Guest Post: Stochastic Simulation of Hypothetical MH370 End-of-Flight Scenarios

Brock graph

Note: This paper was prepared by noteworthy independent MH370 investigator Brock McEwen, who set out to answer the timely question: has the soon-to-be-concluded search of 60,000 square kilometers of seabed in the southern Indian Ocean falsified the flight scenario proposed by the Australia Transport Safety Board and endorsed by unaffiliated researchers belonging to the Independent Group? Through extensive modeling, McEwen calculates that the search has covered more than 99 percent of the potential endpoints predicted by ATSB and IG flightpath scenarios. He concludes that “if both the Inmarsat signal data and its interpretation by search officials is valid, then the search should have turned up wreckage by now. This offers strong circumstantial evidence that either the Inmarsat data or its interpretation is invalid. Accordingly – unless search officials know something we don’t – the announced decision to spend another year searching around the improbable edges of a discredited theory (while this paper does not address surface debris, its absence likewise serves as strong Bayesian counter-evidence) is an extremely poor one.” — Jeff Wise

Executive Summary
This paper stochastically models MH370 end-of flight dynamics, and finds no evidence to support last month’s announced continuation of status quo search strategies. This finding is consistent with a documented pattern of decisions by MH370 search leaders which make no sense. A rigorous, independent audit of both the Inmarsat data’s entire chain of custody and MH370 search leadership is recommended.

In a joint statement issued April 16, 2015, Malaysian Transport Minister Liow Tiong Lai, Australian Deputy Prime Minister Warren Truss and Chinese Transport Minister Yang Chuantang indicated the search for Malaysian Airlines Flight 370 (MH370) would continue to focus on the deep-sea search in the Southern Indian Ocean (SIO):

“Should the aircraft not be found within the current search area, ministers agreed to extend the search by an additional 60,000 square kilometres to bring the search area to 120,000 square kilometres and thereby cover the entire highest probability area identified by expert analysis,” they said in a joint statement. “Ministers recognise the additional search area may take up to a year to complete given the adverse weather conditions in the upcoming winter months.”

In committing to this extension – described graphically as a modest search zone expansion in all four directions – search leaders implicitly assume that

  1. the Inmarsat data is sufficiently accurate and precise to permit interpretation by investigators AND
  2. the conclusion they drew (flight to fuel exhaustion, followed by pilotless spiral to impact) is correct, YET
  3. the wreckage this conclusion predicts remains outside the areas already searched by side-scan

This paper tests this hypothesis, via stochastic simulation of plausible post-fuel exhaustion flight paths.

You can read the full paper here.


13 thoughts on “Guest Post: Stochastic Simulation of Hypothetical MH370 End-of-Flight Scenarios”

  1. Brock – if the spiral goes in the bin so does the AP assumptions. That’s a can of worms and the search is now nakedly politically driven(China). Was it always? Or at least politically contaminated?

  2. Wise quote below:

    “He concludes that “if both the Inmarsat signal data and its interpretation by search officials is valid, then the search should have turned up wreckage by now.”

    The Inmarsat data could be perfect valid as well as the interpretation of that data. What places the aircraft in the SIO is the assumption by both the IG and the ATSB analysts of a constant heading AP flight mode. If heading changes are allowed there are many places (not just the SIO) that can satisfy both the BTO and BFO as currently interpreted.

  3. @DennisW: I have to agree. It is my opinion that if the search in the SIO is to continue, it is more valuable to recommend limits to the search area by progressively removing assumptions about navigational mode, turn times, and other constraints, rather than continuing to try to determine the end point with increased precision.

  4. I’d like to see an informed analysis of the likelihood that MH370 would be found if it was within the area searched. Brock seems to assume that there is a 0% chance of the debris being overlooked. AF447 ultimately was found in an area already searched, and that search was easy compared to this one.

  5. @Matty: I’m just your friendly neighbourhood auditor. The April decision – like so many before it – is not supported by the actual data. Why this is so – and who is to blame – are matters for professional investigative journalists to unearth. Hopefully soon.

    @Dennis/Victor: that readers would interpret “interpreted” to include (wrong) assumptions was my (wrong) assumption. Will clarify in final draft.

    @Bruce: see discussion in “How Crazy am I..” pp. 20-something. To the extent this is a material risk, doesn’t that FURTHER counter-indicate a decision to search ANOTHER 60K km^2?

  6. @Matty:


    “That’s a can of worms and the search is now nakedly politically driven (China). Was it always?”

    I’m curious: in a discussion of the political dimensions here (and there are many), how does it include China but not Australia, which is the lead on the search?

  7. Nihonmama – As a political nerd based here who reads every utterance daily I’m pretty sure that the Australians and Malaysians were ready to drop it, China had other ideas, Malaysians got their chain jerked and the Australians were in a bad spot if they pulled out. Domestically it’s PR gold for the Chinese govt.

    At this point the search is extended without any clear rationale or direction, they are just going wider.

  8. @Matty:

    “Domestically it’s PR gold for the Chinese govt.”

    So would that PR play be a remediation effort? Because let’s recall how China has treated the Chinese next of kin:

    Posted October 3, 2014 at 7:51 PM
    “MH370 What We Know Now”
    (URLs redacted):


    Food for thought.

    The Chinese government, which was SO vocal early on, and critical of the Malaysians’ handling of the investigation — then turns on Chinese relatives of MH370 pax.

    The MH370 related protests in China weren’t directed at the Chinese authorities. So what changed?”

    Add let’s add that China:

    – forbade media outlets in China from reporting on MH370 independently.

    – extended no support or financial assistance to the Chinese next of kin who returned to Malaysia to protest back in February (some were sleeping on the street).

    “At this point the search is extended without any clear rationale or direction, they are just going wider.”

    So who’d benefit by writing (or causing to be written) a check for a boondoggle?

  9. Nihonmama – For the cost of sending a few ships down to the search area the Chinese govt can project the nonsense that they actually care about their people. All they really care about is social order, and the optics of ending the search earlier rather than later were not good. That’s one angle.

    Continuing the search also the suspends the official MH370 narrative right where it is. I see that Ghyslain Wattleros is of the view that his govt(France) knows something. He bases that on what they haven’t said or done so far. You could throw in a few other govts as well and as I’ve said umpteen times – it doesn’t look right.

  10. I don’t think the new search area of is based on any new or complex analysis of the data. The original June 2014 ATSB analysis stated the “priority area” area to be 50nm in width based on “Loss of control study, tolerance of the arc, balance of uncertainties…”. At that time the length of the “priority area” was 350nm/650km (which gave a total of, used for the original contracts as stated in the June report). Later work in the October report extended the length of the search area to 1100km which meant the width had to be thinned down to give the area budgeted for the first phase. If the original width is reinstated to “…cover the entire highest probability area identified by expert analysis…” (April 2015 Tripartite Meeting statement), that gives a total area of I don’t see any reason why the loss of control study conclusions will have been devalued since last year.

    Earlier this year Martin Dolan said “we would move further north, but we’d also have to contemplate whether we would go wider” so that might indicate that the Northern edge of the zone could move 200km to take in lower statistical significance potential routes, or perhaps in combination with an adjustment to the Southern edge, for a final total of An increase of width of 10nm beyond the 50nm would have the same effect.

  11. @Richard: thanks – final draft will clarify that, technically, my conclusions apply only to a specific subset of the “status quo theory”:

    – roughly an 18:28-18:40 FMT,
    – limited path circuity,
    – near-cruising speed/altitude until fuel exhaustion, and
    – pilotless spiral to impact thereafter

    Were this extension to 120K km^2 the only decision being audited, I would have to cut the ATSB far more slack than the IG in my conclusions and recommendations, because a charitable read of their search zone dimensions is that they’ve simply been comparatively PRUDENT, as you suggest. Perhaps there WAS path circuity near Sumatra – we the general public certainly can’t yet rule that out.

    But at this point, I admit to being incapable of viewing this decision in isolation. Indeed, I feel this decision SHOULD be seen as part of a PATTERN of obfuscation, deception, and baffling search decisions. In THIS context, the ATSB(‘s ventriloquist) has earned a much SHORTER leash than have its IG counterparts.

    I hope I’m wrong.

  12. I recently tweeted out a final version of the above report:

    Post-publication test #1: I extended model to 16 minutes (from 13) to verify truncation wasn’t affecting results (it wasn’t: “within search area” still rounds to 99%, “outside” to 1%, “inside” to 0%)

    Post-pub test #2: As a final sensitivity test, I doubled the volatility parameters governing the amount of random noise in each of heading, vertical speed, and horizontal speed – ALL AT ONCE [note: this would disperse the distribution of turn radii and minutes aloft well beyond levels suggested by IG (2-15 nmi radius, 4-13 minutes) and ATSB (low bank angle, up to 12 minutes) flight sim results, so I feel this calibration OVERDOES the noise element considerably.] The modeled probability of impact within the search zone is STILL roughly 95%.

  13. @Nihonmama and @Matty

    First post here.

    I agree that there may be political influence on the search. Considering how China is rattling its sabre around South China Sea (building bases on disputed islands) and its rapid expansion of its naval fleet in the past couple of years, I wonder if the insistence of continued search represents not a PR exercise but an attempt to collect bathymetric data in areas that it usually wouldn’t have access to for military purposes. Just my 2 cents as a noob.

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.