Updated MH370 Study Reiterates Seabed Search Uncertainties

map_mh370_figure_0516_en_a74ba7fb33

In an earlier post I described research conducted at the GEOMAR-Helmholtz Institute for Ocean Research in Kiel which suggested that, based on reverse-drift analysis of the Rénion flaperon, its starting point most likely lay in the tropical latitudes of the southern Indian Ocean, far north of the current seabed search area.

Today the same scientists published an update of their research, with a press release available here and the full report here. The upshot can be seen in the chart above, which shows the probability distribution of where the piece likely began its journey to Réunion island. Once again the authors have concluded that the greater part of the probability (98.7 %) lies far north of the seabed search area, shown as a white rectangle. The study’s authors suggest that their results might justify a shift of the search area:

The Australian search authorities are aware of this report. “Whether or not these new results will be used to facilitate the last few months of the ongoing search for MH370 is not clear,” Arne Biastoch summarizes.

One of the refinements included in the new study is that while the authors continued to assume that there was no direct wind effect on the flaperon (it being presumed to be floating essentially flush with the surface), they have included for the first time an effect called Stokes Drift, which results from wind-generated waves:

“In our recent calculations we included more physical processes in order to simulate the drift more realistically,” Prof. Biastoch explains. “In particular the drift induced by wind generated ocean waves is now included,” Biastoch continues. “Even though we use state-of-the-art modelling systems, representing the ocean currents in the Indian Ocean quite well, all simulations naturally contain limitations. Our investigation is one important piece of the puzzle in finding MH370.”

As a result of the new calculations the possible source region of the flaperon was refined, and “While it is shifted a bit southward from the initial study done last September, our basic result that most particles originate from a region north of the current search area remains unchanged,” states Dr. Durgadoo.

So should Australian search officials call a halt to the current search and relocate its ships further north? Actually, I don’t think they should. If the GEOMAR scientists are correct and MH370 did crash into the ocean west of Exmouth, the plane must have been following a low and curving trajectory of the kind that is not supported by any simple autopilot mode. That is to say, the plane would have been either conscious control the entire time or flying along a series of arbitrary user-defined waypoints.

The latter seems extraordinarily unlikely. First, we would have to surmise that whoever was in control of the plane decided to fly a basically random path, and to choose a cumbersome way of doing so, entering by hand pairs of latitude-longitude coordinates. This would be bizarre behavior, to say the least. Furthermore, as explained in the DSTG report issued last December, it is extremely unlikely that a randomly chosen set of slow segments would happen to match the ping rings. Instead, random sequences are only likely to match if they conform to a fast-and-straight flight to the south: in other words, if they end up in the current search area.

The former is problematic for the same reasons, and for an additional one as well. If the plane was under conscious control until the bitter end, then we cannot assume that, as in the unpiloted scenario, it spiraled into the sea once its fuel ran out. Instead, the conscious pilot might have chose to hold it into a glide far beyond the seventh arc. We have no reasonable expectation, therefore, that a narrow search along the seventh arc would yield the wreckage.

315 thoughts on “Updated MH370 Study Reiterates Seabed Search Uncertainties”

  1. @jeffwise said, “Furthermore, as explained in the DSTG report issued last TK [?], it is extremely unlikely that a randomly chosen set of slow segments would happen to perfectly match the ping rings. Instead, random sequences are only likely to match if they conform to a fast-and-straight flight to the south–in other words, if they end up in the current search area.”

    This is a misconception that I have unsuccessfully tried to dispel. When calculating possible paths between 18:02 and 00:19, the DSTG uses prior distributions for the number of turns and number of accelerations that peak at 1 turn. This is shown in this figure from the DSTG report:

    https://www.dropbox.com/s/x7aq75jyoxi50o3/Histogram%20of%20number%20of%20manoeuvers.png?dl=0

    These distributions are independent of any BFO or BTO measurements. Therefore, the posterior distribution of end points will reflect this prior distribution of manoeuvers, i.e., there will be a higher probability of end points for paths that require fewer manoeuvers.

    I believe this is a fundamental flaw in the analysis. The preference for straight paths was implicitly included in the model. As I have said before, the DSTG report was written to impress, not to enlighten. A more cynical view is that the authors were motivated to justify prior assumptions that led to the current search area.

  2. @VictorI, I know that we disagree on this point, but I believe the DSTG report is correct. It’s true that only paths with multiple turns have a chance of fitting the slow, curving paths to the northerly parts of the 7th arc. But with each additional turn, there is a multiplication in the range of possible paths that are wrong–so the odds of getting that exact sequence of segments with the correct direction and length decrease. It’s like adding more combinations onto a combination lock.

    On top of which is the entirely different issue, of why someone would be making random turns over the ocean in the middle of the night.

    @Brock, Do you have any interest in doing a stochastic analysis of random multi-segment flights to the SIO?

  3. @jeffwise:
    The DSTG model uses a form of randomness (Ornstein-Uhlenbeck) biased towards zero. You are also focussed on randomness. You have to consider the possibility that the path was not programmed or planned, yet not entirely random, but somehow biased off-zero. For example, if someone not familiar with navigation had taken control of the airplane after 18:25, expecting to arrive somewhere, and making adjustments when it did not occur.

  4. @Gysbreght: Yes. The DSTG assumes the manoeuvers are stochastic with a bias towards zero. I can’t imagine any scenario in which manoeuvers would be random. If a pilot was in control, the path would reflect the intent. If no pilot was in control, there would be no manoeuvers.

    If the prior distribution was peaked at 6 manoeuvers, for instance, the peak in the end point distribution would favor a curve path further north on the arc.

  5. @Gysbreght, If someone took a took a course of action about which we know nothing, isn’t that equivalent from our perspective to being random? Randomness after all is merely an expression of an observer’s ignorance of a system.

  6. Still this study also points to most probable area’s with there selected model trajectories to ~30-32S and ~95-100E.

    With assuming no wind effect on the flaperon it contradicts the clear observations from the Meteo report in which also two possible mixed buoyancy states are mentioned.

    Combining the two study’s could therefore imo prefer the more southern area’s mentioned if a mixed state buoyancy is assumed with a relatively fast also wind driven first trajectory to the north and a slower only current and Stokes- effect driven trajectory to the west.

    Imo this could avoid a lot of the complications mentioned in Jeff’s new article.

  7. @jeffwise: Let’s suppose we had no radar data after 17:21. How well would those a priori distributions have worked to predict the path between 17:21 and 18:22? Was the path random in that interval? Prior behavior would predict many more manoeuvers if the pilot was in control.

  8. Reply to posts from the previous thread.

    @AM2: cc @GortoZ,

    I was originally a believer in the Inmarsat data, and the work of the IG, but not the ghost flight / fuel exhausted / final spiral dive.

    I had always believed it was a meticulously planned, brilliantly executed, perfect “vanishing”, ending in the deliberate ditch. I did a lot of work on it, for over 18 months, until Feb/Mar 2016.

    Around that time, I began to seriously doubt the whole Inmarsat / Mallacca Radar / DSTO(DSTG) Baysian / SIO story, beginning with a rethink on:-
    (1a) the initially “haphazard” air searches in the SIO,
    (1b) the use of a P8, obviously to keep an eye on the PLA IL-76’s,
    (1c) the subsequent grounding of the PLA IL-76’s on “fake” bad weather forecasts,
    (1d) the abrupt end of the air search, “ordered” by Malaysia (supposedly),
    (2) the abrupt “creation” of the JACC fronted by Angus Houston,
    (3) the saga of the fake Sonar Pings, particularly in light of the very public and very heavy involvement of HMS Tireless, and “the” US Navy “Captain” (research both, and then read between the lines).
    (4) the justification of the Inmarsat data by the less than credible Malaysian Mallacca Strait radar track, to make the “connection” to the 3F-1 pings.
    (5) the mounting concern that:
    (a) the provenance of the individual debris items so far found is questionable,
    (b) the examination of that debris by the “authorities” has been questionable, and
    (c) the drift studies, both forward and reverse, “do not support” any SIO terminus, with any significant degree of confidence.
    (6) the fact that Boeing, FAA and NTSB have shown absolutely no interest whatsoever in the whole thing, and the NSA has “classified” anything they knew. In short, the “stunning silence” of what would “normally” be all of US “interested parties”, in any other T7 crash, anywhere in the world, “smells” in this case.

    All of this, now leads me to think, that the whole “search” has been “managed” at the “highest levels”, as a deception.

  9. @jeffwise
    Unsystematic (arbitrary) does not have to equate with ignorance of the system if the plan (or lack of) was only done a different way, random may not be the best descriptive word as “expression of an observer’s ignorance of a system”. One can know what they are doing and choose to be random.

  10. “@VictorI, I know that we disagree on this point, but I believe the DSTG report is correct.”

    @Jeff, the DSTG report is a model. Monte Carlo or the more direct (and in my view more appropriate) data fitting approaches used earlier by the IG, are likewise models. All incorporate flight dynamic assumptions, and they are all “correct” from the standpoint of being consistent with the known constraints and assumptions.

    Like you, I do believe the search should continue as planned simply to properly check that box. There is no point in making what amounts to a knee jerk reaction to drift analyses.

    I do not share your view of a segmented (with respect to speed and heading) Northern path. How hard is it to spin the knobs on the autopilot to set any desired speed and heading? The path you describe as “random in the middle of the night” is logically consistent with a stalling tactic associated with the progress or lack of progress of negotiations going on elsewhere.

    The “elephants in the room” for me are:

    1> Why did the PIC allow the aircraft to run out of fuel when the fuel remaining instrumentation is highly accurate?

    2> Why was no distress call made and position reported when it was clear that a water landing was imminent?

    It is very hard to reconcile the callous nature of the above inactions with any motive except a deranged person intent on committing suicide.

  11. When it rains, it pours: Duncan Steel and Richard Godfrey have also published drift analysis today:

    http://www.duncansteel.com/archives/2631

    Re: Geomar: I am glad Geomar has continued to pursue its drift modelling. They seem to have done a lot of excellent, in-depth research into Stokes effect. FYI: care should be taken to distinguish between unconditional distributions (e.g. Figure 4) and those which ASSUME impact occurred within the ATSB’s wide area (most of the subsequent graphics, including the key summary chart). This is not, in my opinion, a safe assumption. I will see if I can get Jonathan to send me “Stokes at 50%” sensitivity applied to Figure 4.

    Re: stochastic paths: I worry that any stochastic analysis NOT biased toward minimization of turns would struggle to add incremental value over, say, a survey of aviation experts asked to brainstorm plausible paths as reverse-engineered from a more northerly impact point and the BTO arcs. The stochastic runs would simply throw up hundreds of “spaghetti against the wall” twisty paths. Since these would then be winnowed down to a few of the least implausible by aviation experts, I think we’d end up at much the same answer.

    Worse still: these least implausible paths – as derived by either method – will still possess what I consider to be a problematic coincidence: that at least two things are cancelling out perfectly, in order to lay down a set of BTO arcs which falsely appear to indicate a straight path at cruising speed. At minimum, we are talking about a severe left turn between Arcs 5 and 6 – with turn timing and angle by fluke precisely offsetting, so as not to “disturb” arcs 6 and 7.

    And I can guarantee we will struggle to get an impact point out of this exercise sufficiently northeast to justify empty Oz shorelines. I’ll post Dr. Pattiaratchi’s work later today.

    I’m not refusing to attempt stochastic path analysis; just laying out my initial thoughts on potential limitations.

  12. @Brock: If you are going to perform a stochastic analysis and (implicitly) bias it towards straight paths, then just cut to the chase and model the path as one with no manoeuvers, which is what the IG and others have previously done.

    And there is no reason to attempt a stochastic path analysis. It has already been done in great detail in the DSTG report.

  13. @ventus45
    Fully agree.

    I urge all those interested in helping solve this case to re-visit those “early days” as there are clear indications to deceptive behaviour from (Prime) Government sources around:
    1. The fragmented, inconclusive air surface search.
    2. The order to cease the surface search.
    3. Initial acoustic pings.
    We have been “lead down the garden path” and are now going through the final motions of (fruitless) detection/recovery to abandonment.
    In the meantime, 239 poor innocent souls are missing.

  14. If the starting point of the flaperon “most likely lay in the tropical latitudes of the southern Indian Ocean, far north of the current seabed search area”, but the aircraft still flown for several hours, it is possible that for an unknown cause the aircraft lost the flaperon during flight at that point and continued his path, but not so far as the current search area because of the added drag caused by the missing flaperon on the right wing, the elevated flaperon / ailerons on the left wing to maintain the wings horizontal, the added angle of attack to compensate for the loss of lift, yaw control, etc. Does someone know if the 777 can still maintain it’s wing horizontal and it’s altitude in this miserable configuration? Then the wreckage must be North of the current search area but South of starting point of the flaperon on the sea, somewhere along or near the arc…

  15. @Jeff

    Of course, your latest topic raises the much debated subject of whether “reiterate” is a proper word. Most linguists attribute the form to hypercorrection.

    Heisenberg, Godel, and Chomsky walk into a bar…

  16. From the full report’s summary:

    “Recently, several other pieces of debris, some confirmed to belong to MH370, have been collected around the coasts of Mozambique, South Africa, and Mauritius. Unless specific information regarding when these objects made landfall become available, considering these objects will not help to further refine the analysis. However, that the debris were found in the southwestern Indian Ocean does corroborate the results presented here, especially given the general oceanic circulation of the region.”

    Now compare that to what the news paper reported:

    “matches perfectly”

    Was that a case of journalistic license or is it a case of scientists backpaddeling?

    It would interesting to compare the original February report, with the two updated reports, on this matter.

  17. @DennisW – your “elephants in the room”
    also can be pinned to a diversion North or NW.

  18. @Trond,

    Your commented is quite interesting.

    If the flaperon was planted by one of the known search parties in the early days, I wonder which one statistically matches the drift patterns the best?

    In other words, we DO have a series of starting points and one endpoint if we assume that the flaperon was planted by a search vessel. Which makes a whole lot more sense than a random trip to the SIO to plant a flaperon. If it was planted, perhaps the expectation was that such a large floating part would be found sooner.

  19. JS,

    Re: “If the flaperon was planted by one of the known search parties in the early days…”

    France?

  20. @Oleksandr,

    I had China in mind but it could have been anybody that brought a boat out in March-April 2014.

    Which curiously excludes Russia.

  21. JS,

    It was a sarcastic joke; I hope you are not upset. The flaperon, initial satellite snapshots, a few French nationals onboard (linkage to Syria), mysterious visit of the French judge to Malaysia, good political ties with China, finally drift study. For someone all this might be sufficient to develop a new conspiracy theory.

  22. @jeffwise:
    “First, we would have to surmise that whoever was in control of the plane decided to fly a basically random path, and to choose a cumbersome way of doing so, entering by hand pairs of latitude-longitude coordinates. This would be bizarre behavior, to say the least”

    Jeff, that is not necessary. Just three waypoints within the six ping rings from 19:40 through to 00:19 are required to give straight flightpaths and a constant speed to a ditching to the North of Christmas Island and just below Java. It fits all the ping rings like a glove.

  23. @Trond ,JS
    That makes real sense ,totally agree with you.
    @Ventus45 With you all the way, this is a mega deception .

  24. @Oleksandr,

    Not at all. Only the joke flew right over my head.

    I’m thinking that the problem with the “flaperon plant” theories is that they would require somebody to obtain and then deliver the flaperon to the SIO, and then hope that it was found a year later.

    But Trond got me thinking that it could have been planted very early on, with the intention of being found very early on, while the finding on Reunion was purely accidental.

    For that matter, it could even have been recovered early on and then lost – either never quite pulled from the water, or pulled and thrown back in, or pulled and moved before being thrown back.

    Main point was if the drift analysis was limited to known ship movements as starting points, something might stand out.

  25. @Freddie: Please provide speeds, tracks, coordinates, waypoints, and whatever else we need to properly evaluate your proposed path.

  26. @Freddie: I was about to ask the same as VictorI just did. Waypoints, speed, and time at first waypoint would satisfy me for starters.

  27. @all

    People posting comments here who are giving ANY consideration to the Flaperon being ‘planted’, or otherwise arriving artificially (nefariously) on the shore of Renunion are SERIOUSLY out in the weeds. The myriad complexities involved in such a caper, not to mention ends justifying means…it’s well and good truly delusional. Yes, delusional.

    The entire notion is so ridiculous that words will not suffice. I do know some wonderful therapists should reality ever haphazardly intervene.

  28. @Matt, What is so complex about putting a piece of debris on a beach? Especially when compared with the complexity of making a 777 vanish without a trace?

  29. “The flaperon drifted from where they discovered the pings in the ocean.”

    I have noticed that too, but in this case we have now a scenario where a ship dropped or towed a flight recorder and planted the flaperon, for the purpose of misguiding the search operations in that area instead of another (more Southern?) place. Then there are lot of chances that the wreckage is “near” the current search area, or somewhere between these two areas (unless we consider the Northern route).

  30. The shaded area extends to around 350km of Exmouth(North West Cape). A bit more fuel and they would have made the coast for a daylight arrival. If a terror attack, did they stuff it up?

  31. @jeffwise:
    “What is so complex about putting a piece of debris on a beach? Especially when compared with the complexity of making a 777 vanish without a trace?”

    If directly “putting a piece of debris on a beach” the complexity lies in the marine life attached to that piece, but there is nothing complex left if that piece was instead planted far in the middle of the ocean.

  32. @Matty – Perth:
    When I posted somewhere a possible “flight 714 to Sydney” scenario 2 years ago I was treated as tinfoil hat…

  33. Marc – It would explain all of the politics. Failed terror plot, everyone tip toes around Malaysian sensitivities for ongoing cooperation due to the fact that it would become a sudden focus for investigation. Powers would need to know in a hurry how extensive the cell was. Similar to ventus45 I often felt like we weren’t watching the main game.

  34. Marc – That is….if it was suspected and not much was known. The politics fits.

  35. @JS:
    “it could have been anybody that brought a boat out in March-April 2014.”

    The flaperon may have been dropped from an aircraft, in this case the impact with water would make the flaperon look even more authentic…

  36. While I’ve yet to hear back from Dr. Charitha Pattiaratchi re: permission to post his full set of forward drift result charts, I’m reasonably comfortable he won’t object to my posting them here anyway:

    1) When he e-mailed them to me, he made no mention of any restriction on use/publication

    2) I already published 4 of the 25 – spanning over half his tested range – last December in my comparative analysis: s32 & s39 in the UWA section, s25 & s35 as underlays in the Deltares section

    3) I’m sure this forum’s denizens will check with Chari before using in their own (not-for-profit) work – and that any such use would include proper attribution

    Here is the .rar file I received from Chari (so needs to be unzipped):

    https://drive.google.com/file/d/0B-r3yuaF2p72bTJiZ1g3QlM2TnM/view?usp=sharing

    Comment: according to the UWA runs, Western Australian shores seem unlikely to have been missed entirely unless impact is VERY far north in the wide area – equator-ward of roughly s25.

  37. @Marc @Matty – Perth

    Terrorist plot gone wrong seems like a real possibility. However, if the goal of Malaysia (or any government involved) was to cover-up this incident why wouldn’t Malaysia pin Shah for the deed? All they needed to produce was either a falsified 1.suicide note or 2.a list of grievances showing him as a political extremist. Easier yet, whatever they “found” could also conveniently point to ditching a plane in the SIO. Majority of the public and media would have dismissed the case as closed after a fruitless discussion about mental health issues in aviation.

  38. Sounds like Russia who had all access to MH17 parts & debris anyone could have planted these pieces dropped from a plane, sub, or ship! I still don’t believe 100% this plane is at the bottom of the ocean, I pray for the families it is safe on an island somewhere!

  39. @jeffwise. Jeff, taking you back to your last post on the Meteo report, “This report was not officially released to the public, as it is part of a criminal terrorism case.” I may be out of touch. Has it come up what case this is please and do we know what connection that has to MH370?

  40. @David, The criminal investigation is into the disappearance of MH370. The Malaysians have considered the disappearance a criminal case since the very early days and the French seem to be following suit. I don’t know if it has been handed to a terrorism judge because they believe it was taken over by third parties or this is the result of some default procedure.

    It’s been said before but perhaps bears repeating — many of us, including me, tend to refer to a pilot-hijacking scenario as a “suicide” scenario but it’s really a murder-suicide, with 238 murder victims and 1 suicide, so more accurately we should call it a “murder” scenario.

  41. KT – if it was suspected terrorism Malaysia might need it quiet to not destroy confidence in MAS and themselves. It would be completely embarrassing nationally, and the end of MAS.

  42. @Brock, Thanks for sending that link, super interesting! Big take-home for me is that if the debris starts south of 29 South (around mhlc17.png in that set of images) then it’s not likely to get to Réunion and it’s quite likely to hit Cape Leeuwin, whereas north of that it’s less likely to hit Cape Leeuwin and quite likely to get to Réunion.

    I think this is would be very useful addition to the store of general knowledge and would be happy to post a version of it to my blog if Dr Pattiaratchi is amenable.

  43. Great article, @JeffWise !

    ‘First, we would have to surmise that whoever was in control of the plane decided to fly a basically random path, and to choose a cumbersome way of doing so, entering by hand pairs of latitude-longitude coordinates. This would be bizarre behavior, to say the least.’

    You always grab my attention discussing possible scenarios. As always keep up the good work. Lastly, where do you stand now in terms of debris being planted, this new post notwithstanding.

    Thanks a bunch jG

  44. @jG, Thanks! Personally, I do still think that the striking lack of marine life on the African debris, as well as the seeming contradictions between the flaperon’s buoyancy and the distribution of barnacles upon it, as well as the apparent youth of the barnacles living on the Rodrigues debris, are best explained as evidence of planting. I think that the contradiction between drift models and the end-points indicated by BTO analysis also point in this direction. However, I know that many do not share this perspective and so I think it’s important to discuss things like drift analysis in a scenario-neutral way, as I try to do in this post.

  45. @Brock
    Dr. Charitha Pattiaratchi’s charts only go to July 2015.
    It would appear that only items starting north of say 30 south “could” reach Africa.
    Since “items” were found in Africa 6 to 8 months later, can you ask him to extend his model run for another year, to end July 2016 (when the search will be finished) ?
    In addition, a large amount of the “south of 30 south” drift goes east, under Australia, towards Tasmania and Macquarie Island.
    So, could you ask him to expand the east-west spread of his slides, to include the African coast and Tasmania (at least).

  46. @jeffwise. Thanks Jeff. How the flaperon investigation would bear on a murder/suicide investigation is unclear.

    Taken to its extreme, if the criminal investigation is a general impediment to information release it may not be possible to publish a cause-and-remedial-action final report until the criminal investigation is concluded. I wonder what view ICAO has on such interweaving of priorities, even if not to that extreme?

    The French were quick with Germanwings Flight 9525 information release, a suicide.

Comments are closed.