The above graph is taken from the DSTG book “Bayesian Methods in the Search for MH370, ” page 90. It shows the probability distribution of MH370’s endpoint in the southern Indian Ocean based on analysis of the different autopilot modes available to whoever was in charge of the plane during its final six hours. It was published earlier this year and so represents contemporary understanding of these issues. As you can see, the DSTG estimated that the probability that the plane hit the 7th arc north of 34 degrees south longitude is effectively zero.
I interviewed Neil Gordon, lead author of the paper, on August 11. At that time, he told me that experts within the official search had already determined that the BFO values at 0:19 indicated that the plane was in a steep descent, on the order of 15,000 feet per minute.
Such a rate of descent would necessarily indicate that the plane could not have hit the ocean very far from the 7th arc. Nevertheless, Fugro Equator, which was still conducting its broad towfish scan of the search area at the time, spent most of its time searching the area on the inside edge of the search zone in the main area, between 37.5 and 35 degrees south latitude, about 25 nautical miles inside the 7th arc. At no point between the time of our interview and the end of the towfish scan in October did Equator scan anywhere north of 34 degrees south.
Shortly thereafter, the ATSB hosted a meeting of the experts it had consulted in the course of the investigation, and the result of their discussion was published on December 20 of this year as “MH370 – First Principles Review.” This document confirms what Gordon told me, that the group believed that the BFO data meant that the plane had to have been in a steep dive at the time of the final ping. What’s more, the report specified that this implied that the plane could not have flown more than 25 nautical miles from the 7th arc, and indeed most likely impacted the sea within 15 nautical miles.
By the analysis presented above, a conclusion is fairly obvious: the plane must have come to rest somewhere south of 34 degrees south, within 25 nautical miles of the seventh arc. Since this area has already been thoroughly scanned, then the implication is that the plane did not come to rest on the Indian Ocean seabed where the Inmarsat signals indicate it should have.
I would suggest that at this point the search should have been considered completed.
Nevertheless, the “First Principles Review” states on page 15 that the experts’ renewed analysis of the 777 autopilot dynamics indicates that the plane could have crossed the 7th arc “up to 33°S in latitude along the 7th arc.”
Then in the Conclusions section on page 23 the authors describe “a remaining area of high probability between latitudes 32.5°S and 36°S along the 7th arc,” while the accompanying illustration depicts a northern limit at 32.25 degrees south.
In other words, without any explanation, the northern limit of the aircraft’s possible impact point has moved from 34 degrees south in the Bayesian Methods paper in early 2016 to 33 degrees south on page 15 in the “First Principles Review” released at the end of the year. Then eight pages later within the same report the northern limit has moved, again without explanation, a half a degree further north. And half a page later it has moved a quarter of a degree further still.
Is the ATSB sincere in moving the northern limit in this way? If so, I wonder why they did not further search out this area when they had the chance, instead of continuing to scan an area that they apparently had already concluded the plane could not plausibly have reached.
I should point out at this point that the area between 34 south and 35.5 south has been scanned to a total widtch of 37 nautical miles, and the area between 32.5 and 34 has been searched to a total width 23 nautical miles. Thus even if the ATSB’s new northern limits are correct, they still should have found the plane.
As a result of the above I would suggest that:
a) Even though most recent report describes “the need to search an additional area representing approximately 25,000 km²,” the conduct of the ATSB’s search does not suggest that they earnestly believe that the plane could lie in this area. If they did, they could have searched out the highest-probability portions of this area with the time and resources at their disposal. Indeed, they could be searching it right now, as I write this. Obviously they are not.
b) The ATSB knew, in issuing the report, that Malaysia and China would not agree to search the newly suggested area, because it fails to meet the agreed-upon criteria for an extension (“credible new information… that can be used to identify the specific location of the aircraft”). Thus mooting this area would allow them to claim that there remained areas of significant probability that they had been forced to leave unsearched. This, in effect, would allow them to claim that their analysis had been correct but that they had fallen victim to bad luck.
c) The ATSB’s sophisticated mathematical analysis of the Inmarsat data, combined with debris drift analysis and other factors, allowed them to define an area of the southern Indian Ocean in which the plane could plausibly have come to rest. A long, exhaustive and expensive search has determined that it is not there.
d) The ATSB did not fall victim to bad luck. On the contrary, they have demonstrated with great robustness that the Inmarsat data is not compatible with the physical facts of the case.
e) Something is wrong with the Inmarsat data.
@all, VictorI posted this comment in the last thread just as I was about to close it, so I’m reposting it here:
Here is a short video for the simulation that uses the 45S1 data set as the initial conditions and flies for another 27 s. Conditions were light winds of 25 knots. The plane initially starts at 37,654 ft and climbs due to the positive flight path angle imposed as the initial condition. I roll out of the bank and hold the altitude. As the speed continues to decrease, I allow the plane to start descending, reaching 700 fpm and 37,654 ft at the end of the video.
The final altitude is almost exactly the same as the starting altitude, which shows that the result is not at all “remarkable”. After 27 s, the simulation is paused, the altitude manually changed from 37,654 ft to 4,000 ft in the MAP window, and a new flight file is created. The position, bank, heading, and pitch for this new flight file are similar to what is found in the 45S2 file. Considering that I was flying with manual inputs, it is not surprising that the match is not exact. However, the match shows that we can be quite sure that the pilot was manually flying between 45S1 and 45S2 in a very ordinary way and under very ordinary weather conditions.
https://www.dropbox.com/s/6039dgo3hwy8sx6/2016-12-25%20After%20Flameout.wmv?dl=0
Here are some variables from the flight file that was created after pausing and changing the altitude.
Latitude=S045° 07′ 47.9993″
Longitude=E104° 08′ 38.4000″
Altitude=+003999.99
Pitch=-4.3087744394544405
Bank=-0.13368115784497478
Heading=-174.0096132162906
PVelBodyAxis=0
BVelBodyAxis=0
HVelBodyAxis=0
XVelBodyAxis=0
YVelBodyAxis=0
ZVelBodyAxis=325.58327269368777
DynPres=114.38446786895769
AGL=37654.545100655138
@VictorI, Could you paste the same parameters from before the altitude was changed (but after it was already paused)?
@All, Mike Exner tweeted new found Mossel Bay piece might be trailing edge Tanazania flap. https://twitter.com/airlandseaman?lang=en
Here are the same variables from the flight file before the altitude was changed:
Latitude=S045° 07′ 47.7081″
Longitude=E104° 08′ 38.4512″
Altitude=+037653.90
Pitch=-4.3087744937088219
Bank=-0.13368120603826217
Heading=-173.91712176294601
PVelBodyAxis=0.00018614309386330585
BVelBodyAxis=-0.0020271815217738072
HVelBodyAxis=-0.0022713792578914439
XVelBodyAxis=-41.58377886262317
YVelBodyAxis=-54.603579041882462
ZVelBodyAxis=560.31061062745562
DynPres=106.56777021305984
AGL=37654.545100655138
@jeffwise: The failure of the underwater search does NOT mean that “something is wrong with the Inmarsat data” as you assert. It means that the underlying assumptions used in the DSTG model are incorrect. As some of us explained in the days after the DSTG report was published, there are two assumptions that should have been challenged:
1. The vertical speed at 18:40 was zero, which meant that the BFO value could only be matched with a southern track at this time.
2. The a priori distribution of the number of manoeuvers favored one turn.
The results of these two assumptions meant that the highest probability terminus would correspond to a straight path with an FMT before 18:40. It also rendered more “complicated paths” such as those with a loiter, multiple turns, or changes in speed as low probability.
At this point, I don’t think it is politically palatable for the ATSB to revisit the underlying assumptions in the DSTG report. It is also those assumptions that allowed the search area to stay to a “manageable” 120,000 km2. Remove the constraints imposed by those two assumptions, and the search area grows considerably.
@JeffWise
Good summary article. I disagree with the conclusion about the Inmarsat data.
Neil Gordon is one of the inventors of the concept of the particle filter so it is not surprising that it was selected as the approach for defining the 120K km2 zone. I think the assumed dynamic model had issues and this directly influences the probability curve shown in your article. The data may be fine but the analysis flawed.
Nobody is comparing the analysis approach created by Inmarsat with the DSTG approach.
The focus should be a arc 7 crossing. The water impact could be 100 miles for a max range glide. By a pilot. The plunge story in last BFO is questionable.
Someone at ATSB should explain why only the particle filter was used.
If I were committing to spent $200 million and two years on a search I would not have bet everything on one method and one group to define the only approach.
Somewhere in there is a story about Dr. Gordon and Inmarsat and how collegial the multi participant team was at deciding what to do.
@VictorI
What about if the AutoPilot is simply manually shut off, do you still expect True Heading? I am thinking the over-simplified MicroSoft 777-300 model in FS2004 goes to magnetic heading. I do not actually know how to program a discontinuity on that model, so I just turn off the AP when I get to the location in the flight plan that I want to go to manual mode. I can also hold a heading in AutoPilot but that is a magnetic heading in the simple model, it seems to me.
FS2004 (unless updated by the user) has a relatively severe magnetic curve based on 2003. So I had to manually update the magnetic headings to try to match your 180 mag heading paper of June_2016.
@VictorI. Thank you. The dyn press post altitude change is similar to what it had been at 45S1, ie 115.44 – as distinct from 379.55 at 45S2 in the sim data, so apparently the altitude change has not affected it in the same way. Also it is comparatively little changed from the 106.57 before the altitude change. You said earlier, “The readjustment continues while the simulation is paused.” Thus it is the timing of the pause which is behind this?
Going back, I rephrase my question on rpm.
At 4, (45S1) engine ‘Pct rpm” are 0.001. Since simulator rated speed is 29,920rpm, 0.001 of that is the 30rpm I quoted (ie leaving aside the “Pct”).
On engine speed decay you said, “Yves found that it takes 68 seconds from fuel exhaustion to attain the value at 45S1. It takes another 27 seconds to reach the value at 45S2.” I was querying whether his timing of 68 seconds, from about rated speed I assume, was to 30rpm? That looks unrealistic, particularly since one would expect windmilling to be well above it. Moreover, the speed at 5, (45S2) at 0.000007156 of rated, was 0.28 rpm. Did he time such a 27 seconds’ run down, from one turn in 2 seconds to one in 4 minutes, as if on the ground on a windless day?
That is why I supposed I might be using different data or there might be an alternative explanation, such as the “Pct rpm” being measured from a non-zero data base?
As it happens his 27 secs and your trial both took the “about 30 seconds” I described earlier for the 15 deg turn between 45S1 and 45S2, based on a rough average of the turn rates at the two.
All that aside, what I take away from his enterprise is that fuel exhaustion was before 45S1. Maybe fuel can be deleted without the flight file save process? Your paper does not say.
As a stray, which you might have addressed already though not in your paper, that relates how the aircraft model used for the sim data came from “PSS Boeing 777-200LR No VC”. There seems to have been a 200ER model available which ostensibly would have been more suitable if he had flight planning in mind. I assume the LR model had advantages which offset this?
@David,
No matter which tank the APU draws from, when the main engine(s) are operating, about 2/3 of the extra fuel flow is to the main engine(s) due to increased airframe drag, and only about 1/3 is actually used by the APU. The result is the same; for 9M-MRO the right engine runs out of fuel first.
@Jeff Wise,
You said: “What’s more, the report specified that this implied that the plane could not have flown more than 25 nautical miles from the 7th arc, and indeed most likely impacted the sea within 15 nautical miles.”
You also said: “I should point out at this point that the area between 34 south and 35.5 south has been scanned to a total widtch (sic) of 37 nautical miles, and the area between 32.5 and 34 has been searched to a total width 23 nautical miles. Thus even if the ATSB’s new northern limits are correct, they still should have found the plane.”
Using the 25 NM maximum distance from the arc, then by my calculations that is 50 NM total width for the proposed new search zone. Thus the 37 and 23 NM total widths previously scanned to the northeast fall short of the 50 NM total width the ATSB is now recommending. I would therefore disagree with your conclusions that “. . . they still should have found the plane.” and therefore “Something is wrong with the Inmarsat data.”
@Gloria
The video of the energy weapon, if it’s real, resulted in fire/explosions in the targeted drones. Surely there would be evidence of flame/heat damage in the 9M-MRO debris of which there is none. Also traveling in a westerly direction infers an incident much earlier in the flight with a greater fuel load. This would add to the heat damage.
I agree with the’funny shit’ comment from @DennisW.
@SteveBarratt
The debris from what?
Should we look down on children when we have a little knowledge instead of teaching them?
@Jeff Wise
First paragraph; ‘north of 34 degrees south latitude..’
Indeed the ATSB could go searching that area at this moment. And I read before somewhere their mandate for the search is till June next year.
It seems they ran out of money not out of time.
Interesting find Mike Exner twittered.
Obviously another trailing edge piece and a rather big one.
To me it doesn’t seem to be a part of the found outboard flap section though.
A first impression seems to show the bottom skin is carbon which is not the case in the outboard flap section and the already found left wing outboard flap trailing edge piece.
There are also no fastener lines visible like in the left wing trailing edge piece where it broke from the flap.
The boy is holding it up with his left arm and hand. Unless he’s left handed the photos could be mirrored.
If so the tappered impression the piece gives on a small part of its trailing edge suggests it’s from the left wing.
Possibly a piece from a landing flap.
And remarkable again; the piece also seems to be found on the waterline (like the latests Gibson-piece) like it just washed up the shore with no visible biofauling.
@Trond
@Gloria is not on my whacko list
@ all
Happy season greetings!
@ Jeff
I have two logical problems with the spoof to the south (mirroring the north track):
First, if we think about spoof in some way, i find it illogical to only think about this kind of scenario, because IMO then everything should be possible and considered.
Secondly we know, that the MAS ops calls from 1839 and 2313 utc each reset the timer for the interrogation, and i think nearly all are wondering, why only these two calls.
If they had tried it a 50 times in these hours, there would not have been the data log and not these SIO calculations thereof. Thus it is illogical to me to assume someone so sophisticated would rely on any airline/ATC to NOT call the plane more or less non stop…. so a live spoof can be ruled out IMO…
@VictorI: Thank you for the video and the data. I’ll spend some time looking at them.
You said: “Conditions were light winds of 25 knots.” Was the wind from 270 degrees as you said earlier? And is the wind at 4000 ft the same?
@Jeff Wise. The 33 degrees was described on pages 15 and 16 as a result of the aircraft not having been found, this of itself leading to a higher probability area further north.
“.. the probability distribution was updated to reflect the new information of the unsuccessful search which then favoured the other control
modes.” These control modes could have taken the plane “up to” 33 deg.
The reduced search area width resulted from some findings since your Gordon interview, including about the flaps being up and “recent” (Nov 16 announced) simulation findings about the final descent. The ATSB put that to the First Principles Review Meeting at which, “All participants were in general agreement that the distance required to be searched from the arc could be reduced to 25 NM from the 7th arc with a weighting to the west to account for the arc altitude”. This suggests that they see their view as needing endorsement
Drift analysis, quite hot off the press for the Review meeting included in conclusions, p18, “There is a region within the 36-32°S segment of the 7th arc, near 35°S that is most consistent with the following lines of evidence…”.
These conclusions were combined with the new control mode assessment and search widths resulting in a box with four corners. Which two corners you quote depend on personal choice. You have looked at other corners or mid points of the same box, as has the CSIRO. I think the intent of Figure 14 and the Conclusions is clear enough.
If the gamut of international experts, not by any means just the ATSB or DSTG, agree it is worth searching that area are you not bound to submit that finding? Attendees included “representatives … from all of the organisations participating in the Search Strategy Working Group including Australia’s Defence Science and Technology Group (DST Group), Boeing, Thales, Inmarsat, the National Transportation Safety Board of the US, the Air Accidents Investigation Branch of the UK and the Department of Civil Aviation, Malaysia”, plus the CSIRO, Geoscience Australia, Curtin University, Malaysia Airlines and the People’s Republic of China.
You assert the ATSB, “… have demonstrated with great robustness that the Inmarsat data is not compatible with the physical facts of the case”. I think it more accurate to say that the Review has considered new information and come to the conclusion that a new search area is warranted, still within the bounds of the Inmarsat data because there are no better. With a representative at the Review, Inmarsat could make its voice heard.
However the search area current and proposed depends not just on the Inmarsat information. There are also the undisclosed Boeing simulations and the supposed happenings leading to the 7th arc log-on. I would add that a case could be made for more research yet into drift modelling, since the flaperon models used were pretty basic. This could include, for example, experimenting with the part flap, which has a clear water line, including refloating to see how much its moisture had entered the skin during its voyage).
To me the question hanging is what is the probability that the aircraft is in the new search area ‘husk’ if say the probability of the aircraft being within the box is allocated a 1. Just glancing at Figure 14 I would hazard about 25%. If roughly right, that puts it in the possible, not probable zone. Naturally though the next-of-kin will be keen to see even that taken up.
The only reason I see that the ATSB conclusion can be seen as other than impartial is that they did not invite other expertise (without anything to defend) to participate in the review. That at least has a questionable look. If they were constrained from so doing they should have said so and why.
@Jeff
As I understand it, the 2016 DSTG ‘Book’ on Bayesian methods treats the autopilot (AP) mode selection as for any of the priors in the model, in that there is a model. For example, the prior model for the number of turns is shown in figure 7.3. The Book does not state the prior for the AP model but the recent First Principles review says they were given equal probability. The particle filter then selects the paths generated by the various model inputs that best fit the data and results in the posterior distributions – that for the number of turns is in figure 10.4. The posterior for the AP mode is not given, but figure 10.5 shows a very constant control angle with no increase in spread later in the flight, suggesting the constant magnetic heading/track modes made little contribution to the final prediction. Simply, the data suggests those modes were relatively unlikely, in the same way it suggests multiple turns were unlikely.
However, the initial assumption of equal probability for each AP mode is not a reflection of reality. Whatever happened to MH370 did not randomly select the AP mode (in the absence of a bizarre electronics failure that did exactly that), just there is no a priori knowledge of the mode in operation. With the search information to hand (southern destinations ruled out), the rerun of the model allows the other AP modes to make a stronger contribution to the solution and hence the distribution moves north. The original distribution (in Jeffs’s post) is no longer applicable.
Of course, there is a limit. This is not the same as saying that if 99.5% of a probability distribution has been searched unsuccessfully then there is a 100% chance that the target is in the final 0.5% (and hence no end is achieved), which is the way some of the coverage has been written. If none of the possible AP modes are consistent with the data, then a hard limit is reached.
Of course, the AP model may just be inapplicable, in that the aircraft did not fly as if it were under AP control. I would agree that the search is very close to that point now.
I observe that the favored route of Dr. Bobby Ulich (35S) ultimately agrees well with the favored route of Inmarsat (34.7S).
According to Inmarsat, MH370 was travelling at 800kph after NILAM. That would require a considerable deceleration, as if for a landing attempt.
800kph, from NILAM to IGOGU (on N571) turning slightly early towards ANOKO on a heading of ~192degrees, positions the plane pretty precisely over ANOKO at 2:43…
when a radio request for emergency landing (at Banda Aceh WITT) w/ explanation would be expected and required…
the alleged “cabin disintegrating” call could account for the ATSB’s proposed scenario, of cabin depressurization & hypoxia of PAX & crew…
and the corresponding landing attempt could account for Inmarsat’s reduced TAS of 930kph –> 800kph…
Thus, the 2:43 “cabin disintegrating” call comports well with the best candidate flight-paths of Inmarsat & Dr. Bobby Ulich, and with an ultimately futile landing attempt scenario…
The 2:43 radio call would have been cut short by the cabin depressurization & hypoxia underlying the ATSB’s scenario…
An ATT HOLD setting is also consistent with deceleration, descent, & landing attempt.
—
Separately from the above authors, I offer that “cabin disintegration” could have been caused by the preceding 80minutes of “pushing the flight envelope”, flying high & fast at maximum TAS and maximum dynamic ram pressure…
perhaps, even, deceleration was achieved by descent, “down the front edge of the flight envelope” at maximum KIAS / dynamic ram pressure…
perhaps due to damage or electro-mechanical failure of wing surfaces…
such that the aircraft descended to 20-25K feet where 320KIAS = Inmarsat’s 800kph…
IOW perhaps a desperate-measures “air braking”, required by limited wing operability, in combination with the previous 80minutes of envelope-pushing flight, induced the fatigue failure of the fuselage, causing the catastrophe
@Jeff Wise. With flap parts now multiple, the part flap I mention is that from Tanzania, held by the ATSB.
@VictorI
“At this point, I don’t think it is politically palatable for the ATSB to revisit the underlying assumptions in the DSTG report.”
you summed it up quite well, they have cornered themselves and now there is no leverage to search other areas on the 7th arc
@jeff wise
@victorl
Just a question is figure 10.3 (bottom) for 00:1929 onwards or just 00:1059. I also noticed posterior probability is not used between 00:1059 and 00:1929 as it is said 00:19 is understood to be unreliable this permits the filter to make superfluous manoeuvres.
I know since then ATSB has looked at the vertical descent scenario during 00:1929 and 00:1937. It’s just at the time Bayesian was adopted to search 00:19 onwards was not understood them and not used is the centre of Bayesian probability starting point for 00:1059 or 00:1929.
@Victor
you said:
“there are two assumptions that should have been challenged:
1. The vertical speed at 18:40 was zero, which meant that the BFO value could only be matched with a southern track at this time.
2. The a priori distribution of the number of manoeuvers favored one turn.”
I would add, and have added many times, that the accuracy of the BFO data has been greatly over estimated by not only the ATSB (and their collaborators – SSWG, DSTG,…) but by independent analysts as well. Figure 5.4 of the DSTG book is ample evidence of this assertion. My experience is that this is a very common error even among highly educated people – that is, a failure to understand the limitations of statistical analysis, in particular the notions of ergodicity and stationarity. Restricting routes to BFO accuracies on the order of a few Hz or less is just plain wrong.
The reality is that the Inmarsat data cannot be used to determine a terminus with sufficient confidence to justify the expense of an underwater search.
More commentary (written well before Jeff’s latest post) here:
http://tmex1.blogspot.com/2016/12/is-something-wrong-with-this-picture.html
@Erik Nelson, There was no “cabin disintegration” call. Speculation about landing attempts at Banda Aceh and the like is not at all well supported. It seems to me that you have not been paying attention to the conversation that has been going on around you.
@David, You wrote, “If the gamut of international experts, not by any means just the ATSB or DSTG, agree it is worth searching that area are you not bound to submit that finding?” The gamut of international experts as represented by the ATSB has time and again over the last two years stated their great confidence in the results of their analysis, and time and again they have been wrong. What’s more, the point I am trying to raise in this post is that there is no evidence that they really believe this latest claim. Neil Gordon told me about their new understanding of the 0:19 BFO value, namely that it means the plane was plummeting at breakneck speed, and therefore the plane should be very close to the 7th arc, when we spoke in August. He also expressed his view that the plane had to be further to the northeast. And yet Equator did not look here. And when it headed out on its last mission, after the “First Principles Review” meeting, it did not carry the towfish rig to search this new area but instead brought the AUV to search an area that the gamut of international experts had already said they were confident the plane was not.
@Michael Helms, Whatever mechanism by which the data was spoofed, assuming it was spoofed, would have returned BFO values that indicated a flight to the south no matter how many incoming calls the plane received.
@Ge Rijn, It seems to me that in one of the pictures there may be barnacles visible near the “pointy end” of the piece. I asked Albie Morkel if he could post more images of the area in question, he replied that he might but appears not to have done so yet. https://www.facebook.com/albie.morkel.393?fref=nf
Parenthetically I have been told that Simon Gunson has ingratiated himself very effectively with the NOK and in particular is quite close with Zaharie’s sister. Blaine Alan Gibson, too, is obviously very close with Grace Nathan and the other NOKs who recently traveled to Madagascar. This is perhaps why the NOK (Ghyslain Watrelos in particular) seem fond of a Diego Garcia/Maldives/US shootdown perspective on the matter. Thus via these efforts the NOK have been neutralized as a force for useful and effective action.
@DrBobbyUlich, Yes, if we accept that the width of the search area should be 25 nm on either side of the arc, then there remains a fair bit left to do north of 34 degrees south; however, the latest report affords “locations closer to the 7th arc a higher likelihood.” So the highest probability portion has already been searched.
@VictorI, Thank you for posting the .flt file data and the video. This is all extremely valuable and I look forward to examining them.
You wrote, “I don’t think it is politically palatable for the ATSB to revisit the underlying assumptions in the DSTG report.” I would argue that the assumption most in need of revisiting is the assumption that the data has not been tampered with. One can argue as to whether or not it seems plausible that sophisticated hijackers tampered with the satcom system, but what seems unequivocal is that the ATSB has not grappled with the issue at all, and so has no informed conception of whether it is plausible or not.
To spend $180 million based on the analysis of a certain body of data, without taking any effort to inquire as to whether that data is sterile, seems to me frankly irresponsible. And given the acknowledged failure of that resulting search, I don’t see how any impartial observer can remain so confident in their assessment of the unlikelihood of tampering that they would still not want the possibility to be examined.
@All, A lot of you feel that it is unlikely that the data was not tampered with. Can anyone tell me why their certainty is so ironclad that they do not feel that tampering should be examined as a possibility for the cause of the search’s failure?
@Jeff
you said:
“Can anyone tell me why their certainty is so ironclad that they do not feel that tampering should be examined as a possibility for the cause of the search’s failure?”
How would you propose to conduct such an examination?A lot of people have looked at the Inmarsat logs of the “accident” flight. Some of them pretty smart. No one has found any evidence of tampering.
It is really no surprise (to me) that the terminus prediction has not been successful. I was advocating this view before a tow fish was put in the water.
@DennisW, You wrote, “A lot of people have looked at the Inmarsat logs… No one has found any evidence of tampering.” Do you have any reference for this assertion?
At any rate, one can scarcely be expected to find evidence of tampering if one is not looking.
The same goes for debris. The presence of barnacles on a portion of the flaperon that has been demonstrated to consistently float high out of the water is hugely problematic, and has been acknowledged as such by the French, but the Australians seem entirely oblivous. And yes, I agree that they are smart people.
“It is difficult to get a man to understand something, when his salary depends upon his not understanding it!” — Upton Sinclair.
Mossel Bay debris update: After looking at more photos of the new Mossel Bay piece, and comparing them to photos of the Tanzania piece, it appears that the Mossel Bay piece probably is a trailing edge piece of a 370 flap, but not necessarily a piece of the inboard 1/3 of the right outboard flap found in Tanzania last July. Instead, it looks more like a piece that came from the flap further out (possibly from the other 2/3 of the right outboard flap). The relatively long straight edge of the Mossel Bay piece is a thick edge that looks like it broke at the spar. The piece recovered in Tanzania does not seem to have a matching “straight edge” in the area where the trailing edge broke off.
@Jeff
“Do you have any reference for this assertion?”
This blog. No one has produced anything credible in this regard nor has the issue been raised by the SSWG or the DSTG.
As far as the barnacles are concerned, I would agree with you if the flaperon was sitting in a salt water swimming pool. Anyone who has stuck a kayak in the ocean knows that they are going to get very wet very soon.
@Jeff Wise
Indeed there seem to be some small barnacles although not clear. Hope more and better pictures emerge soon.
But then maybe it’s not that important. It might have undergone the same proces as the first Mosselbay-piece (RR-engine cowling).
If it’s true the NoK now are holding their hopes on a Maldives/DG/US shootdown scenario due to influences by the ones you named this would be kind of sad.
Altough maybe I have contributed to this scenario too..
Over a year ago I posted this scenario as a fantasy-scenario on your blog. Never ment to be seriously considered though.
To me the most convincing evidence the Inmarsat-data were not spoofed or tampered with are the debris finds and the latest drift-analysis.
If those would clearly conflict the Inmarsat-data there would be true reason to doubt them.
But this is not the case. On the contrary: the debris/drifter-drift-analysis support the Inmarsat-data and the Northern part (and beyond) of the current search area.
@DennisW, You wrote: “nor has the issue been raised by the SSWG or the DSTG.” Exactly. This is the point. Has not been raised.
This is the huge white elephant in the room at this point.
BTW I find it curious that you can appeal to the authority of the SSWG in one breath and declaim in the next that your own personal analysis of the BFO data (or rather your gut feeling about it; I think you’ve admitted to never actually having rolled up your sleeves) shows that the SSWG can’t tell their ass from a hole in the ground when it comes to using the Inmarsat data to define a search area.
The fact is that I have enumerated a number of reasons why we should be suspicious of the integrity of the BFO data, and what hear in response is pooh-poohing and tut-tutting.
Likewise with the barnacle revelation. I am quite sure that the top of your kayak gets wet when you go paddling, but I guarantee it will not support a population of Lepas.
@airlandseaman
Can you please post a link to other photos you’ve got from the Mosselbay-flap piece?
Indeed like the left wing outboard flap trailing edge piece it’s broken off in front of its end spar. But there are no fasteners visible (like in the left wing outboard flap piece).
Don’t you think the thickness of that broken edge suggests a much larger flap it came from?
And can you possibly confirm the photos are not mirrored?
@Jeff
“the SSWG can’t tell their ass from a hole in the ground when it comes to using the Inmarsat data to define a search area.”
Have I not been saying that for some time?
My assertions relative to the BFO data are based on having fought with oscillators over a long career. They are not based on gut feelings. As soon as the DSTG book went on to calculate the mean and variance of the BFO data collected from the 20 previous flights of 9M-MRO I pretty much switched off. It is like computing the mean and variance of a random walk. You can compute a mean and variance for a random walk, but it would be meaningless to do so. You can say that the mean of an ensemble of random walks is zero, and that the variance of this ensemble grows with time. However, there is no way to use this information relative to where any particular random walk will terminate.
I think the water where I go kayaking is too cold to support Lepas. Not sure about that without consulting with a marine biologist.
@TBill asked, “What about if the AutoPilot is simply manually shut off, do you still expect True Heading?”
In a real B777, if the autopilot is disengaged, the plane will fly according to the trim and the wind. Ultimately, a disturbance will lead to a progressive bank angle, a descent, and a crash.
@David: “Thus it is the timing of the pause which is behind this?”
The dynamic pressure calculation is affected by the altitude change, the pause time, and other parameters in ways that are difficult to repeat and study. The fact that the value changes during the pause could mean the value depends on parameters such as processor speed and hardware, including thread processing. An acknowledged expert with decades of FS experience advises us that the values are incorrect and should be ignored.
“That is why I supposed I might be using different data or there might be an alternative explanation, such as the “Pct rpm” being measured from a non-zero data base?”
The coastdown and windmill modes in the FS9/PSS 777 are simplistic and inaccurate. As I already explained, Yves in a clever way recognized that the simplicity of the model allows N2 to be used as a timer after fuel exhaustion. The fact that the engines don’t windmill while flying demonstrate how simplistic that model is. And of course these coastdown experiments were performed while flying.
“Maybe fuel can be deleted without the flight file save process? Your paper does not say.”
Changing parameters during the flight and creating flight files are independent processes. Yes, fuel can be added or removed without creating a flight file.
“There seems to have been a 200ER model available which ostensibly would have been more suitable if he had flight planning in mind.”
I am not aware of a B777-200ER model for either PSS or PMDG. The 200LR would be the closest to the 200ER. That said, I doubt that a pilot would use an FS9 simulated flight to extract accurate flight parameters. There are better, more accurate ways to do this.
@VictorI: You may have overlooked my post of 3:45 AM about wind speed and direction. I’ve since discovered that the first question was superfluous – the caption above the video states the wind as 300 deg, 25 kt. My second question remains: is the wind at 4000 ft the same?
That information would help me to better understand the FLIGHT file data.
@TBill @VictorI
Wasn’t it discussed before and concluded that when a magnetic heading was flown with ‘true’ selected towards a certain waypoint that, with overflying this waypoint and no pilot inputs before or after, the plane would go on flying on this ‘true’ magnetic heading?
And wasn’t it concluded a ‘true’ magnetic heading would compensate for wind-variations but not for magnetic variations?
Hope you clarify to clear my thoughts about this..
“Was the wind from 270 degrees as you said earlier? And is the wind at 4000 ft the same?”
At high altitude, the wind seems to be 270T/25kt and constant for “Fair Weather”. The magnetic variation (per FS9) at 45S1 seems to be around 30 deg, so the wind displayed is 300M/25kt. The wind is less at 4,000 ft. Just after making the change in altitude, I see a wind of 30M/0kt (no wind), but soon after the wind is around 10M/4kt.
@Jeff Wise
For the record I still think you should change this sentence in the beginning of your topic.
It’s absolutely true but it makes no sence IMO:
‘the probability that the plane hit the 7th arc north of 34 degrees south longitude is effectively zero’.
@Ge Rijn: To my knowledge, the navigation mode after reaching a route discontinuity has never been completely resolved, although many (including me) have offered their opinion.
DennisW said, “I would add, and have added many times, that the accuracy of the BFO data has been greatly over estimated by not only the ATSB (and their collaborators – SSWG, DSTG,…)”
My comments were referring to paths generated in the DSTG report where a much larger BFO error was allowed. However, because it was assumed that the vertical speed was zero at 18:40, a late FMT is not permitted, even with the larger allowed BFO error. I think this is the critical assumption that needs to be revisited (but probably won’t be). Compare this to the ATSB report from June 2014, where the BFO at 18:40 was ignored, later FMTs were permitted, and the proposed search area was much further north.
jeffwise asked, “Can anyone tell me why their certainty is so ironclad that they do not feel that tampering should be examined as a possibility for the cause of the search’s failure?”
Is there a possibility of satellite data tampering? Sure. Let’s continue to consider it. But based on what we know today, the probability of pilot suicide ranks much higher than spoofed BFO data. The simulator data alone suggests the captain is a prime suspect. To use your own words, “The simulator data could reasonably be interpreted as evidence he planned a suicide flight, or it could be a freak coincidence.”
@Brian Anderson
Thank you Brian, I suspected that the wind from the east-northeast would affect the post ISBIX flight path, if in constant heading mode. Good to get expert confirmation, though. You are every bit the equal of Victor, in that respect, but don’t tell him that.
It was just that I was struck by the following:
1) The DSTG Bayesian analysis suggested a straight line path into the SIO, similar in many respects to an ANOKO (or IGOGU) ISBIX, 7th arc at S37.5 path, but the big problem with it is that the plane evidently couldn’t have flown that route (as Jeff points out) and Dr Bobby tells us it didn’t have enough fuel to reach that part of the 7th arc, anyway. Something is obviously amiss.
2) The most recent drift studies, described in the ATSB First Principles report, are pointing a finger at the S31.5 to S35.5 zone. Hmmm.
The major underlying problem with the Bayesian model is that it depends heavily on assumed aircraft dynamics (probable cruising speed ) and assumed MRC as provided by Boeing, and is hampered by under-constrained BFO data. It is rather heavily dependent on assumed initial conditions. The location of the 19:41 arc effectively constrains the position of the flight path in the east-west direction, in the vicinity of the Equator (as it happens, constraining it to pass close to, or through ISBIX). Badically, the effect of this is to aim a straight line at S37.5, or thereabouts, if you assume the plane flew in a straight line, and the Bayesian does have to assume this, because the BFO data is to ambiguous to give any indication otherwise, and you depend on Boeing’s estimate of the MRC.
The intriguing thing for me is that the ANOKO/ISBIS/S37.5 path is parallel to the line of the approaching sunrise terminator, so once you are established on that path, it doesn’t matter if you fly at M0.81 or at M0.78, when you get to fuel exhaustion at 00:17, the sun angle is the same, because sunrise occurs at the same time for all points along the flight path. A flight at M0.78 would not reach the 7th arc before the fuel ran out, but this would not have mattered, and Dr Bobby tells us it couldn’t reach it, anyway. So, on the night in question, the plane couldn’t have flown that route.
Which made me speculate that possibly the pilot had planned to fly that great circle route, but for some reason, it didn’t go to plan, something happened to cause the aircraft to deviate, after crossing the 19:41 arc, and it ended up somewhere between S31.5 and S35.5. It might also explain why there appears to have been no one at the controls at fuel exhaustion. The pilot was alive at the FMC, possibly still alive at 19:41, but dead by 00:17.
Interesting, don’t you think?
@VictorI
Next question ofcourse is; what was your opinion? I cann’t remember I read it.
To me (as a non-pilot!) it seems not a difficult thing to figure out. Just do it in flight (or simulated?) and you’ll know.
The difference between ‘true’ magnetic heading or only magnetic heading could be huge over such distances if this happened shortly after FMT I think.
If compensated for wind-variations (‘true’) after a route-discontinuity early on the amount of possible magnetic routes would be reduced.
It would imply though an unresponsive pilot at least just before the route-discontinuity and after.
And that’s still an assumption that’s unlogical to me.
But as all stones have to be turned at least once in this case I won’t submitt to any conclusion yet.
@JeffWise
Good post. As I read it I thought “he’s been reading my mind!” But perhaps it’s merely another example of Synchronicity. I think the ATSB were encouraged/persuaded to cut short their search of the northern zone, due to the gradually forming general consensus that the diversion was a deliberate act, and the path into the SIO was piloted. They are now eyeing the northern end again because by a process of elimination, if it isn’t in the south, it must be in the north. And the drift studies support the northern part. And they have to make an attempt at saving face. Come the middle of January, instead of merely shrugging their shoulders and admitting defeat, they can say “well we should have been allowed to fully search the northern part, after all. We were deflected from our work by outside interference”.
@Ge Rijn: My opinion is that the plane flies on a true track course after a route discontinuity. When the A/P is navigating a great circle path between waypoints, it is essentially flying a true track path in which the target track continuously changes. After completing the last leg, the target track is frozen, so the plane flies along a path of constant true track. The A/P model employed in the FSX/PMDG 777 agrees with this, although this is not definitive proof. It just means that the developers of the software believed the real plane behaved in this manner.
@VictorI: Thanks for the wind information. You’ve been a great help.
@ROB
I’m in my writing-vibe so I’ll respond;)
The main underlying problem IMO with the Bayesian model is that it was only used with data and information then available (before the flaperon find).
It (they actualy) never showed the flexibility to adjust to incoming data later on to allow the Bayesian method to adjust the priority search area.
After the flaperon find it already soon became clear the search area had be more North. Later with all the other debris finds and more acurate drift-studies it became undeniable clear the crash area had to be somewhere North of ~35S.
All this new data and information was never fed into the first Bayesian model I assume.
So then it’s not the Bayesian model that’s wrong but the guys and/or dolls that refused to implicate the new data and information into the model.
I think there it went wrong. A Bayesian model should be an ungoing proces fed with new data all the time. They stuck with their guns it seems and still do. Face value.
And now they hesitatily acknowledge they could have been wrong all the time. At the end of the search. With no options left.
The face of cowards and opportunists in charge.
@Ge Rijn
“The main underlying problem IMO with the Bayesian model is that it was only used with data and information then available (before the flaperon find).”
The flaperon finding was included in the DSTG book.
Looking back over the discussion of the MSFS “Flight” files, I came across this post:
From the paper “Further Analysis of Simulator Data Related to MH370, Victor Iannello and Yves Guillaume, November 29, 2016”:
Supposedly, ZS went through these steps for each of the six flight files that contained the recovered fragments. I think a good question is: what could have been the intention of Zaharie when he created these save-points?
I’m getting confused …………………….