Translation: “Simulations Call MH370 Search Strategy Into Question” (Der Spiegel)

[Note: the following is a translation from an article published by Spiegel Online on Friday. Thanks to @littlefoot (Sabine Lechtenfeld) for her assistance.]

by Christoph Seidler

What happened to the missing Boeing 777 of flight MH370? The aircraft with 239 people on board has been missing since March 8, 2014. As the end of July, a piece of wreckage washed up on the French island of La vunion, giving investigators that the wing flap might help solve the riddle–among other things, because it could show how the piece of debris made its way across the Indian Ocean.  And that, in turn, should provide a clue to at least the general area that the plane went down.

Now computer calculations by German oceanographer suggest that perhaps the search for the Malaysia Airlines aircraft has been carried out in the wrong area. Until now, says Andreas Villwock of the GEOMAR Helmholtz Centre for Ocean Research in Kiel, it’s been assumed that MH370 had crashed far south of the equator at 35 degrees latitude. Computer models by his colleagues show, however, that the debris  “probably comes from the eastern equatorial Indian Ocean,” far from the nothern end of the current search area.

Over the course of several weeks, the Kiel researchers Arne Biastoch and Jonathan Durgadoo used a computer model to reverse the track of the debris across the Indian Ocean. Their model used daily current data that had been obtained in past months by French colleagues. The key question: Which path did the piece of debris take across the Indian Ocean, which scientist liken to the inside a pinball machine because of its chaotic eddies and turbulence?

Researchers understood from the start that a computer simulation can’t calculate a precise location of the crash site, but at best only point to a broad area. This coming Tuesday, they will present their results in detail at a press conference in Kiel. However, this much is already clear: The reseachers’ results are contrary to the Australian search strategy.

The current seabed search is focused on an area of ocean west of the Australian coast, and not in the equatorial region, which the simulations now seem to point to.

A total of 120,000 square kilometers will be scanned, an area as big as Bavaria, Baden-Württemberg and Hessen put together. Two ships of the geotechnical company Fugro are carrying out the task on behalf of the governments of Australia, Malaysia and China. So far, they have examined around 55,000 square kilometers with sonar, without tangible result. The simulation seems to suggest that large parts of the southern region were scoured in vain.

Meanwhile, the wing flap that washed ashore has been examined in a laboratory in Toulouse, France. According to Malaysian reports, the wreckage definitely belongs to MH370. French investigators speak of a “very strong suggestion.” details the wreckage belongs definitely MH370. French investigators spoke of a “very strong presumption”. Tha’s why Australia expresses confidence that the wreck will be found.

Two German scientists believe that so-called goose barnacles could provide clues to the crash location. These colonized the piece of debris as it drifted across the Indian Ocean. Geologist Hans-Georg Herbig and the biologist Philip Schiffer, both from Cologne, have identified the small crustaceans in photos.

Both are experts in the small animals, which are part of the barnacle family. Herbig and Schiffer have compiled the first genetic fingerprints for five barnacle species from different parts of the world’s oceans. They have also found that each subspecies dwells in specific climatic zones dependent on the latitude.

But because the French authorities have provided them no sample material of organisms from the wreckage, they can’t apply their insights to help narrow the search for the crash site. “I’m pretty frustrated because despite making requests through various channels, no one has replied,” Hans-Georg Herbig told SPIEGEL ONLINE. “We have tried many things, but got no response.”

In summary: Researchers from GEOMAR Helmholtz Centre for Ocean Research in Kiel will officially present their findings next Tuesday–however they have already give a first clue: Computer analysis suggests that a a piece of debris from flight MH370 that washed up on La Réunion orginated from the “eastern equatorial Indian Ocean.” This is far from the area where two ships are currently searching for the wreckage of the Malaysia Airlines aircraft.

187 thoughts on “Translation: “Simulations Call MH370 Search Strategy Into Question” (Der Spiegel)”

  1. It is true that paths are easily found that are consistent with all the radar and the BTO data, ending on the 7th arc from S20-S40. However, it is also true that the BFO data has the closest fit IVO S37, and the BFO error grows consistently as you move NE on the 7th. While imperfect, the BFO data is probably 100X better (more accurate indicator) than the best drift model over 16 months in the IO.

  2. @seaman

    The BFO data did not bring us any closer to MH370 for more than a year. Maybe it is time to try drift models!

  3. @ALSM

    It is possible to construct NE paths above 20S that fit the BFO and BTO data just as well as the S37 terminus.

    A very simple family of paths are “bifurcations” from your path that depart on an “equal angle” from a line from your flight path to the sub-satellite point. This has been pointed out before. I did not chose such paths since I wanted to avoid the radar on the Cocos.

    The reality is that the IG/ATSB terminus has collapsed under its own weight some time ago. It is not supported by surface debris (which was not found despite extensive searching). It is not supported by the underwater search. It lacks any plausible motive/causality. It is not consistent with any a’priori drift modeling. At some point in time you just have to give it up.

    If I were to point to a single ATSB/IG (and friends) error, it would be the steadfast refusal to consider motive/causality.

  4. @Dennis,

    “It is possible to construct NE paths…….”

    It would be helpful if you could publish some examples, and include some observations regarding the fit to the BTO and BFO numbers.

  5. DennisW:

    Just about ever statement you made above is not true. I won’t waste time engaing in this type of discussion. It never goes anywhere useful.

  6. @airlandseaman

    Maybe it is mentally much more sound to accept and work with some limited drift data that can openly and globally discussed and revised than to trust on some pings from a very questionable source (the AES is questionable not INMARSAT), that necessarily lead to assumptions reaching far out in the inventory of psychiatric clinics.

    It is not science to grab some data, because they are the only ones we have! Period!

  7. @ALSM

    Not surprised by your comment. What is true is that you never had a scenario. What you have is a spreadsheet. Huge difference.

    While I truly respect your analytical skill, your spreadsheet suffers from a significant negative attribute. Nothing has been found there, and there is not a single thing that supports it.

  8. Dennis: Well, I have to add one more response. You have no idea what you are talking about.

    The breadth of discussion that goes on within the IG, and among smaller subgroups, is far greater than you can imagine. The IG and members thereof have looked in detail at path models that fit hundreds of scenarios ending from 40S to 40N. We have wide ranging debates that consider every angle. You are just dead wrong about your insulting “spreadsheet” description.

    You are also spreading false information about drift models. It is not true that no drift model supports the southern end of the 7th arc search area. In fact, the only official work in this area (CSIRO) is 100% consistent with the current search area, and both Henrik’s and Metron’s drift work supports the 7th arc only slightly NE of most estimates…not 9S.

  9. @ALSM

    The Metron work uses the ATSB probability density as a Bayesian prior. I am surprised you would site it.

    No matter. Let’s quit arguing. It will go nowhere. BTW, I think Hendrik’s drift model is spot on, and said so earlier. It suffers only from a lack of input data.

    I thought Brock’s paper (in particular Fig. 3) on the CSIRO work would generate some pause relative to supporting your terminus 100% (whatever that might mean).

  10. @alsm

    you have yourself acknowledged few months ago that northern than 20S is possible if you remove assumptions about constant speed/altitude

    ATSB relies on AP/cruising speed theory because they suppose the PIC would want to maximise the range to nowhere for no reason at all, which is ridiculous assumption by itself.

  11. This feels more like the X files every day.

    Either the crew was compromised, the crew was responsible or the plane severely malfunctioned. Since the plane continued flying and behaving as if it were under control, a malfunction seems the least likely. This also seems to indicate that the plane was not electronically hijacked and a person was responsible, either the crew or a perpetrator.

    Much about mh370 hinges around the turn. The last ATC contact was the sign off to Malaysian ATC at 1:21. Immediately after that and in the perfect spot (between ATCs) mh370 goes dark at 1:22. At 1:23 the plane initiates a full standard turn and heads west. If this were a controlled emergency you would not expect a steep turn, as this results in the highest risk and the greatest loss of altitude. You would expect the plane to descend and move toward a runway. The pilot had plenty of altitude as a non-powered glide would easily take him 100 nm. But the plane stays on course, only fluctuating in altitude (FL310-350) an indication autopilot was not being used.

    At 1:30 mumbling and static was heard as contact was attempted by another plane. mh370 did not respond. Three minutes after passing beyond Malaysian military radar (2:22) a log-on request was sent to ACARS (2:25). They knew they were in the clear, and they powered back up.

    Who did they contact and what were their demands? I’ve speculated that it was related to the Uighurs. Did they hold off the Sumatran coast (Geomar flaperon location) waiting for a response? Were they shot down under the guise of being terrorists? All we know from Inmarsat is a time and a distance, we don’t know a direction. All of the assumptions have been based on last known speed, heading and altitude. A semi-circular 6 hour suicide flight plan just doesn’t make sense. What do the BFO and BTO look like if we assume a much lower speed? Georesonance (Bay of Bengal and on the 7th arc) might be making more sense along with the sighting by a woman in a passenger plane seeing a jet in the water there.

  12. @Dennis,

    OK, I have had a quick read of your blog, and your analysis. Seems me you have introduced a number of assumptions that are no more than guesses to get to your end point too.

    How did you establish the start point for your calculations at 19.40, and how do account for the time and distance to that point from about 18.25?

    Why did a very experienced pilot overfly the intended destination only to run out of fuel?

    Why the large speed variations after (and likely, before) 19.40.

  13. Based on recent comments between @DennisW and others, I looked more closely at Dennis’ path. I have these comments:

    1. I could quibble with the exact values that Dennis proposes for speeds, track, satellite position, handshake times, etc., but after taking some liberties, I can reconstruct a path that satisfies the BTO and BFO criteria and can get to the 7th arc at about 16S latitude. With more fiddling, I can probably get further north.
    2. To reconstruct a path in this manner, the path must start at a more northern latitude than the last radar point at 18:22 and the effective speed between 18:28 and 19:41 is very low. Therefore, this path requires a “northern loiter” during this time interval.
    3. In order to match the sat call BFO at 18:40, the plane had to be either traveling south at this time or rapidly descending.
    4. This path is similar to paths considered by Henrik and Niels where the radar data was completely ignored.

    So certainly there is a class of solutions involving positions at 19:41 north of the last radar data point that end further north on the arc. Dennis is correct in this respect.

  14. @Flitzer-Flyer: I did not see your post until after I posted mine. I apologize if I repeated some of what you already said.

  15. Dennis, I’ve asked before, but your route implies a set of speeds which in turn may imply a range of altitudes. Are these compatible with fuel exhaustion by 0019, and if not, what would be the explanation for the partial handshake at that time?

  16. @Flitzer

    “Seems me you have introduced a number of assumptions that are no more than guesses to get to your end point too.”

    the same way ATSB did it, AP path to the SIO although being a very good match for BFO values could still be just a sheer coincidence

    “Why did a very experienced pilot overfly the intended destination only to run out of fuel?”

    99% of people assume that pilot had OK from all the crew and passengers to divert from the route while it doesn’t have to be so. My guess is that some conflict ensued after entering IO(likely between the pilots) and if you add possible nav/comms glitch(es) after fiddling with E/E bay CBs(somewhere around FMT point) then you can imagine thousand of scenarios that happened between FMT and 7th arc which made approach to CI very hard for whoever was left flying(think about captain being neutralized and inexperienced F/O getting controls with the help of other crew/passengers).

  17. @Victor, DennisW, Flitzer-Flyer

    What i like about a Christmas Island scenario is there is a clear perceivable motive: asylum in a western nation.

    Thats something that was done very often before in the history of hijackings, where they were the choice for escaping from rogue regimes, maybe many of them orchestrated by western agencies.

    But did you ever consider the circling model of Coppernickus on reddit?

    https://www.reddit.com/r/MH370/comments/3dmynq/looking_at_circlinglanding_solutions/

    This leads us into the CI direction and middle JAva in a very promising way. Any thought about marrying that with your theory DennisW?

  18. @Victor: if the plane loitered at/near Sumatra – and the JIT knew about it – then much of the past year’s search has been a charade, because a few minutes loitering translates into a dramatic shift east in arc7’s SW fuel limit.

    So, if – IF – Geomar has nailed the drift probabilities – and if you’ve nailed the signal data analysis – then the only way to reconcile the flaperon to arc7 (while still holding to the assumption the search has been in good faith) is via path circuity about which the JIT is, to this day, ignorant.

    (You know me – I feel the evidence suggests we’ve been left to chase rainbows, to hide a deeper truth. But I’ve always tried to give establishment/expert theories the benefit of the doubt.)

  19. Well if he really did it on a whim I guess it’s possible he contemplated about landing on Banda Aceh (therefor the loiter) but decided CI was better solution.

  20. @Cosmic Academy, I think I also brought forward this argument in previous threads – seeking asylum in a Western oriented democracy. That would make a lot of sense if the plane either actually made it to CI or ended up in the vicinity of CI. But it didn’t. It overshot CI considerably to the Southeast. And nobody ever tried to make contact and tell that they were in some trouble. And the idea that somehow passengers panickec and brought the plane down doesn’t ring true at all. It’s somewhat unlikely in the first place and wouldn’t make sense at all so late in the process, when they had made it already and only neede to land.

  21. @littlefoot

    we don’t know what happened and who was left flying, nor if he had fully working flight systems

    maybe it was not Captain but F/O who entered E/E bay and tried to confuse the Captain by switching off nav/flight systems/whatever to try to persuade him to open the door? Would that also explain mysterious satcom reboot?

  22. @Brock: At this point, I put little weight on a drift model’s ability to point us in the right direction. Also, saying debris found on La Reunion is most likely to be from a particular location is almost meaningless, in my opinion. I am much more comfortable with the kind of conditional probabilities presented by Henrik, and even that approach is fairly limited with only one piece of debris found. We need more debris to do any meaningful drift analyses. All of the drift analysts acknowledge this.

    But to answer your question more directly, if the plane crashed further up the arc, then there is “missing time” that has to be explained by a loiter. That was the point of a graphic I created many months ago that questioned the search centered around 30S, where I related the supposed (straight) path to missing time of 52 minutes. Another possibility is that the radar data is not correct, and the plane took a more northern path across Thailand.

  23. If GEOMAR theory is right, how could they get wreckage of a boat from Exmouth coast to Madagascar in 8 months?
    The back modeling does not show potential origin close to Exmouth coast

  24. After 21:40 the speed in DennisW’s path (550 km/h) at 8000 m altitude is below holding speed of 240 – 196 kIAS (depending on weight). Fuel exhaustion would not have occurred until well after 00:10 UTC.

  25. @alex: a boat might well be far more sensitive to wind than a near-fully submerged flaperon.

    Hence my question above, seeking a best estimate as to how sensitive this particular object’s drift would have been to wind. (Upon reflection, I think a proper answer would need to come from robust field trials, not crowd sourced estimates.)

  26. @Gysbreght

    we don’t know the exact flight envelope except until IGARI, we can make just rough estimates on fuel

    the plane likely changed the speed several times until FMT and possibly even further

  27. @StevanG:

    From VictorI’s analysis of the radar data we know that the airplane flew at a constant groundspeed corresponding to M.84/FL340 between the turnback after IGARI and the loss of primary radar at 18:22 UTC. We also know that the fuelflow at LRC speed is 12% higher than at holding speed.

    Where did I say that I did ” exactly know fuel levels at 21:40″?

  28. This corresponds to the witness of Mike McKay. I am the same opinion the MH370 crashed in the South China Sea near the IGARI waypoint. The flaperon could drifted to the Reunion Island from South China Sea trough Sunda Strait and follow Equatorial sea currents west of Java.

  29. hmm I missed that admittedly, however I have read somewhere that at 18:22 the plane was actually flying at FL260 judging at radar capabilities and distance so it I guess flew slower at that time… I leave possibility that I’m wrong though

  30. I forgot that the FCOM Holding FF includes 5% additional fuel for holding in a racetrack pattern, so the 12% I mentioned is actually about 17%.

  31. CSIRO publicly admits that their “1.5% of wind” assumption was back-solved so as to achieve the desired result, i.e. to get the flaperon all the way around the horn.

    I’d like to see a “0.0% of wind” version of CSIRO’s latest, to understand the sensitivity of this assumption within this range. I’ve e-mailed them, asking for this.

    I also think we should take CSIRO’s (admittedly very pretty) graphs with a grain of salt until they empirically demonstrate the applicability of that very generic wind assumption to this very particular drifting object.

  32. @all

    I did not run away, just went for a long walk on a beautiful day on the Cali coast. So this post combines science, metaphysics, and personal baggage.

    First, the metaphysics. What can the ISAT data actually tell us? If one looks at the example flight paths in the ATSB reports it is clear that there are very large differences between the flight path predicted by the ISAT data, and the actual flight path logged by ACARS reports. One has to assume that these example flight paths were not selected to be worse case. We have never been provided ISAT data along with accurate positions to form our own conclusions relative to accuracy. For this reason I have always regarded the ISAT data as a feasibility constraint, not an accurate predictor of where to stick a pin in map. Back in the Duncan days the only allowed solutions were fixed speed and heading after the FMT (the location of which is still debated). So I made such a path, and it came out about 500nm North of the IG consensus. No matter. I did not put a whole lot of effort into it, but I suppose I just wanted to bark with the big dogs.

    Relative to starting my latest path at 19:40, it was personal baggage related. At the time I made that post I was extremely annoyed with the radar data, and the antics of many people who were making significant efforts to make sense of it using various overlays – SkyVector etc. I finally said screw it. I was not going to try to unravel something that should be simple if we were given the raw data which we never were given. Mea culpa. I was upset and that is the simple reason why I started where I did. My earlier CI plots included BTO/BFO data before 19:40. At one point I even had the aircraft passing Kate at a location consistent with her observations.

    As far as fuel range, altitude vs speed, and why an experienced captain would run out of fuel. Those are all patchable IMO. Pick an altitude, speed, and heading that makes you happy. All I care about is feasibility. I don’t know who was flying the plane. My assumption is Shah, but it could have been someone else. In any case, the time in the left seat, the anxiety, and the stress would have been significant. None of which are conducive to being “on your game”. Human error can never be eliminated no matter how good you are. It may even be that the fuel consumption accumulation was corrupted by bus switching. I simply don’t know why the aircraft ran out of fuel under the circumstances of my scenario.

    The sea level monitoring data (pointed out to me by StevanG) which can be found on the same blog I linked to came as quite a surprise to me. I readily admit it is weak, but a coincidence like that is difficult to ignore.

    My main point is that we should all remain open minded, and consider all the information available to us. To try to force fit everything to a best fit ISAT model using questionable flight dynamics is a mistake, IMO.

  33. One remark concerning the end of flight scenario you are just discussing concernin g Dennis Ci plot.

    You have to make up your mind. It was either a zombie flight running out of fuel with no soul in control, or somebody was in control of the flight and had no reason at all to run out of fuel. In the latter case the reboot could have different reasons other than running out of fuel. I’m not aware of any hint, that the reboot could have been “only” due to fuel starvation.
    I’m still at a loss to see a pilot, suicide or not, fly for some hours until the fuel load has gone to finally crash somewhere. Top of this inconclusive scenario is to try a smooth landing after fuel starvation.

  34. @RetiredF4

    I can’t make the zombie flight scenario work relative to motive or causality. Particularly considering the aircraft maneuvers prior to the FMT.

    Running out of fuel could be an instrumentation error or a human error. Or, as you point out, it did not happen at the 7th arc. Don’t really know.

    The next shoe to drop, IMO, has to do with the BEA forensics on the flaperon damage. That will be very telling relative to what may have happened at or shortly after 00:19.

  35. The flaperon drift models can only try to answer where the flaperon hit the water.

    It does not necessarily answer the question of where the plane as a whole hit the water.

    For the two to match, the flaperon had to be attached to the plane at or very near the time of final impact of the plane.

    Cheers

    Will

  36. The Drift Analysis Models being used are based on tests of buoys with sea anchors. I cannot imagine that a flaperon floating at the surface without a sea anchor would follow a similar path. It would be interesting to see how the floating AF447 rear stabilizer drifted versus predictions. I think a biological comparison of the barnacles on boat versus those on the flaperon would be a better indicator of where the flaperon started its journey.

    LRC is not a constant speed. It is reduced as the plane gets lighter where less lift is required or, if there was a conscious pilot, he could have increased the flight level. IF AP is used for LRC, as Gysbreght demonstrated using his May 11 graphs, there is only one (or possibly two) altitudes that generate the approximate 5.9 hour endurance after last radar at 18:22. The amount of fuel remaining at any time can be estimated using FCOM’s burn rate tables.

  37. con’t
    Fuel needed to be jettisoned in order to run out in the CI area. There was just about enough at 18:22 to go MEKAR>NILAM>POVUS>NIXUL>YPXM (CI)>Bali

  38. @Victor
    Could you please help with answers to a few questions (apologies if they have been answered before)
    1)In the GEOMAR graphic a region is shown with dashed lines (10N to 35S/ 80E to 110E) and comment in the key “Only model particles originating from this region 16 months before were considered” yet some particles (in orange) are outside this region…why?
    2)I had the same thought as Will – even if the model shows where the flaperon might have drifted from, is it possible that it separated from the plane much earlier in the flight? And how might that affect the flight trajectory especially if on AP?
    3)or is it possible that the flaperon separated from the wreckage after some months and so the 16 mths model may be way out?

    @All.
    We still wait of course for the yes, maybe or no announcement from the French, so the drift models are speculative at best at the moment, especially without further identified debris.
    Lets say we get a yes or a “pretty certain” it came from 9M-MRO/MH370 then I wonder if submerged debris can get lodged somewhere for some time and then move again with seasonal variations in ocean currents – so how accurate is the 16 mths anyway.
    The most useful message I get from the GEOMAR model and from Erik van Sebille’s own model (fig 6b in Brock’s comparison paper) is that there is potential for debris to drift across the equator from N to S and from a wide variety of origins. Why there is a major difference between these two models in that van Sebille’s shows major areas off Exmouth etc. as possible whereas GEOMAR’s doesn’t is unclear – any suggestions?

  39. @Lauren, maybe fuel was jettisoned – over the SCS sea. There were two oilslicks discovered when they started looking for the plane. One didn’t contain jet fuel. But the other one did contain traces of jet fuel and was only 20 nm away from the point of last contact. This area got searched and no plane or debris was found. That’s why nobody thought the oil slick was important or had anything to do with the
    plane. But an oilslick so near the point of last contact containing jet fuel might be worth to be considered. The fuel might’ve got lost unintentionally. But if that wasn’t the case, why might the perps have dumped fuel over the SCS? Maybe as a decoy – so that SAR would be focussed on the SCS for a few days?
    The problem with a CI landing scenario is that there wouldn’t have been a good reason for doing that. The pilot had to kill time in order to arrive at CI after dawn. But jettison fuel only makes sense if the plane was intending to land shortly after the turnaround because of an emergency, or if it was supposed to be flown until the tanks were empty so that the final location in the water could not be given away by an oil slick – like in the SIO ditching scenario.
    To jettison some fuel as a decoy and maybe in order to get rid of excessive fuel would make sense if the perps intended for some reason to take the plane to an area Southeast of CI in order to make a controlled crash or ditch after dawn and the plane was not supposed to be found.
    IIt’s not easy to come up with a complete and plausible scenario where a controlled crash or ditch in the Northeastern region of the Southern 7th arc makes sense.

  40. CosmicAcademy – “It is not science to grab some data, because they are the only ones we have! Period!”

    This is what mathematical dunce me was trying to articulate from early on when everyone was going data mad. I said we are going to need some luck but none were having it. They had what they had to work with but it was their choice to inject the air of certainty, and I’m now of the view there is nothing left in this world that might induce some to step back and question it.

    Drift modeling carries uncertainty, the BFO data carries the possibility that it could be flat wrong.

  41. Littlefoot – I agree. If you believe that the 00:19 contact was due to a restart of the AES after fuel exhaustion of the second engine, it is unlikely that they could have jettisoned fuel in the SCS and still flown 7 hours. However, to have run out of fuel near CI, they would have needed to jettison fuel somewhere along the way or flown in circles at a high speed while at a lower than typical altitude.

  42. The flaperon would not separate at any normal speed. It had to be a high speed…higher than the certified max (330IAS?). If it seperated while in flight, it probably happened within the last 2-3 minutes of flight, or ~00:19. Thus, the main POI and the flaperon POI were probably within a few miles, not hours or before 00:19.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.