. With the absence of debris and pings likely being false, IMHO its high time to do so.

    1. Thank you to drath for the interesting article posted regarding the underwater pingings’ fascinating scientific analysis points.

      Having recently become quite a fan of Rodney Thomson’s posts, I inquired of him regarding the particulars of the Malaysian Insider article by William Meacham. Here is his reply:

      “As for the link . . . posted, there are some interesting things in there, along with some things that can probably be dismissed. My thoughts below.

      “an enormous amount of biotelemetry tracking is going on in the oceans, with pinger transmitters attached to all sorts of marine life from whales to sharks, sea turtles, tuna, seals, swordfish”

      This is true, however the majority of long-term acoustic tags are of the low-power high frequency type (<150dB SL, 62kHz).

      "First and foremost is the signal’s frequency of 33.3khz. This is NOT within the manufacturer's specs of 37.5 +/- 1 for the black box pinger"

      Well the manufacturer (Dukane) has already publicly stated that this transmitted frequency is possible due to a number of factors (temperature, battery life etc). The author can choose not to trust their statements, but they better have some pretty good evidence that the ULB can NOT transmit at 33.3kHz. Also note there has been precedence of ULB transmitting at out-of-spec frequencies, such as Air India 182 (42kHz) and Air France 442 (34kHz in lab testing after recovery however this may have been influenced by the 2 years it sat on the seabed).

      "The range of detection is another major issue"

      I completely agree. This is the only part of the Ocean Shield detections that has me puzzled. I have suggested some possible alternatives, ie transmitters drifting to the south, but that seems unlikely.

      "Scientists from the Woods Hole Oceanographic Institution used a pinger with the same specifications as MH370’s in their study of baleen whales. [2]. They tested the range of detection and found the maximum to be 2.3km. But the distances between the four detections of the “black box pings” were well beyond, at 9.5, 12.3 and 13.6km. These signals clearly cannot be from the pingers on MH370’s voice and data recorders"

      Note that the their measured source level of the pinger was 143dB re 1uPa @1m. Compared with 160 dB of the ULB. I'm not saying that I think that reception of ULB pings at 9.5+km is possible, but I'm saying the author can't point to this paper and say it proves their point.

      "Dr Lee Freitag, one of the scientists in the study that I contacted, expressed scepticism that the pings were coming from the black box, and also confirmed that the frequency of the pinger would not change due to deep sea conditions. [3]"
      I bet Dr Freitag confirmed that the deep ocean would not change the frequency of a signal once transmitted. However, depending on how the signal is GENERATED it may be impacted by temperature, pressure etc.

      "Two other pings were detected during the search. A Chinese ship detected a pulsing signal, reported at 37kHz. And a “fifth ping” was detected by a sonar buoy. These were dismissed as unrelated to the plane, almost certainly attributable to drifting or tracking pingers present in the search area."
      I'd love to know the authors references for that last sentence.

      "One wonders if, in the eagerness to believe and the absence of any other lead, the authorities directing the search have ignored or downplayed evidence that is clearly contrary."
      I believe that in 'the absence of any other lead' you go with what you have got! I mean, where else should they have been searching?

      Now don't get me wrong. As time goes on, and the more I think about it, the more doubts I have over the Ocean Shield detections.

      The signals received are consistent with ULB transmissions, and despite numerous suggestions, I've not come across any better suggestions for alternate sources of the signal (the leading alternate candidate is a 3rd harmonic of a low frequency echo sounder in conjunction with the use of the TPL in a 25-50kHz filtered setting, which seems unlikely).

      The range at which the signals were detected is the source of my doubt. And as they say, you can't beat physics."

      My thanks again to Rodney Thompson for taking time to write his thoughtful replies to my e-mails. My thanks similarly to Duncan and all the contributors to this fascinating blog.

      Michael Molinaro

      1. In terms of alternate sources for the acoustic pings how many of them not only fall into the required frequency range but also have the same pulse repetition rate and pulse length as the Dukane locator beacons?

        From the analysis of brief recordings that have been released of the acoustic pings it seems as if the recordings are a pretty close match in terms of pulse repetition rate (0.9Hz – 1.111s) and pulse length (10ms).

    2. Drath,

      Thanks for the link on the black box frequency discrepancy. This is a stunning report that explains why the 33.3 Hz signal is almost certainly not coming from the plane’s black boxes which emit at 37.5. This difference has bugged me since first hearing about it. In all my studies, never have I seen a natural condition that could shift a mono frequency signal like that. Again, thanks for the article. I hope a lot of the readers on this site get a chance to read it.

      1. Charley, did you see Michael’s response regarding the frequency?

        “First and foremost is the signal’s frequency of 33.3khz. This is NOT within the manufacturer’s specs of 37.5 +/- 1 for the black box pinger”

        Well the manufacturer (Dukane) has already publicly stated that this transmitted frequency is possible due to a number of factors (temperature, battery life etc). The author can choose not to trust their statements, but they better have some pretty good evidence that the ULB can NOT transmit at 33.3kHz. Also note there has been precedence of ULB transmitting at out-of-spec frequencies, such as Air India 182 (42kHz) and Air France 442 (34kHz in lab testing after recovery however this may have been influenced by the 2 years it sat on the seabed).

  87. Regarding BFO Summary:

    We are getting there. I sorta wish we could agree on one arbitrary convention: The sign of Doppler. Both conventions are being used, making results more difficult to compare. Personnally, I like to define Negative Doppler to mean moving apart, and positive Doppler to mean moving together. It’s just the way most people think about the frequency change from normal. Both conventions are right as long as you keep track of all the component signs systematically. Let’s pick one to make comparisons easier.

    (1) based on satellite at assumed zero latitude and fixed altitude and longitude; (I’m betting yes at this point. We should know soon. Honeywell could be a help here to.)
    (2) applied in full or with a weighting factor; (In full with possible turn projection?; see discussion on dynamic tracking here:
    … but note that MCS-4200 does NOT use the Rockwell TX Offset system. Still good background on how the Doppler is affected by A/C dynamics.
    More important for the Rockwell scheme than the one used in MCS-4200.
    (3) applied always or subject to a minimum tolerable threshold; (Maybe, but no economic reason not to make it work all the time.)
    (4) derived from a single aircraft sensor measurement of position and velocity, or a time weighted average; (neither; this is not a “noisy” measurement;
    based on real-time full dynamic IES data; Kalman filter of GPS, airdata, ADHARS, etc. Some small dynamic error for normal turns, but very little static error)
    (5) subject to a deliberate small dither (unlikely); ( NO)
    (6) subject to defined default mode when sensor data is lost or deemed dubious. (by definition there is a default mode, but what is it? Honeywell, are you there?
    Can we figure this out from the projected tracks? Maybe close to that now.)

    Add (7) subject to a small uncompensated transponder LO error, probably less that ±5 Hz; caused mainly by some temperature coefficient driven by solar
    angle and eclipse conditions, as recently discussed in the news.
    Add (8) subject to AES reference oscillator error, believed to be small, probably <2 Hz. Ref is an oven TCXO good to 10^-9 short term, with GPS long term freq. accuracy.
    Add (9) possible assumption that the AES ROC is always zero. Points 1-4 suggest this may be true. It simplifies the calculations, like the fixed satellite assumption simplifies things.

    1. ..and maybe also add to bfo summary points to ponder:
      (10) correction offset applied in granular fashion (eg to nearest multiple of 10hz or something)

    2. From –

      When individually wired to the zero state, configuration pins TP11C and TP11D indicate, respectively, that the inputs designated for the primary and secondary IRSs(MP6A/MP6B and MP6C/MP6D, respectively) are connected to an ARINC 429 source of IRS label 310, 311, 312, 314, 324, 325, and 361 information (although label 361, Inertial Altitude, is not required for SATCOM).


      310 – Latitude
      311 – Longitude
      312 – Ground Speed
      314 – True Heading
      324 – Pitch Attitude
      325 – Roll Attitude
      361 – Inertial Altitude

      Some of the IRS data e.g. pitch and roll are needed specifically for the high-gain antenna configuration where the antenna is steered towards the satellite.

      Unless the SDU differentiates the inertial altitude it won’t have a rate of climb (ROC) to use for it’s Doppler correction (AES TX Offset), also since it only has true heading if there is any wind then the difference between heading and ground track won’t be accounted for.

      Not sure whether they would account for the differences of up to 22.4% that you calculated for the 4 calibration points.

      1. Hi Sean,

        If the AES TX offset were to only consider velocity in true heading direction, but not COG (course over ground) direction, we should expect large errors for that part of doppler contribution.

        I have seen planes landing at significant (heading) angles to the landing strip direction in strong cross winds. Lets assume (for ease of calculation) an angle of thirty degrees, than the offset calcs would include a phantom velocity component normal to the actual (COG) velocity of 50%.

        Coming back to the Inmarsat doppler slide, for D1 it states “LOCATION, heading and speed”. Here location would only make sense in terms of doppler contribution if it were used to derive (by differentiation) COG velocity. Location by itself by definition is stationary in the reference system and would therefore not contribute to AES side doppler.

      2. I have to correct my assertion about “location”. The location of the aircraft is of course essential in working out the direction vector between satellite and AES, this being the vector to project the air craft velocity vector, i.e. heading and speed (or may be COG and VMG?) onto to calculate the LOS speed, hence AES side doppler contribution.

      3. After a good night’s sleep, I had another thought on D1 related to my above posts.

        If the D1 doppler is indeed calculated from “heading and speed” as the inmarsat slide seems to infer, that calculation would have a potentially large error (e.g. 50% phantom normal component at 30deg track to heading angle [likely due to wind]). The geometric-mathematically correct parameter to use for a “good” correction would of course instead be the actual COG/VMG data.

        At first sight, this appears to be rather bad for our attempts to derive possible locations of MH370. However, this error in offset calculation, if present in the BFO data, could be a basis for including the effects of the wind.

        We should be able to mathematically formulate the delta introduced by the difference between the heading-speed vector vs the COG-VMG vector to the doppler and BFO data. I suggest the following vector relationship:
        V.cogvmg = V.headingspeed + V.wind

        After solving that for the wind component we may be able to correlate that to the existing wind profiles in the region and narrow down possible locations.

    3. Hi alsm,

      I struggle to understand your point 9. Could you elaborate how or why points 1-4 suggest that a zero ROC is likely assumed in the TX offset algorithm?

      In point 2, you suspect that “possibly a turn projection” may be applied in the algorithm. I scimmed over the patent graphics and think that the corresponding algorithm would be quite complex in comparison to the inclusion of ROC.

      Sean McLeod posted a list of inputs in a previous reply to your post here:
      Lat, lon, ground speed, true heading, pitch, roll and altitude.

      Sean suggested that the ROC could be derived by differentiating the altitude input. He also asserts, that given that only the true heading is available would mean that the difference between heading and track is not accounted for, if any wind is present.

      It is an interesting exercise to analyse what the processing unit COULD derive from the list of inputs:
      ROC = d/dt(altitude) as per Sean’s suggestion, and also
      ROC = v_ground * s * tan(pitch), where parameter s is a scaling factor expressing the lift efficiency at a given speed, in first approximation s=1, if pitch angle and angle of climb are near equal
      V_ground = d/dt(position(lat, lon))

      Here it bugs me that v_ground is the input. If it were v_air, the algorithm could derive the correct full 3D velocity vector of the aircraft from the inputs. The implication being that the algorithm could providing an accurate TX offset by projecting the actual 3D velocity vector onto the LOS vector (but including the error of assuming a stationary satellite of course, if that proves to be the case). ROC and wind effects would be accounted for in D1.

      I scrolled through the document Sean linked to, but didn’t spot the list, which Sean cites. I will have another closer look. Maybe, it was a transposition error, and it is indeed a v_air input instead of v_ground?

      Sean, if you are reading this, and remeber where in the doc you extracted the list, can you have another look (just in case I fail to find it myself)? You also stated that “altitude is not required for SATCOM”. Is that a quote from the document or your own assertion?


      1. Found it!

        Sean transcribed absolutely correctly from Page 5–63, including the assertion about altitude.

        Sorry for doubting.


      2. Continued searching the document and found this:

        Page A-26, 4.A.(2):
        “The MCS system requires ARINC 429 data for antenna pointing, antennastabilization, and Doppler frequency correction. These requirements are defined inTable A–10. If the aircraft does not have an IRS that supplies this ARINC data, theSCU can be used to supply the data.”

        Page A-27
        Table A–10. ARINC 429 Data Requirements
        This table
        – adds label 313 Track Angle
        – omits label 361 Altitude

        So this Honeywell/Thales document is inconsistent on what data is used to calculate the Doppler frequency correction. I will ponder what the correction algorithm could derive from this data set…


  88. Aside from this great work on the pings and Doppler, I am looking at another area where I believe we could extract some meaningful information out of the scant data we have gathered on this flight..

    The problem with finding this specific aircraft, and figuring out what happened, became a whole lot harder when the airplane stopped providing location information after the last data transmission from the airplane at 17:21:03UT. (last ADS-B and transponder off ~17:21 in most reports)

    >>> After that time, all communications from the plane (via satellite comms, both the ground initiated handshakes and the plane initiated events) contain no location information.

    Remember that the plane WAS transmitting periodic location information
    – transponder/TCAS (up to 17:21), and
    – ADS-B via VHF (up to 17:21), and
    – ACARS via satellite (up to 17:07)

    After 17:21, the satcom system was functioning just fine, and communications were taking place periodically, and as-needed. But, the comm system, apparently, had no location data to send.
    So, even if the flight was streaming location data every second, the ‘data’ would have been blank. Thus, it would have made no difference at all in our current situation.


    The satellite communications subsystem is not located with the primary flight electronics bay below the cockpit. All of the satellite electronics are in rack E11, above the passenger compartment, behind the wings. This location is underneath (or between) the antenna. All flight data is provided to the SDU from the main avionics equipment bay over cables. Individual data bus cables are provided from multiple redundant sources – ADIRS/IRU/INU/INS/GPS (air sensors, inertial reference, GPS location, etc).
    General command and control of the satellite subsystem is through the primary AIMS system, where satellite comms is just one of the available comms systems (along with the VHF and HF radios) managed by it’s communications routing functions.

    The SDU also has direct wiring for voice communications from the flight deck (pilot & co-pilot headsets and O2 masks), and the passenger cabin (intercom phones). Uncertain if these links remained connected and operational, or if anyone attempted to use them.

    Currently looking for methods to determine from the available MH370 information, what other systems were still operational, and/or what other systems were still communicating with the SDU. Thus far, possible scenarios include:
    – Catastrophic damage to the cabling bundles from the front avionics bay to E11 rack (due to physical damage from ?? — fire, explosion, lightning)
    – Catastrophic damage to both left and right AIMS plus all Inertial and GPS subsystems (due to ?? — fire, explosion, lightning)

    >>> Complicating Notes:
    – Total loss of the primary avionics bay, (including control surface sensing and control) would have brought the flight to a rather quick end… So, we are looking for some partial damage scenario
    – Manually flipping off breakers for the VHF/HF radios, transponder and ACARS would not have blocked the satellite from transmitting location information.
    – Disabling the voice comms routing through the AIMS system may (or may not??) have blocked voice comms through the satellite system. At a minimum, manual intervention would have been required to override the automated comms control and directly route voice comms to the sat. (I am still unsure about this.)
    – Loss of the cable bundles to the SDU would NOT have disabled the transponder, ADS-B over VHF, voice communications over VHF & HF, etc.

    The satellite comms system is comprised of:
    – SDU=Satellite Data Unit Honeywell-Thales MCS-4200,
    { HDU= High Speed Data Unit -MH370 does not have, no Switft64}
    – RFU=Radio Frequency Unit,
    – HPA=High Power Amplifier,
    – BSU=Beam Steering Unit, (part of the antenna system)
    – HGA=High Gain Antenna (dual ‘patch’ phased array antenna & controller)

    Curious that the SDU (the one subsystem was can tell continued to work properly), is NOT redundant, since it is not a critical flight system. But, the many redundant subsystems could not talk to it.

  89. Apologies, if this is not accurate or initially was misreported, but concerning the North/south determination by Inmarsat- I seem to remember that at the time the south path was decided, it was not due to them obtaining data (such as BFO) that indicated on its own that the a/c went south, but rather that they obtained the straight data from the signals that they assumed were mh370- Then they looked for the same signals from the same satellite from flights that were operating in real time on both northern trajectories (say KUL to PEK) and flights heading to southern destinations (say KUL to PER) and found that the data from suspected mh370 mirrored the data set from the south-headed flights and was contradictory to the data from north-headed flights. This would imply that there wasn’t actually a need to fully understand the data in hand, or have the data specifically dictate anything. It would only require the comparison between the known data of unknown flight conditions (mh370) and the known data of known real time flight conditions to draw a conclusion. This is obviously not meant to undermine any of the great scientific work that has been done by numerous people on this site regarding the use of BFO’s and ping rings to conclude exactly how Inmarsat came to the southern path conclusion- but if a simple data comparison was done with similar northern and southern bound flights post-loss, than the conclusion of a southern path is able to be reached through simple logic and trend analysis, regardless of the complexity of the technical data.

    Am I wrong on this?


    1. Hi Luke,

      I for one think that what you say makes a lot of sense. So much, in fact, that it is the very reason why I have such a hard time with the BFO data. It should be obvious to everyone that Inmarsat made direct comparisons, and thus that the key feature of the BFO graph is the ‘eye’, the slight downward bend on northern flights and a slight upward bend on southern flights. It should also be completely obvious that there really is no room for a north/south mixup here.

      And now we have simulations that support this story fully, by showing that all possible level paths exhibit the same distinguishing feature. It would say that it is becoming very hard to ignore.


    2. >they obtained the straight data from the signals that they assumed were mh370- Then they looked for the same
      >signals from the same satellite from flights that were operating in real time on both northern trajectories (say KUL to
      >PEK) and flights heading to southern destinations (say KUL to PER) and found that the data from suspected mh370 >mirrored the data set from the south-headed flights

      I was quite interested in the above topic a couple of weeks ago, and doing a web search I found that the reporting in the press was contradictory. Some sources reported this more or less as you state (copied above) and others said that Inmarsat came up with a formula for the doppler and then validated it against doppler data from actual post-mh370 flights headed both north and south.


      P.S. Since the press is notoriously bad in reporting technical matters (unless they are copying verbatim an official press release), I tend to think the latter sounds more plausible, but that’s just my intuition, which may be wrong.

      1. I fully agree that the breakdown here has been in the communication and transparency department.
        You have multiple entities, not responsibly bound to each other, proclaiming and sharing tid-bits of information that the press than regurgitates with its own assumptions- and the end result is a confused state of public affairs.
        However, I do not feel that the order in which the conclusion was derived (determined data and found similar flights to corroborate, or looked at similar flights to find corresponding data) makes any difference to the conclusion of that was reached regarding direction of the aircraft.

        For instance, consider the following thought experiment:

        An observer is in a windowless soundproof shack on the side of a highway that runs east and west. A driver in a car in the highway must drive away from the shack in either direction, however the observer will not know which. The key here is that the terrain is not symmetrical both directions, so to the east lies curves, hills and forest, and to the west is mostly flatland, although there is a thunderstorm brewing. The observer will communicate with the driver in basic conversational speak via radio until the car and driver is to the point to where signal is lost. The driver gives no hints regarding direction taken, and all communications are digitally recorded and analyzed.
        The goal here is for the observer to determine which direction the car went utilizing only the data gathered from the radio recording such as timing of interference, signal strength, decay, and end of all transmissions.

        A competent scientist, preferably one who is familiar with radio waves, topography, signal decay, etc., could likely take the initial recording, analyze all the properties of frequency, signal strength, timing of interference, and rate of decay and compare that with characteristics of the terrain in both directions and come to a likely conclusion of which way the unknown car went. –I believe that this is exactly the predicament that the scientists on this blog are in, having only the sketchy Inmarsat MH370 data points to go on and two possible directions to evaluate.–
        For the thought experiment, however, a more simple layman’s approach would be to subsequently send 5 more cars heading east, and 5 more cars heading west- while maintaining and recording the radio communication with each until it’s lost, and look at the characteristics of the radio data of the ten subsequent recordings. It is inevitable that there will be significant variance of the data trends between the five that headed east, and the five that headed west, due to the terrestrial and atmospheric conditions not being uniform in both directions. The data set from your mystery car WILL MATCH either the east-bound group or the west-bound group, in terms of signal characteristics, sound quality, timing of interference, transmission strength and time until out of range. This means that the layman using 10 additional cars and recordings can reach a definitive conclusion as to which way the mystery car went, regardless of having any specifically advanced knowledge of the properties of radio communications, and frequency wavelengths, and signal decay, etc.
        This, I believe, was the Inmarsat method. Like the above highway conditions, the satellite was not truly geostationary, so it is a fact that the northern and southern paths are not possible to be uniform in characteristics, so this will result in differing data determined by direction. They have, and are still acquiring, the same data, frequency offsets, and hand-shake pings from hundreds of flights moving in all directions in relation to every satellite, and can easily overlay the data characteristics with those gathered from MH370 and see enough of a correlation to make a conclusion.
        Unfortunately, they are the only entity to have this other data to make comparisons to, but I am betting that this simpler method is fairly foolproof in it’s directional determination.

        Also, perhaps this is not giving due credit to the Inmarsat technicians, but if the brilliant scientific minds present here on this forum have spent days and weeks trying to comb out assumptions and determine a definite conclusion of direction based on one set of parameters alone (BOF, LOS angle, ping timing) and have still not reached a certain conclusion- than that is not the method that Inmarsat used.

      2. Inmarsat’s own release on 25th Mar ( ) states that they compared the data logged from MH370’s track with data from SIX other Boeing 777 aircraft flying on the same day in different directions.
        Not same route as MH370, data used for comparison not specified.

  90. Duncan, your blog/forum is inspiring the right people – partly *because* of the chief obfuscating clerks in the Commission of Misinformation acting in accordance with the department of Sovereign Pomposity under the aegis of the Grand Committee of Circumlocution [and perhaps also because of your comment culling - getting easier with time?]. We can only hope that the data from the expensive underwater search extends our knowledge of underwater terrain and deep sea currents one day (if released).

    The BFO model would of course become trivial if Inmarsat were more forthcoming, although there is another organisation that could resolve most issues – Honeywell. They could surely tell us (if the army of worker ants does not dig it up) whether their transmit offset is:
    (1) based on satellite at assumed zero latitude and fixed altitude and longitude;
    (2) applied in full or with a weighting factor;
    (3) applied always or subject to a minimum tolerable threshold;
    (4) derived from a single aircraft sensor measurement of position and velocity, or a time weighted average;
    (5) subject to a deliberate small dither (unlikely);
    (6) subject to defined default mode when sensor data is lost or deemed dubious.

    Another thing that interests me is “what would Feynman have done”. Despite his contributions to computing, I imagine he would have tried to simplify away many of the distracting irrelevant fine detail in order to get a more elegant visualized model of the key processes. I believe Henrik has taken a step on this path by, as I understand him, considering a projection of the problem onto a satellite centered radial system which clearly shows the symmetry of solutions and enables the satellite relevant aspects of a path to be simply profiled on a piece of paper with time on one axis, plotting range and range rate. This is helped by his simplified continuous formula for satellite velocity. Anyway, just looking at the ephemeris data over the presumed time of the flight shows that most of the information could be tossed away and that aircraft range is pretty much insensitive to the satellite variations. The only significant *changing* satellite variable of interest is the z velocity (or equivalently the closely related latitude rate of change) which affects the Doppler and is the symmetry breaking factor.

    Regarding north versus south, I feel a bit sorry for the south paths – nobody is interested in them because they all end in uniformly dull ocean water, whereas the north paths have lots of nearby standard aircraft way-points, exciting unclimbed mountains, remote lakes, mysterious valleys visited rarely by Yak herders, and have strange names and can be enjoyably looked at in Google earth.

    And finally, when the BFO constraint is sorted out, I will be very excited to see what paths emerge, SUBJECT TO WIND, and how well they fit the data, subject to being “simple” and “practically likely” paths as well.

    1. Thanks John.

      I know that many people are following the discussion here and admit little technical background (which is fine: I am hopeless at languages, cannot play any musical instruments, and have yet to read even five per cent of the great novels that I would like to read). For them, as they say, a lot of the engineering discussions etc are bewildering. But your points (1) through (6) above provide a useful summary of the various matters that have made it difficult for some pretty smart people to decipher what the BFO graph might be able to tell us. It’s almost like trying to solve a crossword puzzle without even knowing which squares are black.

  91. Even though the current state of the BFO calculation is still moving, it might be of interest to report on the optimal path along all the feasible ones. At 0:11, that path is placed at (-19.178 103.7), some 300km NW of the recent towing area. The 0:19 point would most likely be closer. The BFO and google map below. Far from perfect, and there is no claim of great accuracy here, but it does represent an optimal solution over the entire 360 degree cicle.



    1. “Even though the current state of the BFO calculation is still moving…” – Please let me just say that other correspondents, not necessarily posting here, seem to have the BFO graph solved. Still a few things to do, and I do not want to discourage others from generating their own solutions (as a check); but what you are seeing here in these comments is not the complete story by any means.

      1. Duncan,

        Thanks for the moderation. The points stated above are my own (and obviously some people over at TMF). I got slightly carried away and did not take the time to make sure all parties were included. Likely because this exercise is proving to be quite exhausting. I wonder how you cope. :-)

        I am certainly acknowledging Mike’s work in particular, and I am having private discussions with him. What other parties that you know about are you referring to?

        My current assement is that the Inmarsat graph is inconsistent, because either the graph represents actual-expected, in which case the northern and southern paths are marked correctly in the diagram and the plane did go south, or it represents expected-actual, in which case the northern and southern paths are marked incorrectly and the plane did in fact go north. What makes it inconsistent is that the BFO report claims that the graph shows expected-actual. What a predicament.


  92. For those interested in the recent developments on the BFO side, here is a compilation which also serves to try to move the conversation over to this post, which does not have 500 posts to wade through. ;-)

    Two weeks ago, richardc10 of TMF laid out an interesting model in this post: The key component is that the correction the aircraft transmitter subtracts from the sending frequency is equal to the doppler from the aircraft to a stationary satellite mean point. This model is very efficient in reducing the doppler as received at the satellite. It does leave the satellite-induce doppler still visible in the signal, which is what makes the difference between the northern and southern routes in the BFO graph from Inmarsat.

    I was struck by two things at this point. One was how well the model reproduces the original BFO. Another was that it resolved an issue I had with the BFO sign. In short, the difference between the northern and southern paths is a fairly small number, and it is very strongly related to that residual satellite motion which, according to richardc10’s model, could actually be present in its entirety in the real BFO data.

    Armed with these observations, I started looking at the doppler contributions again, which led to this post: This shows, together with the model calculations here, that:

    1. T he fixed-satellite model is very likely correct
    2. The BFO should be interpreted as it is (not inverted)
    3. There really is a difference between northern and southern paths, after all.

    On a personal note, I always found it odd that we were not converging to a consesus with the general claim of Inmarsat. I hope this will help that end.

    Regarding the BFO model, there are certainly still kinks to work out, and part of that work is to evaluate plausible origins of an offset that currently appears unexplained. But I think we are much, much closer to the real thing now.

    To further test and experiment with the model, and to see if I could reproduce Victor’s and Mike’s work, I wrote a program to create solutions through the ping rings from all possible angles. The result is found here: This is certainly not conclusive yet, in case it came out that way, but it points at a general working model with that aim. We all want closure.

    Just to point out that some things are still possible, here is a post with a northern path which ends up at Qamdo airport:

    Thanks to Duncan for a great forum, which brings together such a diverse and complete set of expertise. Truly a joy to be part of, in spite of it all.


    1. It’s an curious possibility. The agreement with Inmarsat’s north/south projections is certainly notable. But I see a few problems:

      * The residual in this scenario is basically the projection of satellite velocity on the vector from the satellite to the aircraft (+ a second-order correction because the aircraft assumes the satellite to be up to 1.7 deg. away from where it actually is). It is a slowly varying number that is only sensitive to direction from the satellite to the aircraft, but not to its speed or heading. Therefore, jumps in the real data are hard to explain. I suppose you could argue that there is a time delay in the system and correction at time t is based on known speed and heading at time t-dt. But there’s no reason for a modern GPS system to have a delay that is much larger than a second or two.

      * Satellite velocity is known. Therefore this residual gives us the exact direction from the satellite to the aircraft at all ping times. If ping arcs give us distance and BFOs give us direction, we should know exactly where to look. Judging by the fact that Australians were looking everywhere from 20 S to 45 S and in between, they clearly don’t.

      * At 19:40 UTC, the satellite is at its northernmost point along the figure-8 and I don’t see why richardc10 has significantly different BFOs on north and south tracks at that time.

      I have a couple of others but let’s start with these.

      1. Thanks Eugene. This just to note it’s not a figure-8 in this case, but rather a N-S elongated loop.

      2. Hi Eugene,

        “…Therefore, jumps in the real data are hard to explain.”

        The model is a simplification in that it assumes the two paths to be perfect reflections of each other, which does not necessarily have to be the case, and certainly is not the case at 18:22, when both paths are at the same point on the northern hemisphere. The point is that the further away from the origin the two paths end up, the more similar to a simple reflection they become.

        “If ping arcs give us distance and BFOs give us direction, we should know exactly where to look. Judging by the fact that Australians were looking everywhere from 20 S to 45 S and in between, they clearly don’t.”

        But they *do* know where to look! At first, before the BFO analysis, they only had the ping rings, which gave a best estimate far into the southern part of the ocean, based on assumed straight and level flight path. Several contributors on this blog have come to that conclusion, Victor in particular stated it well in a post long ago. My own calculations say the same thing. Similarly, with the BFO data at hand, there is a particular latitude which best matches both the ping rings and the BFO data, and that puts us either in the current search area or at Qamdo airport.

        “At 19:40 UTC, the satellite is at its northernmost point along the figure-8 and I don’t see why richardc10 has significantly different BFOs on north and south tracks at that time.”

        At 19:40, the two path are still quite far from the mirrorpath ideal, so there is a bias left which explains this. The Inmarsat graph shows the same thing.

        Thanks for your questions!


      3. I have a stupid question arising from ignorance.

        Suppose some of the BFO pings were recorded by the neighbor satellite, the one above the Pacific Ocean.
        Shouldn’t we reverse the BFO of these pings relative to the 3F1 pings from geometrical considerations?
        The idea is that moving towards one satellite is necessarily moving away from the other one.

        The ping ring radii were recorded by one satellite but maybe the BFO uses two of them?

  93. Hi Duncan,

    In a comment on your previous post I stated that a Northern Route is more likely with intentional pilot input and a Southern Route is more likely without intentional pilot input.

    Please see:

    In a separate comment on your previous post Henrik stated “What I would like to be able to say with certainty is: “if the plane flew by itself, it had to go south”.

    Please see:

    Both Henrik and I are saying North = Pilot, South = no Pilot.

    However, Henrik and I made these statements independently and for different reasons. Henrik was studying the Doppler Shift for possible Northern and Southern routes together with limitations his analysis imposes on a Northern route, and, I was studying the latest Ping Ring information, especially the 3 Ping Rings around 18:25:24, 18:26:54 and 18:27:48 and the Malaysian Military Radar end point at 18:22:38.

    Same outcome, different process.


    1. Richard:

      If you and Henrik are correct — “Northern Route is more likely with intentional pilot input and a Southern Route is more likely without intentional pilot input.” — you may have also (inadvertently) identified the motivation for the authorities deciding to double-down on the search in the south. And from this observer’s seat, it’s probably not based on the math.

  94. Overnight I received another reply to my email to AAIB. I indicated that there were very skilled people contributing here who would be willing to help with the analysis.

    Here is the reply – –

    “Thank you for your information regarding flight MH370, which has been forwarded on to the investigation team for assessment. Due to the number of possible wreckage sightings and other enquiries or suggestions by the general public, the team cannot undertake to reply individually to emails or to provide feedback on the information provided. However, your contribution is very much appreciated.”

    At least it is not a total brush-off.

  95. I’m still following this blog and want to express my thanks for the great work. I posted a link to this site in a discussion forum on an aviation site, but my post got also deleted. Keep up the good work.


    1. Would have MH370 been able to fly from waypoint BIDOT to the point at 18:29 UTC, taking into account the plane flew across Malaysia and made a turn west? The maximum speed of the aircraft, according to the internet, is Mach 0.89 (590 mph, 950 km/h, 512 knots) at a cruise altitude of 35,000 ft.

    2. Is a sharp turn at 18:29 UTC either south or north at maximum speed technically possible?

    3. I’m wondering if the radar blip at 18:29 UTC was MH370. The flightaware tracking protocol for MAS30, heading for Istanbul on March 8, tracked the aircraft’s position at 17:33 UTC: 5.1360, 98.7958. The estimated position at 18:29 UTC was 9.4466, 93.8526. MAS30 was in the area also.

    4. Chris McLaughlin from Inmarsat said the plane went back west over the Andaman Islands and then went south: If the plane turned over the Andaman Islands, why is the starting point of the paths in the Malacca Street? Why not shift the starting point and analyze the results? (A scenario for a turn over the Andaman Islands: Person A deviated the plane. Person B – perhaps not an expert pilot or not a pilot at all – seized control of the plane over the Andaman Islands in order to fly it back to KL and failed.)

    5. I found this paper, but I’m not sure it’s helpful: The Effect of Tropospheric Propagation Delay Errors in Airborne GPS Precision Positioning:

    1. Propagation delay due to tropospheric water vapor and liquid water is very familiar territory for me, having been intimately involved in GPS/MET, COSMIC and Radiometrics for more than 2 decades. These delays are significant for precise navigation, such as for approach to landing, but negligible for our 370 purposes.

      Side note: The tech company I remain associated with, Radiometrics, was founded in 1987 specifically to develop a microwave radiometer capable of directly measuring the integrated water vapor for the purpose of improving GPS position determination for geodesy. The company succeeded in reducing the position error to <1 cm over a 50 km baseline using radiometer corrections. Today, the inverse problem is solved and exploited routinely. Systems such as the deployed all over the world use precisely known locations of GPS receivers to measure the water vapor in the atmosphere. By 1995, the precise measurement of atmospheric temperate and water vapor took another major step forward with the launch of the MicroLab-1 satellite, carrying the first GPS radio occultation experiment. I was fortunate to have been there on the front lines.


  96. Regarding: Patent Number: 6,008,758

    In search of additional information about the upper bound on “normal BFO values”, I found the Inmarsat “…frequency tolerance…” defined in the above referenced patent. The patent states in part:

    “To solve this problem, Inmarsat requires that each AES 12 transmit all of its signals to satellite 16 within a defined frequency tolerance (e.g., 100 or 185 HZ).”

    The patent does not elaborate on the conditions for the application of the 100 Hz limit vs. 185 Hz limit. It may be the case that the 100 Hz limit is applied to narrow band channels, such as Inmarsat C, while the more relaxed limit of 185 Hz is applied to wider bandwidth channels, like Inmarsat A and packet data channels. An alternative reason for the two limits might be that older equipment is grandfathered with a 185 Hz limit, and newer equipment (with improved reference oscillator technology) is required to operate within the 100 Hz limit.

    During the period prior to 17:07 UTC, the AES met the tolerance requirement with a comfortable margin assuming the 185 Hz limit, and also met the requirement assuming the 100 Hz limit. But the limit was exceeded during the last three transmissions assuming the 185 Hz limit, and for every transmission after 17:07 assuming the 100 Hz limit. For the transmission at 00:11, when D1+D2 = -353 Hz, the limit was exceeded by almost a factor of 2 for the 185 Hz limit. If the 100 Hz limit applied, then it exceeded the limit by a factor >3.5.

    Of course, this does not prove that the offset was zero after 17:07, but it certainly proves (1) the offset algorithm failed to maintain the carrier frequency within either limit, and (2) the offset must have been at least either 253 or 168 Hz less than it should have been, depending on the applicable limit. Given the actual location, speed and direction of the aircraft for any of the possible areas, it is much more likely that the offset applied after 17:07 was <50 Hz to have produce the observed BFO values. Based on the evidence we have so far, I continue to believe an offset of 0 Hz after 17:07 is the most likely value.

    1. airlandseaman,

      If the C-band doppler is present in the BFO signal, then what about the C-band down converter error? As I understand it, both the up and down converters of the satellite exhibit substantial errors. On the EAFC uplink, these errors are monitored over a period over 24 hours to extract both the mean error (converter error) and the periodic error (doppler shifts). Such an elaborate scheme would not be used unless it mattered.

      According to chapter six of B.R. Elbert’s book “Introduction to Satellite Communication”, the satellite LO is commonly exhibits errors in the 0.1 ppm range, which corrsponds to several hundred hertz.

      The more I read about this, the less I would expect perfect tuning at the point where the BFO is measured. After the steps in the downlink, i.e. after the doppler and converter errors have been masked away, the situation will probably be better. But then we would no see any C-band doppler either.


      1. The inbound C band Doppler is definitely included in the BFO values. It’s the one “D value” that Inmarsat defines unambiguously (D3), and the 16:30 calibration point validates this interpretation within a few Hz.

        The inbound Transponder LO offset is corrected independently at each LES (each LES has a different error) using the L to C band Pilot tracking system. The Inbound Transponder LO error is subtracted at the LES *before the BFO value is sampled*. This is required because the Transponder offset is common to all Transponder relayed signals, but the L band Doppler is different for each. Moreover, this error is much larger than the BFO values.

        The Transponder LO offset probably is on the order of ±10^-7, and varies with sun angle, eclipse, etc., but it is corrected to the 1-5 Hz level by the pilot tracking system. (This is the temperature dependence correction recently discussed in the news.) BTW…this is not an elaborate system. It is really SOP used in virtually all satellite communications systems. BTW…I designed the pilot tracking system for the GOES DCPRS system back in the 70’S, and manufactured around 50 Direct Readout Ground Stations for GOES satellites. So I have some direct experience. Still have a pilot receiver in my basement “museum”.

        I feel confident that the sum of all the errors not represented in D1+D2+D3 are very small…maybe <5 Hz. In the GOES DCPRS system, all the carriers were on narrow band FDMA channels spaced 1.5 kHz and the data rate was only 100 b/s. Thus, precise satellite offset correction was essential. The Pilot receiver swept a ±5000 Hz range, locked on to the pilot and produced an IF output with less than 5 Hz residual error. This was before the first PC was manufactured, so expect much better today!

    2. Hi alsm,

      Your quote from the patent says “e.g., 100 or 185 HZ”, not “i.e., 100 or 185 HZ”. Therefore, you should not read too much into those particular numbers. BTW, I think I read somewhere that the target tolerance is 333Hz, but do not know if that is confirmed information.


    3. airlandseaman,

      “But the limit was exceeded during the last three transmissions assuming the 185 Hz limit, and for every transmission after 17:07 assuming the 100 Hz limit. ”

      Looking at the BFO, is it not the slightest odd that both predicted paths would also exceed those limits? Not if the predictions include the same assumed correction failure, I presume would be the logical answer. But if there was a correlation between this BFO and other BFOs for regular flights -flights which evidently did not have this correction failure – what could that correlation possibly be? Before 17:07, the two paths are the same, so there is nothing to correlate to.

      I am sorry, but this does not make the slightest sense to me. I think the explanation is really very simple: the BFO graph really shows actual – expected doppler. Then, by extracting the C-band portion, most likely with both a constant and the known periodic part, the resulting curves are almost flat, and all lie well within the allowed frequency budget. Sometimes the easy answers are also the right ones.


    4. Here are some Doppler compensation patents:

      # Motorola (formerly)
      Robert Thomas Frederick, Nathan West Miller, Daniel Richard Tayloe
      Method and apparatus for erratic doppler frequency shift compensation – US 5703595 A

      # Honeywell but not in use?
      Neal R FEDORA
      A method and system to correct for Doppler shift in moving nodes of a wireless network – EP 1835637 A2

      # Hughes Electronics Corporation
      Qinchong Liu
      Compensation of dynamic doppler frequency of large range in satellite
      communication systems – US 6058306 A

      # Rockwell Collins
      Thomas B. Campbell
      Method and apparatus for Doppler compensation in a satellite communications system – US 6008758 A

  97. Duncan,

    Two decades and a half ago when I was a subject matter desk officer in a Department of Defense agency, I and my colleagues were often tasked to answer letters from folks who had sent us suggestions on how to improve upon our work. We referred humorously to our responses as “Thank You For Your Interest in National Defense” letters. Suggestions ranged from reintroducing swordsmanship as a required skill for gentlemen officers to reading tea leaves as an intelligence skill. Although preparing these answers was a huge distraction from the focus of our day-to-day work. we always tried to be both polite and informative in our answers. In those days before email and before the internet the volume was manageable as each maintained a file full of “stock” answers to be used as a template for our reply. Today methinks many agencies use “automated responses” sent much like “out of the office” replies to email.

    From your description, the response you received from the JACC looks to me like an “automated response.” Such an automated response does not mean that at least some folks within the JACC are unaware of your work. In fact I would be surprised if they were not closely following what you and your talented associates that regularly blog on your site have been doing. And they likely welcome all the help they can get to point them in the right direction in the search for MH 370.

    However, your Circumlocution Office analogy does indeed have great merit. For example among various staff elements in any agency the “not invented here” syndrome still exists. And the 9/11 Commission Report was a case study in the detrimental effect of “stovepipes” of information, where one government agency did not share with another. And the recent memoir by former Secretary of Defense Robert Gates clearly lays out the detrimental effects of politics and bureaucracy. All of these are at work in influencing the ongoing search for the final resting place of MH 370.

    Given enough facts that point them in the right direction the search team may eventually find MH 370. However the search is not only about the “science” but also about the “art” of informed intuition applied to determination of data and the employment of available technology to the search by a mind like that of Sherlock Holmes.

    I would leave you and your talented associates who bring the skills of many disciplines to the search for MH 370 with this thought: ” Never, never, in nothing great
    or small, large or petty, never give in except to convictions of honour and good sense..” [This remark was made by Winston Churchill in a speech made 29 October 1941 to the boys at Harrow School.The full speech is contained in "The Unrelenting Struggle" (London:Cassell and Boston:Little Brown 1942, and is found on pages 274-76 of the English edition). It may also be found in "The Complete Speeches of Winston S. Churchill," edited by Robert Rhodes James (NY:Bowker and London:Chelsea House 1974)].

    Duncan, keep up the good work! And to all those others who have provided supporting comments towards your continuing efforts, a well deserved and hearty: Hear! Hear!

    1. Thanks for that Phil, appreciated. Lots a good sense and philosophy.
      And, yes, I agree with your observation about the ‘automated response’. One hopes that the people directly involved have paid attention, and indeed have good reasons for continuing to search only in the Indian Ocean.

    2. I agree with Phil Webb saying:

      “In fact I would be surprised if they were not closely following what you and your talented associates that regularly blog on your site have been doing. And they likely welcome all the help they can get to point them in the right direction in the search for MH 370.”

      Based on experience in another “open source” investigation the log files of the “” web server may show accesses from governmental and international agencies maybe national laboratories. People recognize the power of open source methodology to solve difficult technical problems.

  98. ****It would appear that the “audit” is being performed by the original perpetrators. This is not an audit! An audit is performed by an independent body. I know who I can recommend for the audit: the author and commentators of this illustrious web-site.

    Indeed!! We have learned to trust your findings, so grateful for these analysis, from being one who believed she did not have a scientific brain, I feel trained by your imparting of truth. By the way I love the Plum in Cuba street also.

    1. Thanks Catie. I’ve learned a lot too over the past six weeks, but my appreciation of that is tempered by the knowledge of the angst felt by too many over this awful event.
      Let’s get coffee at Plum when we are both thereabouts (I am not back until August). Being a Celtic Kiwi you will know that it’s close to an Irish pub.

  99. For any readers in the UK, following up on a earlier comment, please consider submitting a FOI request to ensure that AAIB comprehend that public concern exists. Use this link as your template for the request: , as described in a comment to Duncan’s previous blog post ( )

    MPs are obligated to serve their constituents: contact yours. While no UK citizen was onboard MH370 many of us, our family, friends and colleagues travel frequently for business and pleasure. A successful outcome for this investigation is critical for our continued high confidence in air transport safety.

    I expect my MP to act on my communication, I expect some leverage resulting from a direct link between his constituncy and the AF447 crash: a local family was eventually able to grieve for their loved one. The UK’s representatives serving on the MoTM/DCA ‘International Investigation Team’ have a responsibility to progress their work openly and expediently for all of us. That ‘leverage’ for continued confidence is key for us all, simple message, please us it.

    1. Well said Don. I encourage all readers in nations other than the UK to do likewise, using whatever powers they may have under their own local laws and political structures.

    2. I suspect that a combination of political pressure (e.g., questions from MPs) and public pressure (e.g., adverse press coverage) might influence Inmarsat to change its apparent position on non-release of raw data. Perhaps you could urge your MP to ask about release of the raw satellite data during departmental Question Time or Prime Minister’s Question Time?


      P.S. I am not a UK citizen, so I don’t have an MP to ask.

  100. Warren,

    I am replying to you here, rather than on Duncan’s previous post, so as to ensure that this comes to your attention.

    The three-ping cluster could be indicative of a change in altitude, a drastic change in direction or a mechanical problem, as so described in various contexts by Duncan, Dr. Kuang and Richard. Indeed, the central purpose of my queries is to explore whether there is, in fact, enough evidence to support a post-diversion ‘event 2′ that compromised the aircraft, or if there is no indication of such an event.

    If we assume that the aircraft was intentionally diverted at IGARI for something other than a mechanical failure, and we assume that the aircraft at one point adopted a terminal flight trajectory taking it either north or south, then an event or events or a process reconciling these two patterns of behavior (1. human directed flight; and 2. non-human directed flight) would be indicated. An intentional diversion together with a spontaneous ‘unintentional’ terminal flight trajectory is difficult to otherwise reconcile.

    In sum, I would like to test whether there is enough evidence to support a three-phased flight trajectory where: Phase I, the aircraft was intentionally diverted to an alternative destination at or around IGARI for reasons other a mechanical failure; Phase II there was some sort of intervention or other causal factor where the intentional diversion was impeded/aborted; and Phase III where the aircraft engaged a terminal, unintentional flight trajectory. Again, a reconciliation of an intentional diversion with a terminal flight trajectory would seem to be indicated.

    Thus, I have been seeking 1. evidentiary support for Malaysia as the intended destination as it is simplest solution (assumption: all aircraft have an intended destination); and 2. evidentiary support for behavioral changes in the aircraft that could be indicative of distress or a change in operational circumstances somewhere along the known flight path post-diversion at IGARI.

    I may have missed something or the hypothesis could simply be invalid, I don’t perceive any evidence that would impeach or discount or provide evidence to the contrary regarding the two event, three-phased flight trajectory that I described above. If the aircraft was diverted due to a mechanical failure at IGARI and the evidence indicates as much, then this hypothesis would immediately test as invalid; likewise in the event of a pilot suicide. Otherwise, it seems that it is rather supported.

    I would venture that your exploration of the early period of the flight from IGARI to c. 18:28 UTC is warranted, as, barring witnesses or any evidence found in the flight recorders appearing, the only people that can better inform the flight trajectory are the Malaysians, and then while it was either a. in their airspace; or b. detected by their primary radar. Such information, were it to exist or be made public, could perhaps in turn better inform the search for the aircraft.

    I am not implying that anything nefarious necessarily had to occur. Rather, I am simply stating that their is a potential source of additional information on the flight, and thus it behooves us to examine with a high level of scrutiny the course of travel, the timeline and the official information concerning the period that the aircraft was over Malaysia – the only confirmed area with potential human witnesses. In my eyes, this is forensics 101.

    As I have stated elsewhere previously, it’s either this dead animal of incompetence over here, or that nasty animal of obfuscation over there; take your pick. Regardless, the opacity regarding what is known about the flight over Malaysia is quite palpable, and thus an examination of the characteristics of the flight in the vicinity is warranted, as you have so been pursuing.

    1. Rand:

      You’re right on the money.

      1. “their is a potential source of additional information on the flight, and thus it behooves us to examine with a high level of scrutiny the course of travel, the timeline and the official information concerning the period that the aircraft was over Malaysia – the only confirmed area with potential human witnesses. In my eyes, this is forensics 101.”

      Eyewitnesses matter. And with respect to Malaysia, there may be up to 11 of them (in or near Kota Baru, Keletan) who saw the ‘bright lights’ (think landing lights) of a low-flying airplane that appeared to be descending.

      Two pieces:

      I’d submit that a close look this information, together with any additional radar or other information that might be obtained from the Malaysians, could prove to be illuminating.

      2. “I have been seeking 1. evidentiary support for Malaysia as the intended destination as it is simplest solution (assumption: all aircraft have an intended destination)”

      IF Malaysia was the intended destination and IF MH370 flew back (west) over peninsular Malaysia, then we should ask why there appears to have been no attempt made to land in Langkawi.

  101. A few comments here:

    * The ATSB seems to be particularly understaffed to perform a task like this. The organisation has approximately (if I remember correctly) about 100 staff, of which about 60 are investigators of various types. This is to cover maritime, air and land based transport safety, including rail. Unfortunately, whilst technically competent and knowledgeable, they are from all appearances under-resourced. They are sending staff to aviation incidents less often, and relying on others to gather information for their investigations. It is likely that these lack of resources contributed to some of the issues related to the investigation of the Pel-Air crash, where thankfully everyone survived. There has since been a parliamentary inquiry into the report, which is rather unique with both sides of politics being scathing of the performance of the organisation. Unfortunately, despite their performance, I still feel that they are the ‘best of a bad bunch’.

    * With regard to the continued southern search, a possible explanation is that there is additional evidence, likely from some type of radar analysis, that further suggests travel south, that countries do not want to divulge.

    * As to what type of radar, there are numerous options. The most obvious is conventional military radar. Having said that, it is possible that some type of Passive Radar is being used. Passive Radar is when reflections of signals from transmitters such as FM Radio or TV are used to track aircraft. There is even a group of GNUradio experimenters working with this, so it is not outside the realms of possibility that one of the countries in the region has the capability


  102. Duncan, beautifully said. Anytime a space scientIst can invoke Dickens its a boon. Your choice of Little Dorrit’s, “The Circumlocution Office,” is so apt, so funny awful in its description and implicit reveal/rebuke of the strange phase(s) the official investigation has taken. Am agog at The Guardian and the Australian based Joint Agency Coordination Centre (JACC) re their non-response to your letters. Perhaps some of us who write for a living can shake this up. Meanwhile reading the work of those on this site is a tremendous thing. You all amaze me. More importantly, your combined efforts may solve what needs to be known: where the plane is and what happened. For the families, for aviation and for all who fly you are doing a great service. The world has been gripped by this story because it is important, and has too many loose ends that have import beyond this tragedy. Thank you, and all the contributors here.

  103. It seems to me that asking JACC to consider the possibility of a northbound track is an exercise in futility, and for a very obvious reason. It’s sort of like calling Sydney police and saying “hey, my Malaysian friend got mugged during a trip to Russia, would you mind helping him recover his property?” They won’t do that, it’s not their jurisdiction and it’s none of their business.

    If MH370 crashed south of 12th southern latitude and east of 75th eastern longitude, the incident took place within Australian jurisdiction and they are empowered to investigate. If it went north, since it’s not their aircraft and it’s not their waters, they can’t and won’t look there. The only ones who can are governments of northern corridor countries (Kazakhstan, Kyrgyzstan, etc) And the only ones who would ask them to look are the Malaysians. (The Chinese might listen to reason, but they would be unlikely to go over Malaysians’ heads in organizing the search of the northern corridor.)

    In addition to that, even a public acknowledgement of the possibility of a north track would be a huge egg in the face of JACC, since the next logical question would be, “why didn’t you look into this before sinking $40 million of taxpayer money in the Indian Ocean?”, and quite possibly some heads would roll. Their best hope right now is to double down and to keep looking in the same place, hoping either that they’ll find it somehow, or that no one else will find it elsewhere.

    1. Thanks Eugene. I largely agree with your sentiments (depressing, is it not?) but differ in some ways. The JACC is not in charge of the investigation: the MoTM is (so far as I am aware). It is just that this meeting is being held in Canberra. One might argue that we are giving the JACC (and indeed the Australian Government) access to an escape clause: Australia has led a costly search in the Indian Ocean that might well be based on false information, because Inmarsat may have made a mistake. Of course that would be an indictment of Australian Government (they spent that much without bothering to check?), but it could be less damaging to them than the alternative.

      Of course there is another reason for me to have emailed the JACC: this means that they cannot claim later not to have known about the many useful analyses and ideas people have made on this website. Much earlier on I sent an email message direct to the Vice Chief of the Australian Defence Force, Air Marshal Mark Binskin. I have also had other people with Defence contacts in Australia send messages to AMSA, JACC, and various others.

      I might also note that I have been in direct contact with a long-term antagonist of the right-wing Australian Federal Government, Phillip Adams, who all Australians will know (and I have appeared on his radio show Late Night Live several times, plus we were both on the Advisory Committee of the Adelaide Festival of Ideas for some years). If anyone would want to make Tony Abbott look like a buffoon, I would expect it to be PA. I asked him to alert the ABC (for which he works) to the alternatives in terms of analysing the satellite information. Nothing has happened. As I have written previously this whole sorry affair, quite apart from being a terrible human tragedy, seems to represent a massive failure by the Fourth Estate. This is not good for the future of us all. It would be far, far better if all involved – from Malaysia Airlines to MoTM to Inmarsat to the AAIB to the NTSB to the ATSB to the Australian PM – were being metaphorically poked with pointed sticks by good journalists; and this is not happening.

      1. ‘good journalists’
        Sadly, lacking in today’s media. The Reithian principles of educate, entertain, inform have been distorted beyond recognition. I’d like to recommend the documentary, McCullin, about Don McCullin’s career as a photo-journalist. A very relevant observation is made by Harold Evans, the editor at the pre-Murdoch era Times (London), about the change of editorial direction demanded by the new Australian owner that rendered Don’s services redundant at the paper. Go watch.

    2. Eugene:

      You cite a very important truth – jurisdiction. And let’s add to that geopolitics, which are never not a factor. So what is needed (to the degree that anyone here is inclined to seriously entertain it), in concert with the very laudable work being done by Duncan and all of the contributors here, is a creative and executable strategy to end-run the Circumlocution Office. Difficult? Extremely. Impossible? Not necessarily. But it will entail some thought about the power dynamic (cultural, economic and political ) in that part of the world — and who would have the motivation (and the means) to exert their leverage. No secret: China has been apoplectic about how the Malaysians have handled MH370 since day one. And while for investigatory protocol (and facing-saving) reasons it is almost assuredly the case that they would not go over the Malaysian’s heads publicly, do not ever discount what they’d be open to do in the back channel. Remember, the Chinese have built oil and gas pipelines in Central Asia. Moreover, and as controversial as this may be (and it is, if for no other reason than the current and bloody events in Ukraine), there is another player that has historically had strong ties with countries in the northern corridor, but now finds itself in a battle with China for influence there – and that is Russia. Now are these the only two countries that would arguably have an incentive to engage in heavy lifting on behalf of a crowdsourced effort to find a missing airplane? Not at all. But to the degree that those here believe (with increasingly good reason, it appears) that MH370 may have crashed or landed along the northern corridor — and are prepared to explore alternative pathways to determine if that in fact is the case — I’d submit that in one way or another, you would need the quiet assistance of one or both of the aforementioned players to advance on the board. Thinking of the 239 souls on that airplane and their families, who have been put through unmitigated hell, the question now, is cui bono. WHO WOULD BENEFIT by providing assistance to those who believe MH370 is somewhere in the northern corridor?

  104. Duncan,

    Cheryl #1 here. Your title to this post is quite clever indeed.

    If it is of any consolation to you, this very morning at 6:30 A.M. New York time, David Soucie, former FAA Inspector and author of “Why Planes Crash” appeared on CNN and stated that he now believes that Inmarsat should release the data so other mathematicians and scientists can review it. Later, he along with another aviation analyst concurred that it could be done is such a way so as not to compromise Inmarsat’s interest or stake in it. But they both agreed that “fresh eyes” (their words) on the data was a good idea. So if nothing else, at least someone is saying this on a global stage.

    1. Hi Original Cheryl #1 ~

      Well, well, well…Mr. Soucie has seen the light! I was shocked when David stated on the world stage that knowing the raw satellite data was “not that important”. I felt that he had betrayed all the people who have put their minds and hearts into helping to find the aircraft. I couldn’t believe he was on the wrong side. So, I’m pleased he’s back on our side. We need advocates for speaking truth to power.

      Yes Cheryl #1, the global stage is huge but it’s also a “small world after all”. They hear us begging for scraps of raw satellite meat and north/south predicted path tidbits. Why else would various media outlets suppress Duncan’s and others’ comments? This is why today’s new topic is so apropos.


  105. Besides the JACC, I believe there is another material source of potential information who should receive an inquiry on behalf of the blog. That is the Curtin University Centre for Marine Science and Technology, to determine whether there is indeed a basis they can confirm for the reports of their role in detecting a transient boom at the time the plane reached the last, half ping arc, which reportedly sent Ocean Shield to its underwater search site.

    Michael Molinaro

      1. There are numerous TV and stage shows which illustrate the Circumlocution Office very well, and it translates into British humour very well. “Yes Minister” springs to mind as one example. I’m not sure if there is a US equivalent.

        It’s time to follow up on my email to the AAIB too. Of course I haven’t received anything since the original acknowledgement.

      2. I have not yet tried a “cold” communication to the CMST, but instead, I wrote to Rodney Thompson, who was one of two software engineers /coworkers who made posts regarding the role of the Curtin U’s CMST. Here is his recent reply to me, relayed on to Duncan Steel with his permission:

        “Hi Michael,

        The suggestion that CMST had detected a possible transient (and associated bearing) from the CTBT Cape Leeuwin station data came from a colleague of mine who mentioned in on return from a meeting (unrelated to MH370) with CMST research fellows.

        They mentioned that CMST had presented to AMSA / ATSB some information regarding a possible transient detection as observed on the Cape Leeuwin CTBT acosutic recordings and on a CMST logger. A day after this presentation the Chinese research vessel Hai Xun 01 began acoustic search in the region to the west of Exmouth (north west of Perth) that would commonly be referred to as the current acoustic search area (80*700km). A day after that Ocean Shield arrived to the north of Hai Xun 01’s position and began its acoustic search with the TPL-25 towed pinger locator.

        It appears this was mostly a coincidence. JACC had revised the search location to the north base on updated Inmarsat analysis and modelling of aircraft performance characteristics to further refine a region on the final ping handshake for the acoustic search.

        I contacted a friend at CMST (the researcher who looked at the CTBT data) to ask them about the supposed transient. They did detect a very low SNR transient on the Cape Leewuin data and it provide and on an acoustic logger deployed off Rottnest Island. They were able to obtain a bearing from the Cape Leeuwin data and used the logger to try and narrow down a region along the bearing, but due to the receiver geometry the area of uncertainty is hundreds of kilometers long. Interestingly the bearing however does not intersect with the location that Ocean Shield detected the suspected pinger signals.

        They attempted to use the Diego Garcia CTBT acoustic array to refine the position but the data there was contaminated significantly by seismic survey to the north west of Australia.

        So in the end it turns out the CTBT data was a bit of a dead end, and not as exciting as I first thought, probably not worth pursuing. Thanks for contacting me first however. I don’t think my employment would be at risk if you had attempted to contact them, but I wouldn’t want CMST to be harassed for information that they may or may not be able to release.

        A direct quote from my friend at CMST:
        “So all we have at the moment is a single line of position to a single acoustic event that may or may not have anything to do with MH370! The line of position passes a long way (many hundreds of km) south of the area they are currently searching with the AUV”



        Rodney Thomson’s posts, and those of those commenting on his submissions at and at are fascinating to scientifically untrained, mere attorney mind. Thank you to Rodney for his blogging efforts, and thank you to Duncan, and all who have contributed on Duncan’s site.

        Michael Molinaro

      3. Thanks Michael. I might note that contained in that commentary from Rodney is the first useful evidence regarding possible acoustic detections that I have seen.

      4. Hi Michael,

        With this line of communication, would it be possible to find out the timestamp of the Cape Leewuin detection?

        From this, we may determine if the 0:19 UTC partial ping corresponds to MH370 stalling/crashing (as the investigators believe), or running out of fuel (and the crash occurring some suitable time later).


    1. Thanks very much, walker599, much appreciated. I just sent an e-mail to mandala499 and invited him to respond to Duncan’s analysis, as well as inquiring why Inmarsat changed its conclusion about its ability to determine MH370’s direction . Hopefully, he may respond here, but if he chooses to respond via e-mail to me, I will relay his opinions.

      1. I just heard back from mandala499. Unfortunately, he is not an Inmarsat insider, but rather works for [an aviation company - redact by DS] in Indonesia. For what it’s worth, he has the following observations: “Duncan Steel’s analysis is sound. I may not agree with it 100% but it does help in bringing sense to a lot of things. The latest information by Inmarsat merely releases the elevation angle data in addition to the pings…”

    2. In a follow up e-mail, I asked Rodney Thomson the following:

      “I wondered, ever since I heard of the CNTBT acoustic station on the Cocos Islands, whether they “heard/recorded” MH370 fly past (it seems to have passed within the 100 km range of West Island abilities to detect a normal flight’s sounds), and whether they (Cocos) also detected the transient “boom” you wrote about that were detected at Cape Leeuwin . Wondered if you by chance have heard anything in this regard.” (I also encouraged him to post his own comments in this blog.) This was his response:

      “The posts over at Duncan Steel seem to be mostly focused on analysis of the Inmarsat AES handshake pings. Whilst I understand the concepts that are being used to measure range and Doppler (It is similar to techniques I use in the acoustic space) I think there is already much more better-informed discussion going on there in the comments. If I were to add my own thoughts I’d likely just reduce the signal to noise ratio there!

      The Cocos Islands ‘acoustic stations’ is actually an airborne acoustic infrasonic sensor. As you mention they typically can detect low frequency noise of aircraft at a moderate range. However it is highly susceptible to local noise conditions (ie wind) and apparently on the evening of the 8th of March it was particularly noisy limiting the effectiveness of the sensor.

      The infrasonic sensor would not detect an impact transient through the air. There is simply not enough energy. One would expect that if a plane crashed fairly close, and still had a lot of fuel to generate an explosion, then this MAY be detected.

      CTBT report on infrasonic detection of MH370:

      I then inquired as to whether the article he attached (an interpretation of an underlying March 11 report posted on this Duncan Steel blog long ago) seems to be potentially limited to commenting on search analysis done analyzing a time that was “too early” and done in the wrong direction, despite the subsequent interpretation being published on April 11. I also asked if he had any inside knowledge regarding Cocos Islands detections. Mr Thomson’s responding comments were:

      “That link I sent you seemed to be an extension of the original CTBT report from the 11th of April in which they mentioned:
      “The Malaysia–Vietnam region is also exhibiting rather fair to poor network performances, for this specific period, due to the distance to the closest station (IS06 at over 1600km) and the relatively high level of noise at this station (typical for station on islands).”

      That report also appeared to only look at the CTBT automated software detections. I’m sure however that after the additional Inmarsat analysis was release suggesting a possible southern transit (and hence possible transit near Cocos Is) that they would have revisited the data for manual inspection.”

      Mr. Thomson thus did not purport to have any first or second hand “inside” knowledge of any acoustic findings from the Cocos Islands, despite his (and his coworker metao’s) reddit posts regarding the CMST role in reacting to a transient boom just before Ocean Shield began to search with the TPL.

      I still haven’t summoned up whatever it takes to communicate directly to the CMST or the CTBTO. Dickens fan that I am, I don’t relish getting the form letter response.

      Michael Molinaro

      1. As of yet there has been no acceptable (to me) evidence of any acoustic detection of any form: infrasound (either from the aircraft in flight, or a shock from a crash); hydroacoustic (e.g. at Cape Leuwin); seismometers (from an impact shock); sonic pings at any frequency from a beacon on the flight recorder assumed to be submerged in the Indian Ocean; SONAR (i.e. an active system on a submersible looking for echoes from a target, just as a RADAR system operates in air or vacuum or dielectric); or any other sound system (although presumably a few people heard the aircraft as it crossed the Malay Peninsula travelling west, and *perhaps* people elsewhere heard it too).

        Thus there is no phenomenon for me to study, at this time. Again, I recognise that others find it infuriating that people who have spent decades using a scientific mindset are hard to convince about various things, but some solid evidence is needed before I will post anything else regarding acoustic interpretations and analysis.

        Sometimes we are pushed to consider unlikely possibilities: Charles Babbage once made the assumption, in investigating the likelihood of biblical miracles having actually occurred, that the chance of rising from the dead is one in 10^12. This has been compared, in its remarkableness, with Arthur Stanley Eddington starting an examination question by asking students to consider “a perfectly spherical elephant, whose mass may be neglected.” Of course no elephant can truly be spherical, because that would require smoothness, and we all know why they are termed pachyderms.

        No more on acoustics, please. I may go back and delete all previous comments on that subject because, to date, the only acoustic signal has been noise.

  106. I think you can be pretty darned certain that under-officers of the CO have read your website and the extraordinary set of analysis by multiple contributors. After all-the CO needs to know what to cirumlocute about. For what it is worth this is one “scientist” who knows enough to know that he can not contribute anything of real use to the discussion but who knows enough to know the value of the broad based interaction between individuals expert in their many fields.

    1. Are there any readers here from Inmarsat or who recently worked for Inmarsat? If so, I recognize that you have your job to worry about and therefore likely do not want to make your identity known, but many of us here are interested in the process followed by the company in transitioning from its initial position that the direction of MH370 could not be determined from the satellite data to an assertion that the plane flew south toward the Indian Ocean. Any insights could prove quite valuable in helping to solve the mystery. If you do not want to post your thoughts anonymously, you could e-mail them to Duncan. Thanks.

      1. i have been following another (massive) blog on, current one is part #60 here:

        the man you are looking for calls himself mandala499. he is an inmarsat insider.
        in reply 87 (link above) he answers on somebody saying:
        “I will still entertain the ideas that the data could be wrong anyway or that the pings are just being used to make whoever is involved think they have gotten away because they think that the searchers are looking in the wrong area… Anybody care to rule both of those options out with 100% certainty?”
        he replies:
        “I can… The pings result is what they can release to us. What they cannot release to us, is interesting. Those who are familiar with Inmarsat network control center would know what I mean.”

        maybe you want to ask him something here:

  107. Hi Duncan,

    From the (automated) response that you received to your email from JACC:

    “in particular, the satellite communication subgroup, who will be visiting the ATSB this week to continue the review of information that will assist in progressing future underwater search planning”

    It would appear that the “audit” is being performed by the original perpetrators. This is not an audit! An audit is performed by an independent body. I know who I can recommend for the audit: the author and commentators of this illustrious web-site.


  108. In yesterday’s JACC press conference, during the Q&A period, Warren Truss said of the ongoing search effort:

    “Well, it will be something of an audit of the information that has been collected since-the-beginning of the search. It’ll also look again at the satellite information that has been accumulated so that we can make sure that it’s been accurately interpreted,”

    We shall see . . .

    – Rick

    1. O Ye of Little Faith:

      “Question: Just on the panel of experts who will conduct the information audit, can you say who those experts are?

      Angus Houston: Well, I’d like to get back to you on that because, as you know, the panel of experts is a who’s who of expertise around the aviation investigation world, and I understand that they will be there, but they will also be assisted by some external people, and I will come back to you on that later.”

      1. Thanks Bruce (I think!)

        This whole notion of “they are experts so they must be right” is actually wrong. It doesn’t matter if you read something on the back of a matchbox, or if it was shouted at you by some mentally-deranged person in the street, or it was told to you by a Professor of Umbugology and Ditchwateristics*: a fact is a fact, and a falsehood is a falsehood.

        There is a story – perhaps apocryphal – of a booklet being produced in the 1930s with the title “100 Scientists Against Einstein” in which the notion of relativity (in its mathematical/physical sense) was argued against; or maybe I should say voted against. The reported response from Einstein was: “If I were wrong, one would be enough.” That is the essence of scientific truth, and progress.

        *Another of Dickens’ inventions; also (as with The Circumlocution Office) likely due to his friendship with Charles Babbage.

  109. Hi Duncan,

    As always, you are spot on. I would love to know what was being said in the Circumlocution Office as the “experts” meet to audit the information and maybe even the data and maybe even the process to derive information from the data. Unfortunately I have little confidence in their expertise or that we will be informed of the outcome.

    Meanwhile your commentators continue in force. I was particularly struck by the following comment from Henrik Rydberg:

    If I start on my southern route at the 18:27:48 UTC point (6.1797N 95.8528E) at the last known speed of MH370 which I calculated at 500.36 knots and match all your ping rings at this constant speed, I arrive at the 00:10:48 UTC point (39.6853S 81.8992E).


    Meanwhile, if MH370 ended up somewhere near 39.6853S 81.8992E, then any evidence in the form of surface debris will more likely wash up on the shores of Australia than be found by starting the search again.


    1. Richard, someone mentioned the streetlight effect a few weeks ago, and I have to wonder if that was a factor in abandoning the original Southern Indian Ocean search area. After the initial satellite picture showing a large piece of debris, the next thing we heard was about the Roaring Forties, how bad the weather was, how rough the seas were, how far away from Perth it was and how full of garbage the ocean is. I never did learn that any photographed or other debris from that initial area had actually been retrieved and inspected, but maybe I missed it.

      I am starting to think that the streetlight effect is having a larger overall impact on the search location, inasmuch as politically the ocean adjacent to Australia is the most convenient place to search. The Australian government can be in charge of the search because it is closest to the most likely southern route endpoints. They can go at it with gusto, even, because success could put major egg on the face of the Indonesian military for failing to detect the flight. (Apparently there is no love lost between Indonesia and Australia. .)

      On the other hand, China is the logical government to be in charge of a northern route search, and China is the logical place to start searching. Do you see how politically this could happen? Can you imagine a more audacious phone call than from Malaysia to ask China to take charge of a northern route search? Would China undertake such a search knowing that just doing so would be an indictment of its radar surveillance? And if it did undertake the search, would China want to find the plane, knowing how embarrassing success would be? And if it did find it, would China want to make that finding public? If China resisted taking charge of such a search, what government would dare interfere or call China out?

      Rather than thinking that the current search strategy is obtuse or worse, I prefer to think they are searching south because the Inmarsat information gave them a 50/50 chance of being right and because as a practical political matter that’s where the light is good. I expect the “audit” to confirm that they are on the right track, to reiterate that the AF447 search took two years even though they found debris right away, and for the search methodology to include going back over the undersea areas already searched, since that’s what worked for AF 447.

      1. Thanks Bruce.

        Of course there is a way to crowdsource a search of the northern perhaps-feasible endpoints, and that comes from freely available satellite imagery. Dr Kuang’s first posts involved such satellite imagery.

        Admittedly we are looking for something only a few hundred metres across (i.e. an aircraft crash at high speed), but there is plenty of satellite imagery available at circa 10-metre resolution (i.e. pixel size) which in principle would allow a before-and-after type search. That is, people looking for changes – assumedly in remote areas far from where people on the ground would have ‘noticed’ the crash – in satellite imagery obtained before the presumed crash, and after it.

        On the one hand one would like the time difference before-and-after to be as short as possible (so as to reduce the chance of other events changing the scenes, such as landslips in the mountains); on the other hand one would want the times of the two images to be compared to be seasonally-similar (e.g. images obtained mid-March this year and mid-March 2013).

        This would be a big task, but not impossible. I’d suggest starting along a band 10 km each side of the 00:11 UTC ping ring and then working outward from there (i.e. more distant from the ring). We would also like to narrow down the longitudes on which to concentrate. Personally I favour a higher speed for the aircraft and therefore longitudes further west, whereas VictorI’s analysis suggests a lower speed (circa 250 knots) and an end point further east, in Tibet; but I have more to do so as to firm up (or exclude) my own ideas.

        On this broad topic: does anyone know whether a search has yet been conducted of Dr Kuang’s suggested crash site in the Beshtash Valley?

Comments are closed.