MH370: analysis of where to look…

Posted in General, Inmarsat at 2:23 pm by timfarrar

Last week’s Wall St Journal article and my blog post highlighted that the MH370 search area was poised to move to the southwest, and yesterday this shift was confirmed by Inmarsat.

Although the location of this new search area has not yet been released, the independent team that has been analyzing the publicly available data felt it was appropriate to provide a statement, given below, with our best estimate of the highest probability (but not the only possible) location for a potential search. In this way, we hope to provide information which can clearly be seen to be completely independent of any locations that might be published by the search team in the near future.

The statement is as follows:

Shortly after the disappearance of MH370 on March 8th, an informal group of people with diverse technical backgrounds came together on-line to discuss the event and analyze the specific technical information that had been released, with the individuals sharing reference material and their experience with aircraft and satellite systems. While there remain a number of uncertainties and some disagreements as to the interpretation of aspects of the data, our best estimates of a location of the aircraft at 00:11UT (the last ping ring) cluster in the Indian Ocean near 36.02S, 88.57E. This location is consistent with an average ground speed of approximately 470 kts and the wind conditions at the time. The exact location is dependent on specific assumptions as to the flight path before 18:38UT. The range of locations, based on reasonable variations in the earlier flight path result in the cluster of results shown. We recommend that the search for MH370 be focused in this area.

We welcome any additional information that can be released to us by the accident investigation team that would allow us to refine our models and our predictions. We offer to work directly with the investigation team, to share our work, to collaborate on further work, or to contribute in any way that can aid the investigation. Additional information relating to our analysis will be posted on http://duncansteel.com and http://blog.tmfassociates.com. A report of the assumptions and approaches used to calculate the estimated location is being prepared and will be published to these web sites in the near future.

The following individuals have agreed to be publicly identified with this statement, to represent the larger collective that has contributed to this work, and to make themselves available to assist with the investigation in any constructive way. Other members prefer to remain anonymous, but their contributions are gratefully acknowledged. We prefer that contact be made through the organizations who have published this statement.

Brian Anderson, BE: Havelock North, New Zealand;
Sid Bennett, MEE: Chicago, Illinois, USA;
Curon Davies, MA: Swansea, UK;
Michael Exner, MEE: Colorado, USA;
Tim Farrar, PhD: Menlo Park, California, USA;
Richard Godfrey, BSc: Frankfurt, Germany;
Bill Holland, BSEE: Cary, North Carolina, USA;
Geoff Hyman, MSc: London, UK;
Victor Iannello, ScD: Roanoke, Virginia, USA;
Duncan Steel, PhD: Wellington, New Zealand.


  1. enjineerin said,

    June 17, 2014 at 2:29 pm

    For those pulling out their microscope, here are the coordinates of the five data points shown on the map.

    These specific points came from different mathematical models and different assumptions about the flight prior to turning South after 18:28UT. The spread across the data points is about 125 miles (200km).

    A 36° 30′ S 88° 37′ E
    B 35° 46′ S 89° 06′ E
    C 36° 47′ S 87° 11′ E
    D 35° 25′ S 89° 49′ E
    E 34° 45′ S 88° 11′ E


  2. Richard said,

    June 17, 2014 at 2:34 pm

    Hi Tim,

    On the BBC2 TV Horizon Programme aired tonight in the UK at 21:00 local time, the Inmarsat people interviewed concluded that the end point of MH370 was on the final arc at around 28S.

    This would mean a point 28S 98.1E.

    This is still 1,265 km to the North East of the groups conclusion in your post above of 36.02S 88.57E.

    It would mean that Inmarsat believe the speed of MH 370 from the a point at 18:39:55 UTC of 6.85N 95.06E was on average 381 knots and a track bearing 175 degrees.


  3. James A said,

    June 18, 2014 at 6:26 am

    If the grouping of these calculated possible crash sites are so close wouldn’t it be more expedient to send satellites across this path and capture some images of the area. It would seem to me that this would cost a lot less then dispatching a large amount of ships to “manually” search for this plane.

    I am not an expert this is just my personal opinion.

    James A

  4. jlangdale said,

    June 18, 2014 at 8:40 am

    Please state your assumptions about the autopilot/FMS heading mode TRK/HDG when you state your search area based on “wind conditions at the time.” is it not the case that if the plane was flying on waypoints FMS or on AP TRK heading mode, that the flight computer would have accounted for wind? Are you then making alternate assumptions? How are your wind corrections factoring into your calculation?

  5. Skwosh said,

    June 18, 2014 at 9:20 am

    Howdy Tim. Very pleased to see you (and your group) haven’t given up on this. Like you I have also been trundling away in the background on this, and it looks to me like your results may broadly be in line with my own – which makes me feel a bit less out on a limb.

    Sorry for the long post – though it has been partly provoked by the BBC Horizon documentary that was on the (British) telly last night – and from which I will first quote some short extracts relating particularly to the BFO analysis (everything in square brackets is me adding context):

    ——– 32:39 ——–:
    Voiceover: “Chris Ashton [of Inmarsat] was trying to work out if the aircraft had gone north or south…”
    Chris Ashton: “Of course the big question is – which route was taken?”
    Voiceover: “There was one more avenue to explore in the electronic handshakes – a second piece of data – the frequency at which the signal for the aircraft arrived at the ground station.”
    [Much footage of Chris Ashton gazing thoughtfully at multiple tellies bearing columns of figures, squiggly graphs and so forth]
    Chris Ashton: “We had a northern and a southern route that were so very different – so very far apart – the frequency information at that stage was something that was probably going to be good enough to discriminate between those two routes.”
    [Further voiceover and graphics drawing our attention to the fact that the satellite moves, references to the Doppler effect, and an animation showing two developing aircraft tracks along with two developing lines plotted on separate graphs (similar to the published BFO plots) showing the two plots diverging as the two aircraft tracks turn respectively north and south. We then switch to pictures of Chris Ashton drawing mathematical symbols on a piece of glass...]
    Voiceover: “The calculations were incredibly complex, and there was no guarantee of success.”
    Chris Ashton: “We’d attempted this calculation two or three times and abandoned it… we were working at it for a long time and not getting a good match between the measured data and the predicted data.”
    Voiceover: “But then came a break…”
    Chris Ashton: “We’d been working on the Doppler analysis all day long at the end of a week of investigating and collecting data. Quite late on the Friday night – at eight o’clock in the evening – suddenly the graphs matched, the data worked, the calculation was solved.”
    Voiceover: “Chris had eliminated a hemisphere, but he didn’t yet know which.”
    Chris Ashton: “… we’d got the calculation to work – and then I checked to see which of the flight paths it was – and we then identified it was in fact the southern route.”

    ——– 42:33 ——–:
    Voiceover: “… but the batteries that power the locator beacon [of the flight recorder] only last for about thirty days. In the southern Indian Ocean – in an area dubbed as close to nowhere as it’s possible to be – the race to find the black box was on. Aircraft and ships from eight nations were scouring six hundred and twenty thousand square miles of ocean. Could anything be done to help target the search?”
    Voiceover: “Back in London Inmarsat had been scrutinising probable flight paths MH370 could have taken, and what they had discovered was astonishing. By modelling a flight with a constant speed and a constant heading – consistent with the plane being flown by auto pilot – they had found one flight path that lined up with all their data.”
    Chris Ashton: “We can identify the path that matches exactly with all those frequency measurements and with the timing measurements, and lands on the final arc at a particular location which then gives us a sort of hot-spot area on the final arc where we believe the most likely area is.”

    ——– 56:11 ——–:
    Chris Ashton: “My thoughts were that they [Ocean Shield] were probably going to traverse down the final arc to go over our hotspot area, but of course they found their ping detection very early on.”
    Voiceover: “Ocean Shield could not ignore the detections it had heard – the search entered a new phase [the ocean floor search]…”

    ——– 57:35 ——–:
    Chris Ashton: “It [the ocean floor search] was by no means an un-realistic location, but it was further to the north east than our area of highest probability.”
    Voiceover: “Perhaps the best place to look for MH370 was always further to the south – Inmarsat’s hotspot on the final arc where their data says MH370 is most likely to have crashed – is here…”
    [A diagram shows a location on the final arc appearing to emphasise that 'here' is definitely not where the acoustic pings were heard – though there is a "Not to scale" caption.]

    ——– 58:41 ——–:
    Closing captions: “Horizon understands Inmarsat’s hotspot on the final arc is around 28 degrees south. It is an area yet to be searched.”

    My thoughts:

    (1) Part of how the TV programme was promoted before broadcast was the idea that Inmarsat’s current best hot-spot location has not yet been searched – and they certainly seemed to be emphasising this heavily at the end.

    (2) At one point above, the voiceover says “… by modelling a flight with a constant speed and a constant heading – consistent with the plane being flown by auto pilot – they had found one flight path that lined up with all their data”. This completely flummoxes me – and has been flummoxing me for weeks on end – those low-speed Inmarsat ‘high probability’ *curvy* paths are absolutely *not* constant-heading paths – as far as I can see there is *no* way that the phrase ‘constant heading ‘ can be usefully applied to these paths. Now, in the documentary, we get this odd phrase “…constant heading consistent with the plane being flown by auto pilot…”, and the way the voiceover phrases it is as if “consistent with the plane being flown by auto pilot” is an important qualifier – but this is not explored further.

    (3) As observed by yourself and others there seems to be no justification – in the absence of further explanation – for simply drawing constant speed lines between the ping-rings and calling this a ‘likely’ or ‘plausible’ track. My own calculations have convinced me that constant speed tracks (which are in general highly curved – and thus very *non* constant-heading) can be consistent with the BFO (particularly if you only worry about matching the BFO values after the re-start) – and that the particular Inmarsat constant speed curvy tracks that have been published *do* seem to match the BFO well – so if you want a track that (1) is constant speed and (2) fits the ping rings and (3) fits the BFO for the last five full handshakes, then, yes, the Inmarsat constant speed curvy tracks fit the bill. However – unless there is a good explanation as to *why* the aircraft’s heading would be changing so much during flight then surely these Inmarsat curvy tracks are of questionable importance. As I have said before, maybe there *is* a good explanation as to why the heading would be constantly changing – and maybe that is all to do with modelling the auto-pilot (hence the “consistent with the plane being flown by auto-pilot” qualification of the voiceover in the documentary) but no one seems to be coming forward with an explanation for this – and instead we have this cognitively dissonant presentation of these very obviously *not* constant heading tracks being described as “constant heading” – which is more than a bit vexing.
    I think we could be forgiven for concluding that in the absence of further information it just looks like the Inmarsat analysis is naive. For sure this may well *not* be the case – but there doesn’t appear to be much evidence in the public domain at the moment to refute this conclusion – it just looks like they drew constant speed lines between the ping-rings and picked the speed that best matched the BFO. It would be great if we could have some more information to reassure us that there was rather more to it than that – which I’m almost sure there must have been!

    (4) For what it’s worth, my fairly basic modelling (which doesn’t require heading changes to coincide with the ping-rings crossings) seem to agree more or less with your own group’s work – though my best-fit tracks I think probably go further east than yours before turning and then go more or less due-south (though actually the result in terms of the end point isn’t that sensitive to where the turn happens) – I end up quite a bit further East than you (91-92 degrees, though I am still working with the old pre-data-release ‘elevation-angle photo’ based ping-rings – my effort so far concentrating on trying to get something simple and robust rather than something precise). Basically – what I find is that – if you overlap the sets of tracks (all of which must be consistent with the rings) which (1) have more or less constant speed (2) have more or less constant heading (after the turn) and (3) have a good fit to all the BFO values (apart from during the re-start)… then the *only* tracks that can satisfy *all* three of these requirements at the same time are – as I sated above – roughly due south from a turn just before the Andamans travelling at around 470~500 kt. Anyhow – I’ll be very interested to see your write up – and I may do one of my own.

    Thing is though – in a way, where I (and probably your group also) are now – it looks a bit like where Inmarsat were *before* they suddenly decided to go all-out slow and curvy, and further east – which makes me think that there may well be some important stuff that they know/realise that we still (or at least I) don’t!

  6. thegarin said,

    June 18, 2014 at 10:14 am

    Hello, MH370 resque team.

    The first question occurs – why the MA can not use another Boeing for the natural experiment? Try to start test flying with similar satellite position – use the same trajectory – detect all pings BTO and BFO. Five or ten flyings and your give aproxymately correct path of MH370. The cost of this test can be less than the cost of another methods.

    Also – there is very correct message about using waypoints by autopilot placed in comments upper. It can be “magic” trick.

  7. enjineerin said,

    June 18, 2014 at 12:05 pm

    Re: autopilot/FMS heading mode TRK/HDG
    Each model is different with regards to wind effects and other assumptions. The individuals are each writing up their specific details. Intention is to post additional information here or on on duncansteel.com as it becomes available.

    (Precision of any wind estimates is fairly limited, and somewhat generic across various sections of the flight path. I believe that most models looked at average wind across a segment, and the time spent in that segment, and adjusted the end point by an appropriate distance.)

    Overall, there was slightly more wind through the Indian Ocean section then was occurring in the Straits. None of the winds were particularly strong.

    I favor a specific set of conditions… autopilot maintaining magnetic heading. Likely just holding the last heading after some unknown ‘final maneuver’ between 18:25 and 18:39UT. With throttle control maintaining a constant airspeed and altitude. (at least up until 00:11UT)
    But, others will likely have variations.

    Much effort was spent looking for waypoint combinations. Some of those were interesting, but none was a great fit.


  8. Alex Siew said,

    June 19, 2014 at 12:04 am


    Bill, u said in the previous thread ” The BTO provides the necessary evidence that the plane was traveling hundreds of miles per hour…”

    I humbly disagree. There was one thing traveling hundreds of miles that night (during the 6 pings) and that was the satellite. As an example, from 22.40 UTC to 00.11 UTC, the satellite traveled a distance (z) of nearly 400 km. See Duncan’s post on his blog on March 26th.

    At 16.30 UTC, the satellite was 830 km from the equatorial plane. At 19.36 UTC it reached its northern apex and started its descent downwards. At 22.40 UTC, its distance from the equatorial plane was 835 km. In effect, the satellite had gone a full ‘loop’ from 16.30 UTC to 22.40 UTC.(see Duncan’s post on March 25th for a graphic illustration). If the plane had been travelling hundreds of miles per hour during the 6 pings, the BTO for 22.40 UTC would be very different from the BTO for 16.30 (since the satellite was more or less at the same position in the sky at those 2 times). But it is a fact the BTOs for these 2 points in time, MORE THAN 6 HOURS APART, are approximately the same at 14540 and 14920.

    I have posted a more detailed comment in the preceding thread on MH370 on this blog addressed to u, with more figures showing how the BFOs and BTOs for the 6 pings were merely reflecting the velocity and movement of the satellite during that period. If i have erred in my analysis, feel free to correct me with a detailed rebuttal.

  9. Skwosh said,

    June 19, 2014 at 3:13 am

    Correction: Curse of dyslexia! When I said in my earlier comment that my tracks probably go further east than yours before turning south I should have said they go further *west* before turning south (as in nearer the Andamans before turning south).

  10. DaggerDirk said,

    June 19, 2014 at 7:17 pm

    There are a few factors that have never been “factored in” to the MH370 mystery. Casting aside all conspiracy theories, there is a logical explanation that’s around 99% complete and coherent for the MH370 scenario. Take the time to read into the following articles:
    The More Likely and most logical) explanation for the MH370 tragedy:

    1. http://tinyurl.com/lrhentv main theory and precedent

    2. http://tinyurl.com/ksugyh2 prior document

    3. http://tinyurl.com/mwnfn3s (SU-GBP final report)

    4. http://tinyurl.com/ngjdd8m cover doc

    5. http://tinyurl.com/q4dgor4

    6. http://tinyurl.com/mh370-mostLikelyCause

    7. http://tinyurl.com/ngjdd8m

  11. enjineerin said,

    June 19, 2014 at 7:41 pm

    I had chosen to not reply to, or comment on, your posts.

    Your theory does not in any way fit the facts as I understand them.

    On multiple occasions, you have not paid much attention when corrected on factual errors, and erroneous assumptions.

    You are entitled to your opinions. But, I will not be drawn into further discussion on them.

  12. Alex Siew said,

    June 19, 2014 at 10:40 pm

    To enjineerin,


    I think your reply shows u do not have an answer to the very pertinent question that was posed by my previous comment: why if indeed the plane was still flying during the 6 pings, the BTOs and the BFOs merely reflect the velocity and movement of the satellite during that period.

    I was not stating an opinion in my previous comment, i was stating facts. These facts as to the position of the satellite (z in km) at various times were obtained from Duncan’s post on March 26th entitled ” Positions And Velocities Of Inmarsat-3F1 During The Flight Of MH370″. The picture of the path of the satellite during that period is found in Duncan’s post on March 25th entitled ” The Locations of Inmarsat-3F1 When Pinging MH370″. The numbers as to the BTOs are from the 47 page document released by the investigation team late May.

    I was not asking for your view on my “theory”. I was asking if u have an explanation as to why the BTOs for 2 points in time, 6 hours apart, when the satellite was more or less in the same position, are around the same, if as u said, the plane was flying hundreds of miles per hour during that period.

    If u don’t have an explanation for that, why not just say so instead of evading the question by shooting the questioner.

  13. Skwosh said,

    June 20, 2014 at 3:09 am

    Alex – the distance from the satellite to the aircraft was probably about the same at both 16:30 UTC and 22:40 UTC – the BTO only corresponds to the shortest distance from the aircraft to the satellite at that instant, not how far the aircraft has travelled. The fact that the BTO values are more or less the same does not mean that the aircraft must have been in the same *place* at those two times (though it do not exclude this either) – the two BTO values being more or less the same is interpreted to just means that the aircraft was at roughly the same *distance* from the satellite.

    Think of holding a fixed length of string tied to a post so that it’s taught – you can be standing *anywhere* on a circle around the post and still be exactly the same *distance* from the post – this is why the BTO values correspond to rings – and if you look at any of the many diagrams of the BTO rings plotted on the surface of the earth you’ll see that the 22:40 ring does indeed come quite close to Kuala Lumpur (among many many other places).

  14. Alex Siew said,

    June 20, 2014 at 11:59 pm


    I agree with what u said. The fact the BTO values for 2 points in time are the same (when the satellite happens to be in the same position for those 2 point in time) simply means the distance between the aircraft and the satellite for those 2 times are the same. There are 2 possibilities:

    1. The aircraft was not moving ie stationary between the 2 points in time, or

    2. The aircraft was still flying but coincidentally the distance between the aircraft and the satellite was the same for those 2 points in time, ie at point A in time it was at some point on the arc while at point B in time it was still on the arc but at another point on the arc.

    Any person of science would want to examine the available data to rule out the first possibility ( the aircraft was stationary) before concluding that the aircraft was still moving. However, as exemplified by Bill’s reaction, there is a steadfast refusal on the part of many on this blog and on Duncan’s blog to even consider the first possibility.

    This refusal is puzzling considering:

    1. The fact that the BFOs for the 6 pings correlate or were tracking the velocity of the satellite. ( BFO= fixed offset of around 90 + satellite velocity in knots). Thus at 2 different points in time when the velocity of the satellite happens to be around the same, at 18.25 UTC (49 knots) and 20.41 UTC (47 plus knots), the BFOs were around the same, at 143 and 141.

    2. The fact that the BTOs correlate or were tracking the movement of the satellite. The BTOs for 16.30 UTC and 22.40UTC, previously cited, is merely one example.

    3. The graph of the movement and velocity of the satellite matches the graph for both the BTOs and BFOs. For the BFOs, we see a dip at 19.40 UTC, when the satellite was at its slowest at around its northern apex. Similarly for the BTOs, the BTO for 19.40 was the lowest, before progressively increasing thereafter (indicating, among other things, that the plane was north of the satellite at all times).

    4. The fact that the Inmarsat’s model for Doppler correction assumes a stationary satellite directly over the equator.

    5. The fact that Inmarsat’s “BTO model” likewise was based on the assumption that the satellite was stationary over the equator.

    Thus just looking at the data on the BFOs and BTOs alone, there are 2 possibilities:

    A. The plane was flying during the 6 pings, or

    B. The plane had already crashed early on, and the differences in the BFOs and the BTOs were caused by the velocity and movement of the satellite, firstly northwards at decreasing speeds for around 370km between 16.30 UTC and 19.36 UTC, and thereafter southwards at increasing speeds for approximately 770km between 19.36 UTC and 00.11 UTC.

  15. Skwosh said,

    June 21, 2014 at 5:51 am


    You say: “Any person of science would want to examine the available data to rule out the first possibility (the aircraft was stationary) before concluding that the aircraft was still moving. However [...] there is a steadfast refusal on the part of many [...] to even consider the first possibility.”

    I think you are being a bit unfair there Alex. I think many people have considered the possibility that the aircraft was stationary along with many other scenarios (a good example of someone considering other possibilities occurs in the recent BBC Horizon documentary where one of the Inmarsat engineers (Alan Schuster-Bruce) is reflecting on when they first realised that data was being transmitted for many hours after the flight was first thought to have been lost – he says: “One of the concerns we had was that this could all be just one big hoax that someone had played on Inmarsat, that the aircraft went down, and that someone at the same time had pretended to be that aircraft.” Now, I know that is not the same as considering that the aircraft was stationary, but it is just another angle on the *core* question here that we are all trying to answer:

    “What possible/likely explanations are there for these transmissions and the timing and frequency values that they have?”

    It is in the nature of the models that people (presumably inside Inmarsat and certainly some of us outside of Inmarsat) have built of the various pieces of equipment involved that a scenario in which the aircraft is stationary *does* *not* *fit* these models – so it is not generally true to say that such a scenario has *not* been considered – many people here are considering all manner of scenarios/tracks/possibilities and then testing them against their models – and *if* a stationary scenario fitted and/or tracks that converged to being stationary fitted then I think I can be reasonably confident in assuming that people would be *more* than happy to report such a result – I really don’t think there is any prejudice against the possibility – it is just that it doesn’t check out with the kinds of models that people are building based on the publicly released data and their understanding of how these systems work under normal circumstances.

    Yes – maybe the systems weren’t working normally (which people have been and are continuing to consider).

    Yes – maybe the models are wrong.

    However, the best any of us can do is to be as open as possible about what we are doing/thinking/calculating/modelling so that everyone else (yourself included) can peer-review our work and find flaws, identify dodgy assumptions and so forth – that is the process of science –science is nothing special – it is something we can all do – and that is what is happening here on these blogs – and it is a truly wonderful thing to see.

    Alex – please *do* keep banging on about your ideas. I for one value your tenacity (and the digging it inspires you to do) and I admire your bravery in standing up to people in the face of arguments-from-authority. There are far too many people in this world who are afraid to speak-up when something does not make sense to them.

    At the same time, I get the general impression that your approach seems to be (1) The aircraft definitely *did* crash soon after take-off and so therefore (2) Anything that is inconsistent with this idea must somehow be wrong.

    This is certainly not the conventional way of approaching problem solving that most people formally trained as engineers and scientists take – and I can understand why it can make such people very frustrated and even angry with you – and I think it is important for you to understand that, and to take it into consideration.

    However, I would also say, that sometimes people who have taken your rather un-compromising approach to problem solving have actually ended up either being right all along, or have either directly or indirectly lead to the problem being solved simply by constantly forcing everyone else to explain to them why they think they are wrong!

    I am a great believer in the usefulness of being made to explain one’s-self – and it can be even more useful if the person you are having to explain yourself to is a sceptic!

    Anyhow – I’m wobbling on (and apologies for being pompous – occasional occupational hazard of being me).

    To the matter in hand:

    BFO: I agree with you that the BFO strongly correlates with just the satellite motion. This turns out to be in line with the current broad-consensus understanding of how the frequency correction works – in this model by far the most significant factor in the BFO is indeed expected to be down to just the satellite motion (the whole idea of the aircraft’s on-board correction machinery being that it is trying to remove as much of the Doppler shift caused by the aircraft as possible) – so – the fact that the BFO looks like it is more or less a reflection of just the satellite’s motion is actually entirely expected – and so part of what a lot of us here are wrestling with is trying to squeeze as much information as we can out of the (very small part) of the BFO that isn’t actually dependent on the satellite motion and which should contain at least some information about the aircraft speed/direction (and thus help to pin-down the final location).

    BTO: Unfortunately I haven’t myself had much time to study the details of the BTO values in the recent Inmarsat data dump and have perhaps rather lazily assumed that others – in who’s abilities I have much confidence – will sort it out and give us all some high quality ping-rings – though I think I have a rough understanding of the issues (I really *must* get around to getting my head around this though). As I understand it the ‘fixed-satellite’ assumption with regard to the BTO is just a kind of ‘accountancy trick’ used to make storing the numbers in the Inmarsat logs take up less space, and it doesn’t actually remove any information (about the satellite’s motion). This is not conceptually ‘the same’ fixed-satellite assumption as employed in the BFO compensation (and, incidentally, it looks like another un-related ‘fixed-satellite’ assumption has been found in another part of the system too!) Anyhow, it *is* interesting to me that you claim to have detected a pattern in the BTO values that corresponds to the satellite motion – but I haven’t engaged sufficiently with the details of the ‘accountancy trick’ to know if that is actually something one might expect or not (perhaps along similar lines to the BFO case) – perhaps someone else could comment on this?

  16. JS said,

    June 21, 2014 at 11:59 am

    While I generally accept that the ping rings yield a distance, and those distances suggest a moving aircraft which could certainly be at the same distance 6 hours apart, I do find it mildly coincidental that the pings at 16:40 and 22:40 are so similar.

    More troubling, to me, is that the minimum ping delay occurs roughly when the satellite appears to be at its northernmost point, a little after 19:40 IIRC. That makes three points of coincidence.

    I also recall that the BTO is an offset, not the true speed of light time. I’m curious as well as to whether this could have been miscalculated based on the assumption of a perfect geosynchronous orbit, or if there is additional compensation occurring. I’d have to think this over but it would be quite interesting if the pings were indeed a mere reflection of the satellite’s motion. Remember, without realizing that the frequency was compensated based on the plane’s flight, the BFO looked like an indicator of ground speed. The compensation scheme changed the game.

    Note that none of this would require a stationary plane – it would just make the pings less useful.

    I’ll leave it to Alex to show a correlation across all the other pings and the satellite position, because if one exists, we might have more work to do.

  17. Alex Siew said,

    June 21, 2014 at 4:41 pm

    @Skwosh, @JS,

    There is really no need for people to get exasperated or angry with someone who takes a different view, especially if that someone has laid out detailed reasons or grounds to support that view. All that the other people have to do ‘to shut the guy up’ is to show him where he has erred, if indeed he has erred.

    1. The BFOs

    The consensus now is that the BFOs for the 6 pings merely reflect the satellite velocity and movement, the plane’s movement (if any) does not show up in the numbers. The explanation given by some ( Henrik for eg) is that any Doppler caused by the plane’s movement had been perfectly compensated by the compensation mechanism. However, it is also a fact the plane’s Doppler showed up in the transmissions prior to the 6 pings.That leads to the question: Why did the plane’s movement and resulting Doppler show up in the pre pings transmissions (ie transmissions at 1.07am and prior) but not for the 6 pings?

    One possible explanation is that some time after 1.07am, the data path link between the AIMS and the SDU was broken, with the result the SDU was no longer getting inputs as to the plane’s position, altitude, heading etc (as suggested by Victorl), and thus the SDU simply ended up assuming the plane was stationary. The alternative explanation is that the plane was in fact stationary during the 6 pings resulting in the absence of any indication in the BFOs of any movement by the plane.

    For the record, the BFOs for the 6 pings (first ping at 18.29 UTC and last at 00.11 UTC) and the satellite velocity ( in knots) during such transmissions are set forth below:

    BFO: 143-111-141-168-204-252
    Vz : 49 – 3 – 47 – 88 -123-160

    So basically BFO= Fixed offset of around 90 + Vz (in knots)

    The slight aberration for the 19.41 UTC ping was probably caused by the eclipse ( between 19.21/22 UTC and 20.23/24 UTC according to Duncan), with Inmarsat indicating as much as reported in the WSJ regarding the effects of the eclipse: “… 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…”.

    2. The BTOs

    The BTOs and z, the vertical distance in km of the satellite from the equatorial plane, for 16.30 UTC and the 6 pings are set forth below:

    BTOs: 14920-12480-11500-11740-12780-14540-18040
    z: 830 – 1155 – 1206 – 1159 – 1032 – 835 – 433

    The following observations can be made:

    1. Where z happened to be around the same, at 16.30 and 22.40 UTC, BTOs were ‘ coincidentally’ also approximately the same.

    2. The smallest value in BTO at 19.40 UTC ‘coincided’ with the largest value in z.

    3. The graph of the BTOs ‘coincidentally’ is exactly the inverse of the graph for z.

    4. The rate of decrease of z ‘coincidentally’ correlates to the rate of increase of the BT0s, in particular for the last few pings ( when the values would have stabilised as compared to the preceding values closer to the time of ‘reboot’ at 18.25 UTC). For eg, for the last 4 pings, every decrease of 1 km in z equated to an increase of around 8 plus microseconds in BTO.

    Can all of the above observations on the BFOs and BTOs be mere coincidences?

  18. JS said,

    June 21, 2014 at 8:33 pm

    Alex – you may have misunderstood me. I’m encouraging your viewpoint, though admittedly there is a lot of headway against it. I’m open minded but I’m also looking for a theory that explains the correlation between z and BTO.

    I’m intrigued by the fact that z and BTO correlate well. I suspected as much once I realized that they both reverse at 19:41. Though I’m shaky at regression lines, I come up with the following:


    Based on my limited understanding, there is pretty tight correlation between z and BTO. Of course, it’s only 7 points, but it is very interesting.

    So, why? This is where I don’t have a good answer. Z is only a small factor in the overall distance between the satellite and a point on the ground. A 100km shift in Z has less than a 1km impact on the distance between the plane and the satellite. That cannot account for the significant changes in the BTO, and certainly not enough to overcome the effects of changes in the plane’s location.

    It seems to me that you are suggesting that it makes no difference where the plane is, the BTO is what it is based on the satellite’s position. But what is it then? How would the satellite even have a number like that?

  19. Alex Siew said,

    June 21, 2014 at 11:10 pm


    It does matter where the plane was, the BTOs indicate that the plane was stationary during all 6 pings, at a location north to the satellite (north at all 6 ping times). That is why BTO was lowest at 19.41 UTC when the satellite was closest to its northern apex (and inversely z was highest). It was not a case of a plane traveling southwards from a stationary satellite, but a case of a satellite traveling southwards (from 19.36 UTC onwards) from a stationary plane.

    While in absolute terms z may only be a ‘small factor’, we have to remember the BTO values are offsets, in other words differential values and not absolute values. It is stated in the 47 page document that ” The BTO is a value (in microseconds) relative to a terminal at a nominal fixed location”. It is the differential value which matters and in this case, the differentials/offsets correlate with z, the movement of the satellite, which can only mean the plane was not moving during that period.

  20. andyo said,

    June 22, 2014 at 6:09 am

    Fantastic stuff. I was wondering why you guys haven’t found the plane yet. Just from a very quick perspective…

    a) The z values relating to distance above or below the equatorial plane going to about 800 km on an geostat orbit of radius 42000km amounts to a change in altitude of some 7km as far as I can tell. Therefore this has a negligible effect on distance measurements for an object ‘under’ the satellite. I would want to factor it in though, because it may have slightly more effect on the distance to aeroplane away from satellite.
    b) Using KISS principle (Keep It Simple, Stupid), even allowing for oscillation around the ecliptic plane, the fact that the ping ring footprints are not in fact concentric at nominally similar distances, but are in fact spaced well over 1000km apart if you crossed them perpendicularly, then the obvious conclusion is that the plane must have been moving.
    c) Are the ping ring maps correct? Surely they couldn’t be THAT wrong. They must be drawn on the planet’s surface taking into account the satellite’s wondering position… Again, if they weren’t taking the satellite’s position into account, using KISS, we’d expect the rings to bunch up (with their centres moving up and down).
    d) Wind speeds will obviously affect the final resting position and possibly substantially. Autopilot would presumably compensate for this. A manual flight wouldn’t. Either way, we cannot tell what the intended destination actually was.
    e) The final resting place has to be somewhere near the ‘final’ (reboot) arc. Obviously.

    It’s got me intrigued, this has. I’ll try and take a look at the numbers when I get a chance.

  21. JS said,

    June 22, 2014 at 6:51 am


    You are suggesting the plane was stationary the entire time? That contradicts the ADS-B data, at the very least. However, we could consider the possibility that the pings aren’t from the plane at all, but a different one parked at a gate all night.

    However, if your theory is viable, then the location of that stationary plane must be on arc, and it must be north of the satellite at 19:41. I believe that is a straightforward calculation as you are simply solving for the constant in microseconds – the difference between the BTO and the absolute value. However, there is no single solution.

    I believe you would end up with a parabola opening to the north and centered at 64.5E. The vertex must be further north than the satellite’s northernmost latitude. The points near the center of the parabola are the locations in which the offset is highest in proportion to the absolute value. The outer points on the parabola are east and west of 64.5E, so pings from those points are less affected by the north-south movement of the satellite. Therefore, those points must also be further north.

    I’m curious to hear your response.

  22. andyo said,

    June 22, 2014 at 7:18 am

    f) Just triangulating the maximal z value yields a surface (i.e. at sea level) divergence of about 125km. Which is nothing in the grand scheme of things (compared to ping ring separation), but’s it’s a shed load if you’re using an underwater search vessel.
    g) If the plane did in fact head south, then it’s perfectly reasonable to show a (straight) course that crosses inner ping circles (i.e. nearer the satellite’s ground position), then crosses the outer ping circles, right back to the starting ring (almost).
    h) Any assumptions on the course, autopilot and velocity (i.e. heading and speed) outside of the primary radar ranges have to be VERY educated guesswork. There does appear to be enough data to at least get a gist of heading. There are finite speed limits (max and min), although minimum depends on whether you believe the aircraft to be airbourne or not. Given the ping ring separations, it’s very hard to see below stall speed held for 7 hours.
    i) I am willing to consider the possibility that I’m wrong (just).

  23. andyo said,

    June 22, 2014 at 7:24 am

    The different plane on a gate all night presumably got ‘bumped’ into a reboot by careless cleaners.

  24. enjineerin said,

    June 22, 2014 at 10:11 am

    Thanks for adding to the discussion.

    Please show how the BTO values can represent a stationary plane.
    Show your reasoning, show your math, and show how the numbers agree with your proposed location.

    The correlation of BTO values to distance from the satellite is well understood and well agreed. This relationship has been used to calculate elevation of the satellite from the airplane’s location, and the ‘ping rings’. All of these show that the plane was in different locations for each subsequent data point (ping). Thus, the plane was moving. Calculating the range of possible distances between data points, it is also straightforward to show that the plane was moving far enough in an hour, that the plane had to be moving fast enough that it must have been flying. (21:41 to 22:41 for example, nearly 450km, ~300 statute miles minimum in one hour.) (Yes, I said ‘must’ excessively…)

    Your theory is critically dependent on some other understanding of the BTO numbers. For me to engage in discussion of your theories, I would need to see that is was plausible.


  25. JS said,

    June 22, 2014 at 10:29 am


    a) The baseline for any distance is the straight line, perpendicular to the earth, at 64.5E/0N. If either the plane or the satellite is 800km from that line, the ping distance should be the same, because the hypotenuse is the same length. However, yes, the maximum z-value is much smaller than the maximum ping ring diameter, so unless we add a co-efficient we can’t just say the BTO is a distance if the plane is stationary.

    g) Of course it’s reasonable, but it’s also reasonable to question how the plane’s flight distances, despite its erratic course, so closely correlate with the satellite’s movement. It’s also reasonable to question how the inflection points on both the satellite’s orbit and the plane’s path occurred around 19:41. Correlation is not causation, of course, but the fact that there is such a correlation merits a closer look to see if the BTO is indeed dependent on the z-coordinate. If it is, then it’s back to the drawing board.

    I’m intrigued by this correlation, but I agree with Bill – it requires a theory to explain why the BTO isn’t a straightforward time value, convertible to a distance value.

    One thing that does interest me on that front is this: the BTO is, as we’ve said, an offset. It’s an offset because, presumably, the absolute value is large, and for data storage and bandwidth purposes, it’s efficient to trim the number by some nominal value before transmitting it. This makes sense to me, but what I don’t understand is why the value is rounded by 20. If the purpose of trimming the value is to preserve detail at the lowest data cost, why is it also discarding the most granular part and yet storing data as microseconds, rounded? Why not store the value as “number of 20-microsecond units?”

    My guess is that the BTO values we have are not actually raw values. They are either 20-us values multiplied by 20 for our consumption, or they are manipulated in some other way. Indeed, if that manipulation is incorrect, nothing can be ruled out. If the BTOs were off by a factor of 20 from the distances to a stationary object, that would be very interesting.

  26. andyo said,

    June 22, 2014 at 12:25 pm

    Thanks for comments on my comments. I’m just trying to state the bleedin’ obvious simply (for public consumption if you will).

    I thought about what I’d written afterwards and accept your wisdom. Obviously the altitude of the sat is constant (at geo stat altitude) but it presumably carves out something akin to a Lisajous curve on the Earth’s surface, which is most likely an elongated figure ’8′ or ideally a straight line. The apparent change in altitude is about 7km for a static point on Earth. I think the apparent change in position at the Earth’s surface is highly relevant for a pin point determination of final resting place.

    I think the BTO correlation is just coincidence, rather like the sun and moon apparent size being almost identical (just coincidence that the ratio of their sizes is identical to the ratio of their distances from earth – zehfernspukschaft as Einstein would have said). I’ll eat something nice that’s edible if it’s not coincidence.

    I thought the offsets were in microseconds. Given a light nanosecond is about one foot (3*10^8m/10^9), then the BTO can be considered to be in 1000feets (i.e. one unit represents approximately 300m of ‘offset’).

    The absolute value must be humongous. Roughly (42,000,000-6,500,000)/300 for a random guess…

    Not sure which way I’d interpret the data if it has been adjusted by a factor of 20. Why can’t people just hand out the raw data unadulterated with simple explanatory notes? It’s no use to Joe public in any form, but it’s no use to number crunchers in any ‘cooked’ form either. It’s a bit like climate change data. The European stuff is raw as you like. The US data has a fudge factor that basically increases temperatures for decades… What?

    Also has anyone determined fully why the northern and southern arcs weren’t joined up? I’ve seen comments about bits being dismissed because of radar data etc, but why not show full arcs with excluded areas (like known radar areas, empty fuel tanks…) I suppose at its most brutal it’s the final southern arc.

    I’m interested to see arcs being described as ‘parabolas’, which may well be the case in extremis, but for the most part, the Earth can be considered spherical and I would have said moreso at those latitudes in question (bulgier at the equator and flatter at the poles). Therefore the ping ring projections on Earth from a satellite point must be circles centred around the point of intersection between satellite and centre of Earth. If the sphericalness or lack of, of Earth has a major impact on position I’d be surprised. If it does, then surely you must factor in the Moon’s gravity and tidal effects… Put it this way, the apparent movement north/south in orbit is far in excess of the apparent movement towards or away from Earth. The tangentialness of the ping ring and the satellite obviously has a bearing on the accuracy of the position, but we’re talking quite a reasonable sized triangle at the final arc, so accuracy is relatively assured. Just how far around though?. Obviously, it’s the question we’re trying to answer.

  27. andyo said,

    June 22, 2014 at 1:00 pm

    Mercator projections necessarily distort the circles (or parabolas) of course.

  28. andyo said,

    June 22, 2014 at 2:46 pm

    If you take the view that the BTO is directly related to distance and why not – ‘c’ is constant, then you have to understand that this is straight line distance between satellite and aircraft which are both moving in parallel ‘orbits’ (not necessarily similar directions). 300m might not sound a lot in relation to a distance of 36,000km (distance between orbits), but this can translate to a much larger horizontal offset distance. If the plane was flown erratically with widely changing altitudes, then we could be in a little bit more trouble (as in this could account for up to say 30 microseconds of BTO). It seems more likely that the plane was flown on autopilot, but that’s a mighty big assumption (although you wouldn’t expect random variations in BTOs, which might be consistent with this data).

  29. andyo said,

    June 22, 2014 at 3:12 pm

    Sorry to be such a pain… just thinking out loud. If the BTO are quantized to 20uS (a veritable eon in computing terms) then you wouldn’t be able to see altitude adjustments. It comes down to ring separation for the final 4 or 5 arcs. Are they near linearly spaced (taking into account Earth’s curvature)? Surely you’ve got straight line autopilot if they are. Nice ping ring spacing and nice BTOs = straight line/autopilot? Static would be same ping ring and nice BTOs… I figure the satellite is doing roughly 80mph (average). Over one hour this would affect a measured distance to a static point by only 2-3uS or so.

  30. Alex Siew said,

    June 23, 2014 at 2:55 am


    I have provided the math and the reasoning in my previous comment of June 21, 2014 at 4.41pm on this blog. I would be grateful if u can take a look at it and correct me if i have erred in my reasoning or in the math.

    The summary u provided on Duncan’s blog on June 1, 2014 at 21.47 pm is compulsory reading for anyone trying to understand the matter of the BTOs and BFOs. On the BTOs, i quote from the summary:

    “….For the BTO, the number is a representation of the difference in round trip time to the plane (earth-sat-airplane-sat-earth) and round trip to a virtual ‘nominal terminal’. There is no indication of the specific ‘nominal terminal’ used but it has been relatively simple to determine from the data at known locations (at the gate, and correlated to ADS-B recorded specific locations). The remaining surprise/uncertainty is that the algorithms used in recording the BTO makes a simplification – the satellite is modeled as a stationary point in the sky, an idealized geostationary satellite. This leaves a slight mathematical ‘error’ term caused by the actual satellite position/motion in the BTO numbers….”.

    Regarding the BFOs, u said:

    “…The BFO is still a messy algorithm to work through from the limited data provided……..To some extent (still being explored if it ALWAYS true) the satellite is assumed to be stationary in an ideal geostationary orbit (like the BTO)……This is one of the MOST significant new facts to emerge from the analysis of the released data – Since the plane is pre-compensating for most of its own motion, the BFO is NOT a direct measure of the airplane movement (speed&heading). Rather the BFO is a residual, comprised of:
    a) the Doppler effects due to the satellite motion relative to the Perth earth station, and b) the Doppler effects due to the satellite motion relative to the airplane, and c) the satellite position off its ideal location….. and d) the Doppler effects due to the vertical velocity of the airplane (rate of climb)…if any…”.

    In a later post on June 2, 2014 at 14.29 u said:

    “… In fact, we are seeing very little of the plane’s relative speed showing up in the BFO numbers….”

    I would submit that likewise, we are seeing very little of the plane’s movement showing up in the BTO numbers, as evidenced by the direct correlation of the BTOs to z.

  31. Alex Siew said,

    June 23, 2014 at 3:04 am


    I am not suggesting the plane was stationary at all times. In comments in previous threads on MH370 on this blog, i have laid out the evidence pointing to a crash some time around 1.46am, not far from BITOD (at the South China Sea) where the plane was last observed by Ho Chi Minh City air traffic controllers on their primary radar.

  32. JS said,

    June 23, 2014 at 5:41 am


    The problem with all of this is that as long as we accept the BTO values as being offsets in microseconds, the range between the longest and shortest times involve thousands of kilometers. That range cannot be explained by the satellite motion alone, nor can it be explained by an early crash, even if we were to assume that the plane was transmitting after it crashed.

    So, either 1) the BTO values are in microseconds and the plane traveled substantial distances, 2) the BTO values have been erroneously converted to microseconds, or 3) the BTO values don’t belong the MH370.

    I consider all three at least remotely possible, but to entertain a crash at a time and location other than the 00:11 ping ring, you must discard the BTO values. If the BTO values are a mere artifact of the satellite’s motion, then you must explain what units the BTO values are in, because they can’t be in microseconds.

    Along these lines, I would start by proposing BTO values that would be observed if the plane were stationary the entire time. I suspect they are off by a factor of 5 compared to the published BTO values, which is curious considering the 20us rounding. But that’s off the top of my head – I’d encourage you to create the list of 7 absolute values and 7 BTO values for a stationary plane and let’s see what they look like.

  33. Alex Siew said,

    June 23, 2014 at 6:27 am


    The BTOs and BFOs are not real values, in the absolute sense. I can do no better than to quote from Bill’s summary again:

    ” …Both the BTO and BFO values are mathematically abstracted numbers that represent time and frequency respectively. Neither, it turns out, are a simple measure of a physical parameter. Instead, they are each the result of measured signal parameters that are then compared to an expected value – the mathematically derived ‘difference’, or residual is what is logged. This is what has made the reverse engineering a technical challenge that remained open to many varied interpretations…”

  34. Alex Siew said,

    June 23, 2014 at 6:44 am


    Or to quote from Duncan (May 28, 2014 at 23.17 on his blog):

    “… To say it even more simply: the final letter in the BTO stands for OFFSET. The BTO is NOT a light/radio signal travel time. It is an offset from a nominal value. It is the changes in the BTO values that are important”

  35. Skwosh said,

    June 23, 2014 at 6:48 am


    OK – trying to put my money where my mouth is. Brain dump follows attempting to explain ‘myself’ in terms of where I am with the BFO and BTO, and why I think what I currently think:

    BFO: We can all more or less agree that there isn’t *much* information about the aircraft’s motion in there – and there is still much discussion of how much use can be made of what little there is. However, I do not think the current consensus is that there is *no* information in there about the aircraft’s motion (including the later hand-shakes) and you, Alex, must yourself surely admit that the BFO numbers are not *fully* explained by the simple linear relationship you have observed (yes.. errors… maybe…). Anyhow, the current consensus BFO model offers a good explanation as to why the BFO would be *mostly* just satellite motion (which is reassuring, because that is indeed what the numbers do look like – *mostly* – but not entirely) and the BFO model also seems to fit the current understanding of the mechanisms of the Doppler compensation hardware on the aircraft and on the ground (which is also reassuring). At the same time, the current BFO model *also* says that there will be *some* information in the BFO numbers that could be used to narrow down the aircraft’s motion, and people are understandably trying to exploit that as much as possible. So I think from the perspective of the BFO there should be room for you to accept that the current model does offer *an* explanation of the BFO numbers, and that this explanation seems to be holding up reasonably well, with no one having come up with any glaringly obvious flaws in it, at least so far (as far as I know anyway). Whether or not this is *the* correct explanation is, for sure, obviously still an open-ish question.

    BTO: I have still not had time to run any numbers myself on the BTO, but I have had a bit of time to review some (though not all) of the comments on Duncan’s blog that were made after the Inmarsat data dump, and I have had a quick ‘look’ at the actual figures. I think I understand the virtual terminal concept – and I must say that I am not at all surprised that it seems (according to your quotes from Bill above) that the round-trip time to the virtual terminal (the factor being subtracted off in the Inmarsat logs) turns out to assume a static satellite. I may have got the wrong end of the stick on this, but it seems to me that it would be (a) actually quite hard-work/overkill to determine and use the true variable satellite motion/position in a fairly complex calculation every time the system needed to make a log entry and (b) even if you did this it wouldn’t actually eliminate variation due to the satellite’s position and motion from the logged BTO values, so you’d still have to work out where the satellite was and how it was moving at the instant of the log entry if you wanted (later) to use the BFO values in the log to calculate a precise satellite to aircraft distance. Anyhow, I probably still need to do some more catching up on exactly where the consensus is on the BTO/virtual-terminal (if indeed there is a consensus), but if my understanding so far is correct then the virtual-terminal concept and the issue of whether or not a fixed or moving satellite model was used when calculating the BTO log values should *not* really have much influence over the amount by which the BTO values vary with respect to each other (in accordance with what @JS says above) and, if the system works as described, and we really are talking about speed-of-light round-trip times… and the units of the BTO are as described… then it is very difficult (for me at least) to see how the variable satellite motion/position could be a *significant* contributing factor to these values (though for sure one would expect it to be in there) – which makes the apparent correlation you claim to have found all the more fascinating. Perhaps it really is just a coincidence. However, almost (though not every) time in my life I can remember saying “perhaps it really is just a coincidence” it – what ever it was – turned out not to be… a coincidence… but… like I say… not always. I certainly find it *very* difficult to come up with any hair-brained ideas as to how the BTO could possibly *not* be what everyone is (quite reasonably assuming) it to be – and this must clearly include Inmarsat who I understand decided to record this value for the express purpose of being able to recover satellite to aircraft distances in cases like this – and they are obviously interpreting it the same way most of us ‘blog’ people seem to be. As has been suggested by others, I guess it is possible that these figures could have been doctored in some misguided way in an attempt to make them more helpful as a public data release – but given that we, the public, are able to use these BTO values to come up with sets of distance-ping-rings that are pretty close to the rings Inmarsat themselves appear to have been using suggests to me that, if there was any doctoring, then it certainly wasn’t a *botched* job – the BTO values we have been given check out in terms of what they are described to be (given a bit of bickering over the interpretation of the ‘virtual terminal’) – and they can certainly be used to generate rings that resemble the ‘Inmarsat’ rings (as in- depending on one or two differences in assumptions they’re not different by more than a few %). Could it be we are all making some kind of massive false-assumption? Perhaps all we have demonstrated is that if we make the same assumptions as Inmarsat then we get the same results… but it is really difficult to see any other explanation for these numbers… which I guess is perhaps best described as a failure of imagination on my part… I guess I must try a bit harder before giving up.

    General point however: Inmarsat have stated several times that they have looked at similar logs (so presumably both BFO and BTO values) for aircraft with similar (the same?) equipment and with known flight paths, and that the numbers check out (as in – presumably – the BFO and BTO figures as interpreted in the same way as they are interpreting them for MH370 also come out right for the known tracks of these other aircraft (not saying they’ve claimed to have re-run possible MH370 routes, just saying they’ve claimed to have looked at consistency of their models as applied to other known flights)). I think this is a very important and relevant point in many of these discussions and it is easy to forget about it (as I often do myself). Now, *if* what they claim is true (and why wouldn’t it be… though as others have said it would be ever so nice to have some of this data – more than one ‘instance’ of anything is always worth having if at all possible – so another instance of a BFO/BTO log along with the corresponding aircraft track would be pretty groovy…) – but – anyhow – if this is true – then if the BTO always comes out the same (so more or less correlating with just the satellite motion) then they would presumably have noticed that? Anyhow – I’ve ranted on too long. Still pondering…

    maybe… it… is… just… a… coincidence…

  36. JS said,

    June 23, 2014 at 6:54 am


    I understand that. But if the BTO is expressed in microseconds, they NECESSARILY exclude a stationary source.

    To support a theory of a plane stationary even over a portion of the six hours absolutely requires us to discount the notion that the BTO units are microseconds, or, to discount the accuracy of the BTOs altogether.

    You can’t fit the existing, microsecond, BTO values with a partially stationary plane. If you see this differently, can you explain it without referencing any previous post? I’ve read them all and none explains how microsecond BTO values of this magnitude can occur on a stationary plane.

  37. JS said,

    June 23, 2014 at 7:37 am

    @Skwosh -

    One obvious problem with the idea that the numbers were logged for the purpose of calculating distance is that they don’t correct for the channel type, nor for the “reboot.” Notice that one channel almost consistently lags the other by 5000us.

    That leads me to suspect that they were merely recording sent and received times and the offset is the difference between the expected and actual values.

    I’m struggling with the coincidence, honestly. On the one hand, the BTOs sure seem to be a microsecond speed-of-light value that matches up as they say it does. On the other hand, we have a flight that zig-zagged for at least part of 6 hours and somehow logged a curve of BTO values that correlates extremely well with the satellite’s ellipse.It would be a coincidence that the plane left a curved BTO graph to begin with, let alone one that matches the satellite so well.

  38. andyo said,

    June 23, 2014 at 9:19 am

    The O of BTO does in fact mean offset, but these offsets are the times between two moving points and a third theoretical virtual FIXED point. If we assume the satellite is fixed, then there are two fixed points and the only variable is the plane’s position, which means the BTO must directly correlate to its distance. When we expand the model to take into account the satellite motion, we determine that the satellite movement and speed is relatively low compared to a plane at cruising speed, therefore MOST of the BTO does in fact relate to changes of the plane’s distance. Not only that, we know precisely what the satellite’s relative speed and position is at all times. We can see that the effect of this movement on BTO is actually quite small. It’s presumably sinusoidal (in the N-S axis), reaching a minima at the N and S inflection points and a maxima as it crosses the equatorial plane. That is, the BTO relates FULLY to plane distance at the inflection points and MOSTLY (say 0.75 given the relative speeds) at the equatorial plane crossing.

    That said, there is no evidence of any sinusoidal change in the BTOs, which given the time correlation means that the plane maybe had a sinusoidal effect on BTO nullifying the satellite movement. which would be a mighty coincidence. I’m starting to be concerned about the BTO/time correlation… However, most of the BTO data relates to the satellite heading back towards the equatorial plane from the northerly inflection point doesn’t it? (If so, then the satellite is just accelerating southwards for most of the time).

  39. Skwosh said,

    June 23, 2014 at 9:38 am


    Yup – see where you’re coming from. For clarification though – in the recent BBC documentary, there is the following (just re-watched – yet again):

    Commentary: “But deep in the architecture of the system, a feature had been added that might offer a clue.”
    [Sequence of reporting of recovery of debris from Air-France flight 447...]
    Commentary: “In 2009 Alan [of Inmarsat] helped with the investigation into the disappearance of Air France flight 447.”
    Alan Schuster-Bruce: “We had a couple of brain-storming sessions at Inmarsat – one of the things we did work out is that we could take some additional measurements in the stations, and that would let us get a determination of the distance from the satellite to the aircraft, and therefore that would give a one dimensional position fix.”
    Commentary: “So after the Air France disaster, Alan decided to keep timing data in the electronic hand-shakes that might help track an aircraft.”
    Alan SB: “I was thinking we might need it one day – might be useful – might not be useful – but – err – I had no idea that it would essentially be the – err – really, the only evidence in town.”
    Commentary: “To try to find MH370, Alan now began analysing the timing data stored in the electronic handshakes…”
    [Received wisdom explanation of round-trip time, ping-rings etc. follows with accompanying graphics of satellite and ping-rings...]

    Now – the thing I like about this documentary (and I do seem to be going on and on about it I know, and quoting from it all the time – so maybe I’m getting a bit obsessed!) but these are interviews with the actual Inmarsat people – presumably *the* *actual* BTO person – Alan Schuster Bruce – (and then later *the* *actual* BFO person too). For sure I do not accept without question everything I see in a BBC documentary – and maybe the Inmarsat people are doing a bit of re-writing history – and/or tying to appear cleverer/more-prescient than they actually were – who knows – but these are the actual people, being actually interviewed, in front of an actual camera, and they know this is going to go out to the world – so I think it is quite a big deal in terms of at least clarifying the official Inmarsat *position* on certain things. So it at least appears that they *claim* to have started logging the timing data specifically for this purpose.

    Not that I’m arguing that this helps very much one way or the other! Just trying to clarify why I said what I said about the logging having been added specifically for this purpose.

  40. andyo said,

    June 23, 2014 at 9:44 am


    This in fact mirrors some of what you said. one axis sinusoidal movement or elliptical… not much difference. Either way, the plane appears to have flown a very similar curve in a different orbit. It IS disconcerting, although, it ought to be possible.

  41. andyo said,

    June 23, 2014 at 9:48 am

    The TV programme in question is readily available on the BBC iPlayer called “Horizon”. Anyone outside the UK, will need a UK VPN or proxy…

  42. andyo said,

    June 23, 2014 at 9:54 am

    Hang on! Satellite accelerating south… plane flying south around more and more curvature of Earth away from satellite. Maybe they nullify one another…

  43. JS said,

    June 23, 2014 at 9:57 am

    I’m doing some math here, and as I understand, 1000 miles of satellite movement is equal in effect to 1000 miles of plane movement. The triangle has the same sides, just upside down. The effect of a spherical earth and the sphere that the satellite moves around in must be factored in, but for practical purposes, it doesn’t matter which side of the triangle sees a change in distance.

    I have just noticed, however, that the distances the satellite travelled along the z-axis are almost exactly half the distances between ping rings. I have to do a few drawings to confirm how this affects distance, but a factor of 2 is stunning.

    It would be very easy to be off by a factor of 2, if round trip times and one-way times were confused.

    In other words, the BTOs could very easily be round-trip times between the satellite and a fixed object, rather than one-way times between a fixed satellite and a moving plane, which is how they were originally presented to us.

    I encourage others to see if they arrive at the same place.

  44. andyo said,

    June 23, 2014 at 10:08 am


    I’d be careful here. The satellite is on the end of a very long line from the centre of the Earth and moving, what, +/- 800km N-S repeated over a sidereal day. The plane is on the end of a much shorter line from the centre of the Earth moving at 900kph in a yet to be determined direction for a few hours.

    The satellite is ‘so far away’ that its slow positional changes have a relatively low effect, surely?

    If the plane was static, then you’d expect the BTOs to change by exactly the small amounts that the satellite movement has on the distance between satellite and static object (i.e. not much).

  45. JS said,

    June 23, 2014 at 10:22 am

    Is that true? What does the center of the Earth have to do with it? The plane and the satellite are at opposite ends of the same line. That line is the hypotenuse of the triangle. The other side of the triangle is indeed a line (or a section of the line) from the center of the Earth to a point z=0, distance 23000 or so.

    But the third side could be at either end with the same effect.

    There are cases where the extreme distance changes the equation – for example 1 degree of latitude is many more miles in z at the satellite’s altitude than the plane’s altitude. But this isn’t one of those cases – we are merely trying to determine the distance between two 3d points.

  46. andyo said,

    June 23, 2014 at 10:22 am


    Sorry, you ARE right. Both objects (sat and plane) are essentially moving in ‘orbits’, so each of their absolute distances from the centre of Earth are constant (disregarding flight altitude changes, which could account for 30uS in BTO).

  47. andyo said,

    June 23, 2014 at 10:25 am

    The sat has a ‘one axis’ simple harmonic motion. It does however maintain its constant distance to Earth, so it reciprocates along an N-S arc of circular orbit (relative to Earth).

  48. JS said,

    June 23, 2014 at 10:48 am

    Andyo – more important is the simple distance in space between the two objects. The same calculations can be done on two objects in deep space – the orbit merely gives us the coordinates.

    As far as the logging goes, I agree that the comments by the staff are telling. However, I think there is a difference between designing a satellite to report time offsets for distance calculations (which wasn’t done) and simply logging available data (which was done.) The latter provides no gaurantee that the data being logged is suitable for distance calculations. Useful? Absolutely. But not designed to be relied on as such.

    That may or may not make a difference.

  49. andyo said,

    June 23, 2014 at 10:57 am

    It’s useful info alright. Agreed, the system wasn’t designed with that in mind, but like Gene says in Apollo 13, “I don’t care what it was designed to do, I just want to know what it can do…”

    We know one of the orbits. We know the exact coordinates on that orbit. We know the time/distance between the the coordinates. We guess at the other orbit (somewhere between 0 and 9km above sea level) and we can derive the final coordinate.

    And thus, we have ping rings. The ring ‘width’ has an error depending on the angle between the tangent at the ring position on Earth with the direction of the satellite (same for the entire ring). And the only other variable is the altitude, which just increases the ring width a bit.

  50. andyo said,

    June 23, 2014 at 10:59 am

    Derive the final loci of coordinates, that is…

  51. Alex Siew said,

    June 23, 2014 at 1:53 pm


    The BTOs are round trip offsets, not single trip offsets. It is stated in the 47 page document (Inmarsat data) as follows:

    “The round trip time for a message is a combination of :

    1. Time from the ground station – satellite – aircraft – satellite- ground station

    2. Processing time…………”.

    Also to once again quote from Bill:

    “…. For the BTO, the number is a representation of the difference in round trip time to the plane (earth-sat-airplane-sat-earth) and round trip time to a virtual ‘nominal terminal’…”

  52. JS said,

    June 23, 2014 at 2:07 pm

    Well, I think we’re left with two options regarding the BTOs.

    1) We accept the BTO as offset values from a one-way trip, and by extension we accept Inmarsat’s published ping rings and wait out the SIO searches.


    2) We hypothesize an alternate definition of the pings, thereby implying that Inmarsat erred. This isn’t too far-fetched, if indeed the published BTO distances are twice the z coordinate at anytime, and especially considering Inmarsat first published rings without factoring the satellite movement at all, IIRC. That alternate definition would go along the lines of the ping times being round trip times to a stationary plane.

    Since we’ve never been given the value of the baseline, nominal terminal distance, the only reason to choose option 1 above is Inmarsat telling us to. I’ll echo others frustration with the lack of raw data once again – without knowing what that baseline was, we have no way to prove that the BTOs we have are one-way or round trip.

    If we were to try to pinpoint a stationary plane yielding these BTOs, it would again fall on an arc of sorts, only centered due north of 64.5E/0N, and opening northward. How far north depends on the guess for the nominal terminal.

    Can any of the above be proved false?

  53. Alex Siew said,

    June 23, 2014 at 2:50 pm


    I stopped watching the Horizon program after the first several minutes as i could not bear to continue watching falsehoods being disseminated.

    There was this guy that came on early in the program to speak about primary radar coverage showing a map of Malaysia and the south of Vietnam with small circles on the map purporting to be the coverage of various primary radars for that area. His point/argument was that the area where MH370 had supposedly disappeared from (shown on the map as a dot around IGARI) was not covered by primary radar. The circle shown on his map extended only a short distance from the Malaysian north east coast with the edge of coverage displayed some distance from IGARI.

    The truth is MH370 did not disappear at IGARI but only the secondary radar label disappeared from the radar screens when the plane was at IGARI at 1.21am, with both Malaysian ATC and Ho Chi Minh City ATC continuing to track the plane (now just a radar blip) on their respective primary radar until 1.30am up to BITOD, which is 37 nm from IGARI.

    I have on several previous occasions both on this blog and on Duncan’s blog set out the statements from both the Malaysian ATC and the Vietnamese ATC on their tracking of the plane on their primary radar up to BITOD/1.30amm and will continue to do so each time i see someone distorting the truth about this critical information.

    On March 8th, this is what The Star, a leading Malaysian newspaper reported:

    “The last signal position of MH370 recorded 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 Centre at the Sama-Sama Hotel here”.

    On March 12th/13th WSJ reported as follows:

    “Azharuddin Abdul Rahman, director general of the Department of Civil Aviation told a news briefing that air-traffic control lost contact with Flight MH370 on its secondary radar system at 1.21am Saturday before losing contact on the primary radar at 1.30am…….. As is standard practice, Malaysian controllers use two radar systems, a primary and a secondary, to monitor their airspace…” .

    Extracts from the schedule of recorded calls between the 2 ATCs on that night, annexed to the MH370 Preliminary Report dated April 9th and stamped ‘Confidential” and only released on May 1st grudgingly after unrelenting criticism of Malaysia’s initial refusal to release the report:

    “1.38.19 am…….. HCM first enquired about MH370, informed KLATCC that verbal contact was not established with MH370 and radar target was last seen at BITOD….

    1.41.21 am……. HCM queried about MH370 again, stating that radar contact was established over IGARI but there was no verbal contact. HCM advised that the observed radar blip disappeared at waypoint BITOD……

    2.18.53 am…… HCM confirmed earlier information that radar contact was lost after BITOD and radio contact was never established….

    The last location of MH370 at BITOD was not mentioned in the main text of the Preliminary Report which only stated the last secondary radar position. The people who prepared the report went to some length to suppress the information about BITOD, as evidenced by this sentence in the report: “At 01.21.13 MYT the radar label for MH370 disappeared from the radar screen at LUMPUR RADAR KLATCC…”. No mention anywhere that the plane had continued to be observed on the radar screen, as a radar blip, for 9 more minutes, until 1.30am.

    8 minutes later, at 1.38am, HCM ATC first called KLATCC.

    To the Independent Group, please do not ignore the foregoing information in your analysis, thank you.

  54. Alex Siew said,

    June 23, 2014 at 2:58 pm


    I think our comments crossed. The offsets are round trip offsets, not single trip offsets. Please see my previous comment on this thanks.

  55. JS said,

    June 23, 2014 at 3:07 pm

    Can you prove that they are round trips? I’m of the impression that they cannot be proved round or one way without knowing the absolute values.

  56. Alex Siew said,

    June 23, 2014 at 5:32 pm


    Please look at my previous comment of June 23, 2014 at 1.53pm above. The BTO was defined by Inmarsat as a ’round trip time’ relative to a nominal terminal, please see the first page of the 47 page Inmarsat data document released in late May.

    Also, Bill who is a member of the Independent Group referred to in the introductory comments in this thread, has also described the BTO as a “representation of the difference in round trip time to the plane (earth-sat-airplane-sat-earth) and round trip time to the nominal terminal…”

    So i do not think there is any question that the BTO is a round trip offset, rather than a one way offset.

  57. Alex Siew said,

    June 23, 2014 at 6:05 pm

    To the Independent Group

    If what JS said is true, that the published distances between the ping radii are a multiple of 2 of the respective distances traveled by the satellite during the corresponding times, i believe the riddle of MH370 has been solved.

    Given the implications, I would urge the Group to take a look at the numbers again to see if there is any merit in what has been discussed in the last several comments on this blog.

  58. JS said,

    June 23, 2014 at 6:38 pm

    Ok, fair enough on the round trip.

    One thing I have to point out – we talk about the absolute round trip time being huge. It’s not, actually. It’s roughly 75,000 km, each way, between Perth, the satellite, and the plane. Round trip, that’s 150,000 km, which is about half a second. In us, that’s about 500,000. The round time to the nominal terminal is in the neighborhood of 497,000, if we believe that the BTOs belong to a moving plane.

    Does anyone have an exact estimate of the fixed portion of this absolute value? It does not appear to me that it’s the trip time to the surface at 64.5E/0N. Does the fixed portion fit for all ping rings? Is there an algebraic solution for the fixed portion that fits all BTOs assuming a stationary plane?

  59. andyo said,

    June 24, 2014 at 3:14 am

    Whoaaah. The round trip time is twice the surface (6,500km) to sat (42,000km) altitude plus the terminal-plane distance. So more like 75,000km in TOTAL. More like 1/4 second. (i.e 250,000 on an absolute BTO). The virtual terminal is Earth based isn’t it? Could be any reference point anywhere though…

  60. andyo said,

    June 24, 2014 at 3:57 am

    Simple observation on ping rings…

    The satellite’s reciprocating movement around it’s circular orbit by approx. +/- 800km translates to an apparent Earth position that reciprocates N-S at 65.5E of about +/- 125km.

    All the ping rings must be centred on this +/- 125km N-S arc on the surface of the planet.

    Regardless of where the ping rings are drawn (assuming they are correct), then surely for a static plane, they must all intersect at that static position. It just ain’t so.

    For @Alex to be right, either the ping rings are centred in the wrong place (i.e. satellite position has been disregarded) or they’re drawn with the wrong radii (i.e. the BTO interpretations are wrong).

  61. andyo said,

    June 24, 2014 at 4:10 am

    To all intents and purpose, the +/-125km surface Earth movement of the satellite is minimal. Obviously it makes a big difference to the search area so it needs to be factored in. It makes relatively little difference to the rough position of the plane relative to the Indian Ocean – potentially out by 250km – although the satellite only traversed part of its arc during the flight.

  62. JS said,

    June 24, 2014 at 6:06 am

    Andyo – I disagree with your round trip math. I get about 38,000 from Perth to the satellite, and about 37,000 from the satellite to the plane. That’s 75,000 each way, or 150,000 round.

    You are correct in stating the ping rings would need to intersect, but that is possible, and in fact gauranteed for any two ping rings with the correct nominal terminal. For example, if the satellite is over a point 125km N at 19:40, and 75km N at 22:40, ping rings of 375 and 425 will intersect at a point 500km N.

    How does one obtain those ping rings? We would need different BTOs, and hence a different nominal terminal. But we were never given a nominal terminal to begin with, were we? One nominal terminal creates BTOs putting the plane in the SIO, while a different one puts a stationary plane in India, while still others put a slow moving plane in the Bay of Bengal.

  63. andyo said,

    June 24, 2014 at 10:08 am

    Geosynchronous orbit is 35,786km above the equator (Wikipedia). Earth’s radius is approximately 6370km

    Altitude of satellite (from centre of Earth) is 42156km

    The satellite cannot see the poles as they are tangentially over the satellite’s horizon. but if the sat could see the poles, they are 6370km above and below the line from sat to centre of Earth.

    Therefore the pole’s distance from the sat is hypotenuse of the triangle given by orbital height (from CENTRE of Earth) and radius value, which is:-


    Now, if you suppose the virtual terminal to be on Earth at the opposite pole, then you get a worst case round trip of:

    satellite to one pole to other pole to satellite, which is:-

    2*43126km + 2*6370km = 98992km

    Absolute best case round trip distance is where the virtual terminal and the plane’s position coincide directly below the satellite, which is twice the height of the satellite above the horizon:-


    So there you have it… round trip, is bounded (if the virtual terminal is on the face of the planet), by

    71572km <= round trip distance < 98992km

  64. andyo said,

    June 24, 2014 at 10:09 am

    Above the equator, not horizon… (apologies).

  65. andyo said,

    June 24, 2014 at 10:27 am

    Are you suggesting the round trip is:-
    sat to plane to sat to terminal to sat?
    Not the shortest triangular path?

  66. JS said,

    June 24, 2014 at 1:52 pm

    Round trip is Perth to satellite to plane, back to satellite, back to Perth.

    Ground station initiates a ping, timer starts, sends to satellite. Satellite sends to plane and waits for a response. Plane responds to satellite. Satellite relays to ground. Timer stops.

    Basically, every round trip involves 4 trips at least 35,800 km, and up to 43,126km. More accurately, the two trips to Perth are nearly constant distance, and the beam won’t quite reach the poles.

  67. Alex Siew said,

    June 24, 2014 at 7:54 pm


    It is stated in the 47 page Inmarsat data document released in late May as follows:

    “Understanding the Burst Timing Offset (BTO) values:

    - The round trip time for a message is a combination of :

    1. Time from the ground station-satellite-aircraft-satellite-ground station
    2. Processing time within the ground station, satellite and aircraft terminal, which are constant

    - The BTO is a value (in microseconds) relative to a terminal at a nominal fixed location. Only R-channel messages are used.

    - The BTO therefore allows the determination of the distance between the satellite and the aircraft. It does not provide the actual aircraft location. “

  68. JS said,

    June 24, 2014 at 10:02 pm

    That’s correct Alex, except for one detail (not yours, Inmarsat’s.)

    Changing the nominal terminal, and thus the baseline time delay, causes different results for BTOs and ping rings. Consider, for example, if the time subtracted for the nominal terminal was off by 5000us. The ping rings would move dramatically.

    There is no way to detect an error here. We have BTOs, but no nominal terminal value and no absolute value. With two out of three, we can confirm the numbers. With one out of three, we can’t.

  69. Alex Siew said,

    June 25, 2014 at 11:50 pm


    Yes,agreed. I don’t think Inmarsat have the ping time absolute values but they have the nominal terminal’s location but for some reason, they are not telling. Richard c10 seems to think the nominal terminal would be around 718km above the earth surface (at 64.5 E 0 N). This does not look like a number one would pick as some fixed reference point.

    I am still hoping that besides u, andyo and Skwosh, others with the requisite mathematical skills will comment on the apparent correlation between the BFOs/BTOs and the movement of the satellite.

  70. andyo said,

    June 26, 2014 at 2:01 am

    @Alex & @JS
    I stand corrected.

    My calculations would change somewhat. I can’t even say that my calculation bounds half the distance anymore…

    Why wouldn’t you consider a virtual terminal to be at 64.5E, 0N? It’s directly under the ‘nominal’ satellite position.

    An altitude of 718km sounds odd, but it ‘almost’ rounds the VT to sat distance to around 35,000km…

    Like I said, the VT can be any fixed 3D reference point. An assumption that it is on the Earth’s surface is just a random guess. I’d have thought a negative altitude of 718km would give better triangles.

    I see that assumptions are now being made (by the powers that be) that the plane flew south on Autopilot at fixed altitude, which doesn’t look overly unreasonable from the data to hand. Apparently, this gives a final ping ring width of just 58km. I think I’d want to bound it wider, at the expense of search time. No point looking into too tight an area.

  71. rw said,

    June 26, 2014 at 12:49 pm

    The difference in BTO between 21:41 and 22:41 is 1.76ms. The round trip path , at the speed of light, needs to be 528km longer, so around 250km difference on the path between plane and satellite. So just in 2D, assuming the planes distance from the satelites vertical projection to the surface is 4000km, I draw 2 triangles with common height 35000km and bases 4km and 4+d km. If the differnce in the hypotenuses is 250km then I get that the plane should have travelled(d) at least 1800km in this 1hr. Either I screwed up or the BTO numbers make no sense.
    An 800km N/S movement of the satellite only accounts for 10km on the path, and I don’t think any other simplifications can explain this away – but will stand corrected.

  72. Alex Siew said,

    June 26, 2014 at 10:27 pm

    To Bill and other members of the Independent Group,

    According to the just released report from ATSB ” MH370 – Definition of Underwater Search Areas” dated June 26, 2014, the ” final primary radar fix occurred at 1822 (Figure 2)…”. This reference is obviously to the now infamous unidentified blip at MEKAR at such time.

    In a preceding thread on MH370 on this blog and on Duncan’s blog, this blip has been shown to be UAE343 and not MH370. The following statements were from Don aka GuardedDon who i understand is one of the members of the Independent Group:

    1. On this blog, under the thread ” Understanding the ‘satellite ping’ conclusion”, on April 28, 2014 at 3.14pm, in reply to ‘seanmcleod at el’:

    ” This possibly does refute the RMAF/TUDM MoTM released radar data. I took the pretty picture that was shown to the MH370 relatives in Beijing purporting that a fuzzy green element in the image was the radar return from Mh370. I scaled and sheared the image to fit a snapshot of waypoints, paths and coastline at Skyvector, then I took the FR24 log of UAE343 and composited that with the other two images. I maintained some transparency to confirm the alignment. That final composite was imported as an overlay to Google earth. See http//i.imgur.com/NHbLF46.

    MH370 and UAE343 could not have been in the same place at the same time. The excursion around Sumatra has never been credible to me….. There have been been many inconsistencies in the MoTM’s presentations since 8th March……. I dread to think what they have presented to ICAO as the first interim report……” .

    2. On Duncan’s blog under the thread “Developments In Investigations of the Route taken by Mh370″ posted by Duncan on April 27, 2014, with Don’s comment posted on April 29th at 10.35am:

    ” I ‘ve composited UAE343′s track onto my previous overlay of the RMAF-MoTM radar track that alleges to show MH370 flying out past Sabang/N. Sumatra.

    That shows UAE343 at the position RMAF-MoTM stated MH370 had reached at 18.22 UTC.

    I ‘ve checked the FlightRadar24 report a number of times to verify, the area coverage for UAE343 is accessible at this URL……”.

    The above comments were prompted by reports by Dr Kuang a few days earlier, translations of which were reposted on Duncan’s blog by LGHamiltonUSA, suggesting that the misrepresentation by the Malaysian authorities that the blip was MH370 could have been due to ‘negligence’ but more likely it was intentional.

  73. Alex Siew said,

    June 26, 2014 at 11:23 pm

    The following is stated in the ASTB’s report:

    ” In the case of MH370:

    -The aircraft departed from Kuala Lumpur on March 7 2014 at 1641

    -The final automatically transmitted position from the aircraft occurred at 1707

    -No radio notification of a problem was received from the crew

    -No radio communications were received from the crew after 1719

    -The final ATC (secondary) radar fix occurred at 1722

    -At 1725 the aircraft deviated from the flight-planned route

    -The final primary radar fix occurred at 1822…..”

    “Radar data showed that after take off MH370 tracked in accordance with its flight-planned route to waypoint IGARI and then turned right towards waypint BITOD. Secondary radar was lost shortly afterwards. Primary radar then showed that MH370 deviated from its flight-planned route…”.

    The foregoing is a distortion of the evidence.

    1. Radio communications on the emergency frequency were heard from MH370 at 1.30am (1830 UTC) by the pilot of MH088 who had been asked by Vietnamese ATC to contact MH370 after MH370 did not report on the radio to Vietnamese ATC at the point of handover. This was confirmed by an official statement by the authorities on the official DCA website on March 9th that “… we last heard from the plane at 1.30am….”.

    2. MH370 did not deviate at 1725 UTC (1.25am) from its flight-planned route which was to IGARI then to BITOD but had continued to fly or glide to BITOD until 1.30am. See the statements by the Malaysian DCA/ATC on March 8th and in the days that followed that Malaysian ATC had tracked the plane on their primary radar until 1.30am and the statements by the Vietnamese ATC as recorded in the schedule to the MH370 Preliminary Report that MH370 as a ‘radar blip’ (ie on primary radar) was last seen at BITOD/disappeared after BITOD.

    3. Just like the main text of the Preliminary Report, there is no mention in the ASTB’s report that MH370 was tracked by both Malaysian ATC and Vietnamese ATC on their primary radar until 1.30am/BITOD. (ie for 9 minutes after loss of secondary radar signal at 1.21am at IGARI).

    4. As regards the statement that ‘ no radio notification of a problem was received from the crew”, there were several reports by Taiwanese portals on March 8th of an SOS call from MH370 having been picked by the US 7th Fleet at 1.43am, with the US Embassy cited as the source of the information. These reports have not been confirmed/verified but at the same time (a) the American authorities have not come forward to deny these reports (b) the portals have not retracted these reports and (c) the investigation team have not denied these reports either.

    5. The purported ‘final primary radar fix’ at 1822 UTC was actually UAE343, as has been shown by various people on this blog, on Duncan’s blog and on PPRUNE.

  74. Alex Siew said,

    June 27, 2014 at 12:15 am

    It is also stated in the ASTB’s report:

    ” In the case of MH370, there were multiple redundant communications systems fitted to the aircraft (3 x VHF radios, 2 x HF radios, SATCOM system, 2 x ATC transponders)”.

    Compared the above to the following description of the communication systems reportedly by a 777 pilot, regularly quoted on the net:

    “There are many ways to fly the 777 and there are safety layers and redundancies built into the airplane….. There are so many communication systems on the airplane: 3 VHF radios, 2 SATCOM systems, 2 HF radio systems, plus transponders….”.

    Which leads to the question: was there just 1 SDU on MH370 or were there 2? The ASTB’s report is vague (intentionally so in my view) in merely describing it as “SATCOM system”, but as i have been arguing for the last few months, the evidence shows MH370 actually had a dual SATCOM system involving 2 SDUs, on a master slave basis, with the master SDU linked to the high gain antenna subsystem and the slave SDU linked to the low gain antenna, and with the slave SDU/low gain antenna only coming on in the event of failure of the master SDU/high gain antenna subsystem:

    1. MH370 had both a low gain antenna (shaped like a shark fin) located on top of the rear fuselage AND dual side mounted high gain antenna located above the rear door.

    2. The MCS 4000/7000 or 4200/7200 series of SDUs manufactured by Honeywell are primarily geared towards a dual SDU/ dual antenna configuration on a master slave basis as described below.

    3. All indications are that the SATCOM system/s on MH370 were fitted by Honeywell.

    4. This dual configuration, with the slave SDU/low gain antenna as a backup, is provided under ARINC 741, the industry standard for SATCOM systems’ design.

  75. JS said,

    June 27, 2014 at 6:59 pm

    I think I could be a little off on the idea that 800km at satellite altitude is equivalent to 800km on the surface, because the satellite is maintaining a relatively constant altitude regardless of its z-coordinate.

    My concern remains, though. The BTO values we are looking at are 1 of 3 variables in an equation. The other 2 are 1) the absolute time delay, and 2) the value used for the nominal terminal.

    In this situation, there is no way to confirm that the BTOs mean anything in particular, because the BTOs themselves do not indicate a distance, only a part of an unknown distance. The only real constraint I see is that the maximum possible variation in the BTOs is 45,840us, because that is the difference between the vertical point directly below the satellite and the distance to the ring at the edge of the earth perpendicular to that vertical line. Theoretically, though, these BTO values could fit anywhere in the hemisphere.

    @Alex – I’m guessing from your post that you are no longer (or maybe never were) convinced that the plane even turned left from the Gulf of Thailand?

    If so, what is your theory on where it went?

  76. Skwosh said,

    June 28, 2014 at 2:54 am

    Agree about the geometry, but I think it’s great that you and @andyo have had a good old go at kicking the tires of the BTO and trying to tease-out the assumptions – personally I still can’t think of a simple alternative interpretation and/or failure-mode that could result in the apparent range of BTO values either – but, of course, the co-incidence is still sitting there – like an inaccessible smug cat. I’ve consigned it to the ‘try not to worry about it at the moment’ list for the time being while I plug the numbers from the new ATSB report into my horrifying monstrosity of a track-fitting algorithm – not that this will prove anything particularly – but I want to try it anyhow!

  77. Alex Siew said,

    June 28, 2014 at 4:21 pm


    I tried to post a reply to u yesterday but it seems it did not go through, so i will try again here.

    The evidence shows the plane crashed at the South China Sea. The plane suffered an electrical failure at IGARI at 1.21am and from such point at 35,000 ft and at its reported glide ratio of 18:1, it would have glided (ie without engine power) for approximately another 25 minutes for approximately 100 nm before crashing. The crash time would thus be around 1.46am and the crash site around 100nm from IGARI, most likely on a 40 degree path, the direction the plane was flying ( IGARI to BITOD) at the time of the electrical failure.

    Everything went off at 1.21am just after passing IGARI, the lights literally went out (the secondary radar signal, ADS-B, Satcom, etc). I have in previous comments on this blog laid out the evidence that this electrical failure was caused by positive lightning striking MH370 at that point in time.

    Although all signals from the plane ceased at 1.21am, both Malaysian ATC and Vietnamese ATC continued to track the plane on their primary radar which did not require signals from the plane. The head of Malaysian DCA said on March 8th and in the days that followed that week that Malaysian ATC had tracked the plane on ATC primary radar up to 1.30am. Vietnamese ATC is on record in the schedule to the Preliminary Report that they tracked the plane as a radar blip (ie on primary radar) all the way to BITOD. Both ATCs said the plane then disappeared.

    All the signals going off could only mean electrical failure, whether caused by human intervention or otherwise. Also, the fact the plane disappeared from primary radar after BITOD can only mean it had gone from 35000 ft at IGARI to a lower altitude at BITOD to cause it to then lose primary radar coverage, another sign the plane had lost power and was only drifting from IGARI onwards.

    Coincidentally or not, we have several reports in the Chinese language press on March 8th citing the US Embassy as the source, of the US 7th Fleet having picked up an SOS call from Mh370 at 1.43am, with the pilots reported to have yelled that they had to land ie ditch the plane. These reports have not been confirmed but neither have they been denied or retracted.

    There is simply no evidence to show the plane turned west off IGARI. The evidence shows instead the plane continued on to BITOD. The plane could not have turned west off BITOD and still make it to MEKAR at 2.22am as that would mean flying above its maximum speed. There were so many airports/ATCs at the border region between Malaysia and Thailand, yet no one saw Mh370 or an unidentified blip crossing over Malaysia. The blip at MEKAR has been shown by various people to be UAE343. If there is a recording of radar whether ATC primary radar or military primary radar, showing a plane turning west and crossing Malaysia that night, the recording would have been produced and CNN would have been playing it every night for the last 3 months as their ‘breaking news’.

  78. Alex Siew said,

    June 28, 2014 at 6:57 pm


    I omitted to mention in my previous comment that MH370 had traveled at a speed of 240 to 250 knots from the point secondary radar signal ceased at 1.21am just after IGARI, to BITOD at 1.30am. 240 to 250 knots is the reported gliding speed for a B777 (see for eg the comments of Dave Whittington who is trying to recreate MH370′s flight on a simulator, on Duncan’s blog) ie the speed of a B777 if it has lost all power.

    According to the Preliminary Report, MH370 passed over IGARI at 1.21.04. 9 seconds later at 1.21.13 ” the radar label for MH370 disappeared from the radar screen at LUMPUR RADAR KLATCC”. The last reported altitude and speed for MH370 was 35,000 ft and 471 knots at IGARI.

    BITOD is 37 nm from IGARI. In that 9 seconds from IGARI at 471 knots, MH370 would have traveled 1.18 nm. That leaves 35.82 nm which was covered in 8 minutes 47 seconds which works out to a speed of 244.69 knots.

    So at IGARI we have 471 knots at 35000 ft. From slightly after IGARI to BITOD we have an average speed of 244.69 knots and a drop in altitude from 35000 ft to a level below primary radar coverage after BITOD. Dave has estimated a drop rate of 1400 pm for the B777 and 25 minutes in the air from 35000 ft. That means at BITOD the plane would have been around 22,700 ft in altitude.

    If the plane had continued to drop at that rate, it would have crashed at 1.46am (putting aside the wind factor which was reportedly a North Easterly) .

  79. Alex Siew said,

    June 28, 2014 at 7:25 pm


    As shown in my previous comments, there is considerable evidence to show MH370 had crashed around 1.46am ( or slightly earlier given the prevailing NE head wind).

    That is what makes the report about the SOS call at 1.43am so intriguing. According to these reports which were all on March 8th and in the Chinese language with the news portals mainly based in Taiwan to my limited knowledge, the US Embassy (either in Taipeh or Beijing, not clear) reported that the US military base at Utapao at Thailand had picked up an SOS call from MH370 at 2.43am, with the pilots heard on the call yelling that the cabin was disintegrating and that they had to do an emergency landing ie ditch.

    At first blush, the time of 2.43am does not fit into an early crash timeline and the military base at Utapao might have been out of VHF range from where we think MH370 was at that time ( not far from BITOD).

    On reflection, as regards the time, the US 7th Fleet of which the base at Utapao is a part, is on Japan time, as it is based in Japan. Japan time is one hour ahead of Malaysian time, meaning 2.43am US 7th Fleet time was actually 1.43am Malaysian time or 17.43 UTC.

    As regards whether the Utapao base could have picked up the call, I note Duncan’s comments on his blog that radio signals could conceivably travel much further than its generally given range. Also there is the possibility the call was made on HF which has a much greater range than VHF.

    More likely however is that the call was actually picked up by a US ship at the South China Sea but the US Embassy attributed the pick up to the known base at Utapao to avoid disclosing the location of the ship at that point in time (quite likely in Vietnamese waters) and having to explain as to what the ship was doing there at that time.

    The USS Pinckney happened to be at the South China Sea at that time. ( The USS Kidd was also at the S C Sea but further away). The following is an extract from a media release from the 7th Fleet on March 8th, 9.13am:

    “USS Pinckney (DDG 91), an Arleigh Burke- class guided missile destroyer, is en route to the southern coast of Vietnam to aid in the search efforts of the missing Malaysia Airlines flight MH370, March 8.

    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 and carries two MH-60R helicopters equipped for search and rescue…”

  80. Alex Siew said,

    June 28, 2014 at 7:48 pm

    @JS, @ Skwosh,

    Back to the numbers, the BTOs and BFOs.

    I was wondering whether the fact that the distances between the ping radii are observed to be around twice the distance traveled by the satellite on the z axis for such corresponding ping times, could be because the distance traveled by the satellite was not taken into account (as Inmarsat had assumed a stationary satellite in their analysis) on BOTH SIDES of the equation.

    In the ATSB report, the equation is given as:

    “Range (satellite to aircraft)=C (BTO-bias)/2 – Range (satellite to Perth LES)”

    In other words, for any period of time, the satellite was moving further away from the plane ( crashed not far from BITOD let’s say) and at the same time moving the equivalent distance closer to the Perth ground station.

  81. Skwosh said,

    June 29, 2014 at 7:27 am

    @Alex – my short answer – and please don’t take this the wrong way – is no.

    As I say – at least so far – while I acknowledge the existence of the co-incidence, and find it disturbing/intriguing, I can not as yet think of any explanation for it either involving a stationary aircraft or indeed any other ‘non-conventional’ scenario. The problem for me is not the principle – that some oversight may have been made and/or that the equipment may have been operating in an unusual way – the problem I have is with the actual values and magnitude of the BTO numbers in any of the scenarios I have thought about so far – the variation in the BTO values that was actually recorded is just *too* *large* to be accounted for by any simple ‘non-conventional’ scenario I can think of in which the BTO could be attributed primarily to just the satellite’s position or motion.

  82. Alex Siew said,

    June 29, 2014 at 7:01 pm


    There is ‘more than meets the eye’ regarding the BTOs. It is stated at page 55 of the ATSB report:

    “Each power up sequence starts with a Log-on Request message which has been found to have a fixed offset of 4600us relative to the LLA message exchange by inspecting historical data for this aircraft terminal. The subsequent messages during the Log-on sequence have variable delay, and so are not helpful in this analysis. This means that the BTO data for 18.25.34 and 00.19.37 should be ignored, but that corrected BTO values of 12520 and 18400us may be derived from the Log-on Request messages at 18.25.27 and 00.19.29 respectively.”

    As u would be aware, i have been arguing that the SDU that stirred to life at 18.25 was the backup/slave SDU and not the master/primary SDU (the master SDU would have been knocked out with the rest of the avionics at 1.21am if the theory of MH370 being hit by positive lightning at 1.21am is correct). The BTO for the LLA messages at 18.25.34 and 00.19.37 which Inmarsat want us to ignore are 51700 and 49660 respectively. However the BTO for the LLA at 16.00.13, the very first entry in the 47 page Inmarsat data log document, is 14820 which is around the same as the subsequent messages. Two questions would arise:

    1. Why did the LLA at 16.00.13 have an apparent normal value but the LLA for 18.25.34 and 00.19.37 are way above trend, have ‘variable delay’ and ‘should be ignored’.

    2. Why did Inmarsat not disclose the first Log-on Request when the plane first powered up at KL (the first of the missing 6 lines), which BTO would show whether if indeed there was an offset of 4600us for this log on request as well.

    As i have pointed out in previous comments, all data that can show whether the SDU/antenna used to transmit the post 1707 UTC pings was the same as for the earlier transmissions, have been taken out from the 47 page Inmarsat document including:

    1. The first 6 lines, showing the first Log-on request when the plane was at KL

    2. The Rx (dBm) which would show the signal strength of the various transmissions (which would be lower for the later pings if transmitted using the backup SDU operating on battery power through the backup low gain antenna)

    3. The related values for C/No (noise) and BER (error rate) ( which would be different if the later pings came from a different SDU/antenna).

  83. JS said,

    June 29, 2014 at 7:09 pm

    I have a slightly different problem with the BTOs, which for the moment doesn’t address the moving/stationary debate.

    Based on my reading of the ATSB report, it appears to me that NEITHER an absolute time delay NOR a nominal or bias value was known prior to the incident.

    In other words, it is my understanding that the ground station was recording some numbers in the 5000-20000us range, but curiously rounded to 20us. Those numbers were subtracted from the calculated time for a signal based on the plane’s known location at certain times. The remainder was then used as the nominal or bias value. That value was used to convert stored BTO values at unknown locations into absolute trip times and ultimately ping rings.

    To me, this is like subtracting the number of dog barks from a car’s license plate and then adding it the next time the dog barks to predict the next car’s license plate. It can be done, and it may even be right, sometimes, but it’s a correlation without causation.

    I find it very odd that the value is stored as a multiple of 20. That strikes me as a manipulation to “fit” the model. I further find it odd that the absolute value was trimmed to save space – this was done post-Y2K, and by my calculation the saved space is about 12 bits per signal, vs. 24 bits untrimmed. Daily, this likely amounts to a few MB of data. This level of conservation seems odd for a feature tacked on since 2009, and especially considering once the plane lands safely the storage can be purged.

    But the most striking thing is that the method used at the time of storing the data – the act of subtracting some large value to get a small value – does not seem to be understood by Inmarsat. Are they really unaware of the presumed constant that is subtracted? They arrive at 495ms trip time as the nominal terminal, which is in agreement of my 497ms guess. But they had to obtain this, not from the source code that stored the BTOs they have, but from “calibrations.” They made it store the BTO as an offset. How do they not know where the number came from?

    Please advise if I have misinterpreted any key details here.

    As for the coincidence problem, I’d like to see what the correlation between the BTOs and the known locations looks like. I think I’d be quite suspicious if the correlation between known locations and BTOs was weaker than the correlation between BTOs and satellite position. Same for the converse. I’ll try this out.

  84. Bruce Lamon said,

    June 29, 2014 at 7:14 pm

    @Alex, what do you make of the BTO/BFO flight path verification claims in the ATSB report p. 31?

  85. Skwosh said,

    June 30, 2014 at 12:14 am

    I don’t think the presentation in the report necessarily means they *don’t* know exactly how their logging works – even if they do know the exact value of the (presumably) constant time that is being subtracted from the round-trip there would still be un-known (also presumed constant) time delays in the system that have to be calibrated-out using known positions, so from the point of view of keeping the presentation in the report simple and straightforward it is perhaps reasonable to just lump all these constants into one and gloss over the details of the logging process. I agree with your general concerns about correlation without causation though – so I hope very much that they *do* know what their system is actually doing when it records these numbers – or that if they don’t then someone takes the time to go and look at the code (surely they must have done this…?) and verifies that it is actually doing what it is expected to be doing. On the other hand, there is also the assertion that they’ve looked at the BTOs from other flights, and that these generate the correct distances – so, even in the absence of full understanding that is a very good check to have performed – though there is the subtle point that this does not guarantee that the timings for MH370 *must* therefore correspond to its true distances from the satellite as there seems to be an implicit assumption that the (latterly) recorded BTOs from MH370 were being generated by a-system-sate/equipment that was in the same condition of continuous/uninterrupted full working order as the corresponding system/equipment during the calibration/verification flights.

  86. Skwosh said,

    June 30, 2014 at 12:55 am


    1. I think the idea here is that different message types can take different amounts of time for the system to process before it responds, and that perhaps also some message types may take longer/variable times the first time they happen after a cold start (analogous maybe to it taking some time for a computer to load a bit of software after a re-start etc.) I agree that this does sound a bit like cherry picking – and it is very important to look out for cherry picking of data in these sorts of situations – but to me, in this case, it sounds at the moment both plausible and innocent. If I was in charge – and thankfully I’m not – and I had infinite resources – then, if it were possible, I’d have someone (or myself) try to dig into the hardware and the code and verify how and why this variability exists – if only because so much is resting on these numbers (maybe they *have* done this already?) Anyhow, it would also have been great if we also had the extra example of the initial log-on too, but…

    2. Maybe they just don’t have this information – is it possible the logging process has some sort of latency and misses the first few un-solicited transmissions when a completely new session starts? If that was true it would be a bit duff. However, having said that, the logging didn’t miss the initial log-on transmissions at the 18:25 – but I’m not sure to what extent the system had ‘given up’ on the session at that point because the session hadn’t ended properly – so it was probably still listening-out/logging at that time.

    I think that like you Alex – and probably many others (presumably also Inmarsat and the various official investigators) – we are all interested in the possibility that something may have *changed* commencing from the 18:25 log-on. I agree that if there were any signal strength information (or indeed any more information of *any* kind) it would be great if it could be released. I certainly remember from very early on there were reports of phrases like “these pings were very weak” – and it has been argued that if transmissions from the aircraft were from some sort of directional/’steerable’ antenna then weakness in these transmissions could be indicative of – say – some confusion in the aircraft’s state of knowledge of where it was and what its orientation was. However, I can also see why there may be reticence to release the full data as it may be that it is very susceptible to misinterpretation and that such misinterpretation might not be addressable without having to get into details of parts of the investigation that they don’t-want-to/can’t release into the public domain at the moment. I don’t know.

  87. Alex Siew said,

    June 30, 2014 at 7:23 pm

    @JS, @Skwosh

    Sorry if this is a silly question but i am wondering why there would be a BTO (ie an offset representing expected versus actual value) for the period when the plane was stationary at KL. The plane was not moving, the position of the plane was known. Any ‘offset’ could only have come from the movement of the satellite, assuming the model assumed a stationary satellite?

    And how does one reconcile the concept of ‘nominal position of aircraft’ or ‘terminal at nominal fixed location’ with the concept of ‘bias’?

  88. Alex Siew said,

    June 30, 2014 at 9:56 pm

    @JS, @Skwosh,

    Just to refresh everyone’s recollection about the sequence of events leading to Inmarsat’s northern southern arc analysis, here is an extract from a report from the New York Times on March 22nd :

    ” Chris McLaughlin, a vice president at Inmarsat, the satellite communications firm, said technicians pulled the logs of all transmissions from the plane within four hours of its disappearance. Then, after a day without the sign of the plane, they began scouring the company’s databases for any trace of Flight 370.

    ‘We decided to go have another look at our network to see if there was any data we had missed,’ Mr McLaughlin said. It turned out there was. Inmarsat technicians identified what appeared to be a series of fleeting ‘pings’ between Flight 370, a satellite over the Indian Ocean and a ground station in Perth, Australia.

    The signals – seven of them transmitted at one-hour intervals – were an important clue, because they could have come only from an antenna receiving power from the plane itself. But while they carried a unique code identifying the aircraft as MH370, the signals contained no positioning or other data that could indicate where the plane was when it sent them.

    By Sunday afternoon, a team of Inmarsat engineers set to work using the principles of trigonometry to determine the distance between the plane and the satellite at the time of each ping, and to calculate two rough flight paths. The plane, they concluded, had turned again. But it may have then traveled in more or less a straight line, heading north over countries likely to have picked it up on radar, or south toward the Indian ocean and Antartica……”

    The satellite, as we all know by now, headed north before making a turn (at 1936 UTC) and then traveled southwards in more or less a straight line, which would be in the direction of the Indian Ocean and Antartica.

    After several additional months of analysis, the conclusion remains basically the same, that the most likely path of the plane was a turn southwards with the plane by 19.41 traveling on more or less a straight line to the Indian Ocean/Antartica , as exemplified by the Analysis A in the ATSB report.

    Thus if Inmarsat’s analysis is correct, both the plane and the satellite made a turn at around the same time (1936-1941 UTC) and then traveled in the same straight line southwards towards the Indian Ocean for the next several hours, with the plane traveling 2 km for every 1 km traveled by the satellite and with the BFO of the pings tracking the velocity of the satellite at each ping time.

    Yes, it could just all be a coincidence, a mighty coincidence but a coincidence nevertheless.

  89. Alex Siew said,

    June 30, 2014 at 11:26 pm

    @Bruce Lamon,

    First of all, a note of appreciation for posting those photos of MH370 on Duncan’s blog. Those photos show conclusively that MH370 (9M-MRO) had 2 satellite antennas from the very beginning, a low gain shark fin shaped antenna on top of the rear fuselage and dual 45 degree side mounted high gain antenna above the ‘middle’ door ( can be seen as a patch in the photos with high resolution).

    This configuration, low gain on top and dual side mounted high gain, with each of the low gain and high gain antenna linked to a different SDU (ie 2 SDUs in total) is provided under ARINC 741 and is the first configuration shown in the Honeywell manual for the MCS series which incidentally does not show any single SDU configuration.

    Regarding your question, i note that the predicted path is ‘from BTO/BFO values’. I am of the view that a path can be predicted from BFO values which are well understood (provided the SDU is functioning normally) but the jury is still out as to what these BTOs really represent and i will reserve judgement until there is more clarity whether from Inmarsat or from others.

    Also u would recall we also had ‘predicted tracks’ when the Doppler analysis was announced on March 25th but 3 months later we are none the wiser about how these paths were calculated and what they really mean.

    If Inmarsat want people to believe that they are able to verify ‘path prediction analysis techniques’ they should disclose all the data for those flights and the calculations and not just show some diagrams which are meaningless without such data.

    Regarding the BFOs for MH370, the formula for the BFO values for the earlier transmissions (1707 UTC and prior) cannot explain the BFO values for the subsequent transmissions (the pings). As i understand it, the subsequent values reversed engineered would mean speeds above the maximum speed of a 777, an observation first made by Victorl who has theorized about a disabling event subsequent to 1707 UTC causing a break in the data path between the AIMS and the Satcom terminal, with the SDU possibly not getting inputs as to the altitude, heading and speed of the plane from that point onwards.

  90. Skwosh said,

    July 1, 2014 at 7:31 am

    @Alex – In my experience the term ‘offset’ can have a wide range of meanings in general technical usage – more or less anything that’s measured or expressed relative to something else. In the case of the BFO there was a lot of talk of ‘difference between measured and *expected*’ but I think that was specific to the case of the BFO – so the fact that the ‘offset’ in the case of the BFO was a measurement relative to something ‘expected’ just happened to be the case with the BFO – whereas with the BTO the offset is – sort of – a measurement relative to a fixed-position ‘nominal terminal’ and there isn’t really any kind of ‘expected’ BTO value at all, at any time (the whole notion of ‘expected-ness’ doesn’t really apply to the BTO). There seems to be a general agreement now that the simplest interpretation of what the BTO measurement was seems to check out very well – and gives a consistent answer compatible with a non-moving aircraft and a moving satellite during the initial pre-takeoff period when the aircraft was indeed static – so all of the (really quite simple) assumptions that appear to have been made about the BTO so far certainly seem to be self consistent – and – as I think it is important to always remember – are claimed to also work with measurements from other known flights.

  91. JS said,

    July 1, 2014 at 8:08 am

    @Skwosh -

    All good points. I agree that there’s a possibility they used a known value for the nominal terminal, and just used measured data to calibrate it. It’s also possible data from other flights was used. 

    However, the published explanation doesn’t really fit the first notion, and the data doesn’t fit the second. In fact, the data doesn’t fit much of anything. 

    First, if a known constant was used, and the “bias” represented equipment processing times, I believe that fact would have been emphasized in the explanation  in the ATSB report. Why lump the constant in with the variable, when the constant is roughly 495000 and the variable is on the order of +/-100?

    Anybody reading the explanation and understanding it also understands that the speed of light is constant. They aren’t obscuring anything here. The fact that they presented this as a single bias value suggests to me that they are being truthful and it’s the only number they have – that the nominal terminal is unknown. Furthermore, even if the data they showed was used as an example, the average bias of 495679us from that data seems to be the one that went into actual use.

    The story with the data gets worse. While they list 17 points, those 17 comprise of only 3 distinct locations. Somewhere in the list the plane was moving, so some of the points must have been estimated. In any case, the calibration was done on a mere 3 locations a mere 30km or so apart. No values are listed for a BTO from the plane a day earlier in a distant location, nor from any other planes. On the surface, the sole test of the BTOs for distance estimation consists of 17 points in roughly the same spot. 

    Finally, the data itself is awful. The correlation between the times and location is effectively ZERO: R_Squared=.000000002. The variation in the predicted distance is greater at a single location than between locations. The rounding by 20 is no help, either. There is nothing in these 17 points that should lead anyone to conclude that this information can be used to calculate ping rings for the remainder of the flight. 

    So I see two possibilities here.

    1. Inmarsat/ATSB published the worst data they had, but despite outcry over transparency, held back the good data that makes a compelling case for using these values to determine position, or

    2. Inmarsat/ATSB published the best data they had, and it makes no compelling case for distance determination but they did it anyway. 

    I don’t even know which one is believable. 

  92. JS said,

    July 1, 2014 at 9:39 am

    Skwosh – my reply was to an earlier post of yours. It is still largely applicable to your latest but if it appears slightly disconnected, that’s why.

  93. Skwosh said,

    July 1, 2014 at 2:53 pm

    @JS: Agree very much that it is odd that the explanation in the ATSB document does not introduce the concept of a separate component of the bias distinct from the “physical delays in the system” – it *does* strike me as strange, and this is partly why I am still worrying a bit about the interpretation of the BTO too.

    If all I’d ever seen was the explanation in the ATSB BTO appendix (so if I’d not seen mention of anything ‘nominal’) then I’d likely be wondering why on earth the BTO values don’t look anything like round-trip times, and why the ‘bias’ is so huge.

    However, on re-reading earlier bits of the ATSB document yet again, there is also this (p18 – so *not* in the BTO appendix bit):

    “For system efficiency and for the satellite communication to remain reliable, aircraft R-Channel transmissions are in time slots referenced to the P-Channel as received by the aircraft. The BTO is a measure of how long from the start of that time slot the transmission is received. This is essentially the delay between when the transmission was expected (given a nominal position of the aircraft) and when it actually arrives and is caused by the distance of the aircraft from the satellite (Figure 16)…”

    So – contrary to my missive to Alex above (apologies Alex!) – there is that word again – *expected*!

    I think the above is still basically compatible with the idea that they just record the round-trip time less a constant, but it has none the less thrown me into pondering again (a bit worried about the “…referenced to the P-Channel as received by the aircraft…” bit).

    I also take your wonderfully ruthless point about the lack of convincingness of the calibration data if it is taken in *isolation* from any physical interpretation of what the numbers are supposed to be measurements of – though to be fair they *do* make a clear assertion in the appendix of their physical interpretation of the BTO numbers (bias + light round-trip time) – but if it weren’t for the claim that the same analysis/calibration applied to other known aircraft tracks appears to give the right distances then I would be worried.

    I salute your reasoned caution/scepticism, and your discipline in applying yourself only to what we have *actually* been presented with so far, rather than what we (well – me anyway) may be assuming exists, but which hasn’t actually been presented/confirmed yet!

  94. Alex Siew said,

    July 1, 2014 at 10:02 pm

    Just some thoughts. There must have been something not quite normal about the raw data for the handshakes, for Inmarsat to have missed them when they first compiled and sent satellite data to SITA. The raw data for the last log-on request at 00.19 UTC (initially reported as a ‘partial ping’) was detected even later.

    All of which may suggest these bursts of radio waves could have been very weak or faint.

    I think it is a pity that unlike previously where there was pressure from various quarters for the release of the Inmarsat data which resulted in the release of the 47 page document, it has gone all quiet on this front. May i humbly suggest to the Independent Group not to let up on this and to continue to press for the release of the full data, both formatted and in raw form. The latter may be indecipherable to non Inmarsat staff but its release will at least serve to ensure the formatted data have not been fudged.
    Human nature is such that it would be perfectly understandable for Inmarsat to have released only the data that support their conclusions while withholding other data.

  95. andyo said,

    July 2, 2014 at 12:30 pm

    Been busy…

    A few quick thoughts on further parts of the thread.

    Even assuming the plane went over BITOD after IGARI, which is a relatively small distance, a turning to the right is at complete variance with all other ‘evidence’ presented to date (i.e. turned left and flew back over the peninsula).

    The problem I have with an emergency at IGARI/BITOD and a crash within an hour after that is that the pings kept on going for hours… Noone’s actually claiming that the pings aren’t from MH370 are they, so any crash site in the vicinity (150km) of IGARI/BITOD seems most unlikely. Also, sea searches occurred within hours around that sort of area yielding precisely zip. It seems too far fetched to assume a crash before 2a.m.

    BTO’s being quantized or divided (essentially the same thing). Quantizing actually means losing precision by discarding low order bits, whereas dividing does a similar thing (it can discard fractional bits), but shifts down the absolute magnitude of a number (quantizing doesn’t).

    Divided by 20 (assuming that they have been), BTOs become 8 bit (byte) values, which is essentially the smallest unit of storage in most systems (binary values are stored in single bits and usually packed 8 into a byte).

    The absolute magnitude of the BTO (offset) if divided by 20, would be nominally a 12 bit value. Noone stores things in 12 bit ‘fields’ if they can help it. Either you round UP to the next logical bit field size, so 16 bits or possibly even higher depending on the bus width of the actual system (32 bits wouldn’t be unreasonable). So a quantized BTO fitting 8 bits is ‘nice’ inasmuch as it’s a small value. It’s crap inasmuch as it’s rather imprecise – equating to 6km on a round trip (which corresponds to a greater distance on the surface of the planet).

    The absolute distances are in the order of 20 bits, so would logically be stored (and manipulated internally) as 32 bit values. This is why a) an offset is used and b) why a quantized offset might be used. To some extent, we’re lucky we’ve even got an 8 bit value to play with.

  96. andyo said,

    July 2, 2014 at 1:28 pm

    I’d forget my head if it wasn’t screwed on…
    Either you round up to 16 or 32 bits or you quantize down to 8 bits. If storage is an issue then you go down. It’s probably just a freeby as in there was probably an unused byte of record space available.

  97. JS said,

    July 2, 2014 at 4:12 pm

    Andyo – I agree with the theory you’ve presented, but it doesn’t entirely work here.

    The absolute minimum range of the BTO, assuming a Perth ground station, and allowing for a plane anywhere in the satellite’s visible hemisphere, is 45,000us or so. If we round to 20, that still requires storage of a number from 0-2250. That does not fit in 8 bits.

    I agree that nobody on the ground in their right mind would use 12 bits. I can’t say the same for a satellite launched 20 years ago, but these times seem to be handled entirely on the ground, and only since 2009. So these are most likely 16 bit integers, with 32 bit integers the next step larger.

    However, 16 bit integers WILL store the unrounded value range of 0-45,000. That’s the max variation in an “offset” value. Only by allowing for a MOVING ground station do we exceed the capacity of a 16 bit integer, and then only slightly. Any fixed location yields microsecond variations significantly less than 65,536.

    Meanwhile, the total trip time, without an offset but with rounding, will ALSO fit in a 16 bit integer.

    So, in summary, assuming the availability of either 16 or 32 bit integers, we can do any of the following:

    A) 16 bits, offset value rounded.
    B) 16 bits, offset value unrounded.
    C) 16 bits, absolute value rounded.
    D) 32 bits, absolute value unrounded.

    No 8 bit option is available, no matter what we do.

    So the question is, why choose option A, when better options B or C are available at the same cost, and D might be available at only slightly higher cost? The combination of rounding AND trimming makes no sense whatsoever. These are presumably not script kiddies writing code for a satellite – they should know the ranges they are dealing with and model accordingly.

    I’m not sure what to make of this, though, but for the time being, none of the public data supports the notion that these BTOs are anything meaningful.

  98. Alex Siew said,

    July 2, 2014 at 7:05 pm


    There is absolutely no evidence to show the plane had turned west or back from IGARI whether at 1725 UTC or otherwise. Perhaps u can enlighten me on what i might have missed.

    I did not say the plane turned right after IGARI or BITOD. The plane continued to be tracked from IGARI all the BITOD up to 1.30am on the primary radar of both ATCs, after which it disappeared from said primary radar. IGARI to BITOD was part of the original flight path of MH370.

    Contrary to what Professor Stupples said on the Horizon program, there was no ‘blindspot’ in terms of ATC primary radar coverage in the boundary area between Malaysian and Vietnamese airspaces. Professor Stupples was either ignorant of the ATCs’ primary radar capability and what both ATCs said they saw that night on their respective primary radar ( 1.30am/BITOD) or he was lying.

    Likewise the people who produced the Four Corners program were either ignorant about the plane being tracked by both ATCs’ primary radar to BITOD up to 1.30am, or they were deliberately disseminating false information.

    Even a 6 year old kid will understand that the location where and the time when a plane was last observed is of critical importance and would be the starting point of any search, if that plane happened to subsequently disappear.

    But in the case of MH370, all we hear, watch and read is this falsehood about the plane having disappeared at IGARI at 1.21am. That this is deliberate there can be no doubt, great care has been taken in the drafting of the Preliminary Report and the ATSB report for example, to qualify the time and place of disappearance by reference only to secondary radar.

    Yes, it is true the plane’s label (from SSR) disappeared from the radar screens at both ATCs at 1.21am but the plane remained on those radar screens as a radar blip until 1.30am all the way to BITOD, after which it disappeared from the screens.

    Vietnamese ATC called KLATC at 1.38am, 8 minutes later. Prior to that, after the plane had failed to radio in to Vietnamese ATC as part of the handover, Vietnamese ATC had repeatedly tried to contact the plane on 120.9 the ATC assigned frequency and on the emergency 121.5 frequency but without success. In addition to their own attempts at radio contact, Vietnamese ATC had also asked at least one other MH plane to contact MH370, MH088. The pilot of MH088 reported that he finally managed to make contact at just after 1.30am and heard mumblings from who he thought was the co-pilot (as well as static/interference) and on March 9th at 2.28am, slightly more than 24 hours later, the Malaysian authorities confirmed this contact by issuing a statement on the DCA MH370 website that ” …it has been more than 24 hours since we last heard from the plane at 1.30am…” .

    So the statements made on both the Horizon and Four Corners programs that there was a lag of 17 minutes before Vietnamese ATC acted, again were either made in ignorance or were intentional attempts to mislead.

  99. Alex Siew said,

    July 2, 2014 at 8:28 pm


    The brief ‘search’ at the South China Sea was shambolic and on the part of the Malaysians, based on a false premise.

    The head of Malaysian DCA, the person in charge of the search efforts in the first week, said on March 8th and in the days that followed that week, that the plane had disappeared from ATC secondary radar at 1.21am and from ATC primary radar at 1.30am. Yet from the outset, he gave the coordinates of the focus of the search as 6.55 N and 103.343 E which is ever so slightly just before IGARI (6.56 N 103.35 E):

    From the The Star, which together with the New Straits Times, the two leading (and government owned) newspapers in Malaysia, on March 8th:

    ” The last signal position of MH370 on the Dept 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 the reporters at the Coordinating Cenrte at the Sama-Sama Hotel here. He said Malaysian search and rescue operations for the missing plane would concentrate in a location off the Vietnamese coast with coordinates latitude 0655N and longitude 103343E…”.

    Why give coordinates of just before IGARI when the plane had passed IGARI at 1.21am and continued for another 9 minutes to BITOD?

    Could it be because Malaysia wanted to retain control of the search under applicable ICAO rules?

    IGARI for some reason is stated to be Singapore airspace(?) . Past IGARI would be Vietnamese airspace. Did Malaysia purposely give those coordinates so that the plane was seen to have disappeared while still in Malaysian airspace and thus giving control of the search to Malaysia?

    Thus from the outset the Malaysians did not want to know anything about BITOD and never searched there. By the next day, Sunday, we have word of the rumour that the plane had crossed back to the Straits of Malacca on the other coast and Malaysia started diverting resources there. If u look at the news reports for those few days, Malaysia did not seem serious at any time about searching the South China Sea.

    By Wednesday, the only other country allocating real resources to the search, Vietnam, had more or less called off the search, after days of reports in the press about the plane having turned back (Vietnam also complained bitterly that they had yet to meet their Malaysian counterparts to coordinate the search). From what i can gather, Vietnam also never searched the area to the north east or east of BITOD where the plane had likely crashed. The Vietnamese to their credit interviewed the Kiwi who said he saw a burning object and apparently were planning to search the area identified by the Kiwi, but then Malaysia announced the search at South China Sea was being called off (March 15th) and that was the end of the search for everyone at the South China Sea.

    Interestingly, USS Pinckney which was the first ship (from any country) to rush to the ‘vicinity of the missing jet’, did not take further part in the search after the search at the South China Sea was called off. Instead of going over to the Straits of Malacca where the search had continued, the ship docked at Singapore for ‘maintenance’ (The ship personnel also reportedly continued searching, while making their way to Singapore, despite the search at the South China Sea having been called off) .

    I would love to have a chance to speak to the people who were on USS Pinckney that night on March 8th about what some of them might have heard or picked up that fateful night.

  100. Alex Siew said,

    July 2, 2014 at 8:43 pm


    Or what certain of the ship personnel might have seen on the state of the art AEGIS AN/SPY-1D(V) multifunction radar with “enhanced electronic countermeasures and more effective capability in littoral environments”

  101. Alex Siew said,

    July 3, 2014 at 12:30 am


    Is it possible that the plane had crashed at the South China Sea soon after 1.43am and somehow the Satcom terminal continued to transmit those pings until 8.19am?

    The answer is yes, provided (a) the Satcom terminal and at least 1 Satcom antenna survived the crash (b) there was a source of power for the terminal to transmit those pings and (c) the Satcom terminal and surviving antenna remained or floated above water as part of the upper rear fuselage after the crash.

    1. The Satcom terminal on MH370 was fitted by Honeywell. From the link provided by Bruce Lamon on Duncan’s blog showing over a hundred pictures of the plane that was MH370, MH370 had both a low gain antenna on top of the rear fuselage (shark fin shaped) and dual side mounted high gain antenna located above the door (the one after the wings) between the middle and rear fuselage (shaped like a rectangular or square patch).

    2. The presence of 2 Satcom antennas means MH370 was fitted with a dual Satcom configuration. This dual configuration is provided under ARINC 741, the industry standard for Satcom terminals (first generation). Under this dual configuration, each of the antennas would be linked to an SDU, meaning 2 SDUs in total. In the Honeywell manual for the MCS series, the master/slave SDU configuration was described to be the ‘logical and potentially dynamic distinction’ whereby the slave SDU would be linked to the low gain antenna as a ‘standby backup system for low rate packet-mode data services, ready to take over as master in case the original master fails’.

    3. The low gain antenna acting as backup in case the high gain antenna fails, is expressly provided under ARINC 741. In this backup arrangement, when the high gain antenna or the master SDU fails, the slave SDU will take over and signals will be transmitted through the low gain antenna instead. In addition, under this backup system, when the switch is made to the backup, the signals will go directly from the backup SDU to the low gain antenna, without going through ie bypassing the High Power Amplifier and Beam Steering Units through the use of a ‘mute’ function. In the Honeywell manual, the Avionics Block Diagram is at page 1-19/20 as Figure 1.4 while the configuration is at page 2-22 as Figure 2.4, both involving 2 SDUs, a SDU no.1 and a SDU no.2.

    4. The Satcom terminal for Boeing 777s is located above the overhead luggage compartment at rack E11 (which rack is behind the wings) according to the manuals published online.

    5. The SDUs from Honeywell contain an internal battery or batteries which main purpose would appear to be to power nonvolatile memory.

    6. From literature on the net, the Satcom system is said to be connected to the ‘hot battery bus’ in addition to the normal AC or DC power.

    7. It is not yet known (to outsiders like me) whether in addition to the abovestated sources of power, the Satcom terminal on Mh370 also had a separate battery backup system whether on a standalone basis or as part of a larger integrated system with some other avionics. The pictures of the power inputs to the SDUs and the use of the term ‘primary power’ in the Honeywell manual indicate that the MCS series of SDUs can be linked to a secondary power source as an option. On this critical point, i quote from a commenter on Duncan’s blog also called Bill:

    ” 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”.

    8. It is now almost 4 months after the plane’s disappearance and u will note that the authorities have kept a deafening silence as to the details of the Satcom system on board MH370. No word even as to the manufacturer, let alone the model or the configuration. No description whatsoever in the Preliminary Report, the 47 page Inmarsat data log or in any of the ATSB’s reports. It is as if these details are taboo.

    9. I have on previous occasions pointed out that if the pings had been sent by the backup SDU operating on some sort of battery power and transmitted through the backup low gain antenna, the signal strength, the noise level and the bit error rate for these pings will be different than for those earlier transmissions (1707 UTC and prior) that were transmitted on regular AC power while the plane was still flying. Coincidentally or not, Inmarsat took out these data from the 47 page document.

    10. I have also pointed out that if those pings were transmitted by the backup SDU, the log-ons for this SDU would be on a different class (Class 1) as compared to the first log on when the plane was powered up at KL (Class 3). Coincidentally or not, the first 6 or so lines showing the first log on when the plane was at KL, have been taken out from the 47 page document.

    11. There are other signs that the SDU transmitting the pings was not the same as the SDU which transmitted the earlier transmissions. For example the LLA for the first log on at KL showed a normal reading while the LLAs for the 2 subsequent log-ons showed much above trend values which according to the ATSB report have ‘variable delay’ and ‘should be ignored’. The ATSB report also maintained the log-on request for 1825UTC had an offset of 4600us, we will just have to wait to see if the first log on request at 1600UTC, which has been taken out from the 47 page document, also showed such an offset.

    12. In previous comments in the main original thread on MH370 on this blog (the one with over 900 comments) i have laid out the evidence to show MH370 was hit by positive lightning at 1.21am just past IGARI, which caused the plane to suffer a catastrophic electrical failure. Briefly, a pilot flying in the vicinity reported seeing lightning in the area where MH370 was flying through. Lightning in clear or slightly cloudy skies, the prevailing weather conditions at that time, indicate positive lightning and a lightning strike at an altitude of 35000 ft again would point to the positive type of lightning which can exceed 300,000 Amperes as compared with the much more common negative lightning strikes which seldom exceed 20,000 Amperes.

    13. Commercial aircraft like the Boeing 777 are protected against lightning only up to 200,000 Amperes.

    14. The Boeing 777 is a fly by wire plane and although many of its systems have double or even triple redundancy, there is no redundancy in terms of the wires used to connect all these systems.

    15. The Boeing 777 was the first Boeing commercial plane to utilise significant amounts of composite materials. The cabin floor beams for example are made of composite materials. Composite materials while lighter and therefore cost saving in terms of fuel, are much less conductive than aluminium the original choice of material for planes.

    16. So what would have happened to MH370 if it was indeed hit by positive lightning of a power exceeding 200,000 Amperes?

    17. The wires would have been ‘fried’. In addition there would have intense electromagnetisation of electrical circuits and radio equipment. Anything that was operating at the time of the strike would have been knocked out. The secondary radar transponder, the ADS-B, the master SDU/ high gain antenna subsystem, the engines ( the latter as evidenced by the dramatic slowing of the plane to a gliding speed and the loss of altitude, between IGARI and BITOD).

    18. But MH370 had a backup SDU and a backup low gain antenna which were designed to come on only after the master SDU or high gain antenna system fails.

    19. This backup SDU did not come on immediately after the master SDU was knocked out. That is understandable given the ‘turmoil’ the plane would have been under from 1.21am onwards, not to mention the crash some time after 1.43am.

    20. The last ACARS was at 1.07am or 1707 UTC. At 2.07am or 1807 UTC, the satellite would have started pinging the plane.

    21. It is entirely plausible that the continuous pinging from the satellite ‘stirred’ the backup SDU to life which then attempted to log on at 2.25am or 1825 UTC.

    22. By some time after 8am, the battery power powering the backup SDU was nearing exhaustion and some time before 8.19am the SDU went off then revived briefly one last time at 8.19am before konking out for good soon thereafter.

    23. The scenario described above is consistent with the BTOs and BFOs. I have shown in previous comments the direct correlation between the BTOs and BFOs with the movement and velocity of the satellite for the period of the pings, which can only mean the plane was stationary at all such times.

    24. We shall know if indeed this is what happened, once the data on the signal strength, noise and error rate on the pings, as well as the data on the first log-on, are released.

  102. JS said,

    July 3, 2014 at 12:09 pm

    @Skwosh – I’m a day behind. Thanks for the encouraging comments. 

    @Alex – I agree that there is correlation between the BTOs, BFOs, and satellite position.

    I’d even go a step further, and suggest that the proposed flight path heading of approximately due south also corresponds to, guess what? The satellite’s heading at the same time. To my knowledge, nobody has yet suggested that 180 is a default autopilot heading, though the question has been asked repeatedly.

    But… these correlations don’t mean that the plane was stationary. At the moment, they don’t mean anything. They might, but so far there is no well-developed hypothesis explaining the correlation as anything but a coincidence. 

    The data and the correlations are suspect, but someone has to propose an explanation that people can get on board with. I don’t have one. My best shot is that the BTO is not logging the data they think it’s logging. 

  103. Alex Siew said,

    July 3, 2014 at 7:40 pm

    @JS, @Skwosh

    In the 47 page document: “The BTO is a value (in microseconds) relative to a terminal at a nominal fixed location. Only R-Channel messages are used”

    In the latest ATSB report (June 26,2014) at page 18:”…..R-Channel transmissions are in time slots referenced to the P-Channel as received by the aircraft. The BTO is a measure of how long from the start of that time slot the transmission is received. This is essentially the delay between when the transmission was expected (given a nominal position of the aircraft) and when it actually arrives and is caused by the distance of the aircraft from the satellite.(Figure 16). The BTO was only a relatively recent addition to the ground stations data set. It was added at the suggestion of the satellite operator following the AF447 accident to assist in geo-locating an aircraft”.

    At page 54 of the same report: ” The BTO measurement comprises two components: a bias component caused by fixed delays in the system, plus a variable component caused by the time taken by the outbound radio wave to pass from the GES to the aircraft and the inbound radio wave to make the return journey.”

    In ATSB report dated May 26, 2014: ” The BTO is a measure of the time taken for a transmission round trip (ground station to satellite to aircraft and back) and allows a calculation of the distance between the satellite and the aircraft.”

    In one of Tim’s early posts quoting GuardedDon: ” The GES transmits to the AES over the P channel and receives over the R channel. The initial response burst on the R channel is the timing datum transmitted by the AES [plus or minus] 300 us of receiving the incoming frame on the P channel”.

  104. JS said,

    July 4, 2014 at 7:13 am

    @Alex – I’ve read all of that several times. Taken at face value, it basically puts the plane in the SIO. However, if we think there’s something wrong with the explanation, that might allow an alternate location.

    The lack of clarity on the “nominal position,” the poor correlation of the data used to demonstrate the process, the rounding of the BTOs by 20, the rounding of the KLIA location by 5km or so, and the strange choice to limit the BTO storage to 16 bits but then waste most of those bits all has me at least considering that somebody botched this BTO logging.

    If it’s not botched, though, I don’t think there can be much disagreement over the ping distances.

    Let me hear your ideas if they differ.

  105. JS said,

    July 4, 2014 at 7:17 am

    Also note that the satellite position was estimated, to the nearest 5 minutes. While perhaps trivial, you would think they would at least calculate the exact position of the satellite for every handshake. That looks like laziness to me, but we have an awful lot of rounding going on.

  106. andyo said,

    July 4, 2014 at 8:46 am

    Agreed, missing data points don’t help the cause of robust explanation. If the missing data doesn’t fit the current explanation, then it’s perfectly reasonable to ask why not (and use theories based on different power supplies for example). First off, pre take off pings should show a static MH370. Discarded BTOs should be provided. Second, missing BTO data may support a different explanation.

    Still, for a crashed plane somewhere after BITOD whose tail section is floating and whose antenna is transmitting on backup power should still produce a bunch of intersecting ping rings. As I said before, it just ain’t so.

    Lots of estimates, roundings and guesses doesn’t mean the math is wrong, it just means the error is huge. It’s quantifiable though. Those ping rings have a very defined and ever increasing width.

  107. andyo said,

    July 4, 2014 at 8:47 am

    Also, 5 minute resolution on sat position isn’t very accurate, but it’s a relatively small inaccuracy.

  108. JS said,

    July 4, 2014 at 3:49 pm

    @andyo – agreed that the inaccuracies just cause bigger errors, but in the case of the satellite positions it also introduces the possibility of using the wrong data.

    Assuming the author(s) of the report had an equation for the satellite’s position, they chose to plug in only these 5-minute times, and those were copied over to the Table 2 calculations. If they DIDN’T have the equation, that means they got the positions from another source. In both cases, they’ve copied data, introducing the possibility that they have the wrong data. (For example, if they pulled the positions from another source, is it current?)

    If they had the correct equation, or an accurate program for calculating the correct values, why not simply use the exact times? Here we have a case where the accurate values require less effort than the inaccurate values.

    That points to the usual suspects of incompetence, confirmation bias, and/or obfuscation. My suspicion is that the “remarkably consistent” bias values, which already correlate very poorly with distance in the current sample, become “remarkably less consistent” when accurate values are plugged in.

    So while putting inaccurate values into the BTO-to-distance equation merely yields inaccurate results, if the equation itself is based on inaccurate values, it becomes completely invalid. Absent a test of the equation at some point other than KLIA, an equation from poorly correlating inputs is meaningless.

  109. Alex Siew said,

    July 7, 2014 at 8:39 pm

    The correlation can only mean 1 of 2 things:

    1. The aircraft was stationary during the pings ie the plane had crashed prior to the first ping, or

    2. It is just a coincidence.

    Assuming it is merely a coincidence and the plane somehow had continued to fly on for nearly 7 hours after it disappeared from ATC primary radar at around 1.30am after passing BITOD, so how?

    It should be obvious to anyone who has followed this saga of the BTOs and the BFOs, that it is impossible to pinpoint the crash site, just with the Inmarsat data. All models of potential flight paths need a starting point and we do not have a starting point. Is it really that productive to concentrate on such modelling when at the end of the day all we can probably say is that if the plane had been here at this starting point in time, then it could have ended at point A but if the plane had been there at this starting point in time, it could have ended at point B instead.

    Would it be more productive for those searching for the truth as to what happened to the plane, to first check out the possibility that the plane had crashed early on, before concluding that the correlation is just a coincidence. If the plane had crashed early on, the search zone would be dramatically narrowed down to a relatively small and shallow patch of sea at the South China Sea and chances of finding the plane would be relatively high.

    Regardless of whether one thinks the plane had crashed early or had flown on, everyone would agree that it is critical Inmarsat be compelled to release the data that they are withholding namely

    1. the missing first several entries relating to the first log on request when the plane was powered up at KL

    2. all the missing data fields for all entries

    3. the location of the ‘nominal terminal’

    There is no reason for Inmarsat to withhold such data ( other than to block the truth from coming out) so if the Independent Group working with the media can make the point that this data have not been released but should be released (similar to the campaign that resulted in the release of the 47 page data log), Inmarsat may have no choice but to release all the data (ie all entries with all data fields).

  110. Alex Siew said,

    July 8, 2014 at 9:09 pm

    Several people have commented on Duncan’s blog that the ATSB report omits any analysis of the period between 1822 and 1941 UTC, with the various scenarios outlined in the report essentially commencing from 19.41.

    Several people have theorized about where MH370 could have flown for the period prior to 1941 UTC, some say west, some say north west, some say circling, some say north.

    Prior to 1941 UTC ( 1936 to be exact), the satellite was traveling northwards before turning south at 1936 UTC. The fact that people are struggling to explain the flightpath of the plane prior to 19.41 UTC and that most models ‘require’ the plane to go west or north or north west first or to otherwise burn up time before a turn south at around the 1941 ping, is another sign that the BTOs/BFOs were merely reflecting the movement of the satellite.

    From 19.41 UTC, the satellite was traveling in a straight line southwards in the direction of Antartica and that is exactly the path that Chris Ashton and the other Inmarsat people concluded the plane had flown on, working from a stationary satellite model/assumption by the look of it, on March 10th after less than 2 days of analysis, as reported by New York Times.

    Thus we now have ‘truncated’ diagrams of flight paths, all starting from around 19.41 UTC, with a straight line southwards, on or close to a180 degree heading. The flight path of the plane for the period prior to 1941 UTC apparently is irrelevant to the analysis.

  111. Alex Siew said,

    July 9, 2014 at 8:54 pm

    This is going to be my last comment on this thread, so please bear with me.

    I quote from Sid Bennett, a member of the Independent Group from his comment on Duncan’s blog on July 8th at 10.48:

    ” One of the necessities in doing a dispassionate assessment of the data is to recognize where assumptions have been made and to question them until we believe that the possibilities have been exhausted”.


    That the plane was still in the air during the pings is merely an assumption. Yes we have an SDU from the plane responding to the pinging from the satellite until after 8am but those faint 6 or 7 hourly pings could have been emitted from the SDU via the low gain antenna even after a crash into the ocean, as long as the upper rear fuselage (where the SDU and low gain antenna were located) remained afloat after the crash.

    Those hourly handshakes are no different from the handshakes between cell towers and our handphones. As long as our handphones have some battery power from somewhere, these handshakes will keep going and none of us will even notice. Likewise, as long as the SDU had some sort of battery power from somewhere (and the upper rear fuselage stayed afloat), the SDU could emit these basic level hourly signals in response to the pinging from the satellite.

    So which is more probable, the backup SDU and low gain antenna surviving a crash with the upper rear fuselage remaining afloat after the crash OR a Boeing 777 with 239 on board continuing to fly on for 7 hours (the last 2 or so hours in broad daylight) in complete silence and undetected by all the air traffic controls of all the countries in the region all of which have primary radar and by all the military bases, ships and aircraft out there all of which also have primary radar, not to mention all the satellites in the sky and the various detection stations of the global nuclear organization some of which are based off Perth.

    To those who say the latter is more probable, please draw on a map the 7 hour flight path of the plane and ask yourself if it is at all possible that the plane could have flown so without being detected, especially if the theory is that the plane was flying on autopilot with all on board unconscious or dead from hypoxia.


    This is also merely an assumption as the changes over time in the BTO and BFO values could simply have been due to the movement of the satellite relative to a stationary ie crashed plane.

    To the Independent Group, please continue to question these 2 assumptions until the Group believes ‘the possibilities have been exhausted’. And if any of these possibilities cannot be ruled out without looking at the data being withheld by Inmarsat, please endeavour to obtain such data, in your collective quest to uncover the truth behind the disappearance of MH370.

  112. Bruce Lamon said,

    July 11, 2014 at 12:07 pm

    Alex, may I coax you out of retirement long enough to answer a question?

    “as long as the upper rear fuselage (where the SDU and low gain antenna were located) remained afloat after the crash.”

    What precedential or theoretical basis do you have for thinking this a realistic possibility all the way to 00:19?

  113. Alex Siew said,

    July 12, 2014 at 8:28 am


    I am a layman, i have no special expertise about what could or is likely to happen to a plane upon a crash or ditch into the sea.

    According to manuals online, the Satcom terminal for a 777 is located at rack E11, to the rear of the wings, above the overhead luggage compartment. The low gain antenna is on top of the rear fuselage. So basically the terminal is just underneath the upper rear fuselage with the antenna on top of that part of the fuselage.

    Could MH370 have survived the ditching more or less intact and in such form floated until after 8am? I do not think so.

    Could the plane have broken up upon impact and the portion of the rear fuselage where the Satcom terminal and the low gain antenna were located, remained afloat for several hours after the ditching? Yes and why not.

    In the hijacked Ethiopian Air Flight 961, a 767 crashed into the Indian Ocean albeit close to a beach. The plane broke up upon impact but the tail and rear fuselage did not sink but remained afloat in an upright position. Admittedly being close to a beach, the waters there would have been relatively shallow.

    In the Miracle on the Hudson, the plane remained intact and although submerged partially, did not sink. Again the Hudson river is shallow waters, and is just a river.

    The captain of MH370 was an experienced pilot, but ditching in the dark in open sea with the plane devoid of all electrical power and with the cabin already disintegrating if the SOS call is true, most probably would have resulted in the plane breaking up upon impact.

    The thing about the early crash theory is that it can be proved or disproved. If the plane was still flying during the hourly handshakes, the signal strength, noise and bit error rate for the pings would not be any different from the earlier transmissions as all transmissions from the start to end would be from the master SDU on primary power from the plane’s engines, with all transmissions having gone through the High Power Amplifier, Beam Steering Units and the High Gain Antenna.

    If the plane had crashed early, the only way the pings could have been transmitted would have been through the backup low gain antenna, with the backup SDU operating on battery power sending the signals directly to the low gain antenna ie bypassing the HPA and the BSUs (as the backup system was designed to do). The signal strength of the pings in this scenario would be significantly lower than for the earlier transmissions and the noise and bit error rate would also show a difference. Also, the log-ons after the crash would be on a lower level as compared with the first log on when the plane was powered up at KL, since the backup system is intended only for ‘low rate packet data’.

    The data on the signal strength, noise, bit error rate and the first log on were part of the original Inmarsat data log (shown on CNN I believe) but were later taken out and the 47 page document released in late May does not contain this data.

    So if this theory of an early crash is not true, Inmarsat can disprove it by releasing the data mentioned above.

    I should point out by the way that the BFOs and BTOs for the pings correlate to the movement of the satellite, which in my humble view, can only mean the plane was not moving during the pings ie it had crashed already. U may have noticed that not a single person has come forward to dispute the correlation, whether on this blog or on Duncan’s blog.

    Finally, a backup SDU operating on battery power and coming on only after the crash with the master SDU having been knocked out in the outage that occurred at 1.21am, would explain the behaviour of the SDU from1825 UTC onwards.

    The ‘two power interruptions’ theory in the ATSB’s report makes no sense whatsoever. Cheryl on Duncan’s blog, is not the first and will not be the last person ‘who does not get it’ about the ATSB’s theory. Apparently the ATSB thinks someone pulled all the plugs on the plane at 1721 UTC but reconnected the SDU and only the SDU and nothing else at 1825 UTC, just so that the SDU can do the handshakes for the next 6 hours. Perhaps this person who must have been very smart to be able to cause all the communication systems to cease at the same moment in time, had been rendered not so smart and hypoxic by 1825 UTC

  114. JS said,

    July 12, 2014 at 9:16 am

    Alex, you are correct that the correlation has never been discounted or explained, and it is glaring. The near perfect correlation between the satellite’s z position and the BTOs, along with the non-existent correlation between the BTOs and the plane’s distance on page 55, leads me to conclude that there’s something wrong with the numbers.

    I don’t believe it proves (or disproves) a stationary plane, though.

    My current theory is that the logging is missing the delay between the plane and the satellite. In other words, only the satellite to ground delay is logged. Is this possible? I’m not sure. It would depend on whether the “ping” really made a round trip, or whether the satellite has some ability to queue the message. For example, if the satellite acknowledged the ground station’s ping immediately, and then sent the ping to the plane, you’d have a ground-satellite delay that could be recorded. At one layer of the network model, you may have a single round trip, but at a lower layer there may be multiple trips.

    The fact that the ATSB has apparently withheld the “nominal terminal” value aused drives a brand new one for the example on page 55 suggests to me that a decent amount of manipulation is going on.

    As for the reboot, I’d like to point out that whatever the reason for the shutdown was, it had been overcome by the reboot time. That’s true for mechanical and human intervention.

    I see three possibilities for the reboot:

    1. The SDU was needed for its intended purpose (communicating or navigating) by an unauthorized crew, but never used or was used in a way we don’t yet understand.
    2. The SDU was used to cover the trail (yes, by spoofing location data.)
    3. The SDU was in the hands of a reclaiming, non-pilot crew that did not know how to properly turn everything on, but believed they needed it.

    There is a fourth, that I consider impossible: the original crew debugged an earlier problem and turned it back on. This doesn’t work for me – at that point they would have landed without it.

  115. Alex Siew said,

    July 12, 2014 at 8:06 pm


    A decent amount of manipulation going on is probably putting it mildly.

    The whole thing reeks of retrospective or after the fact self serving ‘analysis’. Instead of a simple matter of having a pre-determined formula and recorded absolute values for the pings delays, we have an ‘offset’ relative to a ‘terminal at a nominal fixed location’ where both concepts have not been explained. The points u made in previous comments are all valid, yet i do not know see anyone else whether from the Independent Group or otherwise, posting similar criticism. Why is that so? It has been more than 2 weeks already since the release of the ATSB’s report.

    So now we have sk999 giving one interpretation, Duncan another and Richard Cole yet another, of this nominal terminal. They cannot all be right and they may all be wrong. The only way to go forward is to demand from Inmarsat an explanation regarding this nominal terminal thing rather than forever speculating on what it might mean. This nominal terminal may turn out to be yet another red herring, just like the ‘possible turn’ postulated from the above trend BFO readings for the 1825 UTC log on request or the purported ‘last primary radar fix’ at 1822 UTC.

    There is more than than a little disquiet and even dismay among many following this blog and Duncan’s blog, that this collective effort to seek the truth as to what happened to MH370, seems to be petering out amidst endless mathematical equations. At the risk of incurring yet more wrath from the Independent Group, i hereby urge the Group to always keep in mind that many people are looking to them to lead the way. They have the name recognition, they have the scientific credentials and their voices will be heard, unlike others. They did a fantastic job in getting Inmarsat to release some data and they should take the next logical step of getting Inmarsat to release the rest of the data. Many are hoping to see a concise cogent statement from the Group pointing out all the deficiencies/inconsistencies of what has been released so far, as well as the data that has been withheld, followed up by appearances on media to press the case.

    I apologize if i have overstepped in any way.

  116. Alex Siew said,

    July 12, 2014 at 8:39 pm


    On the point about the reboot, i have to disagree.

    The fact is MH370 had 2 Satcom antennas, a high gain antenna (Aerospace Ball Airlink dual side mounted conformal HGA) and a low gain antenna on top of the rear fuselage (shark fin in shape). U can see both antennas in the pictures of the plane that was MH370 posted on the net ( see for eg the link provided by Bruce Lamon on Duncan’s blog).

    Dual antenna configuration means dual SDUs, with a primary SDU linked to the HGA subsystem and the secondary SDU linked to the low gain antenna which acts as a backup. This is provided under ARINC 741. In the Honeywell manual, the 2 SDUs are described to be in a master/slave configuration, with the slave SDU only coming on as a low level backup after the master has failed.

    Basically everything that required electrical power went off at 1.21am, for eg the secondary radar transponder, the ADS-B transponder, the Satcom terminal. Only one thing ‘came back on’, at 1825 UTC, an SDU. In previous comments, i have laid out the case that this SDU was the backup SDU operating on battery power and only at the most basic/backup level, transmitting those 6 or 7 ‘i am here’ signals to the satellite in response to the pinging by the satellite. The rest of the plane never revived, including the master/primary SDU and the HGA subsystem.

  117. JS said,

    July 12, 2014 at 11:33 pm

    @Alex – I wouldn’t worry about the wrath of the IG :-) At times, you (and I, certainly) may be outside of the framework within which they have developed theories. I don’t think they are opposed to outside ideas, but at the same time, everybody has to pick a starting point in their assumptions. Theirs, as far as I can tell, are based on the BTOs being valid, and the SDU compensating normally. They have also put considerably more mathematical effort into the problem than I have, so quite a bit of credit is due. And hey, this is Mr. Farrar’s blog, after all, so now is a good time to thank him for hosting it, even if only the contrarians are actively posting.

    Back to the reboot, I get what you’re saying, sort of. My characterization of it is based on the news reporting of the reboot as being intentional. That could be wrong, but if it’s wrong, you have an hour and 20 minutes or so of downtime to explain – if the backup kicked in, why not earlier? Presumably, you are not suggesting that it took a crash to fire up the backup power, are you?

    I think it’s very interesting – the reboot was being discussed here and on Duncan’s site days before the news report called it “new evidence.” I saw it myself in the CSV file – the glaringly long BTOs, among other things. One take on the news reporting it as new evidence, when it clearly wasn’t, is that many folks on blogs had basically stumbled on a very critical detail in the investigation. The reboot seems to be one center of attention in the investigation.

    It would be ironic, of course, if there was a sudden departure from the ping rings in the next few days, now that we’ve tried, and succeeded, at showing a correlation where one shouldn’t be, and a non-correlation where one should be. I won’t hold my breath.

  118. sk999 said,

    July 13, 2014 at 11:59 am

    The BTO turns out to be quite simple. It is given by equation (1) on p. 54 of the ATSB document. It is the round-trip travel time between Perth and the AES (wherever it may be on Earth) with a “bias” removed and quantized in units of 20 microseconds. What is the “bias”? It is a big number that is just small enough so that the BTO is always positive. Why units of 20 microseconds? If you look at the range of numbers that any possible station on Earth (up to 81 degrees latitude) can generate, it all fits neatly within 10 bits. Remember, Inmarsat is recording this value for all round-trip communications for which it can be measured, so keeping the storage down is important.

    Can we test this equation? Yes. Table 2 gives the BTO values for a half hour period while MH370 is parked at the gate. The tabulated BTO values actually increase by about 100 microseconds during this time, presumably due to the motion of the satellite. Since we know the position of the satellite (from data at celestrak.com) and the location of Perth and Kuala Lumpur (from, say, skyvector.com), we can predict what this change should be. I get 104 microseconds; Inmarsat’s table says it should be 103. Another test is to compare the BTO at 16:30, when MH370 backs out of the gate, and 17:07, the time of the last ACARS message. We know the location of MH370 at the latter time from both ADS-B data on flightradar24 and from Table 5 of the report. (I prefer the report, because flightradar24′s timing can be off.) I predict a BTO change of 690 microseconds over that gap. The observed change is 693.

    Inmarsat’s original attempt to explain the BTO in terms of a “nominal terminal” may have been done with the intention of trying to be helpful, but it just caused confusion. The ATSB report has clarifed the situation.

    Suppose MH370 went kerplunk into the ocean at BITOD at 17:30 and communicated using its backup transmitter for the next 5+ hours. Our BTO bias calibration would no longer be valid, but we could still predict how much the BTO would change over time. If one resests the bias so the BTO is 0 at 18:28, then the BTO should increase to 50 microseconds at 19:41 and slowly go down to -470 microseconds at 24:19. In fact, the tabulated BTO has the opposite behavior. It drops to -966 microseconds at 19:41 and climbs up to 5930 microseconds at 24:19.

    Conclusion: MH370 was never afloat in the South China Sea.

  119. Alex Siew said,

    July 13, 2014 at 4:20 pm


    On your interpretation of what the BTO represents, i am hoping Duncan and Richard Cole will provide a response, since they each seemed to have a different interpretation.

    I would however like to point out one thing, u are assuming the BTO formula for the earlier transmissions ( 1707 UTC and prior) applies equally to the later transmissions, the pings.

    As one of the people who regularly posted on the original main thread on MH370 on this blog ( the one with over 900 comments), u would recall i had quite early on suggested with regard to the discussion on the BFO that (a) the BFO readings were tracking the movement and velocity of the satellite during the period of the pings and thus the plane must have crashed by then, and (b) a different formula may be applicable for the earlier transmissions as compared with the pings, to explain the BFO readings.

    U would also recall i was repeatedly admonished by among others @airlandseaman who repeatedly maintained that the BFOs for the pings conclusively show that the plane was still flying during that time, and gave changing LOS speeds for the plane for the period of the pings.

    Fast forward some weeks later, starting with the observation by Victorl that the BFO values for the pings cannot be taken at face value as reversed engineered, the speed of the plane would exceed its maximum speed for some of the BFO readings. The main thread on MH370 on this blog more or less ended on a consensus that the SDU must have been operating on a degraded or default basis for the period of the pings (further to a disabling event prior to the first ping, as theorized by Victorl).

    After the main thread stopped being active, people migrated to Duncan’s blog. In due course, the observation was made by Rob, Henrik and Victorl among others, that the BFO readings for the pings were indeed tracking the satellite’s movement and that the plane’s movement does not show up in those numbers. That i believe is still the consensus, as recent summaries by Bill would indicate.

    So what was once a sure thing, that the changing and large values of the BFOs mean the plane was still flying, became much less of a certainty. The reason given for the absence of the plane’s movement in the BFO for the pings was that the plane’s Doppler had been somehow perfectly compensated. However, the absence of the plane’s movement in the BFO (for the pings) can equally be explained by a stationary plane ie the plane in fact was not moving during the pings.

    Now we have the BTOs. Once again we have values which change over the period of the pings and large changes in that, just like for the BFOs. One can be forgiven for feeling a little bit of deja vu.

    As was the case for the BFOs, the BTOs also appear to be tracking the movement of the satellite (please see previous comments from me and @JS on this, earlier in this thread).

    So for some reason, both the BFOs and the BTOs were tracking the movement of the satellite during the period of the pings. Only 2 conclusions are possible, it is merely a coincidence OR the plane was stationary during the period of the pings. The non mathematical evidence would suggest it is the latter:

    1. the power failure suffered by MH370 at 1.21am when all systems requiring electrical power ceased, the secondary radar transponder, the ADS-B transponder, the Satcom terminal etc (that MH370 had suffered a power ‘interruption’ is now acknowledged in the ATSB’s report).

    2. the further evidence of a power failure from the plane’s change of speed from 471 knots at IGARI to an average speed of 244 knots from IGARI to BITOD, which latter speed happens to be the gliding (ie unpowered) speed of a 777 (see my previous comments).

    3. the further evidence of a power failure from the plane’s loss of altitude from 35000 ft at IGARI to a level below primary radar coverage after BITOD ( taking Dave Whittington’s figures of a drop of 1400 to 1700 ft pm, the plane would have been around slightly below 20000 ft to 22000 ft then and bearing in mind Professor Stupples’ estimate of 6000m as the minimum height for primary radar coverage) (see my previous comments)

    4. the reports all on March 8th of an SOS call from MH370 at 1.43am picked up by the US 7th Fleet, with the pilots heard on the call yelling they had to land ie ditch. Again taking Dave’s drop numbers of 1400 to 1700, from IGARI at 1.21am MH370 could have stayed in the air for around 21 to 25 minutes ie the crash would be around 1.42am to 1.46am.

    5. The complete absence of any evidence that MH370 was still in the air thereafter, with zero radar evidence and not a single air traffic controller from Malaysia or any other country coming forward to say he or she saw a blip that could be MH370, crossing back over Malaysia or flying off somewhere else. They could not all have been sleeping on the job and even if they all were, the radar coverage would have been recorded and thus available for review after the fact. Contrary to popular belief, absence of evidence is evidence of absence.

  120. sk999 said,

    July 13, 2014 at 6:58 pm

    Alex Siew writes:

    “On your interpretation of what the BTO represents …”

    I am not “interpreting” anything. Equation (1) of the ATSB report says everything. Duncan wrote down exactly the same equation even before the report came out – some of us were just slow catching up.

    “… u are assuming the BTO formula for the earlier transmissions ( 1707 UTC and prior) applies equally to the later transmissions, the pings.”

    I do not assume anything. The BTO formula applies to ALL transmissions from SATCOM receivers and transmitters, because that is how Inmarsat records the data. The only thing in the formula that can change from one SATCOM system to the next is the value of the bias, and I made it very clear that I do NOT assume it was the same between the earlier and later transmissions.

    “… the BTOs also appear to be tracking the movement of the satellite …”

    The tabulated BTOs in the signal logs most definitively do NOT track the movement of the satellite. That was the entire point of my previous message, which Alex has chosen to ignore completely. When the satellite motion said the BTO should go up, instead it went down. When the satllite motion said the BTO should go down, instead it went up. Furthermore, the BTO changed by vastly more than what the satellite motion said it should.

    Let’s put it in more familiar terms. The round-trip distance recorded by the BTO during the time that, according to Alex’s viewpoint, the tail of the plane was a floating boat hazard in the South China Sea, should have changed by 97 miles due to satellite motion during the time after 18:28. Instead, the round-trip distance, as recorded by the BTOs, changed by 1284 miles. 97 miles I can drive there and back in a day. 1284 miles – that’s a 2-day trip just to go one way. They aren’t the same.

    The BFO is irrelevant to the present discussion of BTOs. The remainder of Alex’s post is a repeat of material already presented ad nauseam in over 40 posts of his in the comments section to this blog article and are also irrelevant to the discussion of BTOs.

  121. JS said,

    July 13, 2014 at 8:24 pm

    @sk999 – Alex may have his own “style” of posting, and I’m not onboard with the floating SDU myself, but some of your points, with all due respect, are unsupported by any data.

    The BTOs shown on page 54-55 absolutely do not correlate with known distances at the time. You picked one large value to make your conclusion, but the other valued defeat it. You would obtain better correlation if you threw 17 darts at a blank wall, because presumably your arm would get tired and the latter darts would diverge from the earlier ones. Worse, the satellite distances aren’t even correct – they are 5 minute estimates, and the airport location is rounded by 5km or so. You might say, well, this just increase the error, but if you don’t have any real correlation, your formula isn’t right to begin with, and the errors are irrelevant.

    As for the rounding by 20us, who uses a 10-bit integer? This storage was implemented in 2009 or 2010. The data is unneeded once a plane lands, as once it lands it’s location is known, and this was implemented to assist with MISSING planes. So realistically, the data can be dumped daily, as long as there is no plane crash in the news. The extra data logged is a few megabytes daily, stored on the ground, for ALL flights. So who in 2009 went to the extreme trouble to program the storage of a 10-bit integer? And, if they did, should we even trust the value from such an unconventional endeavor?

    Next, the 7 BTOs from unknown locations correlate SO well to the satellite’s position that one should wonder not whether the plane was stationary somewhere, but whether the ground station was logging the wrong values altogether. Otherwise, we have a coincidence in which the plane’s distance is closest to the satellite when it’s at it’s northernmost, and farthest when it’s at it’s southernmost, and so well correlated that one could predict a BTO within about 60us at all 7 points. Nobody has yet disputed or explained this correlation. But we have a jagged flight path that generates ping times that correlate almost perfectly with an elliptical satellite movement, and it’s just a coincidence?

    The key point that Alex and I seem to agree on is that the data used to demonstrate the calculation of the “nominal terminal/bias,” is about the worst data they could have chosen for that purpose. To me, they are either hiding or missing better data. Either causes me to discard the BTO values altogether, for now. We all asked for data and methodology, and we got an example using rounded numbers that show no connection to their known distance. We got a bias of 495,679us – derived from the junk data, but the real one, if anyone even bothered to calculate it, remains a secret?

  122. JS said,

    July 13, 2014 at 8:41 pm

    One more point:

    “When the satellite motion said the BTO should go up, instead it went down. When the satllite motion said the BTO should go down, instead it went up. ”

    This does not mean the satellite motion and the BTOs don’t correlate. It just means they have opposite signs. There could be many reasons for this – mostly an inverted sign somewhere or the use of an absolute value, or the squaring of a term. If the stored BTO was tracking satellite distance, it could be distance from anti-Perth, for example.

    In simpler terms, if you found a perfect correlation between the amount of pressure you apply to the gas pedal, and the car’s speed, but the car’s speed is going down, not up with more pressure, it doesn’t mean there isn’t a correlation or a causation. It means you’re stepping on the wrong pedal.

  123. Alex Siew said,

    July 13, 2014 at 11:39 pm


    About your question as to why it took the SDU more than an hour to ‘show up’.

    Under normal circumstances, the switch to the backup SDU would have been immediate. But if my theory of MH370 having hit by positive lightning exceeding 200,000 Amperes is correct, then the circumstances were anything but normal. The wires would have been fried and there would have been intense electromagnetization of electrical circuits and all things electrical and electronic.

    As an aside, MH370 was showing all the signs of having been zapped:

    1. A complete electrical failure

    2. The difficulty in establishing radio contact, and the static/interference heard the only time radio contact was established (at 1.30am with the MH88 on the emergency frequency)

    3. What appeared to be St Elmo’s fire on the plane, observed by the Kiwi on the oil rig

    4. The pilots of MH370 heard on the SOS call yelling that the cabin was disintegrating ( the cabin floor beams of the 777 are made of composite materials and thus be at risk of shedding from a particularly powerful lightning strike- see the 1999 glider plane incident where the plane made from composite materials disintegrated in mid air after getting hit by positive lightning).

    The SDU transmitted a log on request at 2.25am but it is not known when this backup SDU, which had an internal battery or batteries, actually ‘came on’ (when the switch to the backup system occurred). What could have prompted this log on request at this time?

    The last ACARS transmission was at 17.07.48 (1.07.48am). One hour later more or less, at around 2.07am, the satellite would have started pinging the plane. I believe the repeated pinging from the satellite ‘activated’ this backup SDU and eventually prompted this log on request at 2.25am. In previous comments, i have set out the potential sources of power for the SDU to be able to transmit these basic level 6 or 7 pings in response to the satellite pinging.

  124. Alex Siew said,

    July 14, 2014 at 12:11 am


    I omitted to mention that there is a gap in between 18.05.59 and 18.25.27 UTC in the 47 page Inmarsat data log. I believe there would have been ‘handshake requests’ from the satellite prior to 18.25, similar to the handshake requests later at 19.41, 20.41, 21.41, 22.41, 00.10, 01.15, 01.16 and 01.16.

    As a comparison to the Inmarsat data log, the investigation report for Swiss Air Flight 111 contains (in a side technical report) a detailed description/explanation of the transmissions from the Satcom system (also a Honeywell SDU) for that flight.

  125. JS said,

    July 15, 2014 at 12:54 am

    Alex – would you mind restating the backup power available to the SDU? I’m struggling with this. I can understand it having enough power to protect it from switches between engine and APU and shore power, but that’s momentary. Does it have a battery designed for 6-8 hours of no power, and if so, for what purpose? It would not be expected to survive a crash, like (maybe) a beacon, and certainly nobody would expect the plane to be flown for 6 hours without electrical power. So I can’t see this as part of the design, but accidental, if this is possible at all. What level of battery power are we talking about?

    The one thing that makes me think of a hijacking here is a comment from Cheryl #1 over on Duncan’s blog. I’m not sure if she was getting at exactly the same point, but the way I read it was that the disabling of the inflight entertainment map was the iintended goal.

    In order:

    1. Crew says goodnight to KL and leaves Malaysian airspace and radar range.
    2. SDU is shut down to disable the seatback map, so passengers cannot detect a deviation in the course.
    3. Plane crosses peninsula at a narrow spot with spotty radar.
    4. Plane heads for the nearest radar-free area.
    5. Unknown operators ensure that reactivated SDU either does not enable seatback maps, or is otherwise in control of passengers.
    6. SDU is needed for some unknown purpose, and is restarted once plane is clear of radar.

    The fact that the SDU was needed, I think, whether its shutdown was intentional or otherwise, is very interesting, and I think more considering that a statement about it hit the news.

    The shutdown/reboot at the fringes of radar coverage is highly coincidental – enough for me to think they were intentionally timed. On that subject, I know there is at least some doubt about the radar sighting at 2:22 or so. But as I understand, there was a turn prior to that at a waypoint that was consistent with a plane coming from MH370′s location at 1:21, and not consistent with other flights like the Emirates or Singapore flights. So at least some of the radar is undisputed and puts MH370 over the Strait, no?

  126. sk999 said,

    July 15, 2014 at 4:05 pm

    Hi JS, You write:

    “The BTOs shown on page 54-55 absolutely do not correlate with known distances at the time.”

    I am unclear what you mean. Those BTO’s come at a time when MH370 is parked at the gate (or possibly being pushed back but that would only be a short distance). There is a large (unknown bias) that must be removed from the data, which means we cannot know the absolute position of the gate, but once you do that, there is a very clear trend of increase in BTO with time. It was before the plane began taxiing (which we know from the ATC transcripts.) Thus, for our purposes, the plane was at one location this entire time. Why the increase in BTO? The satellite was moving. We know where it was. We know the location of Kuala Lumpur. We know the location of the ground station in Perth. We can calculate the impact of that motion on the BTO. They match. Not a very accurate test, but enough to show that something happened, and we understand why it happened.

    “Worse, the satellite distances aren’t even correct – they are 5 minute estimates, and the airport location is rounded by 5km or so.”

    I don’t use the satellite distances in the table. I calculate my own from first principles based on two independently determined sets of orbital elements. I also do not use their airport location – better values are easily found elsewhere. The plane won’t be at that exact location, but it will amount to a fixed offset, which will be absorbed in the bias value. (In fact, I don’t use anything from that table at all! I take the BTO values from the Signal Logs and other information from various non-Inmarsat sources.)

    “As for the rounding by 20us, who uses a 10-bit integer?”

    This was a bit of conjecture on my part and ancillary to the rest of the BTO discussion. It’s not that they would use 10-bit integers, but rather that it would be a 10-bit field embedded with other information (e.g., channel number, etc.) inside a 16 bit or whatever value. Who knows – the BTO measurements were retrofitted to an existing system, and maybe there was an existing 10-bit field that was obsolete or unused, so the BTO could be added without worrying about changing anything else. Changing a system is WORK, and changing an existing, functioning 24×7 system so it doesn’t break is a lot of work. I could be wrong.

    “And, if they did, should we even trust the value from such an unconventional endeavor?’

    Why not? Just because they don’t do things the way you would?

    “Next, the 7 BTOs from unknown locations correlate SO well to the satellite’s position …”

    Could you quantify that? I did the calculation and didn’t see any correlation at all.

    “Otherwise, we have a coincidence in which the plane’s distance is closest to the satellite when it’s at it’s northernmost, and farthest when it’s at it’s southernmost,”

    The BTO is the round-trip between Perth, the satellite, the AES, back to the satellite, and back to Perth. The Perth-satellite distance dominates and is getting longer while the satellite-aircraft distance is getting shorter. Your “correlation” is backwards.

    “The key point that Alex and I seem to agree on is that the data used to demonstrate the calculation of the ‘nominal terminal/bias,’ is about the worst data they could have chosen for that purpose.”

    Really? I did a statistical analysis of those data. After removing a constant bias and the satellite motion (which is calculable and also clearly present in the data), the rms. of the residuals is 29 microseconds, with a kurtosis of 1.8 and negligible skew for all 54 data points in the logs from before 16:30. The worst residual is 2.5 sigma. That’s an exceedingly well-behaved data set. If only they could all be that way! Or perhaps you have some magical statistical test that shows otherwise?

    “But as I understand, there was a turn prior to that at a waypoint that was consistent with a plane coming from MH370′s location at 1:21, and not consistent with other flights like the Emirates or Singapore flights. So at least some of the radar is undisputed and puts MH370 over the Strait, no? ”

    That is very much my understanding as well. I have read (not to be trusted) that 3 radar saw MH370. We have read of at least two – Thai, Butterworth. These are military radar – understandable that we have gleaned little information about or from them. The third must have been the ATC primary radar at Kota Bharu. A civilian radar. Not enough range to reach IGARI, but plenty good enough to see a plane coming back practically overhead. Or perhaps it wasn’t functioning that night (a rare event but one to be expected from time to time.) Thoughts?

  127. JS said,

    July 15, 2014 at 8:56 pm

    Slightly out of order, but let me try to answer some of your comments. Thanks for replying, btw, and with clarity.

    10 bits – It’s conjecture on my part that nobody who knew what they were doing would choose a 10 bit integer, I suppose. But the value is being timed on the ground and stored on the ground, so it’s my opinion that it would be unnecessary to store it across byte boundaries or inside another record. Yes, maybe there was space at the end of the record full of Boolean values and only 1 byte and 2 bits from another remained. I have a hard time believing that.

    Now for the 17 BTOs. As I understand, BTO + bias = Round Trip Time, or maybe 2BTO but it shouldn’t matter w/r/t correlation. Based on that equation, a 20us increase in total time should always yield a 20us increase in BTO. If I try to plot BTO on one axis, and total time on another, they should form a line, because they should correlate perfectly, since they are part of the same equation. If I do a simple least squares, I should get an R_squared value close to 1. I don’t. I get .0000000something. Admittedly, the sample is small, but that should further disqualify these 17 points as calibration/validation for the BTOs. Indeed, it appears most of the IG is using a bias around 495,636, not the 495,679 that the ATSB published. Feel free to correct me here, because my stat background is awful, but based on these 17 values, it doesn’t look like we can say BTO correlates with total trip time. We could accept errors, but with no other data we could also say that there’s no connection. (I fully understand that these are claimed to be offsets, and if they are indeed, then it’s just a matter of finding the error. But I’m suggesting that these are erroneous numbers altogether, not valid numbers with a quantifiable error.)

    As for the other 7 BTOs, I *can* fit them to a line, as the dependent variable, with the satellite’s Z coordinate as the independent. The R_Squared is near 1. I don’t have an explanation for this, but as I’ve been saying, we have a plane flying a jagged route over time, but generating BTOs that fall pretty neatly on a sine curve with the same characteristics as the satellite’s orbit. It may not have anything to do with Z directly, but for example, the satellite’s speed, which is itself coupled to the Z coordinate. The BTOs are lowest right around the time when the satellite’s relative speed (relative to 0N 64.5E, or possibly Perth, but not the plane) is lowest.

    So I guess my next questions for you are:

    1. Can you fit the 17 BTOs to a line, as a dependent variable of total round trip distance?
    2. Can you NOT fit the 7 BTOs to a line with some aspect of the satellite’s motion of position along the Z-axis as the dependent variable?
    3. Do you believe that either 1 or 2 are unnecessary?

  128. Alex Siew said,

    July 15, 2014 at 9:13 pm

    @JS, @sk999

    According to the Preliminary Report, MH370 passed over IGARI at 1.21.04 and the secondary radar signal from MH370 ceased at 1.21.13.

    There is no evidence to show MH370 had turned back or turned west from IGARI. The evidence instead shows that from IGARI MH370 had continued on its flight path to the next waypoint BITOD which is 37nm away, although at a much reduced speed ( around 244 knots) and losing altitude at all times ( around 1500 ft pm), reaching BITOD at around1.30am, after which it disappeared. This is the evidence from the air traffic control at KL and the air traffic control at Ho Chi Minh City, both of which had continued to track MH370 on their respective ATC primary radar those critical 9 minutes, from 1.21am to 1.30am, all the way to BITOD. I have in a previous comment on this thread addressed to @Skwosh, on June 23, 2014 at 2.50pm, set out the statements from the Malaysian DCA and from their Vietnamese counterparts, on what they saw that night on their ATC primary radar. Please do go and take a look at those statements.

    MH370 could not have turned west off BITOD either after 1.30am and still be the blip at MEKAR at 2.22am as that would mean it would have to fly above the maximum speed of a 777.

    This blip at 2.22am at MEKAR has been shown on this blog, on Duncan’s blog and on PPRUNE to be another plane, UAE343, including by Don Thompson, an unnamed member of the Independent Group. I have reposted Don’s analysis/comments on this issue in a previous comment on this thread on June 26, 2014 at 10.27pm. Please do take a look at what Don had said regarding this blip.

    There were many airports/ATC in the border region between Thailand and Malaysia. Not a single air traffic controller has come forward to say he or she saw a blip coming off from IGARI or BITOD and crossing over Malaysia to VAMPI or MEKAR or whatever.

    Not a single radar track or recording has been produced to show any blip off IGARI or BITOD, or any blip crossing over Malaysia.

    There is simply no evidence to show MH370 had ever turned west or back.

  129. JS said,

    July 15, 2014 at 9:32 pm

    Ok, sorry to double post, but I have to point something out here:

    “The BTO is the round-trip between Perth, the satellite, the AES, back to the satellite, and back to Perth. The Perth-satellite distance dominates and is getting longer while the satellite-aircraft distance is getting shorter. Your “correlation” is backwards.”

    Your statement is based on the assumption that the BTOs are in fact what they are claimed to be. In a sense, I find that somewhat circular, because you are stating that my correlation is backwards based on your definition of the BTOs, but it is precisely that definition that I am challenging. So yes, my correlation is backwards, but only because you’ve framed it within an assumption that may be backwards or flawed.

    A backwards correlation, though, is still a correlation. It is not necessary that the sign of the dependent variable be the same as the sign of the independent variable – the regression line can have a negative slope, no? It may make explaining the correlation harder, but it doesn’t invalidate the correlation. In some cases, the sign inversion may be an artifact of the coordinate system – it could be “degrees latitude in the hemisphere of the ground station.” It could be a reversed Z coordinate system. Supposedly, Inmarsat had the wrong latitude for Perth, so there would be some precedent for sign reversals here (especially… if the sign reversal was done to compensate for an Anti-Perth error.)

  130. Bruce Lamon said,

    July 15, 2014 at 9:57 pm

    Once I looked into the commercial flight schedules for the civilian airports to which MH370 might have tried to divert, including Kota Bharu and Langkawi, and found that the latest scheduled landing or departure at any of them was around 2300 MYT. From that I speculated that by 0130 those airports were probably closed and their primary and secondary radar had been turned off for the night, and that’s why there was no identification of MH370 while crossing the peninsula or graphic published of suspected MH370 blips like there was for the Straits.

  131. Alex Siew said,

    July 16, 2014 at 12:23 am


    On the point about the battery.

    It is disgraceful 4 months after the plane disappeared and with the search at the Indian Ocean premised solely on the pings, the authorities still have not said anything about the Satcom equipment that emitted the pings. I have in previous comments made the case that this non disclosure is intentional.

    1. From the pictures of the plane that was MH370 on the net, we can see the plane had both a high gain antenna and a low gain antenna. The HGA was an Aerospace Ball Airlink dual side mounted antenna, located above the door between the wings and the rear, visible as a patch in the pictures with higher resolution. The LGA was located on top of the rear fuselage, shaped like a fin.

    2. Dual antennas usually means dual SDUs. According to literature on the net, Honeywell is said to have developed a ‘dual integrated ACARS system’ for the 777. Also ARINC 741, the industry standard for Satcom terminals, expressly provides for the option of installing an LGA as a backup to the primary HGA subsystem. The Honeywell manual for the MCS series 4200/7200 only provides dual antenna dual SDU configurations, with the SDU1/HGA + SDU2/LGA configuration described in the manual as the ‘original dual ARINC architecture’. Finally, there was a press release from Honeywell some years back saying that the Satcom equipment on Malaysia Airlines B 777 was fitted by Honeywell. So all indications are that the Satcom terminal on Mh370 was from Honeywell and was comprised of 2 SDUs. If i have to guess, i think the SDUs were from the MCS 4000/7000 series.

    2. In the Honeywell manual, the master/slave configuration for the 2 SDUs was described as the ‘logical’ configuration whereby the slave SDU would only come on after the master SDU or the HGA has failed. This backup system is also provided by ARINC 741. Basically when the primary system fails, the backup system takes over and in such a situation, the signals go from the slave SDU to the LGA, without having to go through the High Power Amplifier and the Beam Steering Units (which are bypassed through the use of a mute function). Which would mean (at least to me) that the power required to transmit the pings would have been significantly less than for the normal higher level signals transmitted the usual way through the master SDU-HPA-BSU-HGA.

    3. The SDU for the MCS series contains an internal battery. See for eg at page 4-5/4-6 Figure 4-2 of the Honeywell manual for the MCS 4200/7200 series.

    4. However the manual does not give any details for this battery (other than to indicate the battery has to be replaced within a certain time) and i have not been able to find anything on the net about this battery.

    5. Could this battery have been the source of power for those pings? The onus is not on me to prove that this battery was the source of power for the pings but the onus is on the authorities which have concluded solely from the pings that the plane was still flying, to prove that the pings could not have been so powered.

    6. Apart from this internal battery, there could have been other sources of power for the SDU, even after a crash into the sea.

    7. Firstly, the ACARS system is said to be connected to the ‘hot battery bus’. The 777 has 2 backup batteries, the main one up front and the APU backup battery at rack E10 just next to the where the Satcom system is located. Being a layman, i do not know how everything would have been hooked up and i can only hazard a guess as to whether all these connections would have been severed if the plane had broken up upon impact (which it most probably did). I do note however that batteries can work under water.

    8. In addition, there is the question whether the Satcom system had a battery backup arrangement in the nature of an UPS, whether on a standalone basis or as part of a larger group of avionics, and if so, the location of this external dedicated battery. This is a legitimate question given the redundancy requirements for avionics and the need for ‘transient protection’.

    9. It is clear from the Honeywell manual, both from the use of the term ‘primary power’ in the context of power supply and from the pictures of the power inputs to the SDU, that the MCS series provides the option of hooking up the SDU to secondary/redundant sources of power.

    10. As noted earlier, the authorities have kept mum about the Satcom equipment. They have also taken out from the original data log, the data on the signal strength, noise and bit error rate. The signal strength in particular would show whether the pings were transmitted on regular AC or DC power generated from the plane’s engines which would mean the plane was still flying then, or from a secondary much weaker source of power which would indicate otherwise.

    11. Finally, please also note the following (a) the pings were reported to be ‘faint’ or ‘fleeting’, (b) the pings were missed by Inmarsat when they first compiled and forwarded the satellite data to Sita and were only discovered a day later after Inmarsat staff reportedly went back to take another look at the data, and (c) the last log on request at 8.19am, initially reported as a ‘partial ping’, was discovered even later, all of which suggest the pings/log on requests were of a signal strength much below normal.

    12. Will the authorities be so helpful as to release the withheld data any time soon? To use the same expression as u did, i won’t hold my breath.

  132. Alex Siew said,

    July 16, 2014 at 1:36 am

    @Bruce Lamon

    Bruce, there were other airports which would have been opened for eg Penang in Malaysia. In addition, there were military bases in that reqion which would also have primary radar for eg the Five Nations Base in Butterworth. If MH370 had flown back across Malaysia these primary radar would have picked it up. Yes, some people may have been sleeping but surely not all. And even if some had been sleeping on the job, as long as the radar is switched on it would have been picked up and recorded for after the fact review.

    Also, both ATCs in KL and HCMC were tracking the plane on their primary radar until it disappeared at around 1.30am. A plane disappearing from radar firstly from SSR (at 1.21am) and then PSR (at 1.30am ) and also not radioing in or responding to radio calls for 9 long minutes (until the contact with MH88 at 1.30am), would have been a rare occurrence and both Malaysian and Vietnamese ATCs would have been on the lookout for the plane. By 1.38am the 2 ATCs were already on the phone. So it is inconceivable for KLATC to have missed the plane turning back and then crossing over Malaysia unless the plane had flown below primary radar range for the entire trip from BITOD to MEKAR. Penang was and is a big town, a lot of people would still be out in the streets at 2.00am on a Friday night and MH370 could not have flown at low altitudes past Penang without anyone noticing.

    Other than the blip at MEKAR at 2.22am, there is nothing. No one really believes that this blip was MH370 but no one wants to come out and say it is bogus because without this blip, the people doing all this modelling would not even know where to start.

    Can i take this opportunity to ask u and LGHamiltonUSA something. It is stated in the Preliminary Report: ” It was later established that the transmissions from the [ACARS] through satellite communication system occurred at regular intervals starting before MH370 departed Kuala Lumpur, Malaysia at time 12.56.08 MYT and with the last communication occurred at 01.07.49 MYT”

    What does the time ’12.56.08 MYT’ signify to u?

    Could it be that this was the time of the log off from the previous flight? Would u know what previous flight could have such a log off time?

  133. JS said,

    July 16, 2014 at 9:08 am

    @Alex – wasn’t that it’s local departure time? Only wilh a 12 instead of 00 or 24? I read it as saying signals were received at regular intervals, starting before its departure time of 12:56am. I don’t read it as saying signals were received prior to departure, all the way back to 12:56pm. Although both are probably true anyway, I think they mixed up the time and am/pm.

    My key question about the battery system is “how long does it last?” I’m aware that it existed. There is no “business case,” so to speak, for battery power to an SDU beyond a few minutes. Longer times would require either more weight or more expensive replacements. I’m not saying its impossible, but it’s unlikely.

    I think you already have your answer, though. In effect, you are asking “are you sure the batteries wouldn’t last 6 hours?” Your target audience would be the manufacturer(s). They are apparently quite sure at this point. Now, they could be wrong, but you are asking them a question they already think they know the answer to. You won’t get a different answer from them, even if they are wrong. Nor will answering it be high priority, since even from a lay perspective, there’s no logical basis for batteries lasting so much longer than necessary.

    To go further with the floating SDU theory, you should probably be getting answers from independent sources. What model is the battery? What is the expected draw of the SDU in an alert but not transmitting state? If you had numbers at least approaching 6 hours of battery life, then at least the question might get a second look.

  134. Bruce Lamon said,

    July 16, 2014 at 12:17 pm

    Alex, I am sorry especially in view of your responsiveness to my questions (thanks!) to have no answer to yours about 12.56.08 MYT. If I had to guess, I’d agree with JS. There was R-Channel communication with the satellite at that time. http://www.dca.gov.my/mainpage/MH370%20Data%20Communication%20Logs.pdf at 29.

  135. sk999 said,

    July 16, 2014 at 6:28 pm

    JS writes:

    “…based on these 17 values, it doesn’t look like we can say BTO correlates with total trip time.”

    The R-squared test is easy to calculate but doesn’t say anything about the significance of a correlation. I never use it. A Spearman rank correlation test is far better. By the way, there are 54 data points in this time interval, and Inmarsat picked out only 17 (presumably to keep the table reasonably sized). Running the Spearman test on the full set of 54, I find a correlation coefficient of around 0.5 (not very big) but with a probabilty of only one in a million that it is due to chance. The correlation is very real. The data are not good enough to show that the correlation is linear with time, but they are consistent with it.

    Regarding correlations of the later BTO points with satellite Z position, since the BTO measures the timing of a signal going through the satellite, there will always be a correlation between BTO and the position of the satellite as it moves. It is the line-of-sight distance that we really want. Let’s address the real question – how much did the distance from the satellite to BITOD change during the last 7 data points (18:28 to 24:19)? Here’s the answer – 3 km, 20 microseconds in the round trip. 2 miles. It had negligible impact on the BTO. One can do the calculation with pencil and paper and a pocket calculator.

    Alex Siew writes:

    “This is the evidence from the air traffic control at KL and the air traffic control at Ho Chi Minh City, both of which had continued to track MH370 on their respective ATC primary radar those critical 9 minutes,”

    Here is a link to a document that gives the range of all ATC primary radar in Malaysia:


    Every one of the primary radars has a range of 50-60 nautical miles. The nearest radar to BITOD is at Kota Bharu. The distance from Kota Bharu to BITOD is 127 nm. That’s over twice the range. The nearest Vietnam ATC radar is at Camau. It is 131 nm away. Can you tell us which ATC radar was tracking MH370 and how it acquired such super-sensitive powers?

  136. Alex Siew said,

    July 16, 2014 at 6:59 pm


    I hear what u are saying.

    Whether it is theoretically possible or not, no one is going to admit now, that the pings could have been powered by battery power. But what if

    (a) the signal strength of the pings ( which were transmitted over a period of 6 hours) was significantly lower than for the earlier transmissions? and

    (b) the signal strength, just among the pings, show a gradual decline.

    If the answer to the above questions are in the affirmative, the conclusion surely can be drawn that the pings were powered by battery power. It may not constitute conclusive proof that the plane was stationary during the pings but it will prove the pings were not powered by the plane’s regular AC or DC power and thus render untenable the critical assumption underlying this whole investigation and search in the Indian Ocean, that the pings necessarily mean the plane was still ‘powered up’ and flying.

    As i have said many times before, Inmarsat can prove me wrong by releasing the data on the signal strength (and on the C/No and BER), if indeed there was no discernible difference between such data for the pings as compared to the earlier transmissions. The fact that Inmarsat have so far refused to do so, even after a 4 month search has turned up absolutely nothing, speaks for itself. Just what justification could there possibly be, to withhold this data, with the bodies of these 239 people continuing to languish in the sea and the suffering and anguish of the families.

  137. Alex Siew said,

    July 16, 2014 at 7:31 pm


    On the point about primary radar, those are very good questions but unfortunately u have addressed them to the wrong person.

    Those questions should be answered by the Malaysian ATC and the Vietnamese ATC since they are on record as having said that their primary radar tracked the plane up to 1.30am and to BITOD respectively. The statements by Vietnamese ATC were made in the early hours of March 8th and recorded in the calls between the 2 ATCs (see Preliminary Report), the statements on the part of the Malaysians were by the head of their DCA, who incidentally was the person in charge of the search in the first week, on March 8th and repeated in the following days that week including in the presence of hundreds of journalists and broadcasted to the world. Please see my previous comments for details of those statements.

    Incidentally, Professor Stupples, speaking in the context of the primary radar of the Malaysian military, mentioned a range of 402 km. And there were some reports of the Malaysian DCA and military being located in the same building or something to that effect.

    This primary radar tracking by the 2 ATCs would have been recorded. Malaysia has reportedly sealed this recording. See the report by the Straits Times. I have no knowledge whether Vietnamese took the step of preserving the radar recording for that night. In any case of a missing airplane the evidence of the ATCs would be at the forefront of the investigation but as is obvious to many already, the usual rules do not apply when it comes to MH370.

  138. Alex Siew said,

    July 16, 2014 at 11:03 pm


    “Regarding correlation of the later BTO points with satellite Z position, since the BTO measures the timing of a signal going through the satellite, there will always be a correlation between the BTO and the position of the satellite as it moves.”

    U addressed the above part of your comment to @JS, I am sure @JS will respond and i do not have any mathematical or engineering background, but the point of the correlation is that if the plane was still flying during the pings, then there should not be a direct correlation between the movement of the satellite and the BTO (unless the plane was flying in exactly the same manner as the satellite at all such times).

    If i am not mistaken, each km of movement in the z axis by the satellite corresponded to 8 plus microseconds in BTO (especially true for the latter pings). Also as observed by @JS, the respective distances between the ping radii are roughly twice the distance traveled by the satellite on the z axis (again especially true for the latter pings).

    The correlation also extends to the BFO, with BFO for the pings = fixed offset of around 90 + satellite velocity in knots (with a small aberration for the 19.41 UTC ping which aberration was probably caused by the eclipse as Inmarsat themselves indicated from a report by WSJ). Where the BFO was more or less the same, for the first and third ping, the satellite velocity also happened to be around the same. To my limited knowledge, there was no such correlation for the earlier transmissions.

    The graphs for the BTO, BFO and the satellite movement/velocity all have the same shape (for the pings), with a dip in both BTO and BFO seen for the second ping at 19.41 UTC when the satellite happened to be around its northern apex (at 19.36 UTC) with z inversely peaking at such time.

    I am not the only one having a problem accepting that these are all coincidences, see for eg the comment of a @Steve on Duncan’s blog in the latest thread.

  139. JS said,

    July 16, 2014 at 11:39 pm

    @Alex – CopperNickus on Reddit also noted a correlation between BTO and altitude. But altitude roughly increases with time, to a point, so I suspect he and I were seeing the same thing.

    @sk999 – I feel like you’re jumping into a more advanced statistical analysis when the results of a simpler one aren’t to your liking. A series of point ls should fit a line when the dependent variable is truly dependent on the independent variable, and they should not fit a line when the proposed dependent variable is actually independent.

    The one-in-a-million probability of a correlation when the plane is sitting still is misleading, and so is your sample selection. I got less than .00001 on the 17. You are claiming that those 17 were chosen to “keep the table short.” So the data chosen by ATSB or Inmarsat doesn’t work, we add a few points back in, and that’s not a sampling bias?

    In any case, I get that .000001 or something on the 17, and you got .5 on the 54. We should have gotten .999 for crying out loud. We have a short time interval, and a set of BTOs in a narrow range, and a stationary plane. There should be a stronger correlation than .5. The BTOs should barely move.

    On the later BTOs, you are saying that they correlate because the BTO is part of the round trip. That, again, is the assumption I am challenging. You are using a circular proof – the BTO is part of the round trip, so there’s correlation, and since there’s correlation, the BTOs must be part of the round trip. Yet the correlation is almost perfect against the Z coordinate. What are the chances that there is correlation above .9 between the satellite’s position and the BTOs, despite the satellite movement being much smaller than the plane’s movement, and the satellite’s motion having an even smaller impact on the total trip time.

    I will rerun my numbers, again with a least squares because I think the BTOs should either correlate with a line (for the original 17) or absolutely not correlate with a line (the last 7, against the Z coordinate.) I’ll post the results in the next day or so, with the hope that at least we can agree on numbers, if not interpretations.

  140. Alex Siew said,

    July 23, 2014 at 12:28 am


    I have had the opportunity to do a little research on the question u posed about what sort of (primary) ATC radar was tracking the plane to 1.30am/BITOD as claimed.

    The DCA document u referred to listed only the ‘terminal radar approach control’ primary radar (aka TRACON) of the various airports. This kind of primary radar typically has a range of 50 to 60 nm.

    However, in addition to TRACON, ATCs have another form of primary radar with much greater surveillance range called ‘en-route long range’ or ‘air route service radar’ (ARSR). This type of primary radar has a range of between 180 to 300 nm, is typically operated in conjunction with the country’s military/air defense and located at the country’s perimeter or borders. Thus the USA have several ARSR-4s (or the latest version) at its borders.

    Malaysia appears to be no different in having joint military/ATC long range primary radar, as the commercial director of Malaysia Airlines Dunleavy alluded to in saying the DCA and military radar were in the same building or words to that effect. The MATSMP radar system in Malaysia has been described as follows: “…a very modern system with Raster Scan Colour Display, fully integrated radar data called MRT- Multi Radar tracking systems [where radar data from various stations] are piped down to the ATCC Centre at Subang to form a real time complete picture of all the air traffic flying over the skies of Peninsular Malaysia…”.

    Where are these long range primary radar located in Malaysia? From a quick search on the internet, there is such a radar at the airforce base at Butterworths/Penang at the north west (headquarters of the five nations defense arrangement) and another such radar at the north east coast at the airforce base at Gong Kedak (located at the border of Kelantan and Terengganu) where Malaysia keeps its fleet of Sukhoi Su-30 MKMs.

    According to former Deputy Prime Minister Anwar Ibrahim who is now the leader of the Opposition, the radar at Gong Kedak was supplied by Marconi and its range would extend even to the Indian Ocean on the other side of Malaysia. I do not know the exact range but it most likely would not be less than 200nm.

    The latest radar supplied to Malaysia, the GM 400 from Thales has a range of over 400km or 220nm.

  141. andyo said,

    July 27, 2014 at 2:51 pm

    which seems to suggest that it’s a perfectly reasonable assertion that MH370 was tracked back across the Malaysian peninsula from ICARI/BITOD. Th ranges are just sufficient.

Leave a Comment

You must be logged in to post a comment.