Ghyslain Wattrelos, who lost family members traveling on MH370
The article that follows, translated from French, appeared on the website 20 Minutes with AFP:
While all other countries have stopped investigating, France is not giving up. The investigators in charge of the case of the disappearance of flight MH370 of Malaysia Airlines hope to soon go to the United States where crucial investigations must move forward, Ghyslain Wattrelos, who lost four family members in the disaster, announced Thursday.
On March 8, 2014, 239 people disappeared after taking off from Kuala Lumpur aboard a Boeing bound for Beijing. Other than some debris that seem to belong to it were recovered in the Indian Ocean, no trace of the 239 passengers has ever been found. Ghyslain Wattrelos has notably lost his wife and two of his children in the plane’s disappearance.
On Thursday, he was received with his lawyer Marie Dosé by the judge of instruction in charge of the judicial inquiry opened in France, to take stock of the investigations and the avenues to explore by the investigators.
Questions for the FBI and Boeing
Among the priorities that mobilize the investigators, a trip to the United States is “back on the agenda”. A previous trip was cancelled in September, even though it was part of an international rogatory commission launched in October 2017, they announced at a press conference Thursday. According to Dosé, this trip had to be canceled, as the US authorities opposed “confidentiality clauses” and then the “industrial secret” of the manufacturer Boeing.
“We are a little angry and now we want to say stop, it is time that the United States really cooperate on this issue,” responded Ghyslain Wattrelos. “It is necessary to go there because there are three entities that hold important information for understanding what happened on this flight,” he continued. Starting with Boeing and the FBI, even if the investigators seem to have obtained assurances from the intelligence agency that they could be received, he said.
A hacking of Satcom?
But attention is now also focused on a third entity, a company uncovered by investigators. The challenge, according to Ghyslain Wattrelos, is whether it sells software capable of reprogramming or even hacking the Satcom, the antenna that communicates to the Inmarsat satellite signal from the aircraft. “The essential trail is the Inmarsat data. Either they are wrong or they have been hacked,” he says. However, these satellite data are essential to better understand the trajectory of the aircraft.”
The release of the Malaysian investigation report in July had dampened Ghyslain Wattrelos’ hopes, but since then French investigators have suggested new theories to explore. In particular, the investigators found “inconsistencies” in the Malaysian investigation’s official report, and the presence of “curious” passengers, whom “we should continue to investigate”. Among them is a Malaysian traveler with a troubling profile: he was seated under the Satcom module, and proved to be an expert in aeronautics, according to Ghyslain Wattrelos and his lawyer.
[End of story]
It’s not clear what additional information the French investigators expect to obtain while in the US. Boeing has cooperated with the Annex 13 investigation team, and is unlikely to provide private French investigators with data that has not already been made public. Meanwhile, the FBI is unlikely to release information on matters related to ongoing or past investigations.
The mysterious “third entity” referred to by Mr Wattrelos that might be selling software capable of maliciously altering SATCOM data is also unknown, although there are a handful of companies in the US and Canada that supply hardware and software for designing, building, and testing parts of the Inmarsat network.
Independent investigators that are studying this mystery are at an impasse. Although the overwhelming consensus is that MH370 did indeed crash in the Southern Indian Ocean, the considerable efforts of official and private investigators have not succeeded in locating the debris field on the seabed. The data we have, notably the satellite data, is imprecise, so additional data is needed to reconstruct the trajectory of the plane.
There is always the chance that during Mr Wattrelos’ visit to the US, some new evidence or insights will be uncovered that help us to better understand the disappearance and to find the plane.
More likely, the existence of helpful new information will be found in Malaysia.
Update on October 23, 2018
The following article, translated from French, was published in the L’Essor, which is the [unofficial, self-described] French journal for the military police.
MH370: Gendarmes waiting for a green light for a trip to the United States
The investigators of the research section (SR) of the Gendarmerie Air Transport (GTA) are waiting for the green light from US authorities to travel to the United States related to the case of the disappearance of the flight MH370 Malaysia airlines, more than four years ago.
The SR of the GTA is responsible for the judicial inquiry, opened in France, on the death of 239 people, including four French, aboard the Boeing who disappeared on March 8, 2014 after taking off from Kuala Lumpur for Beijing. Debris likely to belong to the aircraft have been discovered in the Indian Ocean but no trace of the 239 passengers has ever been found.
The gendarmes work within the framework of an international rogatory commission launched in October 2017 by the French justice. A trip to the United States, scheduled for September, had to be canceled and French investigators are waiting for the green light from the US authorities.
The investigators want to check overseas if a US company sells software capable of reprogramming or even hacking the Satcom. This system is used by airlines to transmit messages about the state of the aircraft in flight and its communications.
[End of story]
This story, if true, suggests that the French judiciary system, and in turn the military police that are assigned to the case, are seriously considering whether the Inmarsat data was corrupted by a malicious intrusion into the SATCOM onboard MH370. Considering that the Inmarsat data is consistent with the aircraft crashing in the SIO near the 7th arc, and that the timing and location of the recovered parts from the aircraft also suggest that the aircraft crashed in the SIO, to doubt the Inmarsat data implies doubting the veracity of the recovered parts. This is the first time a government investigative body is known to be seriously considering a hack of the SATCOM combined with planting of debris.
To say the least, most private investigators believe there are more productive avenues to pursue. However, an honest, competent investigation of any type is to be welcomed, and there is always the chance some new evidence or insights will be uncovered that help us to better understand the disappearance and to find the plane.
@Victor
The French have never published a report on the flaperon damage. Was it contact with the water or flutter that caused the edge damage? What was concluded about the barnacles?
From your post above.
“We are a little angry and now we want to say stop, it is time that the United States really cooperate on this issue,” responded Ghyslain Wattrelos.
The French have been the least cooperative of any nation investigating MH370. Why does Wattrelos point fingers at the US? Typical French BS attitude I suppose.
@DennisW: The Safety Investigation Report did provide us with more details about the French investigation of the flaperon.
That said, there clearly is a contingent in France that suspects that the Five Eyes countries have conspired to hide a shooting down of MH370, and past statements by Mr Wattrelos suggests that he concurs. Mr Wattrelos has noted in the past that the US has controlled all the important data: the simulator data was investigated by the FBI; the analysis of the radar data was conducted with US assistance; the plane manufacturer was US-based Boeing; and Inmarsat is a US defense contractor. He also believes that MH370 was under satellite surveillance by the US, and therefore the US knows where it is.
I don’t think that Mr Wattrelos’ visit to the US will uncover new information, but I’d be happy to be wrong.
@Dennis
The french report on the flaperon is available in full (256 pages) as part of the final Malaysian safety report on MH370.
You will find the report in Appendix 1.12A-1 and 1.12A-2 from page 496 to page 751 of the Malaysian report.
@V
@Victor
If we could meet with FBI ourselves, we would ask if there could be more info recovered from the simulator runs in theory. As you know, much of the data from the runs is missing, and so on. Presumably, if affirmative, that leads to Malaysia needing to approve, which probably would not happen.
@Victor
Mr. Mohammed Kahairul Amri Selamat, a Malaysian, was seated in seat 29A on MH370. This is right under the SDU, which is housed in the E11 rack accessed through a ceiling panel on the left side of the door 3 cross aisle area.
Mr. Selamat worked as an aviation engineer for the Swiss based Execujet Aviation Group. He was 29 years old, married with a toddler and had just bought a new home. He loved to play with remotely controlled model aircraft.
Mr. Subramanian Puspanathan, a Malaysian, was seated next to Mr. Selamat in 29C. He was 34 years old and was a consultant working for Petronas. Just behind in 30C, we find our Iranian on a stolen Austrian passport in the name of Christian Kozel.
Mr. Selamat has been suspected of taking the aircraft over remotely via the SDU by Ghyslain Wattrelos.
@Victor
Yes, I saw the report. Key conclusion (unless I am missing something somewhere else) cut-pasted below. It amounts to no conclusion, and a jab at Boeing. I never did see any detailed results regarding the French barnacle examination.
The significant damage was appraised for the purpose of determining a scenario for separation of the part from the aeroplane. In the absence of data from Boeing, and despite the deterioration of some fracture surfaces, a hypothesis was nevertheless formulated: taking into account the results of the examinations, it appears that the flaperon impacted the water while still attached to the aeroplane and that at the time of the impact it was deflected. A fall simulation for the flaperon with an initial speed corresponding to that of an aeroplane in flight could definitively exclude the loss of the latter in flight. The little data supplied by Boeing did not enable the examination to be progressed by making calculations that would have made it possible to confirm or reject the proposed hypothesis.
@Richard
You said…
An analysis of beachings on mainland Africa (excluding the unknowns surrounding the finds on Madagascar) shows a possible MH370 end point north of 25°S.
Thx. I said I was not sure what to do next, but one of the possibilities is not to search wider at any latitude. My choices are not to search at all (usual conclusion) or to search farther North.
Note: Once again I correctly used “farther” instead of “further”. Ami has beat me up repeatbly over that issue.
Addressing comments from the previous now closed thread.
@Richard
I agree that it is still early in the OI search for San Juan, and I was not intending to suggest it had failed.
However, it is clear that “their” initial “priority areas” have been “cleared”, with no result.
One must therefore presume, since they are now widening the search, that the initial assumptions for defining the initial search box were wrong.
@Victor
The so called “radar information set” can be divided into three.
(1) Lido Slide
(2) Military Radar
(3) Civilian Radar
Re LIDO
we got Lido first, specifically, the enraged Chineese NOK got it at the hotel.
It was never “published” oficially by Malaysia.
It was never verified or explained by Malaysia.
If it was legit, why not ?
I decided, (given the early time line of events) that it was an initial, very clumbsy, very disingenuious, attempt, to “create” something, “anything”, in an attempt to placate the enraged Chinese NOK, into thinking, that it was hijacked “by someone” and “went somewhere”, and that there were still hopes that their relatives could be alive.
I dismiss it outright.
Re Military Radar.
You yourself received promise after promise, from the Malaysians, directly, that it would be released, did you not ?
It never happened.
It never will.
The statements that were made by Malaysian officials from “indications of a possibleturn-back”, to “Palu Perak”, to “10nm beyond Mekar” etc were a farce from the beginning.
Not only that, but at every itteration of the story, from the RMAF chief’s farcical press conference on, with his smugly spoken line, almost with a laugh and a smile “we are not sayying IT IS MH370 – OK” says it all, don’t you think ?.
None of it has ever been substantiated, not by one single skerric of “evidence”.
The military radar data, if it exists at all, and if it supported the Malacca Strait flightpath, would have been released. The fact that it has not, is damning.
All that was ever “presented”, was the ATSB’s presentation of the “ten second data” that Malaysia gave to the DSTG.
Malaysia NEVER PRESENTED IT, IN ANY OFFICIAL STATEMENT OR DOCUMENT.
The SOLE SOURCE, is the Australian ATSB, and in a form that was not verifiable.
Why should we beleive it ?
I dismiss it outright.
Re Civilian Radar
The civilian radar, if it actually is MH370, “fits”, well enough with the log on ATTEMPT of Hamid’s phone, to be credible.
Although the swing NW past Penang Island fits the narrative to lead onto Vampi and N571, it does not go far enough to prove that it actually “continued” on to Vampi.
It suggests, stronly suggests, but DOES NOT PROVE it.
Just to backtrack to the Lido Slide, and the famous white circle.
I think it is probable that the aircraft initially swung NW as indicated, but then swung left to WSW when outside the recorded coverage of the Butterworth TAR.
I have seriously considered modifing my “via Medan route”, to one from a NEW start point just after the end of the civilan radar data.
That would take it over Sumatra through an IndonEsian radar hole if it was below about FL280, and it can still reach the 18:25 / 18:28 arcs from that position.
The fact that it would then not actually be visible to any Indonesian Radar fits with their denials as well, and since I think we all agreed that if the Ache turn was true, it would have been in plain sight to the Sebang Radar, one could support such a modified cross-Sumatra route. I have not firmed up on the idea yet, and have not expended much work on it yet, only a bit of “what if” doodling.
@DennisW
Re Tim Pardi.
My take on that, is that Z’s primary interest was more likely directed at her child, rather than herself.
Perhaps Tim did not accept that, most women would not, and she broke it off. We will never know. She has refused to comment.
Alternatively, I think Z, having planned, and having decided to execute his mission, wanted to “disengage” from the child, before the event, for the child’s sake. The easiest way to do that, would be to precipitate the break from Tim himself, so that Tim could tell the child that Z and her had decided to split, it had nothing to do with the child.
If that were the case, it would explain her steadfast silence best, since she would not want the child to know the awful truth, if it was the truth.
It could be one, or the other, or perhaps even a bit of both.
@TBill
I think Z’s intent was only to be the catalyst to get the ball rolling.
He knew that he could do that, without directly damagining his family’s reputation. He knew that there would be insinuations that it was him, but he was confident that his plan was watertight enough to ensure that there was no “proof” that it was him.
Z knew full well how the political system works in Malaysia.
He knew full well that normal political action would change nothing.
He adopted a position where he decided that a black swan event, that would rock the system, was the only hope.
Razak and Co knew what had happened, by whom, and what the intent was, rite off the bat.
Their problem, was how to play it, how to contain any fallout.
The last thing they would do is come out and say Z wanted them gone and did this deliberately.
Z knew that too, which helps protect his family.
Z was smarter than Razak.
Z “played” Razak at his own game.
@Ventus45: For a number of reasons beyond what you presented, I question the military data at the turnback at IGARI and the data presented to the NOK in Beijing. However, I have no reason to question the civilian radar data nor to question the Inmarsat data. I think any viable reconstructed path at a minimum has to satisfy the civilian radar data and the Inmarsat data. That would appear to be a challenge for your Medan path considering the BFO values at 18:28, unless there was a climb at that time.
Long time lurker, first time poster here…
There is much speculation as to why someone suicidedly intent on crashing an airplane into the ocean might put it into a steep dive followed by a shallower glide from the mid altitudes. As a 777 captain, this indicates to me someone conflicted about their approaching, inevitable death. On the one hand, he would want to end it quickly, without suffering, thus the high speed impact. On the other hand, he would also want to relish the last moments of his life, slowing gliding silently towards the ocean below, doing what he loved the most for the last time. I can imagine a scenario where, once the engines quit, he pointed the nose down until the airspeed built up to a point at which he was afraid that the airframe would break up, and then changing his mind and pulling back into a shallow descent. Whether this scenario fits into your equations is way over my head, but I thank you all for this fascinating blog.
@Victor
I am not questioning the 18:28 Inmarsat data. My path SW from Penang to/via Medan does violate the civilian radar data, but that data did not exist in the public domain when I developed that route. In any case, a BFO indicated climb around 18:28 in my original route would be both sensible (due to the weight reduction by fuel used by that time, for best LRC), AND, facilitated by having comfortably exited Sibolga PSR coverage at FL400, thus preventing detection.
Factoring in the recently released Civilian Radar Data may invalidate a Medan like track, but it does not invalidate a similar track initiated from early in the Penang to Vampi track segment.
Remember the Indonesian denials (in my view, much more credible than anything Malaysia has said) where they stated, at various times, both of:
(a) WE – did not see it where Malaysia says it was, and
(b) WE – did not see it in our airspace.
A sub FL280 path SW over Sumatra, but north of Medan, satisfies both criteria, and should not be ruled out without further study.
@Richard Godfrey.
As you say the Madagascan items in particular could have lain about: by the time the first arrived in Madagascar there had been 5 items recovered in Mozambique and 1 in South Africa, both more distant.
However I do not see the 23˚ and 22˚ ‘peak’ as affected, that being the model’s prediction for Africa as he has defined it. More I would see the observed percentiles coming left.
How much? Of the items that beached in the Africa David Griffin has modelled and which are assessed as ‘likely’ (or more) to come from MH370, I have compared the transit times of those that beached in Madagascar with the those that beached elsewhere. The average of both is about 27 months. Assuming for this exercise that the Madgascan transits ‘should have’ averaged 2 months less than the others the average time overall to reach Africa, would be reduced by 0.7 month, ie the percentiles in Fig 3 would shift left by that. Not that much.
If you apply that to David Griffin’s 29, of which 12 by my count beach in Madagascar, 12/29 X 2 = 0.8, not much different. Taking off the 4 that beached on the islands on the way increases that to 1.3 months, but his Fig 3 gaps are much bigger than that too.
Incidentally I notice that items 22, 23 and 26 in the below are unreported on the Malaysian listing. Maybe it is incomplete?
http://thehuntformh370.info/content/location-mh370-reverse-drift-study-based-debris-found
I have a hard time believing someone hacked the Inmarsat data. In fact, we already know that Zaharie is the one guilty. There’s strong evidence that supports this theory as FBI investigators found a route in his flight simulator one month before MH370 went missing that goes deep into the southern Indian ocean. It is highly unlikely someone else caused this plane to dramatically change course. It must be someone who is very familiar with a Boeing 777 and someone who knows the airspace around the area where the event began. The fact that he charted this course into the Indian ocean strongly suggest he did this. I know we all want to know what exactly happened, but I think there’s a growing consensus that this was a criminal act. Without a shred of doubt, captain Zaharie Ahmed Shah is to be blamed for everything.
In addition to my first message, the fact that the plane has not been found after years of searching doesn’t mean the Inmarsat data is incorrect. We should not forget how big the Indian ocean is. Besides, only the scenario where MH370 crashed into the Indian Ocean, was considered the most likely one. But the possibility of a controlled ditching has never been ruled out, as there’s evidence that shows this theory is also possible.
@JB: Welcome to the blog. I hope you can contribute often.
Your scenario is certainly possible. Whether or not intended, a glide also makes it harder to find the plane. But even there was a glide, the recovered debris indicates that the flight likely ended in a high energy impact, which in your scenario would indicate another change of heart.
@David
Are you really trying to say the time to Madagascar is the same as the time to mainland Africa?
@Richard Godfrey. To me the effect of late discovery of Madagascan beachings that you raise will bear not on the model predictions but the timing of the percentiles of the observed landings.
Assuming the average time to Madagascar will be X months less than to the African continent, whereas currently they average about the same, then the effect on the average time will be just a proportion of X. I though this provided an insight as to consequences to observed (actual) beaching percentiles.
On reflection this is too general.Tomorrow I will redraw the 50th and 90th percentiles of the likely-and-above MH370 beachings in Africa, including Madagascar, having reduced the transit times of those Madagascan beachings by an arbitrary 2 months, and post the outcome.
@David
You are mixing 2 completely different errors in David Griffin’s results:
1. Madagascar beachings are of unreliable timing for the reasons given and should be excluded.
2. Madagascar is not part of mainland Africa and should be excluded.
Madagascar arrivals can be much more than your arbitrary 2 months before mainland Africa arrivals.
Please try and remain scientific! Arbitrary is not scientific!!
@Robert
I believe your criticism above is aimed at Ghyslain Wattrelos who is pursuing several less likely scenarios, such as (1) shoot down of MH370 and (2) remote control of MH370. I agree with you, but as Victor points out, maybe something useful could come out of the approach even though we would take a different approach.
France is certainly in a unique position to be an impartial 3rd party, whereas US, UK, OZ, Malaysia, Boeing are more directly involved and have interests to protect. But overall, international law apparently gives Malaysia the right to call the shots on accident cause, they say they are clueless and have no interest in figuring out what happened.
@David
From the central east coast of Madagascar to the east coast of Mozambique at the same latitude is 1,460 km.
@TBill
No, not specifically. But the first scenario (MH370 being shot down by a flying object can directly be ruled out. It would definitely not explain why the aircraft kept flying for more than 6 hours after contact was lost. So, scenario one is one that is extremely unlikely. A controlled ditching is not the most likely scenario, but it is one that could explain why the aircraft still hasn’t been found. Moreover, the damage on both the flaperon and the right outboard flap actually suggest that they were both extended, which could only be done by someone in the cockpit. You know, the basic conclusion is that we desperately long for the analysis of Inmarsat. Only that way, the IG group and other people trying to locate the airplane can check whether the assumptions used in their calculations were justified/correct.
@robert said: Moreover, the damage on both the flaperon and the right outboard flap actually suggest that they were both extended…
Actually, the ATSB investigators believe the witness marks clearly indicate that the flaps were not extended.
You know, the basic conclusion is that we desperately long for the analysis of Inmarsat.
The analysis of Inmarsat was published in Sept 2014, and the assumptions that went into that analysis have been checked long ago. Most independent investigators believe the location of the 7th arc is correct, but the assumptions that produced the particular point of impact are in question.
@Victor
The investigators of ATSB do indeed believe they weren’t.
But some other investigators, including Larry Vance, do believe they were. So, no consensus about this one.
About the analysis, where can I find it?
@Robert
You stated “Moreover, the damage on both the flaperon and the right outboard flap actually suggest that they were both extended, which could only be done by someone in the cockpit.”
Please explain how you come to the opposite conclusion of the official ATSB report.
@robert: On this one, the ATSB is many times more credible than Larry Vance. The ATSB investigators had the flap in their possession, and had the help of Boeing. Larry was working from photographs, and also made some errors (such as believing the flaps were actuated by hydraulic cylinders rather than ball screws).
The details of the ATSB’s analysis can be found in the ATSB report “MH370–Search and Debris Examination Update” from November 2, 2016.
@Richard Godfrey:
Peter Foley from the ATSB said in a documentary that the visual evidence on the flaperon suggest it was extended (4.40).
https://www.youtube.com/watch?v=Dmw0evr6uvI&t=276s
That’s why, because the ATSB had said this. And Martin Dolan, also from the ATSB, had said in ’60 minutes’ that there’s evidence that supports both of these theories (piloted till the end and the crash scenario).
@robert: When Peter Foley made that statement about the flaperon, I don’t think the investigation of the flap was completed. As for Martin Dolan’s statement, a controlled glide does not necessarily mean the flaps were extended, which also requires at least one engine or the APU to be running.
@robert
Are you really Robin Stevens by chance?
(1) The Flaperon cannot be extended. The Flaperon is not designed to be extended like a Flap. It moves up and down like a Aileron. Have a look at the videos of how a Flaperon works, posted on this web site. We have been through this discussion many times on this web site!
The Flap (not the Flaperon) was not extended according to the ATSB analysis.
(2) In previous comments, you obviously missed, I have stated that I am investigating both pilot input at the end of flight and no pilot input at the end of flight.
@Richard Godfrey
Ha, I see! No, I’m not Robin Stevens.
I’m just curious about this mysterious flight.
The thing that bothers me the most, besides the fact that authorities still haven’t managed to find the plane, is that the Malaysian governments still try to protect the captain. If I may ask, do you guys believe he’s innocent? What’s your position on this? I look very forward to reading the results of your investigation.
@robert
ZS is the perpetrator, in my view.
Means, motive and opportunity.
I can understand Mr Wattrelos is to believe anything that make him less desperate, but if it is true that French judicial are thinking about a hack of the Satcom, I am ashamed to be French… Are they also thinking about digging in Kazakhstan? Even the BEA “say” the plane is in Indian Ocean (https://www.bea.aero/en/investigation-reports/notified-events/detail/event/disparition-en-mer/)
@Richard Godfrey. What we are discussing is David Griffin’s Fig 3 and its relevance to a future northern search. The model it is based on predicts arrivals in Africa from various possible crash sites and includes Madagascar. If you exclude Madagascar Fig 3 will need to be replaced.
That is the way it is currently, whatever you might like his Africa to include or exclude. It is not an ‘error’. The area considered in its model’s beaching predictions is the same as that considered for its actual arrivals.
My aim has been to investigate its utility and to offer ways around perceived problems with it. So far I have offered two, the first being pruning of arrivals to just those likely to be from MH370, overcoming the objections to unidentified items and non African items being included.
You and others pointed to the distortion of the arrivals’ timing that delays in discoveries there can make. The conclusion drawn by David Griffin is that a crash up north is unlikely unless there were very extensive delays in discoveries all round. While no way has yet been discovered for quantifying what those general delays might have been, clearly they would need to be very large to close the apparent gaps to the predicted arrivals from northern crash sites. Still, there may be a way of adjusting the Madagascan arrivals to correct the extra delay (ie above the general average) which looks to have been evident there.
Hence my second proposal was to reduce the delays specific to Madagascan discoveries.
My proposal was to reduce their average time to discovery by 2 months. I described it as arbitrary because I am unaware of any way of defending that quantification satisfactorily. Qualitatively though, I bear in mind that more direct routes to the south may be possible than all having to dogleg at Madagascar and as you know several items beached in Mozambique quite close to Madagascar.
Besides please bear in mind that 2 months is an average. With the earliest arrival the reduction could be 4 months, with the latest nil.To cover your 1460 km to Mozambique in say 3 months would require a speed of advance of 0.36 knots.
It might have been more constructive had you suggested that I look at 4 months or 6 if that was your preference, rather than concern yourself about whether or not I was remaining scientific.
I have now done as was my intention. It is, and was, principally intended to indicate method. On the way I found that among my 18 items at least ‘likely’ to be from MH370 I had included the vortex generator twice, so there now are 17.
There are 3 results from my visual best fitting of a graph of accumulated landings of those 17 with time. That yields the 50 and 90 percentiles.
The first, in orange below, has no time offset for Madagascan arrivals.
The second, in blue has Madgascan finds brought forward by two months.
The third, red, has the first find at Madagascar being brought forward 4 months, the last nil. It has the most logical type of fit, which I think should reflect an exponential decay in the finds’ rate even if very scattered, as could be expected. I have done this manually and the graphs are unpresentable on line. Anyone else can do the same easily. Their ‘best fits’ might not match mine but I believe the results to be indicative.
https://www.dropbox.com/s/ynle740h0crgv7d/CSIRO%20Fig%203.%20illustrating%20likely%2017%2C%20with%20some%20Madagascan%20discovery%20timings%20adjusted.jpg?dl=0
I should add that assuming the initial finds rate does have an exponential decay leads to a best fit conclusions that the Madagascan spread of corrections should be from 4 to 2 months, average 2.7 months and that some of Mozambique finds between Feb and May 2016 appear to have been delayed some months. However as before, with the variations in drift rate one should expect considerable scatter.
The outcome of my assessment is that the 50% median is brought to the left by 3 months in the red case.
@David
We have written to David Griffin pointing out that including Madagascar in the plot together with mainland Africa causes the peak in the plot at 22°S. There are 2 reasons for this peak, the South Equatorial Current is strong at 22°S and arrivals are on average 174 days earlier on the east coast of Madagascar compared with mainland Africa. We have requested separate plots for mainland Africa and Madagascar.
@Richard G, @David: Thank you for pursuing this. Any progress in reconciling David G’s results with Richard G’s drifter-based results is helpful in determining if a search to the north is justified.
@David @RichardG
I find it helpful to play the .mp4 animations that DavidG posted. The red 50 percentile line that @David drew would indicate that we should focus on the green particles in the animation. They start accumulating at the NE coast of Madagascar already by the end of 2014.
So to have at least two separate figs 3 indeed would be important. It looks like Richard is right that we should be very careful (how) to include the dates the Madagascar finds, especially if Blaine Gibson went to locations where not many people would normally visit.
@Niels
further to … especially if Blaine Gibson went to locations where not many people would normally visit …
I would add… and especially when an item is handed to Blaine by Milson Tovontsoa, Rija Ravolatra, and Eodia Andriamahery, in September 2016 but was supposedly found in February 2016 in Saint Luce, Southern Madagascar …
@All,
It is interesting to compare the Egypt Air 990 rapid descent data with MH370. About 8 seconds after the start of the descent while cruising at FL330, the ROD was 2,500 fpm. Eight seconds after that the ROD was 15,000 fpm. Eight seconds after that the ROD was 36,000 fpm, peaking out near 40,000 fpm in an overspeed condition.
Here are the data:
https://drive.google.com/file/d/1E2YUdHv_ex5_Z3coRw9gYvNgSCcqJWXL/view?usp=sharing
In case there was any doubt, this incident demonstrates that it is possible for a pilot to purposefully descend a cruising airliner (a 767-366ER in this instance) in such a way as to match the 00:19 BFOs. Of course, the EgyptAir flight had thrust from both engines, whereas MH370 had either no thrust or single-engine thrust. That limited thrust will slightly reduce the achievable RODs, but I don’t think it invalidates the comparison.
Another point regarding the Egypt Air 990 comparison is that the throttles were retarded at the beginning of its descent. Thus the thrust contribution would be small (idle thrust), and the comparison with a fuel-exhausted MH370 is valid.
@Richard Godfrey.”We have requested separate plots for mainland Africa and Madagascar.”
Good, as Niels reinforces.
@DrB
Do you have a plot of the debris field for Egypt Air 990 ?
@Richard Godfrey. “Your, “….arrivals are on average 174 days earlier on the east coast of Madagascar compared with mainland Africa.”
Using my 17 the Madagascan average is 841, the mainland 823, diff 18 or2%.
@Victor
I would think drift studies for northern search 22-25 South should include +/- 100 nm or so, matrix like Richard did for 30 South. Actually I’d say +/- 250 nm but that’s probably asking too much.
@TBill
The +/- 250nm amounts to over 8 degrees. Instead you could search a
reasonable width almost to Christmas Island!
@David
You stated “Using my 17 the Madagascan average is 841, the mainland 823, diff 18 or2%.”
Are you saying that the average time to Madagascar is 841 days and the average time to mainland Africa is 823 days?
How can it take longer for floating debris to reach Madagascar, than mainland Africa?
@Victor, @DrB, @Dennis, @Barry, @Niels, @David, @TBill, @Nederland, @Ventus45,
I have now completed my Cluster Analysis for 25 start points at ± 50 NM and ± 100 NM around 24.0°S 102.4793°E on the 7th Arc.
Out of 25 start points near 24°S, 10 end up beaching on Madagascar, 8 reach mainland Africa (from Somalia and Tanzania in the North, to South Africa in the South) and 7 end up mid-ocean off South Africa between 38°S and 39°S heading out into the Southern Atlantic Ocean.
The bifurcation around Madagascar is observed with 12 out of 25 particles heading around the southern tip of Madagascar to Southern Mozambique and South Africa and 4 out of 25 particles heading around the northern tip of Madagascar to Northern Mozambique, Tanzania and Somalia.
The average time of the 25 particles from the start point near the 7th Arc to the north-south line through St. Andre, Reunion was 506 days and the average time of the 8 particles to mainland Africa was 667 days, for those particles that beached. Another 7 particles carried on around South Africa and out into the Southern Atlantic Ocean, taking on average 893 days to reach the Atlantic.
23 out of 25 particles arrived in the Reunion area within 508 days ± 20%, out of which 5 particles arrived in the Reunion area within 508 days ± 10%.
5 out of 25 particles arrived on mainland Africa within 679 days ± 20%, out of which 4 out of 25 arrived on mainland Africa within 679 days ± 10%. I decided to adopt the suggestion from @Dennis to take the average of the 3 fastest items to be discovered on mainland Africa at 679 days.
The results are shown in the graphic at the link below:
https://www.dropbox.com/s/ks4z12hacz7c8br/Array%20End%20Points%20North.pdf?dl=0
The table of results is at the link below:
https://www.dropbox.com/s/o6we5jg5yx2szor/Array%2024.0000S%20102.4793E%20100%20NM.xlsx?dl=0
The 25 individual trajectories are in a folder at the link below:
https://www.dropbox.com/sh/ysi1sq56x8q2tkx/AAC2VCsZKn8eqGP0aJjqdyZea?dl=0
In the previous cluster simulation at 31.6°S, no particles reached mainland Africa in the 679 days ± 20% timeframe and there was no bifurcation around Madagascar.
I posted my previous results for 31.6°S at:
http://mh370.radiantphysics.com/2018/09/07/malaysia-responds-by-releasing-full-message-log/#comment-19228
My conclusion is that a MH370 end point north of 25°S cannot be excluded.
@Victor, @DrB, @Dennis, @Barry, @Niels, @David, @TBill, @Nederland, @Ventus45,
The Flaperon experiences a windage, which is significantly higher than that of undrouged drifters. David Griffin has shown in sea trials near Hobart, Tasmania, with a real Flaperon cut down to the size of the actual Flaperon found in Reunion, that the drift speed is 1.2% of the wind speed, faster than undrogued drifters.
The average wind in the Southern Indian Ocean between 8th March 2014 and 29th July 2015 along the trajectory likely to have been followed by the Flaperon, from the 7th Arc to Reunion, was 17 knots heading in the direction 298°T. The average undrogued drifter speed in the Southern Indian Ocean in 2014 was 0.640 knots. An additional 1.2% of 17 knots equates to an additional 0.204 knots, giving a revised drift speed for the Flaperon of 0.844 knots, which represents a speed uplift of 32% against undrogued drifters.
Please see the link below comparing the results from 25 starting points around 31.6°S and 24.0°S with an artificial drift speed uplift of 15% and 30% for the Flaperon when compared to the results based on undrogued drifters:
https://www.dropbox.com/s/lea8nnu5c3hyv3o/Comparison%20of%20MH370%20End%20Points%20%2B15%25%20and%20%2B30%25.pdf?dl=0
My conclusion is that a MH370 end point north of 25°S still cannot be excluded.
@Richard
“My conclusion is that a MH370 end point north of 25°S still cannot be excluded.”
Thank you for the drift calculations! However, we need to know if you think it is +/- 25nm from Arc7 for the next search. Otherwise we need to search wider and that’s a little different decision.
Update on Oct 23, 2018
In the post above, I added an article that appeared in L’Essor, which is the journal for the French military police. This story, if true, suggests that the French judiciary system, and in turn the military police that are assigned to the case, are seriously considering whether the Inmarsat data was corrupted by a malicious intrusion into the SATCOM onboard MH370.
Considering that the Inmarsat data is consistent with the aircraft crashing in the SIO near the 7th arc, and that the timing and location of the recovered parts from the aircraft also suggest that the aircraft crashed in the SIO, to doubt the Inmarsat data implies doubting the veracity of the recovered parts. This is the first time a government investigative body is known to be seriously considering a hack of the SATCOM combined with planting of debris.
@Victor
It is not clear to me from the traslation above that the French are considering a SATCOM hack that would produce a terminal location outside the SIO that would require debris planting.
@TBill
Searching 3 degrees along the arc (25S-22S) by 8 degrees wide (as you suggested earlier) would be about the same area as 16 degrees along the arc by 1.5 degrees wide (90nm or +/- 45nm). That would take the search along the 7th arc all the way to Java.
@DennisW: I suppose it is possible to slightly manipulate the timing and frequency of the SATCOM transmissions so that the predicted impact site is shifted but the actual impact position still matches the recovered debris.
@Victor
I find it hard to believe that anyone but Wise could believe in debris planting. Maybe another journalist?
@DennisW: French officials seem to be saying that the hacking/planting scenario is possible. The other part of that theory is some of the Five Eyes countries are actively involved in a cover-up. Whether French officials really believe that, or whether they believe there is political value in keeping that theory alive, I don’t know.
@Victor
A cover up on the Malaysian side is very believable, IMO. It may also be possible that US intelligence has info they are not dislosing. Nobody I know will even talk about it (all the good people have retired).
@Victor
You stated “French officials seem to be saying that the hacking/planting scenario is possible.”
The less serious french media have been saying that. Ghyslain Wattrelos has been saying that. Are the french officials really unable to tell Ghyslain Wattrelos, that it is tragic that he lost his wife and two of his children, but this was not a result of a conspiracy theory and there is not a multi government cover up?
As @Dennis says, does anyone apart from Jeff Wise seriously believe that the SATCOM was hacked?
In addition, MH370 was diverted to a location that we do not know and cannot trace, that the Inmarsat data was twisted in real time to look like MH370 ended up in the Southern Indian Ocean. Even with a “HackRF One” from Great Scott Gadgets using a Software Defined Radio running over a L-Band Uplink at 1.647562500 GHz and a C-Band Downlink at 3.616062500 GHz, emulating an IOR-P10500-0-3859 Channel, it would be an impossible task to take over control of MH370 from seat 29A and fly it on a flight path to a predetermined location and at the same time spoof the Inmarsat data in near real time to another flight path and kid us all that MH370 is really in the Southern Indian Ocean.
Then after the plane had crashed (instead of landed – oops how did that happen) at the predetermined location, some secret service agents went and picked out 27 pieces of debris (especially the ones like the Flaperon with hidden identification marks, to make it look real) and then planted them around the Indian Ocean. Then they summoned Johny Begue on 29th July 2015, Schalk Lückhoff on 23rd December 2015, Neels Kruger on 21st March 2016 at the same place (nice touch – finding the same piece twice), Liam Lotter on 27th December 2015, Milson Tovontsoa, Rija Ravolatra, and Eodia Andriamahery to hand over a piece in September 2016 to Blaine Gibson, that they found in February 2016, Blaine Gibson (again) to Mozambique on 28th February 2016, Jean Dominique and Suzy Vitry to Rodrigues on 30th March 2016, an unnamed tourist to Ilot Bernache, Mauritius on 10th May 2016, another tourist to Maçaneta Peninsular, Maputo Bay, Mozambique on 22nd May 2016, a Coast Guard Foot Patrol to Gris-Gris Beach, Mauritius on 24th May 2016, Blaine Gibson (again. again) to Riake Beach, Nosy Boraha Island, Madagascar on 6th June 2016, some fisherman to Kojani Island, Pemba, Tanzania on 20th June 2016, Barry McQuade to Praia de Rocha, Mozambique on 19th August 2016 (missing from the official reported list), Jean Viljoen to Linga, Linga, Mozambique on 26th August 2016 and Blaine Gibson (again, again, again) back to Madagascar in October 2016 (bit risky using the same operative for a 3rd time), …
The script writer was working overtime. Not even Hollywood could invent a story like this.
Just in case:
“L’Essor” is not “the” French journal for military police. It is not an “official” journal. It have a “convention” with the UNPRG (Union nationale des personnels et retraités de la Gendarmerie, which is the 1st association of retirees of the Gendarmerie).
The newspaper is completely independent of the hierarchy of the National Gendarmerie.
Quite all “official” services have the extension .gouv in France (like US .gov).
The very least that should come out of the French visit to the US is that Boeing admits that it inadvertently or knowingly installed a SATCOM system in its 777 aircraft which could be corrupted maliciously.
I don’t know who ultimately would be held responsible in law for this were it proven. But I personally would be fearful of boarding a plane which carried such software especially if it could be hacked mid flight or remotely. The flying public should be made aware.
@Victor
I might assume the France legal system may be bound by the complaints that Ghyslain Wattrelos has filed based on. So at the moment I am not assuming France gov’t concurs with that view.
@Pax Lambda: L’Essor describes itself as The Journal of the Gendarmes. I didn’t mean to imply it was officially recognized as a government channel of communication.
@Richard G, @DennisW, @TBill, @PaxLambda:
First of all, I don’t believe the SATCOM was hacked. I am an observer, trying to make sense of the statements now coming out of France, and hoping others here can help.
I don’t know if the GTA believes a SATCOM hack was possible, or there are political reasons to keep the theory alive, or whether it is simply completing its due diligence in investigating complaints filed by French citizen Ghyslain Wattrelos. I am trying to learn more.
@Victor
I did NOT say you believe the SATCOM was hacked!
I said the idea of a hack, a crash and multiple debris plants makes no sense.
@Richard,
“The script writer was working overtime. Not even Hollywood could invent a story like this.”
Well, I hope that Hollywood pay you handsomely for the complex plot you contrived! No doubt someone will use it.
@Richard Godfrey. “How can it take longer for floating debris to reach Madagascar, than mainland Africa?”
Before posting that I did check it. More or less the same. That is what the data say(s), as I alluded to earlier.
As to an explanation, while discoveries in Madagascar were delayed, so could have been those on the mainland. The flap at Pemba is an obvious example.
Then chuck in the effect of randomness on small numbers, which might be a major ingredient. Don’t know.
Where did your 174 days’ difference come from please?
@RichardG, VictorI
In case a party could steal the aircraft by succesfully hacking the satcom and spoofing the BTO/BFO data, a complicated debris planting scheme would not be needed. The aircraft could have been crashed in the SIO near the position indicated by the spoofed data a few days later.
I feel these kind of scenarios are very unlikely. For example, it would be extremely difficult to produce “meaningful” BTO/BFO data.
However, given the impasse we are in, I understand the need and desire also to take a closer look at satcom hacking vulnerabilities.
@Niels: As I said previously, at this point, I don’t take the hacking scenario seriously. Nor do any of the sincere, informed investigators. Our impasse is related to where to search next that has a reasonable chance of finding the debris field. There is no impasse as to why we haven’t yet found the aircraft.
@Richard G.
It would still be good to know whether or not a crash point, say, 80 nm south from the arc at 31S can be excluded as well.
@Richard Godfrey. About flaperon speed, you say, “David Griffin has shown……. that the drift speed is 1.2% of the wind speed, faster than undrogued drifters.”
I think you will find that what he showed was that the flaperon would drift at 10cm/sec faster than the drifter, which would drift at 1.2% of wind speed, so you might need to take a look at the effect of that on your results.
Also, there is also a subtle difference between “showed” and “found”. “Showed” to me means proved or verified.
@all: @Ventus asked a couple of days ago whether a plot of the debris field for Egypt Air 990 is available.
It seems to me, more generally, there should be some experience with debris fields from crashes over deep ocean.
First, is it possible to generalize the fall out of initially natant debris from such crashes, whether cabin contents, wing sections, or other parts that would flood over a period of a day or a few days?
If so, are there debris components that might have been missed initially, but be recognized on review of the Fugro and OI data?
If MH370 came down a few miles, or even tens of miles, outside the searched areas, AND if prevailing currents crossed from the crash site into the search zone [perhaps 50-50, or slightly better], my sense is that there would be a signature of delayed fallout that would be easily missed on criteria established for the search, yet visible on review.
Any thoughts?
@ikr
I don’t have any thoughts on the debris field other than the search was delayed by some three weeks due to the South China Sea misdirection. My Moffett Field friends tell me that is a long time in the context of a search off the California coast. Don’t know anything about how that might relate to a search at 30S on the 7th arc in the SIO.
@Richard Godfrey,
Thank you for posting your latest drift predictions. It is a thorough job, as usual. My observations are as follows:
1. The “best-fit” flaperon drift speed is slower (= 0% “uplift”, as you called it) from 24S than it is from 31.6S (= 30% uplift). Here I use the number of points in the grid which are within 20% of the Reunion discovery date as a measure of “best fit”. For 24S the number of points is 23 at 0% uplift, 15 at +15%, and 7 at +30%. For 31.6S the number of points is 6 at 0%, 5 at +15%, and 14 at +30%.
2. The best fit at 31.6 S is with the full amount of the extra speed, relative to undrogued drifters, observed by CSIRO with a replica flaperon.
3. The best fit from 24S is with the undrogued drifter speed without applying any part of the extra speed observed by CSIRO.
4. Since the average initial drift from 31.6S is more or less to the NE along the 7th Arc, one would expect it takes noticeably longer to reach Reunion from 31.6 S than from 24S, and your predictions bear this out.
5. Interestingly, the overall best fit (greatest percentage of points within 20%) is from 24S with 0% uplift. I am unsure how to interpret this. I will note that in this case we have only one debris item (the flaperon), and it is quite possible that its track and arrival time are not closely representative of the average of a cloud of particles released from the same location at the same time. Therefore, I think we must be cautious about inferring unwarranted accuracy of the observed/predicted arrival times when using a single item such as the flaperon. In addition, there is some uncertainty in the Reunion arrival date, since there are additional reports of sightings as early as May 2015.
6. If the flaperon actually arrived in May 2015 rather than July 2015 it would seem that a speed SLOWER than undrogued drifters would be required from 24S, whereas for 31.6S the best fit is still likely to be faster than undrogued drifters and thus more consistent with the CSIRO results.
7. While it may not be possible from the flaperon alone to eliminate 24S with any certainty, it does appear that 31.6S is more consistent with the discovery date (especially if it was as early as May 2015).
8. There are a larger number of items found in Madagascar, and this generally helps the reliability of the estimated average arrival date, but many, if not most, of them suffer from unknown arrival times before they were “discovered” (i.e., were given to) Blaine Gibson. So, for Madagascar we probably have even more uncertainty in the observed arrival dates than we have for the flaperon on Reunion.
9. The small difference David noted in the “discovery” dates between Madagascar and the African coast seems to me to be just an artifact of the delayed “discovery” dates of Blaine Gibson’s visit. The actual drift time difference should be much larger, as Richard has pointed out. Thus, we have another reason to think a number of items arrived in Madagascar quite a long time before they were collected from locals by Blaine Gibson.
10. I think, for the purpose of impact latitude discrimination, the best comparison of predicted and observed arrival times is probably coastal Africa. This has better statistics than the single flaperon and may have shorter delays overall between arrival and discovery.
@DrB
you said:
I think, for the purpose of impact latitude discrimination, the best comparison of predicted and observed arrival times is probably coastal Africa. This has better statistics than the single flaperon and may have shorter delays overall between arrival and discovery.
I agree. Although I do have a high regard for the flaperon arrival in May (or so).
@Dennis:
Poor phrasing on my part. I’ll try again. I wasn’t referring to contemporary satellite records of floating debrisRather, I was wondering what the seabed record would look like for initially natant debris that “drowned” over a few hours or days. There should be some sorting [by size, material, and whether floating by virtue of large air spaces [eg tanks] or wettable contents [eg, seat cushions and the like] and you’d expect attenuation with time and dispersion. My question was more: If drift carried the initially floating debris across the searched area, wouldn’t there be seabed debris that was initially missed because small/dispersed/etc. that might be recognizable on review with cognizance of precedents like the Egyptair and AirFrance crashes.
@ikr
I am not able to address your question.
@Richard Godfrey,
On October 8th you said: “Simulated paths from the cluster centred on 32°S show the required bifurcation around Madagascar.”
On October 23rd you said: “In the previous cluster simulation at 31.6°S, no particles reached mainland Africa in the 679 days ± 20% timeframe and there was no bifurcation around Madagascar.”
Which of these statements do you believe is correct, and why the change?
The folder with the 25 tracks from 24S is instructive. Can you post a similar folder for the 25 tracks from 31.6S?
@DrB
The October 8th statement was based on the narrower 9 particle cluster analysis, which showed particles going south of Madagascar from the 2 start points at 32°S on the 7th Arc and at 32°S but 1° of longitude inside the 7th Arc, but on the other hand showed particles going north of Madagascar from the 2 start points at 33°S on the 7th Arc and at 33°S but 1° of longitude inside the 7th Arc.
Yesterday, I was referring to the October 15th statement, which was based on the wider 25 particle cluster analysis, centred on 31.6°S, which did not show bifurcation around Madagascar. 8 out of the 25 particles went around Madagascar to the south, but 0 out of 25 particles went around Madagascar to the north. I subsequently published an Excel with all the results in tabular form, which was addressed to you.
Both of the statements I made were correct. The start points for the 9 cluster analysis are different from the start points for the 25 cluster analysis. The 9 point cluster is ±1° of latitude and longitude. The 25 point cluster is ±50 NM and ±100 NM.
From the 8th October 2018 start point at 33.0000°S 94.2274°E, the track just misses the northern tip of Madagascar and carries on to Northern Mozambique.
From the 15th October 2018 start point at 33.2315°S 94.9738°E, the track hits the northern tip of Madagascar.
However, the temporal granularity of the software simulation is 1 day. So I only check, where the particle is once each day. If the new location is on land, then I stop. If the new location is at sea, then I carry on. Near Madagascar, the drift current is sometimes over 2 knots, so a particle can travel 48 NM in one day. It is theoretically possible that in looking for landfall only once per day, that I miss a beaching when a particle is travelling up or down the coast line.
I will put together a folder of the 25 tracks from 31.6°S and post it later today.
@DrB
Many thanks for you 10 point assessment of my latest drift analysis. I agree with all the points you make.
You stated “I think, for the purpose of impact latitude discrimination, the best comparison of predicted and observed arrival times is probably coastal Africa. This has better statistics than the single flaperon and may have shorter delays overall between arrival and discovery.”
May I underline that in the 24°S cluster, 5 out of 25 particles arrived on mainland Africa within 679 days ± 20%, out of which 4 out of 25 arrived on mainland Africa within 679 days ± 10%. In the 31.6°S cluster, 0 out of 25 particles arrived on mainland Africa within 679 days ± 20%. The mainland Africa arrivals and the bifurcation around Madagascar are strong indicators that 24°S cannot be excluded.
Please note I am not excluding 31.6°S either, but this area has been searched ±22 NM. My recommendation is to also search north of 25° up to 20°S at a search width of ±22 NM.
@Victor
Ghyslain Wattrelos is now claiming that SITA could not connect to MH370 (probably referring to SATCOM calls) but Inmarsat could connect to MH370 (probably referring to the pings). He therefore concludes that either SITA or Inmarsat are not telling the truth!!
Below is a link to a tweet by Ghyslain Wattrelos in French dated 19th October 2018:
https://www.dropbox.com/s/mphertb19ms0q58/Ghyslain%20Wattrelos%2019%20Oct%202018.png?dl=0
@ Nederland
You asked “It would still be good to know whether or not a crash point, say, 80 nm south from the arc at 31S can be excluded as well.”
With an uplift of the Flaperon speed by 30% compared to the undrogued drifters, an end point up to 100 NM south of the 7th Arc at 31°S cannot be excluded. Here is a link to those particular results:
https://www.dropbox.com/s/cp9ax4if57lshmv/Array%20End%20Points%20South%20%2B%2030%25.pdf?dl=0
@TBill
You asked “Thank you for the drift calculations! However, we need to know if you think it is +/- 25nm from Arc7 for the next search. Otherwise we need to search wider and that’s a little different decision.”
My recommendation is to search north of 25° up to 20°S at a search width of ±22 NM.
@Richard G: Regarding the statement that Mr Wattrelos made about either Inmarsat or SITA lying: There is much confusion about the differences between the services and capabilities of ISAT (Inmarsat), SITA, ACARS, and ARINC. Adding to the confusion is that the ISAT data was made available to Malaysia through SITA.
@Victor,
I do not believe that either SITA or Inmarsat was lying.
I think Ghyslain Wattrelos is not correct with his accusation.
@Richard Godfrey: As you know, back before any debris was recovered and we did not have the simulator data from the captain’s computer, I explored possible ways to hack the SATCOM so that the BFO signature of a northern path might look like a southern path. I found that this could be accomplished by changing the satellite inclination parameters (inclination angle and ascending node) stored in the SDU’s System Table.
Based on this, I searched for commercially available software that would allow a change of parameters in the System Table. Although I did find software for configuring the Owners Requirement Table (ORT), I did not find software that could change the parameters in the System Table. That said, I think the software capability to “peek and poke” values in the SDU’s non-volatile memory would not be hard to implement for Honeywell, and probably already exists as proprietary software for troubleshooting and development.
The GTA was responsible for examination of the flaperon recovered at Reunion Island. The GTA is now also pursuing whether the Inmarsat data was deliberately corrupted by hacking into the SATCOM. We’ll have to wait to see if GTA has reason to suspect a malicious intrusion of the SATCOM occurred, other than the seabed search has failed to find the debris field.
@Victor
I remember the analyses well, which I also pursued for a while.
Even if MH370 ended up in Kazakhstan or similar, you still have the problem to plant debris all over the Southern Indian Ocean.
I think it is far simpler to search north of 25° up to 20°S at a search width of ±22 NM.
@Richard Godfrey: I suspect that the GTA is still following a trail that others have long abandoned after additional evidence has surfaced, i.e., the recovered debris and the simulator data. I suspect the GTA questions the authenticity of the additional evidence. It’s not helpful that Mr Wattrelos makes statements that are not true, such as either Inmarsat or ISAT is lying about SATCOM connectivity.
@ALSM, @Victor,
I cannot find Barry McQuade’s discovery of debris at Praia de Rocha, Mozambique anywhere in the Malaysian Safety Report – List of MH370 Debris Items – Table 1.12A.
Has this item been overlooked?
Here is a link to a list of possible MH370 debris items found so far:
https://www.dropbox.com/s/ktiwozxqx5rrxuh/Debris%20Finds%20Overview.xlsx?dl=0
Here are a couple of low resolution pictures of Barry McQuade’s find:
https://www.dropbox.com/s/4wqjbn847j8w4gr/Barry%20McQuade%20Debris%20Top.jpg?dl=0
https://www.dropbox.com/s/5tdp0vsm1ohgept/Barry%20McQuade%20Debris%20Bottom.jpg?dl=0
@Richard Godfrey: I don’t know why Barry McQuade’s find is not officially recognized.
We also have the root of the vortex generator that was found in Madagascar that is being held by the Malagasy authorities as part of the criminal investigation into the assassination of the Malaysian diplomat.
http://mh370.radiantphysics.com/2017/08/28/new-pieces-possibly-from-mh370-blaine-gibson-threatened-in-madagascar/
@Richard
Whatever happened to Bayes? 🙂
So, a back of the envelope calculation based on the assumptions:
1> +/- 25nm search width is correct
2> 100% certainty the aircraft is located between latitudes 20s and 38S
3> latitudes 20S to 38S have equal probability of containing the aircraft (no differentiation based on Inmarsat data)
4> there is a 10% chance the area searched missed the wreckage
25S to 22S represents 3 degrees.
38S to 25S represents 13 degrees.
22S to 20S represents 2 degrees.
So probability of finding aircraft searching 25S to 22S is
0.9*3/(3 + 2 + 0.1*13) = 2.7/6.3 = 43%
Explanation of terms left to right:
90% chance of finding the airicraft in 3 degrees if it is there
3 degrees searched 25S to 22S
2 degrees not searched
13 degrees searched with 10% chance of missing the wreckage
Like you, I say go for it (for the first time I ever said that).
@Victor
Then David Griffin is right, there have been 29 items of debris possibly from MH370 that have been reported locally, out of which 27 were included in the official Malaysian Safety Report.
It makes you wonder, what other items of possible MH370 debris are out there somewhere?
@Richard,
If this piece was discounted by the MY team, or ATSB, it would not have made the list.
I reviewed, again, the images of McQuade’s discovery but there is very little with which to positively attribute it as originating from a 777. The shape of the aluminium honeycomb core is likely the only discriminator for this part’s core which is typical of Hexcel’s Aluminum Flex-Core product.
Panels of aluminium skin + expanded aluminium honeycomb core, while used on the 777 engine pylon, are common on military aircraft.
@DrB
You asked: “The folder with the 25 tracks from 24S is instructive. Can you post a similar folder for the 25 tracks from 31.6S?”
Please find below a link to the folder with the 25 tracks from 31.6°S:
https://www.dropbox.com/sh/pwy8d4axbo2ar4e/AADg6seBPfrMUeJvz3JufMf9a?dl=0
Please find below a table of the results from 31.6°S in Excel:
https://www.dropbox.com/s/kxxl7p227oqbulb/Array%2031.566S%2096.774E%20100%20NM.xlsx?dl=0
Since the topic has come up again, I’ll point out that the script doesn’t need to be quite so cartoonish.
== BEGIN SCRIPT ==
The plane is hijacked, not by a single cartoon supervillian, but rather by a small team. The goal? Something of value on the plane (e.g. undeclared illicit cargo).
Some team members are responsible for flying the plane. They either take over the cockpit, or are aided by someone in the cockpit.
The other team members deal with the satcom to purposefully obscure the actual destination.
The plane lands successfully and its bounty is offloaded. The plane is immediately dismantled and several parts are loaded on a naval vessel. The vessel quietly dumps the parts in the ocean, where they eventually wash up on shore to be randomly discovered.
== END SCRIPT ==
To note further… if the pilot(s) were involved, the simulator runs pointing to the SIO could have been made intentionally so as to match the path intended to be derived from the AES records.
As for the debris, it wouldn’t have to be dropped so far out in the SIO… I suspect anywhere east of Madagascar would suffice.
So now you have the simulator data, debris and satcom trail accounted for.
To note further, related to the original post… if the pilot(s) were involved, the cockpit would not need to be violated at all. Just a number of people with access to the AES… a couple people to tamper with the electronics and perhaps a couple strong people to keep unruly passengers from interfering. Perhaps the people closest to the E11 rack in 29A, 29C, 30C and 27D & 27E are more interesting in this context?
I’m not commenting on probabilities here… just saying.
@Don
Many thanks for putting me right!
As you can see, I am not an expert on aircraft debris.
@Phil
No mention in your script of the Inmarsat data. I assume you meant to say it was spoofed.
The problems I see with your script are:
1> There is no evidence of anything on the plane worth hijacking for.
2> It is easier to get valuable cargo off the plane when it is on the ground in Beijing or Kuala Lumpur.
3> It is not trivial to spoof the Inmarsat data.
4> Dismantled parts are not the same as crash damaged parts.
@Dennis
I was referring to the Inmarsat data when I mentioned the satcom trail and AES records.
1. There would be no record of undeclared, illicit cargo, obviously.
2. The cargo would obviously be destined for someplace other than Beijing or Kuala Lumpur (i.e. wherever the plane actually landed).
3. It is not trivial to spoof the Inmarsat data, no doubt.
4. Dismantled parts are not the same as crash damaged parts, I agree. Has the end of flight scenario been proven conclusively by the parts recovered to date?
@Richard Godfrey,
Thanks for posting the 31.6S maps.
@Phil
1. There would be no record of undeclared, illicit cargo, obviously.
That is exactly what I said. No evidence. You are making it up. The fact that no evidence is expected does not alter the fact that there is no evidence. To show there is no reluctance to consider your theory I will lower my probability of success estimate for 25S to 22S search from 43% to 42.9%.
@TBill
Since I agreed to consider Phil’s theory, it is only fair that I consider your wider glide scenario. Assume the the glide is given a 50-50 liklihood (your suggestion), the new probability of finding the plnce in the 25S to 22S area using a +/- 25nm width becomes:
0.5*3/(3 + 2 + 0.5*13) = 1.5/11.5 = 13%
Not an attractive endeavor. That is one of the reasons for not considering a wider search. It is an activity deal killer.
@Phil: I’ve said before that one big problem with any theory that incorporates an omniscient, omnipotent bad actor with accomplices is it removes the probative value of any evidence we have. Inmarsat data? Spoofed. Recovered debris? Planted. Simulator data? Fabricated to frame the captain. Radar data? Decoy aircraft. We are left with no evidence after 17:21, and an investigation becomes impossible. I suppose that scenario is possible, but do we really want to proceed (or not proceed) on the basis of that exceedingly small probability?
@DennisW
Where did I say 50-50 likihood? I thought I abstained from making that comment earlier today. Anyways, I agree with it.
But your approach is either my way or the highway. If we take the long term view, it is not impossible to consider that wider search may be conducted by somebody someday. If OI comes back they could knock off 20-25 South quite quickly, with time for spot checks elsewhere.
I noticed (when you mentioned Duncan Steel earlier) that some his last posts were suggesting that we do no need to blanket search the whole sea bottom. He was favoring leaving some gasp to cover more ground. If we consider gaps in the current coverage, we could suggest two things: either (1) we missed MH370, or (2) we did not get close to MH370. I am thinking the latter.
@Dennis
I clearly noted it was a script… so yes, I’m making it up. I never claimed there was any evidence of illicit cargo.
To be perfectly clear, I’m not trying to convince you to consider anything… I was merely responding to the relevant thesis of this post and Richard’s overly complicated script idea.
@Victor
Simulator data? Could have legitimately been put there by the Captain, as suggested above.
Radar data? No mention of decoy aircraft from me.
Just spoofed data and some debris tossed overboard in my little script there.
But yeah, of course it gives little basis to proceed on that exceedingly small probability; I’m certainly not suggesting otherwise. Your post is about the alternate possibility apparently being considered by French authorities however, hence my comment. I’m not exactly going on about it, as you know. You and the others have my utmost respect in your scientific efforts.
@Phil: OK. I understand. And I don’t mean to discourage investigators from pursuing what I would consider to be low probability scenarios, as long those investigators are sincerely interested in finding the plane, and don’t have an alternate agenda. This recent story about investigating whether MH370 crashed in the jungles of Cambodia is a great example of a staged media event around a non-sensical scenario that seems to be more of a publicity stunt than any real attempt to find the plane.
@Tbill
You asked:
@DennisW
Where did I say 50-50 likihood? I thought I abstained from making that comment earlier today. Anyways, I agree with it.
On October 17 10:27AM you said:
Since we have been mostly trying to say this was a passive flight/unmanned crash, it is a paradigm shift to try to think of all of this as intentional. But I would say that active flight is just as likely as passive flight at this juncture
Now, you say:
But your approach is either my way or the highway.
A search using a +/-25nm width is the only sensible way forward at this point. I have tried to show that in a number of ways, but you are having none of it.
@Phil
To be perfectly clear, I’m not trying to convince you to consider anything… I was merely responding to the relevant thesis of this post and Richard’s overly complicated script idea.
OK, good. Like TBill says it is either my way or the highway. 🙂
Actually, I have the same feeling as Victor. I like it when alternate scenarios are proposed. It is a good way to test the strenght of your own version of events. I did not intend to come on as critical of you.
@DennisW
I apologize for saying you said my way or the highway”. What you are really saying is more like, nobody would ever fund a wider search, thus we must have a narrow search. But who knows what someone is willing to search in the future?
@TBill
Hey, no biggie. I enjoy the comraderie here enormously.
@all. New drift assessment by David Griffin, including splitting Madagascar off.
http://www.marine.csiro.au/~griffin/MH370/
@DennisW
Dennis, do you have a view on a search width of +/-25nm versus +/-22nm? OI used +/-22nm. Given the number of unknown (and in some cases, unknowable) variables relating to the end-of-flight sequence, I’ve thought that 22nm is overly parsimonious for an uncontrolled descent with no pilot inputs.
@Victor, @DrB, @Dennis, @Barry, @Niels, @David, @TBill, @Nederland, @Ventus45,
Many thanks to @David for informing us of the latest update from David Griffin.
The results for the 3 fastest of the 11 debris items discovered on mainland Africa in Mozambique and South Africa align with the median arrival time from a start latitude of 23.7°S:
https://www.dropbox.com/s/qxhghhow4l8r73a/David%20Griffin%20results%20for%20southern%20mainland%20Africa.pdf?dl=0
The results for the 3 fastest of the 17 debris items discovered on Madagascar, Reunion, Mauritius and Rodrigues appear to have taken 2 months before discovery based on the median arrival time from 23.7°S:
https://www.dropbox.com/s/vii248f3dgihnva/David%20Griffin%20results%20for%20Madagascar%2C%20Reunion%2C%20Mauritius%20and%20Rodrigues.pdf?dl=0
The results for the single debris items discovered on Tanzania appears to have taken 13 months before discovery based on the median arrival time from 23.7°S:
https://www.dropbox.com/s/tdi3lu20e2ubein/David%20Griffin%20results%20for%20Tanzania.pdf?dl=0
Apart from the one outlier in Tanzania, where fisherman found the Outboard Flap in a cave on Kojani Island, the other 28 debris items, possibly or confirmed from MH370 appear to fit a start latitude of around 23.7°S on the 7th Arc.
@Richard
I’ve not felt as positive about the chances of a successful search as I do now.
@Dennis
I agree, we are slowly but surely converging.
The agreement on the data is quite high.
The agreement on the interpretation of the data is still not there yet.
@Richard
@David
The recent drift update article by David Griffin is really good.
I would note three things:
(1) In his conclusions, DavidG is basically saying below 25 South vs. 23.7 South above, so that is one area for collaboration.
(2) When DavidG mentions the reasons why the crash site could be further north than the calculations, he does not mention perhaps the crash distance from Arc7 is farther than we think. In others words, the key assumption here (I am thinking) is that Arc7 is where the plane crashed.
(3) Xmas Island is seemingly ruled out, but this conflicts with Jean-Luc’s explanation that Cyclone Gillian could have pushed debris south
FYI…I sent an email to Barry McQuade’s to find out what he knows about what happened with the debris he found. No word back so far.
@Richard Godfrey
OK, thanks.
@Richard Godfrey,
You said: “The agreement on the data is quite high.”
Perhaps you meant something else, but I would not call the consistency of 23.7S with David Griffin’s predictions particularly high: 0, 2, and 13 months of reporting delay. If you allow such large discrepancies at one latitude, then other latitudes can be matched just as well.
A key consideration is splitting the Africa mainland into north and south components. Your predictions show south arrivals about a year later than north arrivals. This should be accounted for. I was in the process of doing this, but now I will include David Griffin’s new results and post it later.
Your 24S tracks, compared to 31.6S tracks, show a much larger fraction of items going north around Madagascar and somewhat fewer items landing in Madagascar itself. Without considering the timings, your 31.6S predictions have a geographical distribution that better matches the actual debris locations than 24S. Picking a preferred latitude near 24S is only possible using uncertain arrival time estimates, and, even then, it requires assuming large reporting delays.
@DrB
I cannot understand why you think 28 debris items out of 29, at 0 and 2 months reporting delay, is a large discrepancy.
I cannot understand why one outlier, drives your statistical analysis and conclusion.
> Victor Iannello:
> the paper that I co-authored with Yves Guillaume in November 2016
> concluded that the sim data was from a single flight
I read your paper as saying that points 3N + 5N are very likely part of a single flight (red rectangle) and points 10N + 45S1 + 45S2 are also part of a single flight (blue rectangle).
Excuse me if I missed it, but I didn’t find any conclusion in your paper that the “red flight” and “blue flight” are connected.
Looking at table 2, to me it looks like the red and blue flights could be related, but don’t have to be.
I agree with you that “10N, 45S1, and 45S2 are very useful because they definitively link the position of the aircraft in the Andaman Sea with the two positions in the SIO” and this appears rather damning all by itself, but still less damning than if we had proof of all 5 points being part of the same flight, because 2N + 3N + 5N could belong to a standard flight KUL-Europe. If the other 3 points stand on their own, of course they are still suspect, but maybe they can be explained by a bad coincidence. Let’s assume Z wanted to simulate something just anywhere over open water … he had to pick a spot for doing that somewhere.
I don’t say it isn’t suspect. I just think if all 5 points were proven to be linked and together “create[d] a flight that passed over the Malacca Strait to the Andaman Sea and to the SIO in a way that is similar to the flight path that investigators believe was followed by MH370”, as you say, the case would be much more clear-cut.
@DrB
I have a very difficult time with 31S. We have all seen the ISAT movie analytics many times. It has never had a happy ending.
The obvious path in my opinion is to search North from 25S at +/- 25 km.
@all – It is interesting that the Tropic of Capricorn is at 23.5 S Latitude
@Peter Norton: The timestamps of the simulator suggest the flight files were created over a period of about an hour. Here’s what I said in a previous post.
After some email exchanges with the ATSB, I was surprised to learn that the ATSB has additional data values from the recovered flight files that were omitted from the data sets that were provided in the RMP report. In particular, there is a section of the flight files in which the date and time of the simulation session are stored. These data values tie the date of the simulation to February 2, 2014. Also, the time values show that the chronological order of the data sets matches a flight departing KLIA, flying over the Malacca Strait, continuing past the Andaman Islands, turning to the south, and exhausting fuel in the SIO, in the same order that the progressively depleting fuel levels suggest. The time values also indicate that the flight session lasted for about one hour. This confirms that the position and fuel levels were modified during the simulation, just as Yves Guillaume and I had concluded in our paper.
The newly released information regarding the date and duration of the simulator session means that it is almost certain that the recovered data sets were from a single flight session.
@Victor Iannello:
Thank you. Ok, this seems conclusive. Just out of mere curiosity though: Your posting is from 2017. But you wrote “the paper that I co-authored with Yves Guillaume in November 2016 concluded that the sim data was from a single flight. I read your paper but missed this conclusion. Could you point me to where it is located?
@Victor,
Since the sim and MH150 has come up again, and since I can’t remember what time MH150 normally departed KL, I would be interested to know what the typical estimated time at sim point 10N (or DOTEN) would have been for MH150.
@Ventus45
MH150 I believe was normally a 1515 departure flight. The MAS flight schedule is confusing because they were planning to start a second flight (per below). It seems that that schedule change never happened, and MH150 continued at 1515. So I could postulate 3 possible departure times for a rogue flight plan (adding in MH168) but 1515 was the only option in the end.
This apparently did not happen:
“Malaysia Airlines from 28JAN14 to 29MAR14 is adding 2nd daily Kuala Lumpur – Jeddah service, on board Boeing 777-200ER aircraft. Reservation for the new MH168/169 service is now open.
MH150 KUL1145 – 1600JED 772 D
MH168 KUL2230 – 0245+1JED 772 D
MH169 JED0500 – 1835KUL 772 D
MH151 JED1730 – 0705+1KUL 772 D
MH169 begins from 29JAN14 until 30MAR14.
From 30MAR14, MH will operate 1 daily flight as MH168/169, which means the oneWorld member shifting operational schedule to night-time departure from KUL.”
@Victor Iannello
@Peter Norton
Victor, the timestamps of the simulator flight files are almost certainly ‘simulator’ (as opposed to ‘actual’ or ‘real’) time. While generally speaking ‘simulator’ and ‘actual’ time are correlated when a flight file is first created they quickly lose synchronisation as ‘simulator’ time varies from ‘actual’ time. A simulated flight that is commenced on Day X and saved at Time Y can be returned to any time later and continued and the ‘simulator’ timestamps will give the appearance that the ‘flight’ was continuous on Day X and Time Y+. Consequently, it is far from almost certain that the recovered data sets were from a single contemporaneous session. Absent that ‘actual’ file created/saved/modified timestamps we simply can not know whether the recovered data sets were from a single session.
@Richard Godfrey,
One outlier in reporting delay is not driving my conclusions, although 13 months does seem to me quite a long time for the example you showed.
On the issue of arrival timing, I also note that your 24S cluster of 25 points predicts an average of 359 days transit time to Reunion for the flaperon when using the undrogued drifter speed + 30% to match the CSIRO empirical observation. The predicted range is 213-421 days. Your 31.6S cluster predicts an average of 533 days (with a range from 295-893 days), which more closely matches the reporting date of 508 days after 8 March 2014), even if one allows an arrival as early as 2 months prior (May 2015) to the reporting date for a previous sighting. My conclusion is that, when you use the correct drift speed for the flaperon, its arrival is more consistent with 31.6S than 24S. See this chart comparing your two sets of predicted transit times to the estimated actual transit time :
https://drive.google.com/file/d/1tAf93q2KIpuQnnpUx8BHGvUEsmlGUWDZ/view?usp=sharing
You can see in this chart that, for 24S, none of the 25 points have predicted arrival times after 3 May 2015. I would not call that consistent.
Averaging arrival times for items passing both north and south of Madagascar can produce significant error. Using your 24S 25-point predictions, the average predicted N Africa arrival time is 530 days, whereas the average predicted S Africa arrival time is 861 days, which is almost a year later. That is why it is both necessary and useful to separate N Africa landings from S Africa landings.
It is also instructive to compare fractional geographical distributions. Your 24S cluster begins with 25 particles in the cloud. All of these particles pass fairly close to Reunion. Ten land at Madagascar, 11 land in southern mainland Africa, and 4 land in northern mainland Africa. For the 31.6S cloud of 25 particles, all pass fairly close to Reunion, 16 land in Madagascar, 3 land in southern Africa, and 6 do not make landfall within 1200 days. Although none of these 25 examples near 31.6S hit northern Africa, I recall that some did in your previous 9-point cloud predictions near 32S. Although the exact number is uncertain, I think it is safe to say that a very small fraction of the 31.6S items land in Northern Africa. Here I will assume it is roughly 1 out of 25. So, for 31.6S we have 20 landed particles, of which 16 (75%) hit Madagascar, 3 (15%) hit southern Africa, and maybe 1 (roughly 5%) hits northern Africa. For 24S we have 25 landed particles of which 10 (40%) hit Madagascar, 11 (44%) hit southern Africa, and 4 (16%) hit northern Africa.
Next let us figure the fraction of reported debris items in those 3 locations. I count 10 items in Madagascar, 1 in northern Africa, and 5 in southern Africa. Thus, 63% of those finds were in Madagascar, 6% were in northern Africa, and 31% were in southern Africa. Of course, not all areas were equally searched, and some items are easier to find than others. Doubtless these effects skew the percentages a bit, but this comparison is useful because it is independent of the reporting delays. The debris finds indicate that about five times as many debris items were found in southern Africa compared to northern Africa. The 31.6S/32S predictions are consistent with that high ratio of bifurcation, whereas your 24S predictions indicate a lesser ratio of about 2.8. The finds also indicate that 62% of the items in those 3 locations landed in Madagascar. The 31.6S predictions are 75%, whereas for 24S the fraction is 40%. Thus, the 31.6S predictions better match the observed fraction landing in Madagascar.
So, when comparing your predictions for 31.6S versus 24S, I see that 31.6S is a better match for three metrics: flaperon transit time, the ratio of items going south around Madagascar to those going north around Madagascar, and the fraction of landed items (past Reunion) making landfall in Madagascar itself.
In my view these three characteristics may be the most reliable latitude-discriminating observables based on the debris reports. In many cases we know the reporting delays must be quite long (for example, especially so for the three items at Rodrigues/Mauritius, as well as all but the first of the 10 Madagascar items). Thus, time-based discrimination of starting latitude is generally subject to large potential error because of the unknown, and sometimes large, reporting delays. The one exception may be the flaperon, for which the arrival time appears to be known within several months because of the frequent beach visitation by cleaning crews and by the public and also because of the very high visibility and uniqueness of the flaperon itself.
@DennisW,
You said: “I have a very difficult time with 31S. We have all seen the ISAT movie analytics many times. It has never had a happy ending. “
The drift modeling results do not depend on flight path modeling other than starting at an assumed location near the 7th Arc. Drift models are essentially an independent method of predicting crash location (latitude) from route fitting using BTOs and BFOs. Are you suggesting that we ignore what they are telling us? I think we need to go wherever the drift models are shown to be consistent with the debris finds. If this is near 32S, then we have some corroboration. However, I don’t think this is north of 25S, as I have demonstrated in my previous post, based on Richard’s predictions for 24S and 31.6S.
In my opinion we don’t yet know the range of latitudes where good and statistically meaningful matches occur between debris predictions and observations. I have described above three metrics which I think are the best we can do with the debris data we have. Only one, the flaperon arrival in Reunion, depends on transit time. It is a special case. The other two parameters are geographical location ratios – the fraction of the approaching cloud which makes landfall in Madagascar and the ratio of those going south around Madagascar to those going around to the north. These ratios seem to be rather sensitive to the latitude and the track of the debris cloud approaching Madagascar and may provide useful starting latitude discrimination. Richard’s excellent modeling results for two latitudes (24S and 31.6S) provide a glimpse of this possibility. David Griffin has not yet segregated his data to allow these geographical ratio parameters to be determined as a function of starting latitude. I have requested this of him. If he does this, it will be interesting to compare his results with Richard’s results and to see what happens across the whole range of latitudes.
@DrB
On 23rd Oct 2018 you stated “I think, for the purpose of impact latitude discrimination, the best comparison of predicted and observed arrival times is probably coastal Africa. This has better statistics than the single flaperon and may have shorter delays overall between arrival and discovery.”
On 26th Oct 2018 you stated “So, when comparing your predictions for 31.6S versus 24S, I see that 31.6S is a better match for three metrics: flaperon transit time, the ratio of items going south around Madagascar to those going north around Madagascar, and the fraction of landed items (past Reunion) making landfall in Madagascar itself.”
So first it was the timing to mainland Africa, now it is the timing to Reunion, Madagascar and bifurcation around Madagascar. You are obviously changing the criteria to suit your interpretation of the data.
On the other hand, I am trying to fit mainland Africa, Madagascar, bifurcation around Madagascar, bifurcation ration North to South, Reunion, Mauritius and Rodrigues. Discounting one outlier stuck in a cave on Kojani Island for over a year, there is a better overall fit for 24S rather than 31.6S. You criticise the fact that I discount one outlier.
However, I do not exclude either 24S or 31.6S in my conclusion.
You manage to exclude 24S in your conclusion somehow, stating “I don’t think this is north of 25S”.
@DrB
The weight of the evidence is more important than the quantity of the evidence. The Inmarsat BFO data can tell us the aircraft flew South, and came down very rapidly at the end. That is all it can tell us. Take a look at figure 5.4 of “Bayesian Methods…”, and try to convince me otherwise. You keep making the same judgement error over and over again.
To search wider than +/-25km at any latitude is a serious tactical mistake. The drift modeling cannot be used to refine a latitude prediction. Go with what we know for god’s sake, and quit muddying the waters. The only intelligent way to procede is North from 25S at +/-25km.
@Mick Gilbert said: Consequently, it is far from almost certain that the recovered data sets were from a single contemporaneous session. Absent that ‘actual’ file created/saved/modified timestamps we simply can not know whether the recovered data sets were from a single session.
Yes, it’s possible that a single simulated flight occurred over multiple sessions that in total lasted an hour. I don’t see how that changes anything. The simulation data files are very likely part of a single flight.
@Richard G, @DrB: Since we are discussing metrics, and considering single finds as significant, what range of impact latitudes best explain the timing and location of “Roy” and the flap fairing in South Africa?
@Victor @Ventus45 @DennisW @Mick
Simulator Session Spculations:
(1) So on the age old question, why MH370 to Beijing? Another argument might be (per above) that the time schedule for the Jeddah flight kept shifting, and the planned night flight to Jeddah (MH168) apparently never materialized. Other arguments against using the Jeddah flight include that there would be 3 pilots.
(2) Joseph Coleman previously provided FlightAware data for MH150 probably on Feb-2 seems to me it takes a flight path via DUBTA. Thus I wonder if that was the simulator path (versus LAGOG) but I never could prove that from the few saved simulator points.
@Victor
You asked “what range of impact latitudes best explain the timing and location of “Roy” and the flap fairing in South Africa?”
Using David Griffin’s data from his latest simulation at the link below:
http://www.marine.csiro.au/~griffin/MH370/br15_MH370_IOCC_tp3l1p2dp_arc7_3608_25/
The range of impact latitudes that fits the location of “Roy” is 16°S to 24°S and 28°S to 31°S:
https://www.dropbox.com/s/8dzwi5xqg133pc7/David%20Griffin%2023%20Dec%202015%2034.0938S%2022.1499E%20Roy.pdf?dl=0
The range of impact latitudes that fits the location of the Flap Fairing “676EB” is 16°S to 20°S and 28°S to 37°S:
https://www.dropbox.com/s/n6bzmjhvcuc7qkf/David%20Griffin%2027%20Dec%202015%2024.0781S%2035.4998E%20676EB.pdf?dl=0
@DrB
You stated: “The debris finds indicate that about five times as many debris items were found in southern Africa compared to northern Africa.”
David Griffin’s data for the northern area shows 7,707 modelled dates of debris beachings:
https://www.dropbox.com/s/tdi3lu20e2ubein/David%20Griffin%20results%20for%20Tanzania.pdf?dl=0
David Griffin’s data for the southern area shows 1,736 modelled dates of debris beachings:
https://www.dropbox.com/s/qxhghhow4l8r73a/David%20Griffin%20results%20for%20southern%20mainland%20Africa.pdf?dl=0
Although the northern area (1.1 M sq km) is less than half the size of the southern area (2.6 M sq km), and the length of coastline of the northern area (approx. 1,000 km) is less than half the length of coastline of the southern area (approx. 2,500 km), there are 4.4 times more northern beachings.
As David Griffin observed one possibility is that “many items beached in Tanzania, Kenya and Somalia but were either not noticed or not reported.”
What we need to make a meaningful statement is the number of modelled dates of debris beaching for each start latitude, separately depicted over time. The diagrams that David Griffin kindly published include data from all 8 start regions from 8°S to 37°S on one page.
@Richard Godfrey: Thank you for checking.
@RichardG, VictorI
I try to look at the big picture first. I’ve played the .mp4’s David provided several times and to me the green particles seem to fit the overall picture (spatial distribution of finds and rough timing) best. Best, however, is not very good. What is striking is the small number of finds in Tanzania and Kenya, as well as north Madagascar (for all starting latitudes). Not sure yet what that would suggest.
Given the issues with find delays and small number of items it is really troublesome to put any numbers. On the other side these issues also mean that slightly north of S25 probably cannot be excluded. After “splitting off” Madagascar finds and with all uncertainties in find delays it has become clear to me that the area consistent with drift simulations is larger than it seemed before.
Maybe I should add something to “After “splitting off” Madagascar finds and with all uncertainties in find delays it has become clear to me that the area consistent with drift simulations is larger than it seemed before.”:
This is when we are led by what has been found, and ignore the lack of findings in the north. The “area” is based on “along the 7th arc.”
We have to find a good explanation for the lack of finds in the north, otherwise you could as well conclude the opposite: there is no starting latitude on the 7th arc which according to the simulations matches well with the spatial and temporal distribution of the finds.
@Niels: You might expect that if an interesting object washed ashore in Kenya or Tanzania, it might get reported. I think there is little chance of that occurring in Somalia.
So the usefulness of the drift modeling is severely limited by the dispersion of the floating objects, and the uncertainties in the timing and efficiency of reporting of recovered debris.
That’s not very encouraging.
@VictorI
David G. explained to me that Blaine Gibson searched also in the north of Madagascar and found nothing. Maybe we can verify the extent of his search there. It all worries me a bit.
@Niels: I sent Blaine a message. I’ll let you know what he says.
@All: A new article appeared in the French L’Express. The translated version is here.
Some highlights:
1) Ghyslain’s civil suit requested judicial information for a hijacking by a terrorist organization.
2) The French are interested in a Malaysian passenger that they believe is suspicious because he was an aeronautical engineer, he was seated near the SATCOM, and he had knowledge of the electrical “fuse boxes” on the plane. (Yes, you read that correctly.)
3) The French want to know if somebody with technical training could connect to the SATCOM and manipulate the data with a kit sold by a particular American company. (Square Peg Communications sells a GES emulator, but the company is located in Ottawa, Canada. Also, the “kit” would be a large, rack-mounted electronics box that could not be easily carried onboard.)
4) Mr Wattrelos once again claims that the French have never received the Inmarsat data. (The Inmarsat data was released and then analyzed on this blog. The ATSB has verified that they have the same Inmarsat data.)
5) Mr Wattrelos doesn’t understand how the radar systems of 7 countries missed the plane.
6) Mr Wattrelos claims his Facebook account was hacked and some digital files on his computer disappeared.
@Victor
I could not get your translated link to work, but your main link had a translate option, so no problem.
(1) I wonder why Mr. Wattrelos is pursuing such unlikely options, when the truth probably is Malaysia knows and is hiding evidence of pilot involvement? Part of that answer could be that Mr. Wattrelos has to restrict himself to possibilities that are not internal Malaysian issues, since France may have no jurisdiction or desire to intervene in those matters? His sole focus is outside of Malaysia whereas Malaysia is ground zero here.
(2) I think I ran into a interview with areonautics engineer’s father on YouTube, and he was quite convinced his son had no involvement with large planes like B777’s. I am trying to find.
@Victor
6) Mr Wattrelos claims his Facebook account was hacked
Just about all Facebook users can say that. Why even bring it up?
5) Mr Wattrelos doesn’t understand how the radar systems of 7 countries missed the plane.
Nothing new there or unique to Mr. Wattrelos.
4) Mr Wattrelos once again claims that the French have never received the Inmarsat data.
Probably true. I never “received” the data in the true sense of the word “received”. I had to download it just like everyone else. The French think they are special?
Wattrelos is definitely coming across as a whacko.
@Dennis:
“…. Mr Wattrelos claims his Facebook account was hacked
Just about all Facebook users can say that. Why even bring it up?
… Mr Wattrelos doesn’t understand..”
Right on, but elaborate conspiracies are more emotionally comforting than random actions by lone maniacs. [See the Florida demo-bomber….]
@TBill: The link for the translated version of the article is repaired.
@DennisW said: As you say, the Inmarsat data is available for anybody to download from this blog. On top of this, the French have the Inmarsat data, as it was sent to the GTA through an intermediary I know.
@TBill
Thank you for digging up that flight schedule data again.
@all
I am a bit frustrated. The Bayesian probabilities are very clear. As I showed before, with a 10% chance of the sonar missing the wreckage the probability of finding the aircraft in the 25S to 22S zone is ~43%. With a 5% chance of missing the wreckage the probability becomes:
0.95*3/(2 + 3 + 0.05*13) = 2.85/5.65 = ~50.4%.
This is the best opportunity to find the aircraft that we have ever had. Let’s not blow it with a misdirect. The only assumption is that the aircraft is in the +/-25nm band on either side the 7th arc, and the terminus is at 20S or below. I think those assumptions are very good bets.
@Victor @all
If you are following Kevin Rupp @LabRatSR, OI is apparently suspending the submarine search until February, thus making OI return for MH370 search appear unlikely for the upcomimng search season.
@Victor Iannello,
You said: “@Richard G, @DrB: Since we are discussing metrics, and considering single finds as significant, what range of impact latitudes best explain the timing and location of “Roy” and the flap fairing in South Africa?”
Actually the most interesting point to me, and one that I have queried David Griffin about, is that his model has virtually no debris landing that far south. So my answer to your question is that, based on his model, no latitudes explain the debris found there. Richard Godfrey’s model predicts a fair number of landings in South Africa or passing south of there into the Atlantic.
The “Roy” piece is interesting because the first photo of it on the beach shows lots of barnacles. It may be the only piece found in Africa where we are likely to have a very short reporting delay. On the other hand, it is still just one piece, and we don’t know if it was the first piece from a debris cloud or the last piece in the tail of the arrival train. Unfortunately, the time interval from, say, the 10th percentile to the 90th percentile is rather long, comparable to or longer than the difference in transit time from 24S to 32S. This makes it difficult to discriminate starting latitude using a single debris find. The flaperon is an exception because the debris cloud dispersion is considerably compressed in time and space for that shorter drift path to Reunion.
@Victor Iannello
@TBill
Victor, regarding the simulator data files.
I don’t see how that changes anything.
It certainly won’t change anything if you’re mind is already made up. However, the fact that there is no inviolable relationship between the simulator timestamps and real time most certainly raises (or, at the very least, cannot exclude) the possibility that the flight simulation data points represent a number of different and potentially unrelated flights/simulated exercises that were conducted over more than one session.
The simulation data files are very likely part of a single flight.
I disagree. There are at least two very significant impediments to the ‘string around the map pins‘ single flight interpretation of the simulator flight data. The first is that there is evidence that whatever was being simulated up to 5N had nothing to do with what happened at 10N and thereafter. We have previously discussed the the NAV1Active and NAV2Active values in the [Avionics.0] section of the flight files. The NAV1 (116.90, the PUT (Phuket) VOR) and NAV2 (114.10, the VPL (Langkawi) VOR) values at 10N are not consistent with that location. They are, however, consistent with the simulation aircraft being positioned somewhere between 5N and VAMPI. We know from the zero values for XVelBodyAxis, YVelBodyAxis, PVelBodyAxis, BVelBodyAxis and HVelBodyAxis at 10N that the simulation aircraft was most likely ‘dragged’ or otherwise manually repositioned there. Given the NAV1 and NAV2 values it is not unreasonable to assume that it was repositioned there from somewhere between 5N and VAMPI. Accordingly, the simulation aircraft’s attitude at 10N almost certainly represents what the aircraft was doing between 5N and VAMPI.
So what was the simulation aircraft doing between 5N and VAMPI? It was at FL400 and was established in 20° AOB left turn coming through 255.5°. The track to 10N from between 5N and VAMPI is about 300°. Thus, the aircraft was not heading towards 10N, it was turning away from 10N. In fact, if we assume that 2N to 3N to 5N are part of the same simulation exercise, that is a takeoff from Kuala Lumpur with a route up the Straits of Malacca, then somewhere between 5N and VAMPI it started turning around. Further, we know from the sim data that at some point after 5N the simulation aircraft registered +2.2032G (significantly up from the previous high value of +1.4397G) and +0.1453G (significantly down from the previous low value of +0.5958G). In order to achieve +2.2G the aircraft would need to perform a 60° AOB turn or be pulled steeply nose up (or some combination of a steep turn and being pulled steeply nose up); to achieve that very low +0.1453G the aircraft would need to be pushed suddenly nose down or something similar. In other words, the simulation had entailed manoeuvring that gave rise to both relatively high and low G and then the aircraft was subsequently turned around.
People will interpret that as they see fit. One interpretation is that an inflight emergency of some description that required a diversion had been simulated. Coincidentally or otherwise, a not dissimilar exercise forms part of the MAS B777 Full Flight Simulator training. Lesson 3G is a flight from Kuala Lumpur to Chennai with a dual engine failure/stall occurring roughly 40 minutes into the flight. The exercise requires an inflight engine restart and a subsequent diversion to Penang (coincidentally or otherwise, at 5N Nav2 is tuned to 116.20, the VPG (Penang) VOR). In order to practice a heavy weight take-off and to force a fuel jettison on descent to Penang the take-off fuel load is, for a flight to Chennai, an unrealistically high 70,000 kg (coincidentally or otherwise, the simulation flight took off with an estimated 68,424 kg). Coincidentally or otherwise, the Captain was rostered as a Type Rating Examiner for B777 Full Flight Simulator training on 8 February 2014. Let me be clear that there are differences between the simulator data files and FFS Lesson 3G but there are also similarities.
The second significant impediment to the single flight interpretation is that at 45S1 both NAV1 and NAV2 are tuned to 113.40, the BAC (Banda Aceh) VOR. In my experience with the simulation program, if you fly the aircraft directly from 10N to 45S1 you do not achieve those navaid settings at any point. Starting at 10N, the only ways that I have been able to get the BAC VOR tuned on both NAV1 and NAV2 is to either manually tune both receivers in the NAV RAD Page in the FMC or to enter WITT as the destination airport (without entering an approach) and fly there. Now, I am far from proficient in using the PSS B777 Professional flight sim program so there may be something that I am missing – Bill, you might want to have a look at this. However, based on my current experience the simulation aircraft cannot be flown from 10N directly to 45S1 and end up with the 45S1 [Avionics.0] values unless they manually entered (and that strikes me as extraordinarily odd).
It is probably worth noting that in their 3 October 2017 The Operational Search for MH370 report the ATSB raised the possibility that there might have been an intermediate data point between 10N and 45S1 that had not been captured. The 45S1 [Avionics.0] values seem to support that possibility.
Consequently, I contend that there is evidence that the simulation data files are not part of a single, largely or wholly, contiguous flight rather they are more likely to be represent two or three separate simulation exercises. That is not to say that at least one of those ‘exercises’ is not suspicious or potentially nefarious.
@Niels,
You said: “We have to find a good explanation for the lack of finds in the north, otherwise you could as well conclude the opposite: there is no starting latitude on the 7th arc which according to the simulations matches well with the spatial and temporal distribution of the finds.”
I gave that explanation here on October 26th st 1:09 am. Richard’s model predicts, for a 32S starting latitude, a very low percentage of articles landing in Africa north of Madagascar and the majority landing in Madagascar with the remainder landing in Africa after passing south of Madagascar. This matches the observed debris distribution. All starting latitudes north of 25S show a much higher fraction landing in northern Africa. David Griffin’s predictions are in general agreement with Richard’s results for the relative number of landfalls in south and north Africa.
@Richard Godfrey,
You said:”The diagrams that David Griffin kindly published include data from all 8 start regions from 8°S to 37°S on one page.”
David Griffin’s one-page colorized plot is useful qualitatively. I have requested segregated quantitative data. Another issue is that his plot does not distinguish landfalls in Africa from landfalls in Madagascar from 20S-22S. It shows their sum. I have also requested the breakdown.
@Richard Godfrey,
You said: “So first it was the timing to mainland Africa, now it is the timing to Reunion, Madagascar and bifurcation around Madagascar. You are obviously changing the criteria to suit your interpretation of the data.
The old criteria needed changing. The reason is that, with one or two exceptions (the flaperon and Roy, the two pieces with barnacles attached), discriminating starting latitude simply cannot be done effectively using reporting times because of the generally unknown and potentially large delays between arrival and reporting for most of the found articles. The bifurcation at Madagascar is sensitive to starting latitude, and using landfalls integrated over time eliminates the errors from unknown reporting delays.
@Richard Godfrey,
You said: ”You manage to exclude 24S in your conclusion somehow, stating “I don’t think this is north of 25S”.”
The “somehow” is what I showed: that 31.6S matches all three criteria I described better than 24S. In my mind, this makes the probability of 24S relatively low.
@DennisW,
You said: “The drift modeling cannot be used to refine a latitude prediction. Go with what we know for god’s sake, and quit muddying the waters. The only intelligent way to procede is North from 25S at +/-25km.”
I disagree. I believe the debris data can discriminate north/south of 25S.
It seems that if someone disagrees with you they must not be intelligent. I wonder what Ami thinks about that attitude?
In your mind the water is getting muddy; in my mind the water is clearing up.
@DrB
You wrote “Richard’s model predicts, for a 32S starting latitude, a very low percentage of articles landing in Africa north of Madagascar and the majority landing in Madagascar with the remainder landing in Africa after passing south of Madagascar.”
My statement was based on DavidG’s calculations which show considerable beaching probabilities in the north, also for S32 as starting latitude, see his fig. 5b.
@Niels,
I agree that there were probably many more beachings in the north, which have gone unnoticed.
However, I think you are referring to David Griffin’s Fig. 5a, which shows beaching latitudes west of 65E dominantly between 12S and 15S (Northern Mozambique) from start latitudes between 29S and 36S on the 7th Arc.
In Fig. 5b, which shows beaching latitudes on northern mainland Africa west of 45E dominantly at 10S (Tanzania Mozambique border) from a start latitude of 22S on the 7th Arc.
@Victor
You asked “what range of impact latitudes best explain the timing and location of “Roy” and the flap fairing in South Africa?”
I previously answered with reference to David Griffin’s data.
From my own simulation runs, the nearest beaching in location and timing to “Roy” at 34.0938°S 22.1499°E after 655d is the following track from 25.6655°S 101.4563°E:
https://www.dropbox.com/s/cvj8zxi16d1d7fn/Drift%20Map%2025.6655S%20101.4563E%2033.3849S%2027.9161E%20683d.png?dl=0
The nearest beaching in location and timing to “676EB” at 24.0781°S 35.4998°E after 659d is the following track from 23.1672°S 103.8556°E:
https://www.dropbox.com/s/whw38zplfjpqwfv/Drift%20Map%2023.1672S%20103.8556E%2026.2268S%2032.9103E%20727d%20Gyre.png?dl=0
@RichardG
I was refering to fig. 5b, with a focus on S32 as a starting latitude, as this was the discussion point with @DrB
Regarding “Fig. 5a, which shows beaching latitudes west of 65E dominantly between 12S and 15S (Northern Mozambique)”
I think this “peak” between 12S and 15S is actually originating from northern Madagascar and not from Northern Mozambique (by comparing fig. 5a and fig. 5b)
@DrB,
An analysis of the MH370 debris items reported from the various countries both on mainland Africa and the islands in the Indian Ocean is revealing.
South Africa is the only country where all of the MH370 debris items (5 in total) were found and reported by locals.
In Mozambique all of the MH370 debris items (7 in total) were found and reported by tourists (mostly from South Africa) and none by locals.
Countries like Tanzania or Madagascar that have a long coastline would not have the awareness of MH370, nor perhaps the discernment to identify an aircraft debris item, nor perhaps the willingness to report a find. Blaine Gibson says that until he visited Madagascar and raised awareness of MH370, nobody had heard of MH370. On Reunion and Mauritius there was 1 item of MH370 debris, in each country, reported by a local, on Reunion an official beach cleaner and on Mauritius a coast guard foot patrol. As Victor says you might expect Kenya to have some awareness of MH370, but its coastline is shorter and it is a poor country. Somalia with one of the longest coastlines is very poor, racked by civil war, where only 2% of the population have internet and only 10% of the population have secondary education. I agree with Victor that there is little chance of any possible MH370 being reported in Somalia.
Here is a link to a table giving the overview of MH370 debris reported by country and the relevant UN statistics for each country:
https://www.dropbox.com/s/f5f4sxe6eher46a/Debris%20Discoveries%20by%20Country.png?dl=0
@Niels
You make a good point, comparing David Griffin’s Fig. 5a and Fig. 5b.
I had erroneously thought, that Fig. 5b was only the northern mainland Africa beachings, but the graphic shows all beachings west of 45E.
So Fig. 5b shows that the dominant beachings on Tanzania between 4°S and 11°S are from a crash latitude of 22°S on the 7th Arc.
The difference between Fig. 5a and Fig. 5b, as you point, out shows that all the beachings between 12°S and 26°S are on Madagascar, Reunion, Mauritius and Rodrigues from crash latitudes between 23°S and 36°S on the 7th Arc
Where are the beachings on Fig. 5b between 11°S and 27°S on Mozambique (7 reported discoveries)?
Where are the beachings on Fig. 5b between 27°S and 35°S on South Africa (5 reported discoveries)?
Fig. 5a is linked here: https://www.dropbox.com/s/5v3rf6ipp72orp5/David%20Griffin%20Fig.%205a.png?dl=0
Fig. 5b is linked here: https://www.dropbox.com/s/qm2tdh7gjc3gpqy/David%20Griffin%20Fig.%205b.png?dl=0
@DrB
you said:
I disagree. I believe the debris data can discriminate north/south of 25S.
Well, that is certainly a point of disagreement. The fact that latitudes below 25S have been searched to +/-25 nm from the 7th arc is a fact. Nothing has been found there.
What you are advocating is a wider search below 25S. I think that is flawed for several reasons.
1> It is based on your interpretation of the Inmarsat data. Thus far no one’s interpretation of that data has been correct.
2> You are assigning lack debris finds to lack of debris beachings. They are not at all the same thing as people have pointed out above.
3> Searching wider is a very “expensive” strategy. Consider a search of 3 degrees along the 7th arc below 25S at a width of +/- 120 nm. That would involve a seach of 1.5 additional degrees (90 nm) on each side of the 7th arc or a total of 9 square degrees of search area. A search of 3 degrees farther North (Ami would be proud of my correct choice of farther versus further) from 25S at +/- 30 nm is about 3 square degrees of search area.
4> You are discarding one of the two things the BFO data is actually capable of telling us – the plane descended rapidly after fuel exhaustion. Maybe you are not dsicarding it, but rather assuming a pilot recovery from the steep descent. There is no evidence of such a recovery, and the lack of an IFE log-on supports a no recovery conclusion.
Basically, your logic is hanging by a very thin thread.
@Mick Gilbert: You raised points that I’ll address regarding the sim data.
1) The plane was dragged from a position between 5N and 10N while in a turn.
True. So the user was heading towards 255.5° while turning left, and then the simulation was stopped and the plane icon was dragged to 10N. That doesn’t mean that 5N and 10N were not part of the same simulation.
2) The minimum and maximum values of wing loading changed between 5N and 10N.
True. That doesn’t mean that 5N and 10N were not part of the same simulation. It means the user was aggressively working the controls. Now, if the minimum value of wing loading had increased, or the maximum value of wing loading had decreased, or if the maximum value of engine RPM had decreased, that would be persuasive evidence that 5N and 10N were not part of the same simulation. That evidence doesn’t exist.
3) Both VORs were tuned to WITT.
True. That doesn’t mean that 5N and 10N were not part of the same simulation. Depending on the exact route, that occurs in both the FSX PMDG777 and the FS9 PSS777, although I admit that at this point, I am much more familiar with the PMDG777. The VORs that are used seems to be a function of both the current position and the position history, and the VORs used are often the same. That doesn’t mean that 5N and 10N were not part of the same simulation.
If you want to imagine a simulation flight in which there are additional points between the ones we have, that can’t be disproven. Nor can we disprove that within that one hour of simulation time, there were two or more separate simulations that were conducted, and those simulations could have been conducted over the course of multiple computer sessions. However, there is no evidence that occurred.
Based on the evidence we have, I continue to believe the most likely and simplest explanation is the user created a simulation that contemplated a flight up the Malacca Strait with a turn towards the south and ending in fuel exhaustion in the SIO. Considering the similarity with the flight sequence we believe MH370 followed, and considering the timing of this simulated flight session just weeks before the disappearance, that would be an extraordinary coincidence if the simulation data was not somehow related to the disappearance.
I think the bigger question is to understand the true relationship between the simulated data and MH370’s disappearance. I don’t think the simulation session was conducted for practice or for planning, as there are better and easier tools to use for both.
@Niels,
The “big picture” in David Griffin’s Figure 5b is that for starting latitudes north of 23S, most the debris hits northern mainland Africa (and as far north as the equator), a small fraction hits Madagascar, and a very small fraction hits southern mainland Africa. For starting latitudes south of 23S, most of the debris hits Madagascar, a small fraction hits southern Africa, and an even smaller fraction hits northern Africa. Nothing discernible in his Figure 5b hits mainland Africa south of 27S landfall for ALL starting latitudes. This appears inconsistent with the southernmost debris reports, and I have queried Dr. Griffin on this point. Richard Godfrey’s model does predict debris making landfalls in Africa that far south, and it is in general agreement with the bifurcation ratios changing with start latitude near 23S, as shown in Dr. Griffin’s Figure 5b (especially when you account for the fact that Figure 5b is ambiguous from 20S-22S landfall latitude because it commingles Madagascar landfalls and Africa landfalls in this region).
@DennisW,
You said: “It is based on your interpretation of the Inmarsat data. Thus far no one’s interpretation of that data has been correct.”
Actually, we don’t know that at all. We only know that those interpretations which led to predictions within the previously searched areas are most likely wrong. We don’t yet know whether or not any or all interpretations which predict other locations outside the previous search are correct or not. We haven’t searched wider or farther north, so we don’t yet know if any predictions based on satellite data in those regions are correct or not.
You said: “You are assigning lack debris finds to lack of debris beachings. They are not at all the same thing as people have pointed out above.”
I have pointed this out, too. We don’t know the bias introduced by search effectiveness or by reporting delays. For the two items found with barnacles, we can be more certain that the reporting delay is not long.
You said: “Searching wider is a very “expensive” strategy.”
I don’t know anyone else who is recommending a rectangular search area with a width as large as +/- 120 NM, as you used in your example. In my opinion that width is excessive. I would suggest a much smaller width (60-75 NM).
You said: “You are discarding one of the two things the BFO data is actually capable of telling us – the plane descended rapidly after fuel exhaustion. Maybe you are not dsicarding it, but rather assuming a pilot recovery from the steep descent. There is no evidence of such a recovery, and the lack of an IFE log-on supports a no recovery conclusion.”
There are only two facts known about the flight path after 00:19:37: (1) the aircraft crashed in the SIO, and (2) the IFE did not log on circa 00:21:06. There are several possible reasons why the IFE log-on did not occur, the two most likely being either the crash occurred prior to the log-on, or the APU ran out of fuel in the line and the AES lost power at 00:21 with the crash occurring later. At this point we don’t know which occurred. Remember what Duncan Steel used to say: “Absence of evidence is not evidence of absence.” You say “there is no evidence of such a recovery.” I say, there is no evidence (of either a recovery happening or not happening). You say: “the lack of an IFE log-on supports a no recovery conclusion”, but I say there is no evidence to distinguish whether or not a recovery occurred; both are possible and fit the facts.
You said: “Basically, your logic is hanging by a very thin thread.”
I say your “logic” is illogical.
I don’t see a way to draw a conclusion about the probability of pilot recovery with the information we have now. We know that a “suicide descent” can match the 00:19 BFOs (as I have demonstrated using the EgyptAir 990 example). We also know that a simulated unpiloted descent can also match the BFOs (but not at the right time). We also know an unpiloted descent is very unlikely south of 26S; we don’t know anything north of there because it has not been searched. Now, if the Boeing simulations matched all the EOF data (including the timing), then I would concede there is a third piece of significant information consistent with an unpiloted descent, and that would tip the scale in its favor. However, at present there is no explanation of the simulation timing inconsistency.
@DrB
You wrote:”The “big picture” in David Griffin’s Figure 5b is that for starting latitudes north of 23S, most the debris hits northern mainland Africa (and as far north as the equator), a small fraction hits Madagascar, and a very small fraction hits southern mainland Africa. For starting latitudes south of 23S, most of the debris hits Madagascar, a small fraction hits southern Africa, and an even smaller fraction hits northern Africa.”
I find this a bit confusing: are you addressing here fig.5b, 5a or both?
Actually in fig. 5b (beaching on mainland Africa) it is difficult to identify a significant percentage of beaching items south of S12 for all starting latitudes, as also pointed out by @RichardG. I’ll check this with DavidG; it possibly is caused by the definition of colours. Everything below 1.5% appears dark blue.
Based on what I can see in fig. 5b it appears that there are more beaching items predicted in the north of mainland Africa (S15 – equator) than in the south for all starting latitudes. Again, I’ll check this with David Griffin.
@DrB
Ok. Imagine you are calling the shots on the search. What would you do next?
I would search the 7th arc from 25S to 22S at a width of +/- 25nm. That is a step into the tall grass for me.
@DrB
BTW, I tell Ami the same thing I would often tell my boss in my working life. I can only do so much.
Looking forward to your answer on my last query. Be careful with it.
@Mick @Victor
To be honest I am do not look at radio frequencies in Flight Sim as that is not a familiar or past exploration area for me. I would expect what Victor is observing for FSX (MicroSoft Flight Sim 10) might be the same for FS9 (MicroSoft Flight Sim 9). FSX is nearly the same as FS9 except FSX has better graphics resolution. I am assuming the choice of addon 777 model (PMDG vs. PSS) may not ovrride the native radio tuning on the FS9/FSX platform.
Mick’s comment about ATSB suggesting the possibility of another point between 10N and 45S1, gets to the question: Did FBI get all the data they could out of the deleted hard drive files? Or is there more that could be learned with the help of a consultant like Victor who could advise some of the files and file structure we are looking for?
@Victor Iannello
Victor,
On the basis that I don’t believe that the flight sim data has any use at all in refining the search area I’m not going to labour on this issue but I will make a couple of points.
As you would know, not every use of the flight simulator is a complete or even typical flight that starts with the aircraft on the ground followed by a take-off, departure, climb out, etc. Depending on the task at hand, it is common practice to simply go to a previously saved flight file that has the aircraft airborne and then use the menu functions to modify fuel load, weights, etc and the map function to reposition the aircraft to where you want it and modify heading and altitude and then undertake whatever task or exercise you want to simulate. When you do that any data from the old saved flight file that is not modified is ‘inherited’. In other words, unless the user manually modifies the [DateTimeSeason] data, their new flight simulator exercise will appear to be a continuation of the previously saved flight. Accordingly, even though the timestamps will suggest that the current exercise is a continuation of the previous flight/simulation, it most likely is not.
When using the Phoenix B777 Professional flight simulation add-on with Microsoft Flight Simulator 2004 there are at least two different types of .FLT files created, either;
– a comparatively larger file (1397 lines of data organised into 49 data groups), or
– a smaller file (418 lines of data organised into 47 data groups).
The larger files are created using the Save command (;) and contain a data group called [Phoenix Sim Software] which has 946 lines of data specific to the PSS B777 sim. The larger file also contains a 33 line data group called [GPS_Engine].
The smaller file, which is always labelled ‘Previous Flight.FLT’, is created automatically whenever the simulation program is exited. What that means is that every time the simulation program is exited the Previous Flight.FLT file is overwritten. Because the file is overwritten, if the Microsoft System Restore Point function has been set up, the Volume Snapshot Service will create a shadow copy of the old Previous Flight.FLT. Depending on the System Restore Point function setting it is possible that a new shadow copy of the old Previous Flight.FLT file is created every time the program is exited.
At the very least, this raises the possibility that the data files are in fact a series of ‘end of’ simulations that may be a continuous flight but are at least equally likely to be a series of potentially unrelated simulation exercises conducted over different simulation sessions. In my view the recovered data files being versions of Previous Flight.FLT files makes a lot more sense than the user repeatedly saving the simulation session at various points in one session (and naming each saved file differently) and then clumsily deleting the lot. The notion that someone with the wherewithal to build his own flight simulator (and who was looking at building a motion function into it) would not know how to permanently remove data from his computer has never made any sense to me.
Separately, something else that I noticed when using the flight sim software is that when the map function is called up the lat/lon grid gets rescaled depending on the zoom. Interestingly or otherwise, both 10N and 45S1 are very close to lat/lon grid line intersections; 10N with the zoom one level below maximum zoom-out and 45S1 at maximum zoom out. That raises the possibility that there was there’s no particular navigational significance to either point and that the user has just dragged the aircraft icon to a point near a grid line intersection.
Now let me address a couple of the points you made;
‘… the user was heading towards 255.5° while turning left, and then the simulation was stopped and the plane icon was dragged to 10N. That doesn’t mean that 5N and 10N were not part of the same simulation.‘
That’s a curious conclusion although I note that you use the word ‘simulation’ rather than ‘flight’. The aircraft was not only not pointing towards 10N but was actually turning away from it. While that may not mean that 5N and 10N were not part of the same simulation it is evidence that they were most likely not part of the same flight. Accordingly, the old ‘string around the map pins’ representation of the data points as one continuous flight is misleading if not disingenuous.
I also note that back when it was believed that the data showed that the simulation aircraft at 10N was turning south towards 45S1, that was used to advance the notion that the two points were linked. However, now that we have evidence that after 5N the aircraft was turning away from 10N you seem to want to reject that as evidence that the two points are not linked.
‘Both VORs were tuned to WITT. … That doesn’t mean that 5N and 10N were not part of the same simulation.‘.
The VORs were both tuned to the BAC VOR at 45S1; that has nothing to do with the relationship between 5N and 10N. It does however have a bearing on the relationship between 10N and 45S1. If you still have the flight files created for your joint paper with Yves I recommend that you review the [Avionics.0] data group for your 45S1. I suspect that you will find that it does not have the BAC VOR tuned on either NAV1 or NAV2, leave alone both. Rather than me imagining ‘a simulation flight in which there are additional points‘ that should be fairly compelling evidence that there must be at least one additional point. If you want to understand the true relationship between the simulated data and MH370’s disappearance then that might be an important consideration.
@Richard Godfrey. Also DrB, Neils . I note that David Griffin has made a substantial change to his 26th October analysis in finalising his then draft. That change is to his ‘east’, ie Madagascan & other islands beachings’ analysis, 4th para from the bottom, starting “Now considering…..”
While that could be amended further the below addresses his analysis as it now stands.
He alludes to the several months’ gap (note 1) between the modelled and the reported in the ‘east’ 50 percentiles at a 25˚S crash latitude or north. To bring those latitudes to compatibility with the modelled would require a shift of the ‘reported’ 50 percentile line left (“…the delay needs to be 8 months…). That implies the average of the individual reported timings would be reduced by that amount. Maintaining the same spread in ‘reported’ (note 2), that shift also would entail a shift left of the other two percentiles.
The 10th percentile is of passing statistical interest, numbers being small. On the other hand the confidence in the 90th percentile should be the highest of all three, its number of ‘reported’ being highest. However, shifting the ‘reported’ 90th percentile left by even 6 months would lead to compatibility with the modelled 90th percentile just at the extreme northern latitudes of 10˚S and higher. So the fix for one percentile does not suit the other.
Moving north, while you have argued well that a lesser ratio of recognitions to modelled beachings could be expected there, I think it is very questionable that it would reduce the recognitions near enough to zero when the modelled landings there over all crash latitudes have been more than half the total. Moreover as apparent from Fig 5a, as the crash latitude moves north that proportion rises even further.
Besides I think it quite possible that the number of low windage, undrogued buoy equivalents which might have landed there might stand at zero.
As I have advanced previously the one item that did beach, ‘Pemba’, a part of the right outboard flap, has a similar shape and size to the flaperon, so it might not travel directly downwind. Indeed if it drifted trailing edge to wind, as does the flaperon mostly, its shape and right-hand-down (the flaperon floats left-hand-down) should lead it to stray to the right IMO, towards Tanzania. In other words it would not represent one of the low windage/undrogued drifters the CSIRO has modelled (note 3).
As to the accuracy of the model, it differs from many in being based on conditions and currents at the actual time and place and has empirical data in support. Even so it depends on the outcome of inshore and offshore testing that might have been affected by local current eddies and on the addition of increments of deduced speed and direction fixed over finite periods, although minimised; ie tracks are an addition of a series of straight lines rather than curves. David Griffin has mentioned this and I know you are aware of it.
Related to the testing, it supposes that excluding the flaperon, all other items will travel at undrogued drifter speed. That may be so as an average over a large number, though how closely is unknown and as the number decreases confidence that its average is accurate decreases with it. The MH370 items which nominally are drifter-equivalent vary from the engine cowling, item 6, large and curved, to item 11, the IFE monitor frame, small and flat in the water and most likely with less leeway. No-one knows which way up the cowling, as an example, will be in strong conditions and what orientation to the wind and waves/wavelets it will assume.
These MH370 items are likely to be disparate therefore, the spread of speed between them varying with conditions together with their distribution around the mean, which supposedly is represented by an undrogued drifter.
While leeway does not affect the outcome as much as does current, with MH370 it may be that there will be items and conditions where it is appreciable.
Were that not so, tests with the flaperon would confirm it drifts at drifter speed, near enough, though they have not borne that out at all. Yet, a similarly shaped item of comparable size, the flap is assumed to match a drifter. (Aside: The flap and other items should have been released for trials or at least real examples obtained and modified as was the trialled flaperon.)
The effect of the MH370 distribution is unknown so therefore is its consequence, as is that from including items ‘possibly’ from MH370 (note 4), or indeed excluding them as I had done.
An outcome is that considering a solution for 23-24˚S, which I understand to be your focus:
• First, noting the accuracy of this drift analysis is unknown;
• nevertheless taking the outcome as it is and assuming a sensible margin for error to be ± 3 months, from inspection of the diagrams, as an example, shifting the 50th and 90th ‘reported’ percentiles left 4 months in ‘south’ and the 90th percentile 6 months left in ‘north’ would bring 23 -24˚S there into play, and also those north of 20˚S.
• However ‘east’ is incompatible with that, as earlier, because while shifting the 50th percentile left say 6 months would be compatible, shifting the 90th percentile by the same amount makes that incompatible. So the whole fails.
• I have not looked at other crash latitudes as yet for a compatible combination. I hold no high hopes for north. At a glance I see there is a solution below 26˚S for a 1 month left shift of the ‘east’ 50th and 90th by a month (ie a month’s average, not each item). That would require a right shift of a month or two of the ‘north’ 90th percentile, the ‘south’ percentiles staying where they are. That would allow down to 34˚S.
• Of course those results are sensitive to the ± 3 months error margin assumed.
Notes 1 – 4, of more detail, are attached under.
https://www.dropbox.com/s/qnv4e68dgz7rp11/Notes.docx?dl=0
@Mick Gilbert
The nearest waypoint to the 3N simulator location is GOBAS and a simulator heading of 305.3° aligns with waypoint TASEK next. The simulator VOR is tuned actively to VBA at Batu Arung and the standby to the DME and LOC-ILS at Kuantan Airport runway 36, which fits the previous track.
The nearest waypoint to the 5N simulator location is TASEK and a simulator heading of 314.7° aligns with waypoint IDKUT next. The VOR is tuned actively to MDN at Medan and the standby to VPG at Penang, which fits the previous track.
The nearest waypoint to the 10N simulator location is AMVUR and a simulator heading of 255.5° at a 20° Bank angle aligns with a turn towards the Southern Indian Ocean. The VOR is tuned actively to PUT at Phuket and the standby to VPL at Langkawi, which fits the previous track.
At the 45S simulator location the VOR is tuned both actively and on standby to BAC at Banda Aceh, which fits the previous track.
I do not see the discrepancies between 3N, 5N, 10N and 45S positions, headings and VOR tunings that you point out.
@Mick Gilbert: Some more points:
1) If the file Previous Flight.FLT was written over multiple times as you propose, we would only have one flight file in the shadow volume. The shadow volume only contains the blocks of data that were written over at the time the shadow volume was created. The flight file fragments were clearly flight files with different names that were overwritten.
2) The heading and bank at 10N is not evidence that 5N and 10N were not part of the same simulated flight. (And I am not making a distinction between “simulation” and “flight”.) A left turn was in progress after 5N, the simulation was paused, and the plane icon was moved to 10N. The heading may or may not have changed before the save. The flight file was saved. That’s all very easy to do.
3) The fact that both VORs were tuned to the same frequency is not all that strange. It just means that at the time and position the simulation was stopped after 10N, both VORs were autotuned to BAC. That can occur.
So why was a turn to the left initiated after 5N and before 10N? I can think of several valid reasons, all speculative. Why was the simulation even created? I can think of several valid reasons, all speculative. Why were the flight files overwritten? I can think of several valid reasons, all speculative.
Whether we have the correct answers to those questions does not diminish the extraordinary evidence that the simulation files are related to the disappearance. You are clearly not in agreement. I’m open to new evidence or insights. Otherwise, let’s agree to disagree and move on.
@David
I’ve read your paragraph below about five times. The reality is that the reported beachings are what they are. What are you trying to say – the drift models are wrong?
Moving north, while you have argued well that a lesser ratio of recognitions to modelled beachings could be expected there, I think it is very questionable that it would reduce the recognitions near enough to zero when the modelled landings there over all crash latitudes have been more than half the total. Moreover as apparent from Fig 5a, as the crash latitude moves north that proportion rises even further.
@David
I do not understand what you are trying to say.
I do not see any changes in David Griffin’s blog, since it was first published.
I do not understand if you are referring to my Drift Model or CSIRO’s Drift Model.
@Dennis. The crash sites ranged from 8 to 36˚S. More than half the predicted beachings from all those sites were predicted to land in the ‘north’. These came principally from northern crash sites. It follows that were there few reported landings ‘north’ it could be deduced that the actual crash site was not up north. Richard Godfey made an alternative case that the reason for few reportings there was that the reporting rate up there most likely was very low.
I agreed that that could be but I doubted it would be so low that only one was reported ‘north’ where 28 were reported from areas where less than half those aggregate modelled beachings had been predicted. I added the casual observation that if the model had included just the northernmost crash sites the proportion predicted to land north would have been even higher. Had the crash indeed been towards the northern end of the arc and reporting in the ‘north’ sector was non-existent, the ‘evidence’ of no beachings would be misconstrued, the crash seen as being from elsewhere.
You say, “…the reported beachings are what they are.” That is so but the reported beachings need not reflect actual beachings, which is what Richard Godfrey was saying. How under representative they can be is the question.
@Richard Godfrey. I can find no solution based on a 23-24 deg crash latitude which is compatible with David Griffin’s 3 figures, ‘north’, ‘south’ and ‘east’. The principal difficulty is with the Madagascan (“east”) zone figure, when trying to match both 50 and 90 modelled percentiles to the reported beachings while allowing for delay in reporting the beachings. I suspect there will be no crashes in the northern sector which will get through the different hoops in those three figures though there are solutions applicable to the southern sector.
If you think you can find one from the northern sector of the arc well give it a try. Remember that it has to be compatible with both 50 and 90 model percentiles in ‘east’ and ‘south’ and with the 90th in ‘north’.
The paragraph I refer to is not the same as in David Griffin’s original posting, a draft. I have a copy of that but there seems little point in repeating it now. His post was a draft when first you acknowledged my advice about it.
I was referring to David Griffin’s model. I made that clear in the last sentence of my first paragraph.
@Victor
Someone recently encouraged questioning our assumptions. I forget who it was, and I am not going to try to find the post.
But good grief. Between David, DrB, TBill, MickG,… I feel like have walked into the twilight zone. Make it stop.
@Victor
I’d be interested in your speculation for why turn at/near 5N.
@David
You are defending David Griffin’s interpretation of his results, until you are blind to what he is actually saying.
David Griffin is moving progressively to a more northern crash point.
If I have an issue with David Griffin’s results, I write to him directly.
I do need you as an intermediary.
My data is actually not that far different from David Griffin.
My interpretation of the data is still different.
Let us agree to disagree and move on.
Like @Dennis, I am tired of scientists, who cannot follow the data, just their preconceived ideas.
I at least have taken Bobby Ulich’s end point of 31.6S seriously.
My next project is a drift analysis based on Victor’s end point of 21.9666S.
Unlike you, I am keeping an open mind and following the data.
@Dennis
It was me saying we should question our assumptions.
I wish other scientists on this blog would do the same.
@TBill
You asked Victor “I’d be interested in your speculation for why turn at/near 5N.”
I already answered this question and there is no need for speculation.
Just look 7 comments earlier …
“The nearest waypoint to the 5N simulator location is TASEK and a simulator heading of 314.7° aligns with waypoint IDKUT next.”
ZS was using waypoints in his simulator and the small turn at TASEK of 9.4° to the right was because the next waypoint was IDKUT.
@Richard Godfrey: @TBill is referring to the 20° banked turn to the left at 10N. Since we know the position was manually changed just before saving the data set at 10N, and the bank is preserved during that change in position, the left turn had to begin at some point between 5N and 10N, at which time the simulation was paused and the position change made to 10N. That doesn’t mean that 5N and 10N were not part of the same flight simulation.
@David
You stated “I can find no solution based on a 23-24 deg crash latitude which is compatible with David Griffin’s 3 figures, ‘north’, ‘south’ and ‘east’.”
North: Please look at the orange square at beaching latitude 10S (Tanzania) from a crash latitude of 22S in the following link (Fig. 5b) and tell me it is not there:
https://www.dropbox.com/s/qm2tdh7gjc3gpqy/David%20Griffin%20Fig.%205b.png?dl=0
East and South: Please look at the yellow square at beaching latitude 12S (Madagascar) from a crash latitude of 23S. Also at the same crash latitude of 23 S, there are green squares at 13S and 15S, light blue squares at 14S, 16S, 17S, 20S, 21S and 23S in the following link (Fig. 5a) and tell me they are not there:
https://www.dropbox.com/s/5v3rf6ipp72orp5/David%20Griffin%20Fig.%205a.png?dl=0
I am using David Griffin’s data to prove you are wrong.
@Victor
The bank at 5N, to which @TBill was referring, was 0.01386.
@Richard Godfrey: No. We are talking about the left turn that occurred AFTER 5N that was carried forward when the 10N data set was created and saved. That was the subject of the original exchange between @Mick G and me.
@Victor
You were responding to @Mick. You and Mick disagree on the interpretation of the AFTER 5N to 10N move.
I was responding to @TBill, who stated “at or near”. I was giving evidence that “at” is not correct. As you say “AFTER”.
@TBill wanted to know why the move/turn took place. How do we know it was a move then turn, or a turn then move?
Does any of this debate change anything in the fact that ZS ran a simulation ending in the SIO at fuel exhaustion?
@TBill asked: @Victor, I’d be interested in your speculation for why turn at/near 5N.
@Richard Godfrey responded: I already answered this question and there is no need for speculation.
So, if you two are satisfied that all questions are answered, I’ll leave it at that.
@Victor
It was apparent to me that @TBill had not seen my prior comment, relevant to this issue.
However, my comment is certainly not a reason for you to give up following a particular line of inquiry and “leave it at that”.
@Niels,
I should have said Figure 5a, which his text says is “modelled items beaching on African shores”, but whose label says “West Indian Ocean locations.” It’s confusing. I am assuming his text is what he intended.
@DennisW,
The two potential MH370 search areas I think have promise are (1) +/- 25 NM along the 7th Arc from 25S to 20S, and (2) the area outside the previous search and within 75 NM of the Arc from 31.2-32.0 S.
I also believe that the current studies of DG’s and RG’s drift models should be continued. That may lead to a better understanding of the relative probability of the northern and southern arc sectors. If that were to happen, it would factor into setting a priority area.
There is no need to make recommendations now. There is no funding at present and OI has suspended their ARA San Juan search until February. Interesting, isn’t it, that they are stopping in the summertime. There is a parallel with MH370 in that they have completed all the areas that were initially proposed, and nothing has been found. I am assuming they are being paid based on data area, but there is probably a contractual limit on the maximum value. Thus, it may be that suspending the search now and reassessing proposed new areas assures that, whatever is the choice, a new area can be covered within the current contract.
@Richard Godfrey. I think you could do with a break.
@Victor Iannello
1) If the file Previous Flight.FLT was written over multiple times as you propose, we would only have one flight file in the shadow volume. The shadow volume only contains the blocks of data that were written over at the time the shadow volume was created.
Sorry Victor, that’s not correct at all. The Volume Snapshot Service will create a shadow copy of an overwritten file every time a System Restore Point is created. For files with the same name that are overwritten repeatedly the shadow volume contains a history of the versions, not just the penultimate version as you’ve contended. That’s how Windows 7 supports the previous versions function for files. Windows 7 can be set to automatically create System Restore Points either on a trigger condition (such as ‘on start up’ or ‘at idle’) or on a schedule (weekly, daily, even hourly).
I have set up a laptop with Windows 7 (set to automatically create a daily System Restore Point), Microsoft Flight Simulator 2004 and PSS B777 Professional. The shadow volume is populated with multiple versions of the Previous Flight.FLT, one for the last use on any day I have used the flight sim.
The flight file fragments were clearly flight files with different names that were overwritten.
Maybe, maybe not. It is far from clear that we’re looking at flight files with different names. One way to differentiate if they were manually saved files is that they would be large format (1397 lines of data organised into 49 data groups). Unfortunately, the data groups from the recovered files as detailed in the RMP report are common to both manually saved .FLT files and the smaller Previous Flight.FLT files.
I habe asked the ATSB if they can share the structure of the files that were provided to them as they clearly have information (eg [DateTimeSeason]) in those files that is not detailed in the RMP report.
3) The fact that both VORs were tuned to the same frequency is not all that strange. It just means that at the time and position the simulation was stopped after 10N, both VORs were autotuned to BAC. That can occur.
No, it’s not strange and it can happen. My point, however, is that there is no position on a direct 10N to 45S1 track where both VORs will autotune to BAC. I have flown that track in the simulation program and observed the behaviour of the VORs when set to autotune; they don’t tune in BAC. That is why I asked if you still have the flight files created for your joint paper with Yves; if so I recommend that you review the [Avionics.0] data group for your 45S1 amd see what the NAV1ACTIVE and NAV2ACTIVE values are. I suspect that you will find that they remained tuned to 116.90 and 114.10 respectively.
What that means is that there was most likely an intermediate point well to the east of the direct 10N to 45S1 track. That, in turn, calls into question the 10N – NZPG hypothesis.
@Richard Godfrey
How do we know it was a move then turn, or a turn then move?
Richard, to match the data (specifically the zero values for XVelBodyAxis, YVelBodyAxis, PVelBodyAxis, BVelBodyAxis and HVelBodyAxis) the sequence had to be ‘turn and then move’. For it to be the other way around the simulation aircraft would have to have been ‘flown’ at 10N; that would restore true values for the various VelBodyAxis parameters.
What that means is that at some point after 5N (and, based on the fuel burn, almost certainly somewhere before reaching VAMPI) the simulation aircraft was manoeuvred so as to generate quite high and low G values (+2.2032G and +0.1453G respectively) and then placed in a 20° AOB left turn. Based on the difference between the heading at 5N and the 255.5° value and the AOB, the simulation aircraft had most likely been turning left, away from the direction it had previously been travelling and away from 10N, for over a minute when the simulation was either paused or ended/exited.
That’s the problem I have with stringing 5N and 10N together as part of the same flight. The data suggests that shortly after 5N the simulation aircraft was heading somewhere else!
@Mick Gilbert
Interesting thoughts re decoupling 10N from the one flight.
What if the original intention was to cross southern Ache ?
A left turn off B466 soon after Gunip to 255.5 would be heading direct to Mabix.
A slightly later left turn at 44nm past Gunip to 255.5 would be heading direct to Jilat.
Both tracks suggest a strategy to cross southern Ache where the two northern radars suffer terrain shadowing below about FL280.
How would either flight path affect the VOR autotune system, and what might that tell us ?
@Mick Gilbert: You have a BIG misunderstanding about Volume Shadows. You would need to examine the contents of MULTIPLE Volume Shadows to extract multiple versions of a file, as the contents of the Volume Shadow can ONLY restore the file back to the time the Volume Shadow was created. Each Volume Shadow has an associated timestamp. The RMP report makes it clear that the flight file fragments were from a SINGLE Volume Shadow dated February 3, 2014, which means the fragments were NOT from various versions of the same file, as that would require examining the contents of MULTIPLE Volume Shadows. I don’t know how to make that more clear.
If you have configured your laptop to create a System Restore point every day, then you are creating a Volume Shadow every day. If you believe you can obtain multiple past versions of a single file by examining the contents of a single Volume Shadow, you are very wrong.
Here’s a good reference on Volume Shadows, written by a Microsoft Engineer who helped developed the Windows service.
From that article:
* You create a file (version #1)
* You create a restore point
* You change the file (resulting in version #2) — VSS backs up version #1
* A week later, you change the file again (resulting in version #3) — VSS doesn’t back anything up, because it already has version #1 backed up. As a result, you can no longer go back to version #2. You can only go back to version #1 — the one that existed when the restore point was created.
You have ZERO evidence that 5N, 10N, and 45S1 were not all part of the same simulated flight. We don’t know how the plane was flown between those points. Having VOR1 and VOR2 both autotuned to BAC does NOT mean that 5N, 10N, and 45S1 were not part of the same flight. You insist on building theories around data that you imagine is possible but does not exist.
As I said, if you have new data or insights, please present it. Otherwise, move on.
@Ventus45
The left turn couldn’t have occured prior to 5N (N5° 6′ 41.8671″ E98° 35′ 16.6071″). That’s 67 nm past GUNIP so I don’t think that either of your scenarios work. Moreover, the data suggests that the simulation aircraft was at FL400; that doesn’t gel with setting up for terrain shadowing below about FL280.
@All
A Lion Air flight (737) has disappeared shortly after take-off from Jakarta. 188 souls on board. Apparently it was a just delivered 737 (1 month ago approx), though I have not confirmed this. Happened shortly after take-off. Looks like it was climbing as would be expected and something catastrophic (my assumption) took place shortly after reaching 5,000ft. Doesn’t look good.
@David
I appreciate the laborious treatment you have been giving the drift data/studies. Thanks.
@DrB
There is no need to make recommendations now. There is no funding at present and OI has suspended their ARA San Juan search until February. Interesting, isn’t it, that they are stopping in the summertime.
Yes, you are right about that. I was simply trying to goad you, which wisely deflected. It truly is a difficult situation we find oursleves in.
@Victor Iannello
Thank you for the Volume Shadow Copy/System Restore reference. I had misunderstood your initial response regarding multiple copies of the same file. My apologies.
You have ZERO evidence that 5N, 10N … were not all part of the same simulated flight.</
There is the agreed fact that shortly after 5N the aircraft was turning away from its prior north westerly track and was well established in a left turn turning through west-south-west. In other words, it wasn't heading towards 10N any more rather it was turning away from it. To many observers that would pass as evidence that the simulated flight plan to just past 5N did not continue on to 10N.
You have ZERO evidence that … 10N, and 45S1 were not all part of the same simulated flight.
I haven’t said that 10N and 45S1 were not part of the same simulated flight. What I have said is that the NAV1ACTIVE and NAV2ACTIVE values are evidence that there was most likely an intermediate point that was not on the direct track. I would have thought that might be worth exploring, particularly now as the possibility of the crash site being further north seems to be developing.
You insist on building theories around data that you imagine is possible but does not exist.
I’m pretty sure that I don’t. I’m not imagining the data showing a 20° AOB left turn shortly after 5N taking the aircraft away from 10N nor am I imagining the NAV1ACTIVE and NAV2ACTIVE values that are inconsistent with a direct track from 10N to 45S1.
@Mick Gilbert: So I think you now agree that the simulation data sets could NOT have been the same file that was successively over-written, as you previously insisted.
As I have said previously regarding simulation data we have, WE DON’T KNOW HOW THE PLANE WAS FLOWN BETWEEN THESE POINTS. A turn after 5N (even away from 10N) and two VORs autotuned to BAC proves (even if the required path does not align with 10N and 45S1) does NOT prove that 5N, 10N, and 45S1 were not all part of the same simulation.
If the ONLY data sets found in the MK25 Volume Shadow dated February 3, 2014, were 45S1 and 45S2, which showed a plane in the middle of the SIO with no fuel, that alone would be very incriminating. The fact that those two points were alone grouped with other points showing a flight up the Malacca Strait, all created within an hour of simulation time, makes it even more so. If those data points were NOT related to MH370’s disappearance, that would be an EXTRAORDINARY coincidence.
If it seems as though you have worn out my patience, it is because you have.
@Richard
I have a different interpretation of 5N. I presume it was TASEK/-25 to VAMPI or something close to that. I do think that MH150 tended to go on Flight Path P574 thru DUBTA so I am not sure N571 is correct for MH150, but in any case, IDKUT is unexpected as a waypoint for me.
@Victor
I thought Victor was maybe suggesting a sharp turn towards WITT, but I do think if the sim case was for MH150, the aircraft might have jogged a little left to get onto P574.
>>Of course my overall speculation is the simulator case might suggest where MH370 actually flew, crossing Arc7 at about 30 South. The reason for the simulator case might have been to plan the actual flight path, perhaps looking for ways to line up the aircraft without entering waypoints or other possibly incriminating A/P instructions.
@TBill: It’s hard to know what occurred in the simulation, as we are attempting to recreate a simulated flight based on snapshots. There could have been a flight plan entered in the FMC and the plane could have mostly been flying with autopilot engaged in LNAV/VNAV mode. But then after 5N, the A/P could have been disengaged and the user could have accelerated up, accelerated down, and turned right and left. Then, the plane could have returned to the LNAV path, either by flying or manually manipulating the position. Likely, at the time the simulation was stopped after 5N, the plane was in autopilot because the bank was very close to 20°. We also don’t know what the heading was at the time the simulation was stopped because the heading (and altitude) could have been manually changed before the 10N data set was saved.
Info on Lion Air
https://www.airlineratings.com/news/lion-air-crash-plane-problems-previous-flight/
@airlandseaman: Thanks, Mike. With the FDR and CVR likely to be recovered soon, I doubt this one will remain a mystery. My guess is that when the instruments starting giving flaky outputs, there were improper pilot inputs to set and maintain bank, pitch and power.
Victor: I agree. There may be more to the story, but but AF447 crashed due to pilot mistakes following loss of the airspeed indicators. Most here know the story of AF447, but here is a very interesting article about AF447 that is worth a read…and possibly related to Lion Air. http://bit.ly/2zglXT0
@Victor Iannello:
You stated: “the paper that I co-authored with Yves Guillaume in November 2016 concluded that the sim data was from a single flight“
Could you please clarify:
(1) Where is this conclusion in your paper? (Read it, but don’t see it.)
(2) The closest statement to that effect, which I could find, is on p.7: “The data points in the Andaman Sea share some of the same values as the data points in the SIO, suggesting the flight files came from the same simulated flight.” But I don’t even read this sentence as meaning ALL data points are linked. For this interpretation, one would have to geographically include the Malacca Strait as part of the Andaman Sea. However, much to the contrary, you appear to set them apart, distinguishing them as follows:
• page 2: “The position coordinates in Table 1 suggest a flight that departs from KLIA, proceeds up the Malacca Strait towards the Andaman Sea […]”
• page 7: “a user created a simulation in which a B777 is successively positioned in the Malacca Strait, the Andaman Sea, and the SIO.”
• page 8: “it is likely that a user created a simulation on Microsoft Flight Simulator to create a flight that passed over the Malacca Strait to the Andaman Sea and to the SIO”
In each of these cases you distinguish between the Malacca Strait and the Andaman Sea. Therefore I suppose you refer to the data points as follows:
• Malacca Strait: 2N+3N+5N
• Andaman Sea: 10N
• SIO: 45S1 + 45S2
Correct ?
(3) I read your paper as connecting the data points in the Malacca Strait (3N+5N, red rectangle on page 10) with each other and data points 10N+45S1+45S2 (blue rectangle) with each other, but nowhere in your paper could I see you connecting these 2 groups with each other, which would be necessary for concluding that “the sim data was from a single flight” (see question 1).
@Peter Norton: Here are some reasons to believe the simulator data sets 2N, 3N, 5N, 10N, 45S1, and 45S2 represent the same flight simulation. These reasons were presented in the paper I co-authored with Yves Guillaume and in the blog article:
1) The data sets were grouped together in the same Volume Shadow (along with two other points parked at KLIA).
2) The fuel levels decrease monotonically in the order 2N, 3N, 5N, and 10N, with no fuel at 45S1 and 45S2.
3) The altitude increases monotonically in the order 2N, 3N, 5N, and 10N, followed by decreases after fuel exhaustion for 45S1 and 45S2.
4) The minimum and maximum values for wing loads, and the maximum value for engine RPM, are identical for 2N, 3N, and 5N, and are again identical for 10N, 45S1, and 45S2. (The values changed between 5N and 10N.)
5) Between 5N and 10N, the minimum value of wing loading decreased, and the maximum values of wing loading and RPM increased. (If the minimum value had increased, or the maximum values decreased, it would be evidence of different flights.)
6) The timestamps of the data files were in the chronological order 2N, 3N, 5N, 10N, 45S1, and 45S2.
7) The total duration of the flight simulation time between 2N and 45S2 was about one hour.
8) The latitudes increased monotonically towards the north in the order 2N, 3N, 5N, and 10N.
9) The 10N data set shows a banked turn towards the south, and the next points are at 45S in the SIO.
@Victor Iannello:
Thank you for the additional information.
You stated: “the paper that I co-authored with Yves Guillaume in November 2016 concluded that the sim data was from a single flight”.
I read this paper, but cannot find this conclusion anywhere therein. I may have missed it. Where is this conclusion ?
This conclusion doesn’t figure among the 4 conclusions you list on page 7 in the section “conclusions”.
I read your paper as concluding that 2 groups of data points (red box + blue box) are very likely linked. But I don’t see you concluding in your paper (as quoted), that the red flight segment and blue flight segment belong to the same simulation.
Another question:
To your knowledge, have there been any other data points recovered than these 6 ?
@Peter Norton: If you compare the list above with the analysis in the paper, the paper discusses every item in the list except (6) and (7), which were unknown at the time the paper was written. The paper discusses more than item (4). The final words of the paper were: The totality of the evidence suggests that it is likely that a user created a [i.e., ONE] simulation on Microsoft Flight Simulator to create a [i.e., ONE] flight that passed over the Malacca Strait to the Andaman Sea and to the SIO in a way that is similar to the flight path that investigators believe was followed by MH370. That is consistent with the statement the paper that I co-authored with Yves Guillaume in November 2016 concluded that the sim data was from a single [i.e., ONE] flight.
The only simulator data that was recovered from the Volume Shadow on drive MK25 were the data sets 2N, 3N, 5N, 10N, 45S1, 45S2, and two other points with the plane on the ground at KLIA.
@Victor Iannello:
I have seen your final sentence in the paper and I agree that (especially with your added emphasis) it is in line with how you referenced it. I just wasn’t sure how you reached that ultimate conclusion from your 4 conclusions listed on page 7 in the “conclusions” section. That’s the reason, why I insisted.
Sorry for having belaboured the point, but I think it is an important distinction to make, whether or not all 6 points are linked, because the red flight segment alone is a regular MAS route from KUL to Europe and therefore insignificant and there might also be an explanation for the blue flight segment. So only when all 6 points are from the same flight, the simulation becomes particularly damning.
My conclusion (corrections welcome):
• None of your 9 items exclude the possibility of all the 6 data points belonging to the same flight, but none provide any proof either.
• Collectively, the 9 items suggest a certain likelihood of the 6 points belonging to the same flight, but I have no idea how to determine the likelihood.
– If @Dennis’s prediction that we won’t learn anything from the crash proves true,
– and absent other important pieces of evidence emerging,
your interpretation of the flight sim data will become extremely important, as it will be the main piece of evidence incriminating the pilot. At that point it will hinge on the interpretation of the flight sim data, whether or not MH370 will be classed as a crime committed by the Captain.
@Peter Norton: The purpose of this blog is to help solve the mystery of the disappearance of MH370 and to help find the plane, not to build a case to accuse the captain of a crime. There is not enough evidence to reach a guilty verdict. However, even beyond the sim data, the circumstances and evidence of this case point towards the involvement of the captain.
As I said before, even if only data sets 45S1 and 45S2 existed in the Volume Shadow, that would be very incriminating. Even more so with a strong link to 10N. And even more so with the links to data sets 2N, 3N, and 5N. I believe that 9-item list when taken together makes the relationship between all the data sets to be very likely.
An important question to answer is why the simulated flight was created, and why the flight files were saved in the manner they were saved. I don’t think the purpose was to plan and practice the diversion, as there are better tools for doing this.
@Victor Iannello:
In reaction to your comment, and to make sure my remarks are not construed differently: I never implied your work on the sim data were somehow directed against the Captain or anything other than solving the MH370 case.
My point was to
(a) underline its crucial importance if no other evidence emerges
(b) say that others (not you) will use it to class MH370 as pijacking if the crash site is not found or provides no evidence (see @Dennis’s postings).
I agree with the points in your posting.
David says: “A relief. Not the $A drop reducing pages here as I feared.”
Sorry, but what is a “$A drop” ?
(hoping it is G-rated)
@Victor
It is frustrating when you suggest there is no planning value of the Flight Sim. That leaves joyride for the experience of pre-living a suicide mission as the only apparent purpose.
To me the Flight Sim is an important planning tool for a mission like this, assuming there was an intended flight strategy. I have always (borrowed) the hypothesis of a deep ocean hiding place, such as Broken Ridge. So I conduct flights projected onto Google Earth to see the terrain, where I also have waypoints. Does any other tool, even Level D simulator, put the aircraft on Google Earth? I even tried flying into Petronas Towers on Google Earth, but it’s hard because you cannot see forward, only straight down. Of course, we don’t know if ZS had the flight sim connected to show the path on Google Earth…that would be my question.
I would say flight sim work is possibly consistent with the Broken Ridge or other deep zone crash site target hypothesis. I also think it could be consistent with trying to find a way to fly in a “sanitized” manner to say Broken Ridge without entering a waypoint eg; NZPG, which would look incriminating on the flight computer memory chips, assuming they might possibly be recovered.
@Victor
I find myself agreeing with TBill. Not about flying to a deep zone crash site which I think is ridiculous, but about using the simulator in general. Extending your logic leads to the question of why a B777 captain has a B777 simulator at all. ZS obviously enjoyed his toy, and shared his enthusiasm with others. I don’t find it strange that he engaged in some fantasies with it.
@Niu Yunu. The Australian dollar cheapens itself day by day.
(Yes, G. Vulgar but few are offended)
@DennisW: I’m sure the captain loved his simulator. They are also great for practicing certain aspects of aviation, such as difficult departure and arrival procedures for an airfield, or procedures to follow after a failure. I understand completely why he had one.
However, in this case, the user moved the plane along a path, changing position, fuel level, and altitude. What planning or practice do you think he obtained when he conducted the simulation?
I think the issue with the simulator is that there is no conclusive evidence as to what happened to MH370 and why it happened. However the fact that Captain Zaharie plotted this route on his simulator surely this must form part of the evidence in a criminal investigation. Whoever, if anyone is carrying this out. There are other pieces of evidence such as the debris but also evidence known by others which is being kept from the investigation. Phillip Wood was I think the only American citizen on the flight. Let’s hope that M Wattrelos’ efforts in taking the French Air Gendarmerie to America will result in the FBI and the American Government reopening in a very public way the investigation into the plane’s disappearance. One American citizen unaccounted for in this international tragedy is one too many for the Americans,I would have hoped.
@TBill said: It is frustrating when you suggest there is no planning value of the Flight Sim. That leaves joyride for the experience of pre-living a suicide mission as the only apparent purpose.
I have long suspected there was another purpose.
thank you, David
> TBill says:
> on Google Earth it’s hard because you cannot see forward,
> only straight down
to change your view from straight down to forward:
press & hold down middle mouse button + drag mouse towards you
alternatively: click on the top arrow above the eye icon on the navigation aide
@DrB, RichardG
David Griffin has informed us that he has redrawn his figures 5a and 5b, see:
http://www.marine.csiro.au/~griffin/MH370/
The spatial distribution of beached items (predicted and reported) is better visible. It confirms that on mainland Africa the majority of items was expected in the north, see fig. 5b. While Richard tried to provide explanations for the small number of (reported) finds on African coasts, the predicted vs. reported distribution still worries me.
One would for example expect some finds reported from Kenya (Mombasa area), where, similar to the Maputo area (around S25), beaches should be a bit more crowded and quite some tourists are expected to visit.
@VictorI
Thank you for contacting Blaine Gibson, on the issue of the search intensity in the north of Madagascar. I hope he is able to respond in near future.
See fig. 5a, S12 – S15…
@Niels
I noted the private email from David Griffin today.
My statistical analysis of the arrivals vs beachings vs reporting was by country and not by town.
I cannot find any reference to Blaine Gibson’s search intensity in Madagascar in Victor’s current post. Can you point me to the comment I please.
@Niels
From the Stanford encyclopedia of philosophy:
Terms like ‘observation’ and ‘observation reports’ don’t occur nearly as much in scientific as in philosophical writings. In their place, working scientists tend to talk about data.
A very important distinction above. Observations, particularly the random discovery of debris, are not trustworthy. Uncertainty relative to the timing and location of debris in Africa makes these “observations” close to worthless as a qualifier of terminal latitude. You are free to speculate, of course, as to whether or not debris discoveries have any meaning relative to crash latitude. From my perspective they are only useful for concluding the aircraft terminated in the SIO.
What cannot be ignored, however, is the Bayesian interpetation of no debris findings in the underwater search from 38S to 25S. DrB and TBill are prepared to discard the absence of debris finds in the 38S to 25S search due to insufficient search width. That is a lame position based on the last BFO values and a lack of IFE logon. I am having none of it. I prefer science to philosophy.
@RichardG
Please see at
http://mh370.radiantphysics.com/2018/10/20/french-investigation-into-mh370-extends-to-us/#comment-19524
Regarding Mombasa area: I had the wider area in mind (something like 75 – 100 km at both sides of the city) when I wrote the comment; to compare to roughly the stretch from Maputo to Inhambane in Mozambique.
@Victor
However, in this case, the user moved the plane along a path, changing position, fuel level, and altitude. What planning or practice do you think he obtained when he conducted the simulation?
I respond to your question above with a theme similar to my comment to Neils regarding debris findings. That is “who cares”? The value of the sim data lies in the specific coordinates found in that data file and the time of creation. Why that data was created falls in the domain of speculation. The existence (and relevance) of the data is indisputable.
@Niels
Re: DavidG Assumptions
“(1)MH370 did not crash north of 23°S
(2)many items beached in Tanzania, Kenya and Somalia but were either not noticed or not reported, or
(3)MH370 crashed north of 23°S but the debris was not taken north (under the influence of the SE trade winds) as much as happens in our model.
I still feel another assumption is (4) crash happened on Arc7 (vs. some glide distance away)
Re: Are certain Debris from MH370?
It seems like via trace contaminant analysis of the aluminum and other materials in the honeycomb, standard techniques such as neutron activation, authorities ought to be able to determine if some of the unknown debris is likely from MH370 or not (assuming of course Malaysia had any interest, which of course they do not).
@DennisW
I agree that we should be very careful to conclude something from single or few debris finds. However, the global distribution of expected and reported finds I find a bit surprising. Some people here seem to accept the lack of finds in the north just as a fact (the debris was probably there but was not reported). To me it means “trouble” anyway regarding the use of drift analysis: just think of a no-find/no-report as an infinite delay…
Regarding the underwater search: here we are not talking about “observations” but data. For me the quality of this “data”, especially for the OI part of the search has not really been substantiated.
@TBill
If I understood the discussions about aircraft system details correctly the long glide after dive cannot be fully excluded, but based on the final BFOs and lack of IFE logon to me this doesn’t look likely.
@Niels
In addition to the BFO and IFE observations there is the question of motive. Why would any seasoned pilot elect to initiate a steep descent followed by a pull-out and glide?
Glide enthusiasts suggest an attempt by ZS to create confusion relative to the interpretation of the Inmarsat data as an attempt to hide the aircraft debris. In my opinion ZS was clueless relative to the Inmarsat data. There is nothing to suggest otherwise.
Also, the condition of the debris finds is not consistent with a “soft” water entry.
I agree that the quality of the OI search data is not well characterized, and the technology does not have the pedigree of a towed scanning instrument. However, I am also of the opinion that the reflectance of aircraft debris on the sea floor would make the debris very obvious in all cases except a failure to “illuminate” the wreckage.
@Niels
You stated “Some people here seem to accept the lack of finds in the north just as a fact (the debris was probably there but was not reported).” I take that as a snide remark against my attempt to argue certain realities in Africa using UN statistics.
I argued that there was was a 100% track record in South Africa of locals reporting possible MH370 debris finds. Do you disagree?
I argued that there was was a 0% track record in Mozambique of locals reporting possible MH370 debris finds. Do you disagree?
I argued that there was was a 0% track record in Madagascar of locals reporting possible MH370 debris finds, until Blaine arrived and raised awareness. Do you disagree?
I argued there was one find in Tanzania, zero in Kenya and zero in Somalia although possible MH370 debris finds should reach all locations according to both David Griffin and my drift simulations. Do you disagree?
And now you question the quality of the data from Ocean Infinity …
Ocean Infinity deployed the most modern technology there is. Surely you should also question the quality of the Fugro data as well.
In fact, according to you we should all give up and go home!
@DennisW
“Why would any seasoned pilot elect to initiate a steep descent followed by a pull-out and glide?”
Lots of reasons. You seem to be conflating long glide with Larry Vance’s soft ditch, which many reject his approach and hypothesis. Vance could well be wrong about a slow ditch: it might have been a fast ditch. Consider the recent Lion Air crash…implies a higher speed glide without flaps could be devasting to the aircraft upon hitting the water. Alternatively could also be a good way to minimize crash debris, by breaking up the aircraft into several main parts which would sink fast. See the other prior Lion Air water crash where the aircraft broke into two parts.
Admittedly there is lot’s of ambiguity in the MH370 recovered debris and Inmarsat data, but to say a long glide is unlikely seems like an opinion not well founded.
@RichardG
No, it was not meant as a snide remark to you regarding your attempt to explain it. I do understand your points that awareness / access to international news, but also for example level of education, economic situation and infrastructure, could play a role in chances of getting debris recognized and reported. However, if I try to apply this to Kenya (and compare to Mozambique) I don’t understand the situation and I don’t want to step over it easily.
For the rest, I hope I’m free to question the quality of all data. That applies to the quality of debris data (completeness and unknown delays), scan data as well as satellite data. I try to do my part in reducing uncertainties in the data (for example I try to get access to data of previous flights to better understand the BFO error statistics), so no giving up there.
Regarding OI: have you seen any detailed assessment on their data. What about the coverage in rough terrain / shadowing effects?
@DennisW
I hope we agree that, following your own criteria and with the very little we know for certain, the discussion on responsibility and motive is speculative and non-scientific.
@Niels: According to Richard G’s table, no debris was reported by locals in Mozambique, Kenya, and Somalia. Meanwhile, after Blaine Gibson’s discovery of No Step, there were 6 later discoveries in Mozambique, all by non-locals. We also know that Blaine did not search for debris in Kenya and Somalia.
On the face of this evidence, I think it is difficult to conclude that no MH370 debris was beached in Kenya or Somalia.
@TBill
You say “lots of reasons” but fail to name even one. Funny shit.
@Niels
Yes on motive. Responsibility? I have no doubt it was ZS.
@All,
The most striking result in David Griffin’s predictions is the bifurcation of mainland Africa landfalls. His predictions show about 4 times as many items beaching to the north of Madagascar as to the south. The reported finds show about 10 times as many items south compared to north. That is a reported S/N ratio about 40 times larger than his prediction!
I certainly agree that there is a geographical bias in the probability that an item will be reported, and this may explain a significant part of the 40X discrepancy. However, I don’t think it can explain all of it, and that makes me question the validity of his predictions.
I believe Richard Godfrey’s predictions may show a smaller S/N discrepancy, although we don’t yet have his predictions in a similar format as Dr. Griffin’s, so it is not yet possible to make a direct comparison. In general, Richard’s empirical model should do the job quite well in the areas that are well-sampled with global drifter data. It will be interesting to see the S/N ratio his model predicts.
@DennisW,
You said: “DrB and TBill are prepared to discard the absence of debris finds in the 38S to 25S search due to insufficient search width. That is a lame position based on the last BFO values and a lack of IFE logon. I am having none of it. I prefer science to philosophy.”
As far as I know, there are no objective, scientific data and method to discriminate between a glide after 00:19:37 and no glide. All data are consistent with both possibilities. Any determination of their relative probability must therefore be based on subjective reasoning and on assumptions. You have done that and have come to a conclusion (no glide). I have also done that and concluded that both are possible. Your conclusion is not “science” any more than a different conclusion would be. It is a subjective guess. It may be right or it may be wrong, but it’s not objective science.
@DrB
I previously published a comparison to David Griffin’s Fig.3 at the following link:
https://www.dropbox.com/s/47c35xgex2jgd3c/CSIRO%20vs%20RG%20Comparison%20Latitude%20vs%20Time%201%C2%B0%20Step.pdf?dl=0
Do you mean a comparison to David Griffin’s Fig. 5a or Fig. 5b perhaps?
@Niels: Here’s what Blaine Gibson said about his search in the northern part of Madagascar:
I went to the northern tip of Madagascar traveling along the northwest side and found nothing. I went as far north as Antongil Bay on the northeast side and got debris. That’s where the vortex generator was from. The northeast coast is very rocky and difficult to access and I did not go there. The farthest north I went on the northeast side was the northernmost point of Antongil Bay.
@DrB
Lack of reported debris is an observation not data. Recovered and confirmed debris is data (lacking a time stamp). S/N is an inappropriate metric. Observations apply to philosophy and soft “sciences” like climatology and have no place here.
@DrB, @DennisW: It comes down to whether a dive, glide, dive sequence is as probable as a single dive. The BFO values and the debris alone can’t distinguish between those two possibilities.
@Victor
No one with a brain is considering a dive followed by a glide scenario.
Jeff Wise said:
Some other reasons why ghost flight is not tenable:
1) To produce the final BFO values, the plane must have been actively pushed into a dive.
2) In order to fly far enough beyond the seventh arc that its wreckage wasn’t found on the seabed, the plane would have to have subsequent to the dive been actively held in a glide.
Is he right ?
@DennisW: I’m not a big fan of this (dive-glide-dive) scenario, but how do you disprove it?
We had a former B777 pilot (Byron Bailey) that said a pilot would enter a steep descent after fuel exhaustion to produce power from the RAT. Now, the facts are that the airspeed that is required for the RAT to produce power is easily reached with a gentle descent. But, independent of the technical specifications of the RAT, a pilot BELIEVED a high descent was necessary. Perhaps the captain had a similar false impression.
@Perfect Storm:
1) Possibly. In 5 of the 10 Boeing simulations for fuel exhaustion with no pilot input, the rate of downward acceleration matched the value inferred from the final BFO values. To reach this increasing rate of descent, what is required is a large bank, which will eventually develop if the autopilot disengages and there is no pilot input. In the Boeing simulations with a particular electrical configuration, it took >5 minutes for this level of downward acceleration to occur after the autopilot disengaged. On the other hand, based on the assumed startup sequence of the APU and the SATCOM, we believe the final BFO values occurred 2 minutes after the autopilot disengaged. So although the progressive descent rates and time spacing were matched in the simulations, the timing after the autopilot disengaged was not. We can’t be certain that the Boeing simulations represented the entire range of possible configurations, and we can’t be certain that the Boeing simulations accurately represented how all the systems interact after fuel exhaustion. It’s not as though a simulation of this type is easy to validate with experimental data.
2) Possibly. That assumes the plane impacted south of the 25S, and was missed by the search because the search was not wide enough. That’s a possibility. The other possibility is the impact occurred further north along a part of the 7th arc not yet searched.
@DrB
You requested a folder with the 9 tracks from the cluster around 31°S:
https://www.dropbox.com/sh/lchzsey1uxz6sm6/AAB2Pkl8SSKUXOMu-1gFLb2Ca?dl=0
@Victor Iannello:
Thanks. Yes, that corresponds to what I gathered is the consensus around here: that an inactive pilot scenario cannot be excluded yet. To me this means that Jeff’s statement is not accurate, as it claims the inactive pilot scenario impossible.
@DrB
The link below is for my drift results for beachings on mainland Africa west of 42°E from various start latitudes near the 7th Arc in a similar format to David Griffin’s Fig. 5b:
https://www.dropbox.com/s/4a3rtao4l0lljd2/Drift%20Analysis%20Results.pdf?dl=0
@Richard Godfrey,
From a geographical distribution perspective, the simplest meaningful representation is three fractions: the fraction of landfall particles beaching in Madagascar/Reunion/Mauritius/Rodrigues, the fraction beaching in northern mainland Africa, and the fraction beaching in southern mainland Africa, all as a function of crash latitude. That information is contained in David Griffin’s Figures 5a, 5b, and the difference in 5a and 5b. I have requested his numerical values so these quantities can be obtained from his two figures.
@DennisW,
You said: “No one with a brain is considering a dive followed by a glide scenario.”
Ad hominem attacks won’t convince anyone, especially in this forum.
@Richard Godfrey,
Many thanks for posting the 31S tracks. The numbers are, out of 9 tracks, 3 in Madagascar, 1 in northern Africa, 1 in southern Africa, and 4 did not make landfall. Obviously, with only 9 tracks and 5 total landfalls, it’s difficult to draw any conclusion except that the predictions appear qualitatively consistent with the debris reports, taking into account the small number of modeled particles.
@DrB
It was not meant to be an attack merely a metaphor for questionable thinking. We have interacted for a very long time now, and my respect for you is a matter of public record.
What bothers me, enormously actually, is that a wider search is not justified based on data we actually have. That is why I tried to emphasize the difference between data and observations earlier.
A good parallel goes back to my days involved with The Great Randi and CSICOP (Committee for the Scientific Investigation of Claims of the Paranormal). I volunteered in the San Francisco chapter. The “holy grail” of paranormal investigators is the so-called PPO (Permanent Paranormal Object) – that is something (anything) permanent that we can touch and feel and prod that defies a human/earthly explanation. No one has ever found a PPO despite an untold number of seemingly convincing observational experiences.
So it goes with the value of observational references. They cannot and should not be trusted or acted upon.
@DrB
Ad hominem attacks won’t convince anyone, especially in this forum.
I forgot to mention that I am not the least bit interested in convincing anyone of anything. Living in California has completely neutralized my interest in what other people think.
@TBill: Is this what you were looking for in Google Earth ?
@DrB
Sorry, I was using an iPhone and missed your post containing the following statement.
As far as I know, there are no objective, scientific data and method to discriminate between a glide after 00:19:37 and no glide.
The lack of an IFE log-on is objective and scientific. You can “create” many reasons why this log-on did not occur.
1> The APU ran out of fuel.
2> The AES had a spontaneous and random failure.
3> The aircraft orientation was unfavorable.
4> The aircraft crashed into the ocean.
The rate of descent is consistent with 4>. The other reasons fall in the domain of a fabrication to perpetuate an alternate theory.
@DennisW,
The lack of an IFE log-on has multiple plausible explanations that are not “fabrications”. We don’t know their probabilities. You can guess if you want, but don’t expect other people to agree with your guess.
@DrB
That is a dumb and defensive response. You can do better than that. You and TBill should prepare a list. TBill says he has multiple reasons for the glide (although he has not posted any).
@Dennis
Sensical reasons for a glide are neither requisite nor necessarily likely. This was a very disturbed and delusional man who snuffed the life out of 238 human beings, some of whom he most likely had long working relationships with and even friendships.
And you want “rational’ reasons for a possible glide?
I could go on but it would perhaps be wise of you to more deeply ponder the state of mind at play here before dismissing scenarios so cavalierly.
@all -I think it is also important to find possible reasons for Victor’s comment earlier this morning. I don’t know about others but that got me really thinking –
Victor said “However, in this case, the user moved the plane along a path, changing position, fuel level, and altitude. What planning or practice do you think he obtained when he conducted the simulation?”
Thinking about this all day long…
@Peter Norton
Thank you I will give it a try when I get a chance…not sure if that will work when I have an aircraft flying in flight sim. I guess I can toggle back and forth.
@Donald
Being the gatekeeper of the whacko category I inform you that you are IN.
@TBill
I am still waiting for your multiple reasons for the recoevery and glide. Pardon me if I think you are full of shit.
Article quoting Wattrelos and Iannello:
https://www.news.com.au/travel/travel-updates/incidents/french-investigators-uncover-mystery-third-entity-float-new-passenger-theory-in-mh370-probe/news-story/985848ab79b19ec45654f7d9048702f6
@VictorI
I’m grateful to you and Blaine Gibson for this info on the search in northern Madagascar. The high concentration of beached items predicted S12-S16 applies mainly to the E coast, which seems to be unsearched by Blaine in this latitude range. That helps a bit to understand the lack of finds in this latitude range.
Regarding Tanzania and Kenya I’m searching for some tourism statistics, as compared to Mozambique, see for example (select country):
https://www.e-unwto.org/toc/unwtotfb/current
So far it seems tourism levels of the three countries are quite comparable (in the range of 1 – 2 M visits per year).
@DrB
The link below is for my drift results for beachings in the Southern Indian Ocean west of 65°E from various start latitudes near the 7th Arc in a similar format to David Griffin’s Fig. 5a:
https://www.dropbox.com/s/jjstegyp4xh7x7e/Drift%20Analysis%20Results%20West%20of%2065E.pdf?dl=0
The link below is for my drift results for beachings on mainland Africa west of 42°E from various start latitudes near the 7th Arc in a similar format to David Griffin’s Fig. 5b:
https://www.dropbox.com/s/4a3rtao4l0lljd2/Drift%20Analysis%20Results%20West%20of%2042E.pdf?dl=0
@TBill wrote “Does any other tool, even Level D simulator, put the aircraft on Google Earth?”
Modern CGI visuals on Level D simulators do now use scene databases compiled using DEM (digital elevation models) and overlay photography onto a polygon mesh constructed from the DEM. 3D detail in the vicinity of airports will be coded into the scene database. It’s not Google Earth but the principles are similar. I doubt that oceanic bathymetry DEM is exploited, how possibly could that be a benefit to training.
@Donald: Finding reasons for a glide is easier than for the sequence dive-glide-dive.
Comparing the Lion Air crash.
It seems this was a high-speed impact, probably similar to Mh in it’s crash dynamics. I’m stuck by how little floating debris there is. Of note, the largest floating pieces consist of the sliderafts, logo from fin only about 1m2, and the pressure vessels(fire bottles, slide-raft inflators, and Crew O2 bottle). I’m surprised Blaine and co have found as much as they have, perhaps this is partly due to a midair breakup, possibly the failure of the left wing.
@Tim: The last ADS-B data from FR24 indicates the plane was descending at an airspeed of 472 kn (GS=360 kn, VS=-31,000 fpm) at a descent angle of 40°. That represents a very high energy impact.
@DennisW
Some reasons for descent could be:
(1) Recover from stall, whereas in preparation for a glide, the pilot ascended into fuel exhaution.
(2) Operation reasons
(a) to get RAT spinning
(b) to get air for breathing if bleed air lost at altitude
(c) to slosh fuel into fuels lines
(3) Lining out aircraft trajectory for lower power mode due to engine loss
(a) If on one engine, descending fom FL350 to FL200 for single engine ops
(b) if on no engines, lining out for extended glide with no power
(4) Pilot elective choice of maneuver
(a) Purposeful Maneuver, such as getting below clouds to see ocean surface
(b) “Recreational” Maneuver for the heck of it
(5) Preparing for whatever landing configuration was intended, which looks like it could have been a hard glide “ditch” to break up the aircraft in larger pieces.
In general I am envisioning a careful plan followed through to minimize chance of finding aircraft and keeping incriminating data from being discovered, physically as well as electronically on memory chips on computers. So I am thinking pilot may be leaving cockpit during the glide down.
@Niels
Tourists on safari are unlikely to help find an item of debris from MH370.
TBill:
Re:
Some reasons for descent could be:
(1) Recover from stall, whereas in preparation for a glide, the pilot ascended into fuel exhaution. [NO]
(2) Operation reasons
(a) to get RAT spinning [NO]
(b) to get air for breathing if bleed air lost at altitude [NO]
(c) to slosh fuel into fuels lines [NO]
(3) Lining out aircraft trajectory for lower power mode due to engine loss [NO]
(a) If on one engine, descending fom FL350 to FL200 for single engine ops [NO]
(b) if on no engines, lining out for extended glide with no power [NO]
(4) Pilot elective choice of maneuver [NO]
(a) Purposeful Maneuver, such as getting below clouds to see ocean surface [NO]
(b) “Recreational” Maneuver for the heck of it [Possible]
(5) Preparing for whatever landing configuration was intended, which looks like it could have been a hard glide “ditch” to break up the aircraft in larger pieces. [NO]
The most likely reason for the descent, by a large margin IMO, is second engine FE followed by the the type of uncontrolled spiral descent observed in all the EOF simulations. Before searching anywhere new, there should be an audit of all the seabed search data collected to date…Fugro and OI.
@ALSM
I agree passive, uncontrolled descent is one possibility.
I was giving the alternate scenario, list of possible reasons for an active pilot descent.
@DrB
I have long since moved on to 25 cluster points around any location of investigation.
You asked for the tracks from my previous 9 cluster points around 31°S, in addition to the 25 cluster points I had already provided around 31.6°S.
Please do not then complain that there are only 9 points! Actually there are now 25 + 9 = …
It is interesting to compare the FR24 ADS-B data for the final descent of Lion Air LN1610 with the hypothesised descent rate for MH370 as derived from the final few BFO data points.
There has been a deal of debate about the MH370 descent rate. Many comments have challenged the view that a descent rate of say 15,000 ft/min, and an acceleration of perhaps 0.7G to reach that rate is implausible. My discussion paper of April 2015 showed that such a descent rate implied a flight path about 20 degrees below the horizon.
The descent rate over the final few seconds of flight of the Lion Air 737 increased to something like 31,000 ft/min. The granular ADS-B data illustrates the [vertical] acceleration of Lion Air LN1610 over the last few seconds before impact. In just 11 seconds the vertical speed changed from +7,000 to -31,000 ft/min, with almost no change in ground speed, and at an angle below the horizon of about 40 degrees.
That data makes the hypothesised MH370 descent rate seem entirely reasonable.
Now consider the debris field that resulted from an impact following a descent of that nature.
@ALSM
Ocean Infinity owns the data they collected as far as I know. It is not clear that they would entertain a third party audit, and I doubt they would audit it themselves.
@ALSM
Also, I find myself wondering if imaging the sea floor is a reasonable model for “detecting” a debris field. Aircraft wreckage should contain quite a large number of specular reflectors as opposed to normal seafloor terrain which is a diffuse reflector. It is hard for me to imagine missing a debris field unless it was masked (not illuminated) in which case an audit would not yield anything new.
@Richard
Tourists on safari are unlikely to help find an item of debris from MH370.
I agree. It is hard to estimate the probability of a discovery by casual obsevation versus dedicated search. We are charmed by stories of breakthroughs made by tripping across something, but the reality is most of the human breaktroughs are the result of detailed and meticulous research.
I think the Blaine effect is highly underestimated.
@Richard Godfrey,
I don’t recall ever “complaining” about your work. I may disagree on some conclusions, but I think your work is quite useful, as I have said a number of times.
@DennisW,
In addition to specular reflections, the larger proud debris can cast acoustic “shadows” which can aid in detection.
@Richard Godfrey. Also Dr B, Neils. Earlier I posted about crash latitude candidates needing to get through hoops implicit in David Griffin’s ‘north’, ‘east’ and ‘south’ figures.
Below is an attempt to explain this more fully:
https://www.dropbox.com/s/dectotmdb6o4zbr/%281%29c.png?dl=0
and, complementary:
https://www.dropbox.com/s/r6vsuup6qclccnr/%282%29a.png?dl=0
@DennisW
“Ocean Infinity owns the data they collected as far as I know. It is not clear that they would entertain a third party audit, and I doubt they would audit it themselves.”
I was satisfied that OI spent enough time at Broken Ridge, but we seem to lack any follow-up dicussion or documentation of how well they were able to cover the difficult areas.
@DennisW, concerning your comment, “Ocean Infinity owns the data they collected as far as I know.”
According to a joint Ocean Infinity & NF-GEBCO press release 120,000km² of data from Seabed Constructor’s search for MH370 was handed over to The Nippon Foundation-GEBCO Seabed 2030 Project.
I understand that data is maintained by the International Hydrographic Organisation’s (IHO) Data Centre for Digital Bathymetry (DCDB) in Boulder, Colorado, USA. The OI data set is not yet listed on the IHO Data Centre for Digital Bathymetry Viewer. The Geoscience Aus MBES data is available there.
Geosience Australia has published, to the web, the MBES bathymetry DEM data and the detailed seafloor survey data acquired by Fugro’s side scan sonar and the Phoenix International/SL Hydrospheric synthetic aperture sonar.
I used the GA MBES data to generate 3D visualisations of interesting features surveyed by OI’s AUVs and I’ve done some initial scoping for what might be required to visualise the side scan/synthetic aperture data but it’s volumous. I haven’t yet assessed if it’s feasible to render it usefully with open source tools.
@airlandseaman.
Who would be qualified to carry out the audit of the seabed data collected by OI and Fugro and why is this necessary when their remit as I understand it was to find MH370 and not research the seabed for as yet undiscovered data on flora and fauna, and the seabed in general?
(I am sure what Fugro and OI did find has more than covered their costs with the amount of non MH370 data they have collected)
I can see your point though as I think you are saying that they may have missed something which an expert trained underwater plane wreckage investigative team might spot. That is if we are still certain MH370’s resting place is in the areas covered by F and OI.
@Julia wrote that she’s “sure what Fugro and OI did find has more than covered their costs with the amount of non MH370 data they have collected”
Fugro was paid under contract to undertake its survey of the 7th arc and its transits.
Ocean Infinity didn’t not achieve the publicly stated terms in order to collect the success based fee, i.e., to have located the wreckage. Above, I quoted that OI donated the results from their survey to NF-GEBCO’s Seabed 2030 project.
How does your allusion to ulterior motives and funding help?
@Don
I’m not exactly suggesting it does help! Just an observation. Cynical yes but an observation. It’s excellent that OI has donated its findings to the project you mention. I didn’t know that. I do know however that OI did not receive a fee as it didn’t achieve its objective in the time constraint imposed by Malaysia.
I am as keen to understand the disappearance of MH370 as anyone and in doing so, may make the occasional mistake or cynical comment. Better to express one’s views IMO than keep quiet.
@Don
Just to put the record straight my observation in no way suggested that either company had ulterior motives in searching the seabed for MH370. Your words not mine!
I was merely saying that inevitably, valuable research data would be a by product of the search for the plane.
@Victor, @DrB, @Niels, @David,
I find it very interesting that the crossover point in David Griffin’s latest update including a new Fig. 5c is 24°S.
https://www.dropbox.com/s/3095eg1ulfa731m/David%20Griffin%20Fig.%205c.png?dl=0
MH370 floating debris has been found from Mossel Bay, South Africa up to Pemba, Tanzania. This requires a bifurcation around Madagascar. David Griffin’s Fig. 5c shows that a MH370 end point around 24°S is the optimal start latitude for floating debris to go both north and south around Madagascar to beach as far apart as Tanzania and South Africa.
By audit, I meant a detailed check of all the data to identify all the small areas that were not “illuminated” for any of several operational reasons. We know the terrain was difficult in many places. That would be especially problematic for the Fugro towed fish. I seriously doubt that either the Fugro or OI teams missed the debris, if it was illuminated. But how much was never illuminated? I’ve heard estimates as high as 3% from people how should know.
I do not think the audit needs to be conducted by a new team. Andy Sherrell would be my choice to head up the audit.
@airlandseaman
Thanks for this. Very interesting.(I know your reply was meant for others too)
@David,
Your most recent plots show 18 debris reported between 45-65E. I only count 14 (out of the 27 total I am using from the Malaysia MOT list, not including the 2 unassessed items David Griffin later reported).
Dr B. Since I am utilising David Griffin’s data I use his reportings, listed at:
http://www.marine.csiro.au/~griffin/MH370/br15_pwent2d/br15_beaching_map_nonflap_36_08_25.gif
He has redated some in the Malaysian listing having other information about that.
I believe at lease 2, from St Luce, are very doubtful and others at least doubtful. My preference has been to utilise just “likely”, “highly likely”, “almost certain”, and “confirmed” (to be from MH370), amounting to 9 items (I include the vortex generator: my opinion) from his list. I will take a look at what the effect is of reducing to just those and will post anything interesting. That of course includes 4 out of 4 from the outlying islands, though that can obscure the ‘Blaine effect’ in Madagascar somewhat.
I count 15 on the Malaysian list to which he has added the vortex gen (28) plus 29, reported with 28, both unassessed by Malaysia. Also, 24 in fact consisted of 2 items. Hence his 18 total.
@RichardG
I think fig. 5c belongs to fig. 5a. You would need a fig. 5d based on fig. 5b for the bifurcation analysis as you describe it. Anyway,
I’m afraid Dennis is right that there is not much point in comparing reported vs. expected distribution. The “data” is ruined by a lack of systematic search.
@Niels: Using David G’s drift model, the predicted split between north of Madagascar and the rest does not match the location of recovered debris for any impact latitude. Either the model is wrong, or the recovered debris north of Madagascar doesn’t represent in any statistical meaningful way the drift patterns because the discovery efficiency is quite low.
Basically, the drift model predicts that for a large range of latitudes, there should be a lot of debris washed ashore in Madagascar. That’s also the place that a lot of debris was found. That’s also the place that Blaine Gibson and friends (locals and NOK) most thoroughly searched.
I am not seeing an obvious way to use the drift data to discriminate impact latitude along the 7th arc.
@Neils. I butt in. Yes, the reported data are distorted to an unknown extent by some areas not being searched, or competently and delays in reporting. But I notice that the 90th reported and modelled percentiles are comparatively close in many latitudes, leaving out the north one-off.
@VictorI
“I am not seeing an obvious way to use the drift data to discriminate impact latitude along the 7th arc.”
Neither do I. We have learnt a lot over the past months, however it feels a bit that the more we learn the less we can say about possible impact latitude.
@Niels,@Victor
“I am not seeing an obvious way to use the drift data to discriminate impact latitude along the 7th arc.”
I also came to that conclusion a few posts back, but I am less discouraged than I was prior to the OI effort.
If the following assumptions are valid:
1> Wreckage is on 7th arc within +/- 25 nautical miles.
2> Latitudes between 38S and 20S contain the wreckage.
3> Latitudes 38S to 20S are equally likely to contain wreckage.
4> Sonar scanning has a 90% probability of finding the wreckage.
I have heard some pushback on 1>. There might be some disagreement with 2> (Jean-Luc et.al). 3> is simply a math bookkeeping statement. 4> has pedigree from AF447 – I think 4> is very conservative. Overall, I think the assumptions are quite reasonable.
So it is easy to conclude (no Bayes needed) that the probabilty of finding the wreckage in 25S to 20S is
P = 0.9*(1 – 0.1*(13/18)) = 0.835.
Where 0.9 is the probability that if the area contains the wreckage it will be found,
0.1*(13/18) is the probability the wreckage was missed in the scanning of the 13 degrees from 38S to 25S (i.e. the wreckage is in the range of 38S to 25S).
1 – 0.1*(13/18) is the probability that 25S to 20S contains the wreckage.
Never in the four plus years we have been at this endeavor have we been able to make a statement at this level of confidence backed up by reasonable constraints. 5 degrees of arc have a greater than 80% chance of containing the wreckage. I say go for it.
@DennisW,
It is also true that if the debris field is outside 25 NM, the probability of finding it is exactly 0% using the plan you described.
@DrB
It is also true that if the debris field is outside 25 NM, the probability of finding it is exactly 0% using the plan you described.
True. It remains true no matter how wide a search zone one assumes. I cannot internalize the dive and glide scenario from any perspective – data, intent, previous crash histories, pilot protocols,…
@Niels
You stated “The “data” is ruined by a lack of systematic search.”
The only partial systematic search was by Blaine Gibson, who involved the NOK families and locals.
There are 7 items from Riake Beach and 3 items from Antsiraka Beach in Madagascar, but some of the items could have come from elsewhere and brought to Blaine at these locations (as was the item from Saint Luce in South Eastern Madagascar).
Blaine also found one item “No Step” in Mozambique.
Blaine also visited Mauritius and Rodrigues, without making finds, although others were successful.
All Blaine’s search areas were selected on the basis of guidance from Charitha Pattiaratchi and David Griffin.
I agree that the search in South Africa (5 items all found by locals) and Mozambique (7 items all found by visitors) was not systematic. I certainly agree that a systematic search of Tanzania, Kenya and Somalia did not take place.
Who knows what may have resulted if a systematic search of all locations had taken place. We are only talking about 15,545 km of coastline.
However, we have 30 reported items (27 on the Malaysian list and 3 others being held locally). Out of this list, there are 20 either confirmed or likely from MH370. The locations of these 20 items has to fit the drift analysis.
The drift analysis may not be able to point to a single location as the only possible MH370 end point. The drift analysis can give a short list of candidates that are likely as a possible MH370 end point. One such location, which has not yet been searched, is 24°S near the 7th Arc.
@RichardG
Just to make sure: my comment was not in any way meant critical towards those who at least, and as volunteers, took action to search.
Also, to be honest, it is with hindsight that I say that an extensive systematic search of the African coasts would have been beneficial. And to do it properly it would have been a tremendous task, as you already indicated. Possibly, you would need to cover the same area several times.
The main point I want to make is that either the dataset of reported finds is highly incomplete / not representative, or the CSIRO model (incl. assumptions) is wrong. We cannot continue to use both in ways like you were suggesting with your yesterday posting.
My impression is that the main problem is in the dataset. However, an interesting check for the models which could perhaps be carried out, is to compare the latitude where most of the debris hits the Madagascar east coast in your model and in DavidG’s model
@DennisW
I have doubts your assumption 3 is correct. However, part of it comes back to the BFO (error) discussion, where we will not likely come to a common conclusion soon (let alone put any numbers) given the incomplete info we have on BFO error statistics.
I understand, given what we know now, the proposal for a search further north, however honestly speaking my expectations are not very high.
I think it would be wise, as also @ALSM suggested, to first carefully and critically evaluate the previous underwater scan efforts.
@Dan Richter: Stop trying to comment here using different names. You’ve been banned.
@Niels
I actually debated listing assumption 3>. It is actually a commonly made if not explicit assumption used in both the drift analytics as well as the Inmarsat analytics. It is harmless and merely states that no a priori preference is give to specific latitudes.
As far as BFO is concerned I avoid using the term statistics. BFO is very much like a random walk and only ensemble behavior, not useful for a given flight, can be used.
Frustrating …with beach searches….what would be now the chances of finding further wreckage if one were able to search ..and where would you search now.? Still cant believe no organised beach searches ever took place….said this all the way along….imagine if at least some kind of search org was set up engaging countries in the wreckage areas..likely we would have more debris and subsequent drift data now. Throw away view…Z could have played with that EOF ..his mind may have wandered..if it was him all along then I reckon he would have been playing with the aircraft to see what it could do one last time..as a form of distraction from what reality was at that point.
Thanks Victor for allowing me to join your blog. I have followed it with great interest for quite some time after Duncan’s blog effectively stopped. I have never been a blogger before.
My background is research in science and technology. I have written many peer reviewed papers on different subjects from Chemistry and Physics to Engineering. Microwave engineering and Mathematics are mostly hobbies but I have also written a few papers on these subjects.
I have now written up a paper on MH370, which I will publish in a few days. Before the publication comes out, I would like to join your discussions. I think I have made some important observations, which may help locate the plane.
@Viking: Welcome to the blog. New thoughts and insights are welcome. We look forward to reading your paper.
@All,
I never understood why nobody tried to join the temperature information from the clamps into an optimized back tracing of the flaperon (or even just into a much simpler forward calculation). This way I think the problem could have been solved long ago. I have done such joint optimizations in my own field with great success. If done properly it allows removal of numerical instability and chaotic dynamics in many back tracing and inversion problems.
@DennisW,
How can you be sure we can forget latitudes above 20S? After all, the tropical clamps prefer starting their lives at the temperatures common up there.
@DennisW,
I am not quite as skeptical concerning the use of BFO. However, there is a serious pitfall, which I discuss in an appendix to my paper. You may get an impression from this observation:
Look at the Inmarsat plots of the BFO fit, e.g. figure 14 of their 2014 paper. All the last five points of the fit are 5Hz above the measured values. This is truly unusual for a refined model fit unless there is something fundamentally wrong with the measurements or the assumptions in the model.
@jinott: Welcome to the blog.
Unfortunately, Malaysia never has shown much interest in the recovered debris, as those that have tried to send them debris have discovered. It is no surprise that they never organized or coordinated search efforts in Africa.
@jinott
“I reckon he would have been playing with the aircraft to see what it could do”
Well for what it’s worth, a little further above, in my list of 10 possible reasons for an intentional dive at the end, @ALSM would only accepted that one explanation (a manuever for the heck of it).
@Victor
I am not sure about 20S. Extending the range Northward does not alter the conclusion, but makes the resumption of a search less attactive. I suppose the 20S choice shows my bias for a search continuation.
However, my current thinking based largely on Richard’s work is that 20S is a very sensible choice with the most likely terminal location in the 25S to 22S range.
@DennisW: That comment was from @Viking, not from me.
@Viking
I think the path in the Figure 14 you reference was not derived using BFO values. The BFO “fit” merely indicates a strong correlation.
In Section 5.5 (below Figure 14) you find the statement:
Combining the sensitivity data with the measurement accuracy of ±7 Hz indicates that inaccuracy in each individual BFO measurement would correspond to ±28° heading uncertainty and ±9° of latitude uncertainty.
I have long been critical of the tendency on this blog (and elsewhere) to place much too great a weight on the BFO data.
@DennisW,
I presume your latest comment was predominantly for me (sorry, I just joined). At this stage, my argument is that it would strengthen our debris interpretations if we included temperature information obtained from the clamps shells. The beaching pattern alone does not give much latitude discrimination, and therefore a poor basis for limiting ourselves to 22-25S.
@Victor,
Sorry about confusing the commenters. I am pretty sure my answer to Viking would have been the same. 🙂
@DennisW,
I fundamentally agree, Fig. 14 was not predominantly derived from BFO, and one should not put too much weight on BFO. However, it was far from their first attempt, and the deviation is very systematic.
@Viking
You stated “I never understood why nobody tried to join the temperature information from the clamps into an optimized back tracing of the flaperon”.
As you mention you previously read Duncan’s blog, then you will remember this post about barnacles and sea temperatures dated 2nd April 2016:
http://www.duncansteel.com/archives/2536
May I also refer you to an earlier post on Victor’s blog with reference to barnacles and sea temperatures dated 20th July 2018 and 2 papers also published on the subject dated 2016 and 2017:
http://mh370.radiantphysics.com/2018/07/20/godfrey-drift-model-says-mh370-might-have-crashed-further-north-on-arc/
https://www.dropbox.com/s/xijqnc5zk05vizi/Barnacles%20Temperature%20Path.pdf?dl=0
https://www.dropbox.com/s/pveb004xil0egml/Optimum%20Sea%20Temperatures%20for%20Barnacle%20Growth.pdf?dl=0
In fact, barnacles were the subject of numerous comments on many of the posts here.
@Richard,
What I am looking for is not using the temperature curves alone or besides the tracking (indeed that was done already, – sorry i did not express myself clearly). My recommendation is a fully integrated simulation, where both types of information are included simultaneously and at the detailed level.
This is somewhat difficult to code, but extremely powerful.
@All,
On a completely different, non-scientific subject (but potentially relevant for MH370), this was in the news stream yesterday:
https://www.bbc.com/news/business-46062576
I am by no means claiming that the 1MDB scandal was the reason for MH370, but I have a feeling there is some kind of connection. I am not able to quantify, just a gut feeling.
The reason I bring it up is that it will now be investigated in USA. That may provide some new information.
@Viking
You stated “My recommendation is a fully integrated simulation, where both types of information are included simultaneously and at the detailed level.”
That is exactly what I have done with my simulation.
@Richard,
Please do not take this as criticism of your work. I am deeply impressed how much time and dedication you have put into it. You have gone much further than any of the others doing these simulations. However, not to the full detail I was thinking of. Let me give some examples:
1) As far as I can see, you use averaged temperatures, not the detailed temperatures as a function of date on the relevant year.
2) Only some of the simulations include detailed effects of wind.
3) The reference buoys have not necessarily been passing the areas at the relevant time for MH370 debris. However, I do remember that you took away older data (extremely good, since global warming, changing weather patterns and technological improvements change the results).
4) Actively using temperature as (detailed) feedback in the calculation, not just analyse complete tracks.
I know this is difficult and extremely time-consuming, and I am by no means asking you to do all that work. I was thinking about what the official investigators could have done.
I think it would also be necessary to make more detailed measurements on the shells (of course not possible for us, since we do not have any of them in our possession). I would suggest using SIMS, since it gives unprecedented spatial resolution.
@Viking
Here is an example track from 21.1338°S on the 7th Arc, which ends after 520 days in Southern Mozambique at 26.0287°S 32.9411°E. This track passes within 17 NM of Reunion after 356 days. The colour coding gives the drift speed, Yellow is 10 cm/sec (0.19 knots), Orange is 30 cm/sec (0.58 knots) and Red is 50 cm/sec (0.97 knots).
https://www.dropbox.com/s/m8m8jezm7esz6d5/Drift%20Map%2021.1338S%20103.9946E%2026.0287S%2032.9411E%20520d.png?dl=0
Here is the same track but the colour coding gives the sea water temperature, Magenta > 25°C (too hot for barnacles to reproduce), Orange 19°C to 25°C (perfect for barnacle reproduction and growth), Yellow < 19°C (too cold for barnacles to reproduce (please see the paper by Patel, that I reference in the paper Optimum Sea Temperatures for Barnacle Growth).
https://www.dropbox.com/s/3p8mgl3tlz3rzht/Drift%20Map%2021.1338S%20103.9946E%2026.0287S%2032.9411E%20520d%20Temp.png?dl=0
As you can see, for around 50% of the track from the 7th Arc to Reunion, the sea water temperature was too hot for barnacles, but for the last part to Mozambique a debris item should have arrived with some barnacle growth.
@Viking
You stated:
“1) As far as I can see, you use averaged temperatures, not the detailed temperatures as a function of date on the relevant year.
2) Only some of the simulations include detailed effects of wind.
3) The reference buoys have not necessarily been passing the areas at the relevant time for MH370 debris. However, I do remember that you took away older data (extremely good, since global warming, changing weather patterns and technological improvements change the results).
4) Actively using temperature as (detailed) feedback in the calculation, not just analyse complete tracks.”
1) The detailed temperatures used are a function of date, but averaged over recent years for each location along the path.
2) All simulations include the detailed effects of wind.
3) I use a database of 280 undrogued drifter buoys with a full coverage of the Southern Indian Ocean. I recalculate the drift speed and direction on a daily basis with a fine spatial and seasonal granularity.
4) I fundamentally disagree with reverse path simulations as all attempts so far in the academic literature end up with a very large range of drift start points (MH370 end points).
The last 100 days before beaching is sufficient for a colony of barnacles to grow to full size (25mm Capitulum). The flaperon showed evidence of recolonisation. Sexual maturity is reached after 30 to 37 days.
The flaperon arrived after 508 days and the fastest 3 items to mainland Africa arrived after 679 days. The sea water temperature prior to the last 100 days is not going to give you reliable clues, when you use the sea water temperature as detailed feedback in the reverse track calculation.
I wish you luck writing your simulator and look forward to seeing the results.
@Richard
Unfortunately, I am going to disappoint you concerning the writing of a new simulator. I actually did the equivalent of that in my own research field together with a PhD student 16 years ago. It almost cost me a PhD student, since he got so frustrated by the great difficulties that he wanted to give up halfway. However, at the end, it worked fantastic and we managed to beat a group of six Russian mathematicians who had worked twice as long on the same problem.
Their simulator used 24 hours on a super-computer; ours used only 3 minutes on a small PC, and was more precise. When we presented the results, they claimed we were lying. 3 months later, they went completely silent.
If someone would pay me 3 years of research funding I would do it again, but not without funding.
Our computer network will be down for scheduled maintenance during the weekend. I will be back on Monday.
@Viking
Like my daughter, a professor of computer science at Cal Poly San Luis Obispo. I have tried to enlist her help in AI relative to MH370 (she is becoming quite a recongnized expert in that domain, and makes a lot more money consulting, but does the Poly gig for medical benefits which are superb for Cali people sucking on a government teat). In any case, she is (like you) coin operated, and ignores me.
@Viking
You stated “Unfortunately, I am going to disappoint you concerning the writing of a new simulator.”
I am indeed disappointed. Especially when you say you are able to simulate in 3 minutes on a small PC, what for others takes 24 hours on a super-computer. As you say, in this way the problem could have been solved long ago.
@RichardG
I was comparing your “fig 5a,b” with those from DavidG.
Is it correct you have most of the Madagascar beachings in the S16-S18 latitude range?
We should check (also with DavidG) as to me it looks he has most of Madagascar beachings in the S12-S16 range.
This all for starting latitudes in say S36 – S26 range.
@Niels
My peak beachings on Madagascar is at 17°S and 18°S from start latitudes between 19°S and 33°S near the 7th Arc.
There are also quite a few beachings on Madagascar at 19°S and 21°S from start latitudes between 19°S and 31°S near the 7th Arc.
@RichardG
Ok, thank you. I’ve sent DavidG an email.
@Victor
Re: Lion Air
Lately I’ve been watching this pilot’s tutorials on YouTube. This session covers (generically) aircraft out-of-control:
https://www.youtube.com/watch?v=0Z10xIGotPY
Reasons: (1) Weather (2) FoulPlay
(3) Technical Prob at 8-min (e.g; Alaska Air loss due to jack-screw failure)
(4) Pilot error due to lack of airspeed indication at 11:35 min. eg; he feels pilots falsely sense they are pitched-up excessively so there is a tendency to pitch down to the ground.
Contrary to fake news put out on Twitter, regarding the Ocean Infinity search for the submarine off the coast of Argentina, Seabed Constructor returned to Comodoro Rivadavia as usual, not Punta Arenas as stated in Twitter. Seabed Constructor then returned to the search area, where it continues the search. The search has not been suspended until February 2019 as stated in Twitter.
The present location of Seabed Constructor is shown in the following link:
https://www.dropbox.com/s/issotrug7m7uxgb/San%20Juan%20Argentina%2003%20Nov%202018.png?dl=0
@TBill: We’ll know soon exactly what happened in JT610, as either the FDR or CVR has already been recovered. My guess is there was a failure of an airspeed sensor. During the climb to 5,000 ft, the airspeed was well above 250 knots, which should have been the speed restriction below 10,000 ft. Even if the thrust was set to full, the autopilot should have pitched up to keep the speed below 250 knots. The airspeed sensor might have been reading low. Perhaps there were even false stall warnings. Rather than the pilot disconnecting the autopilot and autothrottle and manually setting pitch and power, they might have lowered the nose, oversped, and flown the plane into the sea.
@TBill and Victor,
The video that TBill referenced gives a very good explanation of some of the Human Factors that can deceive a pilot if they are relying on only the “human sensors” to control the plane.
One particular point was made very clearly . . the aircraft can be flown quite safely without any airspeed indication. The standard procedure is to set the pitch [angle] to the appropriate value, and set the power levers to the appropriate position, then if everything else is OK the aircraft will fly straight and level.
On a clear day, as this was, the pitch can be determined quite precisely by the position of the horizon on the windscreen, so even the artificial horizon instrumentation is not required.
@Brian Anderson
RE: “One particular point was made very clearly . . the aircraft can be flown quite safely without any airspeed indication.”
Absolutely. A well trained, competent pilot should be able to fly and land the aircraft quite safely with an unreliable speed indication. The procedures are well defined and the required pitch attitude/thrust tables are readily available in the pilot’s QRH.
@Victor
RE: “My guess is there was a failure of an airspeed sensor. During the climb to 5,000 ft, the airspeed was well above 250 knots, which should have been the speed restriction below 10,000 ft.“
Note that Indonesia does not impose a blanket 250 kt restriction below 10,000 ft. There are speed restrictions for arriving aircraft at busy airports such as Jakarta (annotated on the STAR charts), but there are no such restrictions for departures. That said, it would be unusual for a crew to accelerate to a high speed if they encountered a problem and intended to return for a landing.
@Andrew: Thank you for the speed restriction information. I did look at the B737 departures for the same flight number for the day before and the day after the crash (although they were not MAX-8s), and the IAS was less than 250 knots until 10,000 ft was reached. The speed profile during the climb for the accident flight was quite different.
My guess is an airspeed sensor was reading low, leading to high automated speeds followed by pilot error, possibly to avoid what they thought was a stall. What’s your guess?
I thought I understood what David Griffin’s fig. 5a + 5b. mean, but I just realized I don’t. Can someone help me out ?
(1) The vertical scale at the right side: Looking at the formula I take it this is a percentage scale. So why is it extended to 120% ?
(2) The vertical scale on the left side: I assumed this to be the cumulative probability of beaching latitudes.
– Fig 5b: But why is 24S green when the entire row is largely dark blue ?
– Fig 5b: … whereas 23S is darkblue when the entire row is in a lighter shade of blue than 24S ?
– Fig 5b: Why is 0-12S darkblue when the rows are much hotter than 12-34S ?
– Fig 5a: 12S is the hottest row but is darkblue in the cumulative scale, whereas 17S is red with a much colder row ?
Many thanks
@Niu Yunu,
The vertical scale on the right side is from 0-20%. The colored cells on the left side represent the actual numbers of debris found at each latitude.
@Richard Godfrey,
Per Kevin Rupp, OI will continue the search for ARA San Juan for 12 more days, which will bring them to the limit in their current contract of 60 days. Then Seabed Constructor will go in for a planned dry-dock refurbishment. Then SC may return in February for another 60 search days. That second phase has not yet been confirmed officially.
DrB: “The colored cells on the left side represent the actual numbers of debris found at each latitude.”
Ah, I see. Thank you!
DrB: “The vertical scale on the right side is from 0-20%.”
But it is labelled 0-12 ?
@Victor
RE: “My guess is an airspeed sensor was reading low, leading to high automated speeds followed by pilot error, possibly to avoid what they thought was a stall. What’s your guess?”
I’m reluctant to speculate, but the IAS & ALT DISAGREE fault that occurred during the previous flight obviously suggests there might have been a problem with one of the pitot-static systems. There’s an image of a maintenance log page circulating on the Internet that supposedly shows that fault was rectified by flushing the LH pitot and static air data modules. There is also a report that the aircraft’s speed trim system (STS), a speed stability augmentation system that operates during manual flight, operated in the wrong direction when the IAS disagreement occurred during the previous flight. If the same faults re-occurred during or shortly after take-off, then perhaps the accident was caused by a poorly-handled unreliable speed event, compounded by the STS causing control problems.
@Andrew said: the aircraft’s speed trim system (STS), a speed stability augmentation system that operates during manual flight
The STS is interesting. A pitot probe failure and/or STS issue might explain why the pilots could not gain control while manually flying. From the FCOM:
Speed Trim System
The speed trim system (STS) is a speed stability augmentation system designed to improve flight characteristics during operations with a low gross weight, aft center of gravity and high thrust when the autopilot is not engaged. The purpose of the STS is to return the airplane to a trimmed speed by commanding the stabilizer in a direction opposite the speed change. The STS monitors inputs of stabilizer position, thrust lever position, airspeed and vertical speed and then trims the stabilizer using the autopilot stabilizer trim. As the airplane speed increases or decreases from the trimmed speed, the stabilizer is commanded in the direction to return the airplane to the trimmed speed. This increases control column forces to force the airplane to return to the trimmed speed. As the airplane returns to the trimmed speed, the STS commanded stabilizer movement is removed.
STS operates most frequently during takeoffs, climb and go-arounds.
Conditions
for speed trim operation are listed below:
• Airspeed between 100 KIAS and Mach 0.5
• 10 seconds after takeoff
• 5 seconds following release of trim switches
• Autopilot not engaged
• Sensing of trim requirement
@Victor @Andrew
I take it a B777 does not have a STS system, but a 737MAX does? Perhaps due to higher power engines relative to the size of the aircraft.
@TBill: I have a pretty good understanding how the trim system works in the FBW B777, but only a rudimentary understanding of how it works in the 737, especially since there seems to be a lot of conflicting descriptions of the operation of the STS. Rather than further confuse the issue, I’ll let @Andrew respond.
@Richard Godfrey,
There is a report today that Ocean Infinity’s contract with the Argentine Navy for the ARA San Juan search has been extended 120 days from December until “mid-April” 2019. The extension is based on calendar days, not search days. OI will get $7.5 M USD if they locate the debris field within the contract deadline.
@Niu Yunu,
Look at Figure 5a here:
http://www.marine.csiro.au/~griffin/MH370/
The right hand scale from the current Figure 5a goes from 0 to 20%. I don’t know what you are looking at that shows 12%.
@DrB, @Niu Yunu,
David Griffin changed the scale he was using in recent Fig. 5a and Fig.5b.
On 27th Oct 2018, Fig. 5a looked like this:
https://www.dropbox.com/s/5k1h6pkalflirwc/David%20Griffin%20Fig.%205a%2027%20Oct%202018.png?dl=0
On 1st Nov 2018, Fig. 5a looked like this:
https://www.dropbox.com/s/r25731xpbvjiri1/David%20Griffin%20Fig.%205a%2001%20Nov%202018.png?dl=0
[Comments here are closed. Please continue the discussion under the new article.]