Computer model yields two possible flight paths of MH370 in South Indian Ocean

Martin Vegter
  • Computer model yields two possible flight paths of MH370 in South Indian Ocean Martin Vegter

    I am reading this article on the NY Times, where they show two possible flight paths of the missing flight MH370 in the South Indian Ocean:

    enter image description here

    I am wondering how they can come up with such specific paths, when the only information we have are those two arcs (those arcs are not paths, rather the last satellite ping happened somewhere on the arc).

    How can anybody create a computer model with just this information? Or does it mean that the modelers have more information available than the general public?

  • The "related article" specified in your link has this to say:

    Searching one possible flight path, Australia said it would focus its efforts in a specific stretch of the southern Indian Ocean, using computer models of the plane’s possible flight path that take into account undisclosed satellite data, wind conditions and ocean currents and some assumptions about how fast it was flying and how much fuel it had left.

    So yes, apparently they have more information than the general public does.

  • My personal guess: I think what they've done is to start a line where from where the plane was last known, and then intersected it with the arc where the distance would have been correct at cruise speed with reference to the time.

    Example with just some values. Since the arc radius and time was known, one would be able to draw a line which would show a constant speed. If it was not a constant speed, the line would have to bend left/right to account for it, which does not appear to be the case in the provided image.

    satellite example

    This is based upon the following:

    • 'A signal was sent out every hour'. In other words, they got several data points.
    • Based upon that, they would have points for the arc radius at set times.
    • They might assume that the aircraft has a constant speed. Because of this, they should see if the data points lie on the a straight time, since the only variable is the distance.

    I'm pretty sure that the two lines is based upon the fact that they know the uncertainty at the final position, and in my opinion it suggests that the aircraft flew a constant course from the last known radar fix where it originates.

  • how they can come up with such specific paths

    The correct answer currently is that we (the general public) do not know exactly how these tracks were calculated.

    The specific lines in your graphic were probably invented by graphics artists employed by NYT and based on vague verbal descriptions.


    On March 24th, Inmarsat stated that a new analysis of the records of satellite handshakes with MH370 were able to use doppler effects in the signals together with an analysis of similar satellite handshake data from other Malaysian 777s to determine that MH370 flew into the Indian Ocean and ruled out flight paths to the previously described northern corridor.

    enter image description here
    enter image description here
    From INFORMATION PROVIDED TO MH370 INVESTIGATION BY UK AIR ACCIDENTS INVESTIGATION BRANCH (AAIB) 25/03/14


    On March 21st, a little light was shed on this process by a statement attributed by New Straights Times to Malaysian Defence Minister and acting Transport Minister Datuk Seri Hishammuddin Hussein.

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

    "This type of data is not normally used in investigations of this sort. It is only because we have so little other information to go on in this difficult and unprecedented situation that the data is being used.

    "Upon receiving the raw data, the Malaysian authorities immediately discussed with the US team how this information might be used. The US team and the investigations team then sent the data to the US, where further processing was needed before it could be used.

    "Initial results were received on Thursday, March 13 at approximately 1.30 pm, but it was agreed by the US team and the investigations team that further refinement was needed, so the data was again sent back to the US.

    "The results were received at approximately 2.30 pm on Friday, March 14 and presented to the investigations team at a high-level meeting at 9 pm on Friday. The UK AAIB, who had also been processing this data independently, presented their results - which concurred with ours and those of the US team – at that meeting.

    "The Prime Minister was briefed on this satellite information at 8 am Saturday, March 15 and publicly announced it at the press conference at Saturday lunchtime.

    (my emphasis)

    Obviously the above does not describe the actual calculations involved. However it does confirm that the satellite data had six handshakes (presumably at hourly intervals up to 08:11).


    The graphic below, illustrates one idea for how these track might have been deduced. It is a result of speculation rather than a reporting of information actually provided by the people who made the calculation.

    The graphic suggests that the NTSB calculated possible paths based on data that is not public and that the NTSB were able to assign probabilities to those paths. As far as I know, neither Inmarsat, the NTSB nor the Australian government (or it's agencies) have made public the data or method of calculating these tracks.

    The graphic below suggests several Inmarsat pings were used (the example shows 01:11, 02:11, 04:11, 05:11, 06:11, 07:11 and 08:11 but these are probably speculation).

    So far as is publicly known, there was no ping at 09:11 or later which suggests the flight ended between 08:11 and 09:11. This is broadly consistent with maximum fuel range of the aircraft. The labelled NTSB solutions suggest that NTSB calculated that no other constant speed (and heading?) tracks fit the data.

    enter image description here
    source Scott Henderson

    which says ...

    For a constant speed track, distance between ping arc intersections will be equal.

    NTSB solutions are based on constant speed straight track. Curve in track shown due to map projection

    A later version of this graphic adds a disclaimer saying that the tracks and zones are not accurate and are symbolic only.

    We do know that NTSB sent staff to Malaysia to assist in the search. We know that Inmarsat provided data to the investigating teams.

  • I provided a potential mathematical solution on deriving the flight paths using the ACARS data on 17 March (before the possible flight paths were published by the authorities) on CNN's iReport: http://ireport.cnn.com/docs/DOC-1108312

    The key to finding MH370 without the aid of the transponder data but with regular attempted ‘handshakes’ from the ACARS satellite is by a process known as “vector addition”. If the last handshake with the ACARS satellite took place at 8.11am as reported then depending on the frequency of the previously attempted handshakes (half hourly or hourly, there seems to be some discrepancy in the reported frequency) a set of flight route models can be easily constructed giving a high probably of the final location based on the accurate last known location at 2.40am. Therefore, depending on the frequency of the ACARS handshakes there should be at least 6 and up to 12 vectors that could be mathematically solved.

    Assuming a constant speed for the first probability model, each time the ACARS handshake occurs it gives only a footprint angle on the Earth (as has been shown on the reported 8.11am ACARS ping map; 40 degrees). The next time it handshakes another angle is indicated. At a constant speed only two possible headings are possible to ‘join’ the ACARS isogons (for the plane to get from one ‘angle’ to another). If MH370 was heading in generally the same direction or arcing across the sky, (even if there was some tacking back and forth like a yacht) only one of the possible vector solutions will be valid.

    If the vectors from the last known location at 2.40am were systematically added together the final solution at 8.11am will bisect one of the two arcs provided by the authorities. Subsequent models at different average speeds and taking into account routes that would more effectively avoid radar detection would refine the models to a few high probability scenarios and narrow down the search area considerably.

    It’s simple mathematics but there is no indication the authorities are using it.

    As RedGrittyBrick has pointed out this is almost definitely how the flight path was estimated. The reason for the two flight paths is probably illustrating the two end member solutions based on the accumulated error using this technique (the more times you add one vector to another the more error can be introduced). Hence the two predicted flight paths diverge.

    The reason for the confidence by the authorities in the southern Indian Ocean route is that the mirror set of solutions would need MH370 to pass straight over India and Pakistan; and they have emphatically denied this could have taken place.

    Therefore there is the highest probability that MH370 ended up in the region of the southern Indian Ocean they are currently searching.

    I am somewhat saddened that these probability models were not derived within days MH370 went missing and that at least 7 days were wasted searching in other areas.

Related questions and answers
  • if the plane most likely went along the S or the N arc we see in reports. Unfortunately, only the last ping (at 8:11AM) is available publicly. Here is the basic idea on extracting the information... position at 3:11AM. From the ping at 3:11AM, another circle can be drawn, like the one drawn at 8:11AM. The intersection of the two circles give an arc of most likely 3:11 AM positions. At 4:11AM, we will have a series of circles from max distance and one circle from the ping. These intersections will give a series of possible arcs. This process can be repeated (with increasing complexity

  • I am reading this article on the NY Times, where they show two possible flight paths of the missing flight MH370 in the South Indian Ocean: I am wondering how they can come up with such specific paths, when the only information we have are those two arcs (those arcs are not paths, rather the last satellite ping happened somewhere on the arc). How can anybody create a computer model with just this information? Or does it mean that the modelers have more information available than the general public?

  • WWII. They didn't have flight control computers back then, and the only control complaints I recall them having is that early versions had a tendency to flip over backwards when approaching stall speeds, well, that and the ground effects were pretty strong. But, no mentions of going into flat spins when going into hard maneuvers (that I recall). So how do they control that Y axis on flying wings...How do flying wings, like the B-2 Stealth bomber, actually keep themselves from yawing out of control without a vertical stabilizer? For the record, I assume this has to be a simple mechanics

  • After answering this question on History.SE, I started to wonder if it would be possible to find out even more detail about the plane now that its serial number is known. I have no idea what kind of flight records the US Army Air Corps kept, however. I know most flight logs today are kept by pilot, but I imagine there would be some way to trace what pilots flew a particular plane. I have no idea if this is possible for USAAC trainer planes in the 1930s. Could I get access to these records? If so, how would I go about it? I'm mostly interested in seeing if I can find out more information

  • I am reading in the news that the MH370 flight changed its course by "Computer entry" and that it was programmed into the computer at least 12 minutes before the last voice contact with the plane. How can they know that the change in the computer was done 12 minutes before (while the plane was still flying as on the original course)? And how do they know it was done in the computer, rather than the pilot steering manually?

  • I was looking through my virtual radar logs one of the days and found this "glitchy" ADS-B behavior. I am almost 100% sure that this is not due to my antenna or setup since two independent different radars confirmed this weird behavior from FlightRadar24. Also A/C before and after this one did not exhibit this behavior. Does anybody have any thoughts as to what may be happening??? Why... of occurrence is approximately: 3/16/2014 6:09pm CST I have also verified FlightAware is ALSO showing the same weird glitch. See below "yellow" highlighted airplane: Same A/C from FlightRadar24

  • Looking at the paths of the same flight on two different days, I noticed they flew very different paths. Short Path:1 YAY N184B TOPPS ENE PARCH1 Long Path:1 NIBAX G462 TUMAK UL602 ORSOL UL602... REGHI UN480 ETIKI NATD DOVEY LACKS BERGH L454 OWENZ CAMRN CAMRN4 I asked a similar question in the past. This is what I want to know: What are the reasons which can cause a commercial flight to change its path mid-route? How it is planed and executed? I am sure the captain is the final authority on this, but who else is informed? 1Route Source: FlightAware. P.S.: I could not understand

  • I'm a student pursuing a US Private Pilot License, and recently scheduled my checkride. I've been training in a 1981 Piper Warrior (PA-28-161), but if its annual goes sour I may have to take my club's 1980 Piper Archer (PA-28-181). I have well over §61.109's 40 hours in the Warrior alone, and only ~10 hours in the Archer. I have a separate club checkout and CFI solo endorsement for each... specifically for [II.A.2.] Total time in this make/model and/or approved FFS or FTD (Hrs.) Furthermore, according to the IACRA FAQ ("I'm a Designated Examiner. I noticed a mistake when reviewing the IACRA

  • There are various services that use world-wide Boeing Winds for forecast wind data which can be used to calculate an approximate flight time between two locations. They usually have best case, worst case, and average case for each location, altitude, and date in the future. I have searched and searched Google to no avail. Where can this wind data be found, and how can it be used in a commercial product? For those of you who don't know what the Boeing winds are, I found this description of their software product on am informal message board (not related to Boeing): PC WindTemp

  • Inspired by a discussion in chat. Most GA piston singles are powered by either Lycoming or Continental engines. The engine designs used by both manufacturers are broadly similar (4-cycle, horizontally-opposed, gasoline-powered, air-cooled), and they're both generally available with either carburetors or fuel injection, but I know they're not "identical products". What are some of the differences between the designs used by the two manufacturers, and what practical implications do those choices have for pilots?