Guest Post: Some Observations of the Radar Data for MH370

click on image to expand

by Victor Iannello

[Note: While attention has been focused on the definition identification of the Reunion flaperon, there remain many other elements of the MH370 case that still require careful attention. Here, Victor delves into what the data that has been released, in particular in the Factual Information (FI) report, to clarify what we know about the first hour of flight after the turnback at IGARI. — JW]


This work analyzes the position and time data derived from the publicly-available radar data for MH370. Some of the findings are:

  • After the turn back towards the Malay Peninsula, the flight path recorded by civilian primary surveillance radar (PSR), civilian secondary surveillance radar (SSR), and military radar are consistent with a flight at a Mach number (M) equal to 0.84 at an altitude of FL340.
  • If the aircraft did fly at a steady M = 0.84, then the timestamps for some of the PSR contained in the Factual Information (FI) are offset by about 35 s.
  • After the left turn at around 17:23:38 UTC, the aircraft might have descended from FL350 to FL340 and accelerated from a ground speed of 473 kn to a ground speed of greater than 500 kn.
  • In the FI, the PSR data between 17:47:02 and 17:52:35 UTC is attributed to the radar site at Kota Bharu, but more likely was collected by another radar site. The PSR data between 17:30:37 and 17:44:52 is correctly attributed to Kota Bharu.
  • In the FI, it is stated that Indonesian military radar recorded MH370 as it traveled toward IGARI but not as it traveled back over Malaysia. One explanation is that Indonesian radar site was powered down after midnight, local time.
  • The sharp turn to the left at around 17:23:38 UTC is unexplained, and could be due to either an inaccurate graphical portrayal of the radar track, or crossing radar tracks from two aircraft.
  • The curve in the radar path close to Kota Bharu can be explained by “slant range” due to high altitudes and close distances.
  • The fuel consumption models which assume that MH370 flew near Long Range Cruise (LRC) speeds and at cruising altitudes between 17:07 and 18:22 are likely accurate.

You can read the whole document here.

54 thoughts on “Guest Post: Some Observations of the Radar Data for MH370”

  1. Victor,

    Thank you for the white paper analysis of available radar data. For me, it explains several nagging questions regarding what is arguably the most documented portion of the post-turnback flight.

    I’d have to check the FI again, but I seem to recall the Indonesian military radar may have been referencing the SSR contact only.

    Peanut questions-

    Could Indo’s PSR have been recorded over on a loop before they realized the need to look at them?

    If my math is correct, 35s at 510kn is about 6 miles (10km). Would radar ‘coasting’ explain the 35s discrepancy of the first 3 tracks?

    Regarding the sharp turn, in the June 26,2014 report Fig. 2 “MH370 flight path derived from primary and secondary radar data,” there are two colors shown for the path. Upon first glance, it would appear that the yellow corresponds to the primary radar, and the pink represents the secondary radar – except that there shouldn’t be any ADS-B/SSR after 17:21. Clearly out of range for the PSR at Kota Bahru, IGARI is at a distance of less than 200km from Bukit Puteri. A turnback at 35,000 ft would have been well within range of that military installation. Frustrating there isn’t more data available from 17:21 – 17:30- especially with hints that there is a Thai detection a few minutes earlier.

    The ‘noise’ regarding flying past BITOD, in my opinion, has been lost in translation. I seem to recall was that HCM lost the SSR signal “somewhere beyond BITOD.” From their vantage point, ‘beyond BITOD’ could very well mean towards IGARI.

    I do still wonder what exactly the Kota Bahru fishermen and the villagers in Kelentan province all saw? Seems like a plane at 35,000 feet would be uneventful- especially if you were hundreds of km away, in an opposite direction, of the known flight path of a plane which was thought to have crashed out at sea? I mean- they were RIGHT underneath it at almost exactly the right time- and that wasn’t known for over a week after their police reports were filed.

    Perhaps the ‘two plane concept’ is related to these sightings. I remember a long time ago Mu was almost (halfheartedly) banned for discussing the sharp turn and this very ‘two-planes concept.’ My, how far we’ve come! I also (halfheartedly) miss Rand’s rambles!

    In all seriousness, thank you to JW for hosting us, and to all of those who have, and many who continue, to dedicate their time and energy to try and bring closure to this tragedy.

  2. I echo @orion’s thanks to @Victor and @Jeff – as well as his suggested implications.

    Re: “impossibly sharp left turn”: in simple terms, don’t the pre- and post-turn data come from two different sources?

    If so, experience tells me to search for a data source inconsistency (time OR location, introduced during capture, processing OR portrayal) before contemplating a second plane. Victor, I deduce from your conclusion that you’re keeping an open mind (as usual), but would you agree “inaccurate graphic portrayal” is just one of several possible “data source inconsistencies”?

  3. @orion: Thank you for your interest. I will try to answer some of your questions.

    1) The fact that Indonesia reported seeing MH370 as it flew towards IGARI but not on the return trip would indicate that data was not overwritten on the loop as the earlier data would have been first erased.

    2) I don’t have a good explanation for the 35 s discrepancy. Perhaps there is some estimation of the tracks as you suggest due to missed detections which leads to time or position skew.

    3) My interpretation of Fig 2 in the ATSB report is that pink path is SSR data from Kota Bharu (KB) until the point labeled “Last secondary radar data 17:22” (which I believe is probably closer to the time 17:21:13). After that, it is military radar data until the yellow line begins around 17:30:37. Then, it is the PSR data from KB. I agree that the color scheme chosen for Fig 2 in the ATSB report is confusing.

    4) I do not put a lot of weight on the HCM ATC saying the plane was lost at BITOD, as it might mean it never reached BITOD as they had expected. Also, there might have been the same “coasting” phenomenon.

    5) I wish I could reconcile the eye witness reports around KB with my findings, but I cannot.

    6) The sharp turn as presented in Fig 2 of the ATSB report needs further explanation by the Malaysians. It is disappointing that they would present a radar track that is physically impossible without any explanation. I have some ideas about this anomaly that I decided to omit from this paper but I would like to present at some future time. Hopefully my speculation won’t get me banned!

  4. Brock,

    For the record, I also deem the ‘second-plane theory’ (whatever it is) as extremely unlikely, and it would seem impossible to weave into what is known regarding other parts of the flight.

    For me, it’s only a token manifested out of the frustration of not being able to reconcile the eyewitness reports.

    I would also echo your ‘data source inconsistency’ should also apply to pre-Penang, and post-Penang (Lido) radar tracks, as they are also likely sourced separately (perhaps even from another country).

  5. @Brock

    Forgot to mention, severe altitude change might also be a possibility for the depiction of the sharp turn.

    Could be caused by, or in response, to Mech Failure/Emergency, struggle in the cockpit, aerial maneuver to incapacitate pax, and/or an attempt to evade radar during the turn-around, etc.

  6. @Victor: obvious question (and perhaps due to be covered by your next paper; if so, I’ll wait patiently):

    IF the track immediately post-sharp left turn is not MH370 – and if that non-MH370 track must fly at top speed to meet the signal data at reboot – then how many more dominos would fall?

    For example: will your speculation abandon the ISAT data altogether, or simply connect it to IGARI via a wholly new path?

  7. (Originally posted in previous thread prior to reading here, where it seems a better fit)


    Thanks for the laurels on the sharp turn. But the honour for the two planes theory goes squarely to Ron Black.

    My contribution then was a theory of a manueuvre of a single plane that could explain the sharp corner in the radar trace.

    I am glad to see that Victor now classifies that turn as unexplained and the discussion here is re-examining that conundrum.

    As I said then (many moons ago), if we cannot explain that sharp corner for a single plane, the reminder of the radar track after it would be irreconcilable with MH370.

    I am not saying it isn’t. But in saying it is MH370, we must explain how that corner was generated from a single plane’s path or dismiss it as a graphics artefact.



  8. @Victorl

    Thank you for the information. I have two comments:

    1. It looks like the path of the plane from the first hard left turn to the last hard left turn (FMT—final major turn) is open to interpretation.

    2. Victor, does your information suggest that the plane’s path to the Southern India Ocean may not be correct?

    Thank you.

  9. @Victor, @orion,

    Re potential banning.

    To put you somewhat at ease, I had an email exchange with Jeff seeking clarification on why being redacted and/or risk being banned.

    The main reason was that it was in the context of wild altitude change discussions, which were deemed impossible to match to the high speeds required to reach the later pings. The discussion of it was seen as a distraction from progressing the cause, hence they were banned.

    The attempt of trying to explain the sharp corner was deemed appropriate (JW, please correct me if I msunderstood). I was redacted, but more as collateral damage to the wild altitude discussion shut down.

    Jeff then also said, the fact that no-one seems to be picking up on the sharp corner discussion indicates that it is not a commonly held opinion of being an issue and I should let it rest. Which I mostly did.

    Hence my joy that it it is now finally being discussed. I had a good rest.

    In any case, I sensed an open mind by Jeff about the sharp corner discussion, but he needed to see a wider interest in it. So I think, having a wider interest now, lets go for it. ;o)!



  10. Dennis,

    I share your interest in the fits to other flights that were shown in the June ATSB report.

    The report claimed to have tested their path prediction methods on 9 previous flights of 9MMRO and 87 flights of other aircraft in the air at the same time as MH370. Two examples were shown, both from ‘sister aircraft’ departing the same day, MH021 and MH009, both travelling north west. The errors at the pseudo 7th arc crossings were 270km and 370km respectively.

    It is not clear whether these examples were relatively poor ones chosen to illustrate the uncertainties in the analysis, or in fact among the better matches.

    What always struck me as interesting was that the true aircraft tracks are relatively straight, but the path generation technique used (I assume error minimisation) seems to have created dog legs which diverge from the true tracks at later times. They would in these cases have been better off forcing a model with limited course changes and accepting a presumably poorer match to the BFO.

    Of course scheduled flights fly relatively straight tracks for a reason, and MH370 could in theory have taken almost any route to the SIO 7th arc within the performance bounds of the aircraft. But it is an interesting illustration of the problems of error propagation with a purely data driven (ie error minimisation) approach.

  11. @Brock McEwen: Let’s gather the data before discussing implications of the conclusions. My guess is that, like many other aspects of this incident, the data will be incomplete and inconclusive.

    @MuOne: I agree that we have to be careful not to read too much into the sharp corner. It is possible that somebody that was interpreting the radar data and preparing the figure might have taken undue liberty in the rendering. As you said, this topic has been discussed before. But that is also not a reason to sweep it under the carpet.

    @Joe T: Other than the unexplained sharp turn, the radar analysis suggests that the fuel exhaustion model based on LRC fuel flow rates as applied to a reconstructed path with a terminus in the current search zone is appropriate. That determination was the main reason that I undertook this study. On the whole, the study does not contradict a path to the SIO.

    @MuOne: I was mostly joking about being banned by Jeff for discussing the sharp corner.

  12. This article reports:

    “The contention that MH370 ‘turned back’ over the South China Sea, crossed the Malaysian peninsula, turned again over Penang and headed over the Andaman Sea stems from the analysis of primary radar recordings from the ATC radars at the KUL ACC and at Kota Bahru on the east coast of Malaysia, as well as apparently the air defense radars operated by the Royal Malaysian Air Force (RMAF) south of Kota Bahru at Jerteh, and on Penang Island off the west coast.”

    Thus, two primary ATC and (apparently) two primary military radars tracked MH370 after its diversion. The reason for the hedging on the military radars is unclear.

  13. @Victorl

    Thank you for your comment. Based on my limited reading of your information I was thinking that the Southern India Ocean was still a go. Your comment confirmed it. Thanks again.

  14. @Victor,

    I have two questions:

    You mention a 35s discrepancy of radar track vs reconstructed model path. Is this discrepancy in the same direction for all segment start and end points?

    Does your reconstructed path apply a constant M0.84 from the sharp turn or do you account for acc/dec from pre sharp turn conditions?

    If the answer to the first question is “uniformly advancing the reconstructed path ahead of the radar traces”, that may indicate a retardation or pause in the time stamp spacing of the radar track (vertical 2D projection onto ground of the 3D flight path) at or around the sharp corner.

    Such a compression of time stamps at that sharp corner would be consistent with my “short sharp dive and turn manoeuvre” hypothesis of many moons ago.

    Can such a manoeuvre be accomplished in and account for a 35s discrepancy between modelled and measured time stamped locations later on?

    For reference, here’s a link to my CAD model of the hypothesized manoeuvre:



  15. Now almost exactly 3 weeks since the flaperon showed up. I’d say, if the French had successfully traced it to MH370 there would have been an interim announcement to that effect? In any case, Malaysia’s early call has done it’s job. For much of the media it’s case closed.

  16. @MH,

    Its only one possibility of explaining the sharp corner. Graphics artefact, two planes, etc. may all be valid explanations (or rather only one of them).



  17. Victor, many thanks for analyzing the “Factual Information” presentation of the radar sightings.

    Don pointed out in May of 2014 that the Preliminary Report suffered from “Inaccurate and inconsistent reporting from air traffic surveillance assets: military PSR, ATC SSR, ADS-B Ground Based Terminals, CPDLC and ADS-C. For example, no ‘benchmarks’ provided to describe expected coverage from assets vs positive surveillance of 9M-MRO. No accurate cross
    referencing of Malaysian/Lumpur Area Control and VATM/HCM Area Control sightings.”

    That observation still seems true, but is aggravated in the Factual Information by the lack of cross-referencing to the Thai sightings.

    (Instead, the Factual Information implies but doesn’t quite say that the Thai radar did not see MH370 after 17:21. But to my understanding, it was most likely Thai radar that tracked MH370 in the Straits.)

    Don also pointed out in June 2014 that, “9M-MRO should have been visible to three RMAF radar installations, reporting via two separate Sector Operations Centres, as it crossed the Malaysian Peninsula. 9M-MRO’s further progress westward over the Straits of Malacca and to the north of Sumatra was within the range of two of those three installations. The extreme NW segment of the track would have been visible to the Western Hill radar station only.”

    As far as I know this is still correct (e.g., it seems consistent with Figure 1.1C), but the Factual Information suggests that radar sightings were only intermittent over the Malaysian Peninsula, e.g., the appearances and disappearances described at the bottom of p.3, Fig. 1.1E and Fig. 1.1F.

    For several months after the disappearance, FlightRadar24 posted a recording of the air traffic in that part of Southeast Asia while MH370 was aloft. Is it too much to ask that for each Malaysian military and civilian installation believed to have detected MH370, the recording be identified and posted?

    Would it be too much to ask to have these tracks plotted accurately and legibly on a detailed map?

    Instead we have low resolution trash like Fig.s 1.1A and 1.1B, which without explanation seem inconsistent with the more detailed maps and narrative in the Preliminary Report and which introduce perplexing and undefined concepts like “Coasted” to describe the movement of the aircraft targets.

    Even with Victor’s much-appreciated clarifications, I regard the radar portion of the Factual Information as an obfuscatory piece of drek (in marked contrast to the rest of the report).

    I believe it is designed to help protect Najib and Hishammuddin from political repercussions from the partial or complete failure of Malaysian military radar to track MH370 and thus help secure them in their extraordinarily lucrative government positions.

  18. Bruce Lamon – It’s been bent around political sensitivities from the beginning. Now in France.

  19. @MuOne: I looked at your graphic. While the dive is possible, I don’t see how the turn at the bottom could be achieved without ripping off the wings of a B777.

    In my model, there is an instantaneous change in speed at the sharp corner. The acceleration, which was probably facilitated by the change in altitude of 1000′, might have started before the turn. I don’t think this would change the results very much.

    You can see in Fig 2 of the paper, which is the figure Jeff included above, that the underlying radar position lags the position from my calculated path. However, the offset is variable and disappears altogether for the segment south of Penang.

    When I first started working on this, I allowed for variable speeds along each segment and tried to exactly match the position and times with the PSR data. The resulting speeds were so erratic and unbelievable that I opted for the approach in which I hypothesized the true air speed was constant and the timestamps were off. I suspect that early media reports of erratic speeds were due to similar analyses. For instance, if you look at the speeds printed in the timestamp boxes from the underlying radar data, you see wild variations in speed between 474 kn to 607 kn! Yet, the overall path from the sharp turn to Penang and then from Penang to the last data point at 18:22 is consistent with a fairly constant speed of M0.84 at FL340. Due to inaccuracies in position and timing, the radar system has limited ability to measure speed over short time intervals.

  20. @Bruce Lamon: I did my best to make sense of the incomplete and inconsistent radar data that was presented. I cannot explain why the radar data was intermittent as the military radar heads at Western Hill on Penang Island and Bukit Puteri in Jertih (south of Kota Bharu) should have together captured the aircraft’s position without interruption if the cruising level was a steady FL340. This is what led to early speculation about rapid changes in altitude.

    Many of these discrepancies could be resolved if Malaysia released the raw military data as many of us have wanted.

  21. @M Pat

    I was disappointed in the lack of commentary and stats on the sample flight paths. It was more or less, take a look and draw your own conclusions. They could have done a whole lot more.

    Certainly the examples presented don’t give me a great feeling relative to the accuracy of the method. It was largely for this reason that my initial calculations did not attempt to “zero out” BFO variations, but rather simply support a directional hypothesis. Of course, I took a little heat over it, and had to go back and do some tweaking. Conclusion did not change at all, and it just generated a bit of flack from the SO. (For Joe T. SO is Significant Offer).

  22. @all

    SO = Significant Other

    Spell checkers are the ABS of word processing. Helps the folks who are inept, and annoys anyone with skill.

    For Joe T. ABS = Antilock Brake System

  23. @Victor Thanks for the work in pulling all this info together.

    I can understand Malaysia sitting on the raw data, but what about the Thais and Indonesians?

    The issues with this data, among other things, was why I suggested in a prior thread going back to the beginning and throwing out all assumptions (“10th man analysis”). I still see the same issues, at least some of which Victor has pointed out in his report.

    To the extent that what we are looking at was driven by data from Malaysia, it looks like “Merely corroborative detail, intended to give artistic verisimilitude to an otherwise bald and unconvincing narrative.” (with apologies to W. S. Gilbert).

  24. @Victor,

    Thanks for taking the time to look at the graphic and commenting on it.

    It was never meant to be taken literally, but rather to demonstrate an in-principle 3D path. There are many variations possible on the actual geometry of such a path.

    For example the vertical dive portion could be much shorter or even shrink to zero length. It doesn’t need to be vertical, but could be at an angle. The turn could start via a left roll during the initial downward arc of accelerating descent and continue through the “recovery arc” at the bottom. That recovery arc could be a much gentler (larger radius) arc to bring it back to within air craft performance limits.

    Unfortunately, I lack expertise or knowledge of performance limits to calculate feasible arc radii, so I opted for the “in principle” graphic rather than attempting to create a potential “within performance envelope” graphic.



  25. @MuOne:

    There are two ways to perform the manoeuvre you depicted. The first one is to push into a vertical dive, roll to the left, and pull up to level flight. Starting from M.82 that would take the airplane well over M=1 and require between 20,000 and 30,000 ft of altitude. Alternatively, the airplane could be pulled up into a vertical climb, rolled right, and pushed over to level flight. That would take the airplane to well below stall speed, and require more than 20,000 ft of altitude.

    I’m not saying that either is physically impossible, but …

  26. @Gysbreght,

    I was afraid, or rather expected, someone would say that. Given wild altitude swings are out that would put that hypothesis into the round file.

    Just to humor me, if you will, if such a manouevre had been performed within the physical possibilities, how long woud it take? Is it minutes, seconds?



  27. @Gysbreght,

    Thanks. In other words, it could be in the same order of magnitude as the 35s discrepancy, considering that only a portion of such a manouevre would have a time penalty on progress along the path.



  28. @Gysbreght,

    Sorry, forgot to add, still in the round file, unless somebody could plausibly reconcile wild altitude swings with still reaching the later time stamps of radar track and pings.



  29. @MuOne,

    I must admit that I didn’t quite get how the 35s was derived, since I only looked at the two graphs in the FI that didn’t include the turn past IGARI.

  30. @Lauren (from the older thread): The 473 kn before IGARI was the ground speed (GS). The 498 kn was the true air speed (TAS). The speed before IGARI was not M0.84. I introduced the change in speed at the sharp left turn and maintained that speed until 18:22. If the plane dropped from FL350 to FL340, gravity would assist the engines in this acceleration.

    Your values for LRC Mach number look about right for FL340, although it is hard to tell because I think the interpolation is not quite linear between FL330 and FL350 as the curves seem to flatten around M0.84. I think the values at FL350 might be closer to the actual values. In fact, as a check, I used the values at FL350, which predict speeds between 0.834 and 0.838 over that interval. The value of the optimum M should be relatively independent of temperature, although the true air speed depends on temperature.

  31. @Lauren: Let me rephrase the first paragraph of my response to make it more understandable:

    The 473 kn before IGARI was the ground speed (GS) obtained from the SSR data. After the sharp left hand turn, 498 kn was the postulated true air speed (TAS). The speed before IGARI was not M0.84. I introduced the change in speed at the sharp left turn and maintained the true air speed of 498 kn until 18:22. If the plane dropped from FL350 to FL340, gravity would assist the engines in the acceleration from 473 kn GS.

  32. @Gysbreght,

    From what I understand of Victor’s paper, his model is based on a constant speed and altitude flight path assumption/hypothesis, the model then iteratively arrives at a best match for them to meet the published facts of path, path segments, start point (sharp corner) and end point (18:22) amongst others.

    I now, after Victor’s reply, understand the 35s discrepancy to be the difference of predicted location vs measured location of the first radar track segment.

    My thinking here is that if Victor’s model “looses” the 35s or thereabouts at or around IGARI, i.e. start the model at T=T0 + 35s, the model would/could find a slightly higher speed, which in turn would stretch the predicted path segments on the SW leg, and possibly much better fit each individual segment.

    Or, in reverse, the model could be fixed at the start point of the first radar segment of the SW leg and the 18:22 point, optimise speed and altitude, then back calculate the time required through the turn back and arrive at a dT at/around IGARI.

    That dT could give us some limits as to what could have happened around that sharp turn.



  33. @MuOne: The true air speed is found by pinning the position and time at the sharp turn and at the last radar point. Then, the turns are added to match the path (without trying to match interim times).

    If a 35 s adjustment is added to match the times for the first PSR segments, i.e., eliminate the offsets, then an offset is CREATED south of Penang.

    I cannot find a way to match the times for the shorter PSR segments without varying the air speeds to unrealistic values.

    I have no doubt that at the KL ACC, data appears on screens near real time. After all, you can’t control traffic with a 35 s delay.

    However, how the data is timestamped at the time it is recorded is a different matter. After all, we are really looking at least 3 data sets: civilian SSR, civilian PSR, and military PSR. The clocks for the civilian SSR and military PSR look synced. The clock for the civilian PSR looks 35 s advanced.

    I suspect the last PSR segment, which is attributed to the Kota Bharu civilian radar head, is actually from Butterworth, as it is in range of the head, and that would explain why the offsets seen in the first PSR segments disappear south of Penang.

    It’s just a theory and it’s all confusing. But trying to match the time and position for the 4 PSR segments produced unrealistic speeds. I believe that others have discovered the same thing.

  34. @ Gysbreght, MuOne, Victor, Lauren, others

    Regarding the extreme dive or climb at the IGARI turnaround- wouldn’t either one of those maneuvers have triggered an EHM report, with potentially ‘extreme’ data- if ACARS had been on at the time?

    I would assume an extreme dive would put the plane below the PSR horizon at that distance, so if radar data showing the turnaround does exist, then perhaps the climb scenario might have been the most likely to be detected?

    Another potential explanation for the sharp turn could be that the Fig 2 of the ATSB June report might depict a more ‘raw’ condition, which was subsequently refined- similar to the straightening of the apparent curved path at KB, once the slant angle correction was applied.

    Another glaring inconsistency I noticed between Fig 2 of the June report and Fig 1.1E-F in the FI is that the June report shows a continuous path- without the series of segments and gaps approaching Penang. Interesting to ponder the data refinement process which might lead to that outcome.

  35. @Victor – why does the very first time stamp appear to be off by an hour?

    As for the sharp turn, if the plane had made a slight right well before IGARI, followed by a left just before IGARI, isn’t the turn possible? It seems that if it entered IGARI on a more northerly or northwesterly course, and then reached the 17:30:37 point on a more southerly heading, the turns become slightly more plausible, and coupled with some change in altitude, and topped with a graphical artifact, even more so?

    Basically, can’t the turn be made with a combination of 1) missing data points, 2) a right turn, left turn and then another right, 3) an altitude change, and 4) a graphic error? All seem plausible if the plane was having an emergency. If I recall correctly, the turn at IGARI has been described as either left or right on different occasions.

    @Dennis – nice work with the TLAs!

  36. @orion: I have always assumed that ACARS was disabled about the same time as the transponder.

    A reason for the sharp turn might be that points were missing because the military radar was at the limits of its detection capability. If the data after IGARI was collected from Western Hill on Penang, that would be the case based on a terrain obstruction study that a fellow IG member has performed. (I am seeking permission to share these results.)

    @JS: I cannot explain the data with the ID P1776, which has nomenclature consistent with PSR data, a timestamp that would put it near the first PSR for the first segment from Kota Bharu, yet is positioned at KLIA. I opted to not discuss this in the report because I was not sure it was interesting to anybody but me. I am glad you caught it.

  37. @Victor, very interesting. I had interpreted it as a bad time stamp, but from your explanation I see that it’s far more likely a bad location. But that looks even worse. How did it get pointed at the wrong spot? In order for it to be associated with the flight in the first place it had to be very near the same track as another primary return, right? Or it would interpreted as a different plane?

  38. @Victor,

    Thanks again for your reply.

    You say that an introduced offset at IGARI creates an offset at Penang. Fair enough, but every segment other than Penang has offsets. I don’t see how an offset at Penang would be worse than one NE of KB.

    Eyeballing the graphic (its pretty fuzzy on my iPad, so wasn’t sure if I distinguish black from yellow or red correctly) a tiny stretch of a couple of percent in length, hence speed, would, on average, better fit each individual segment. Would a couple of percent produce unrealistic speeds?

    In my view it would be more plausible to assume constant speed and altitude, which are representative of the overall SW leg, at any of the segment start and end points, or somewhere in the middle of the whole SW leg, rather than near the unexplained sharp corner.

    So instead of pinning the model at the sharp corner, it could be pinned at a representaive mid SW leg point (e.g. mid point in space and time between the first start and last end point) and leave the turn back point floating in time.



  39. An explanation on Long Range Cruise speed: The speed defined by aerodynamic and engine characteristics increases with weight at all altitudes. However, the LRC speed scheduled in the table is subject to a maximum speed of 325 kIAS or M.84, to respect a minimum margin relative to the maximum operating speed Vmo/Mmo of 330 kIAS/M.87.

  40. Victor Iannello posted above: “The sharp turn to the left at around 17:23:38 UTC is unexplained ”

    If instead of a sharp turn a normal turn at a turn rate of 1 deg/s is assumed, it would have to start at about 17:21:50 UTC to reach the point at 17:30:37 with 505 kt GS after the turn.

  41. Here is a question – what portion of the diversion route (up to the 18:22 loss of radar contact) is consistent with having been flown by following waypoints, and which waypoints would have been used? This exercise is made difficult due to the irritating inconsistencies in all the different maps that have been released over time.

    We know, for example, that the last part of the diversion, as depicted by the Lido radar image, could have followed VAMPI-MEKAR and possibly on to NILAM. However, Figure 1.1B, along with the caption, seem to show the route passing 10 NM North of MEKAR. Which, if either, is correct?

    There are no waypoints in skyvector that are 6 NM South of Penang airport; however, there are lots of “fixes” in the navigational database for the airport in that area, any one of which could have served that purpose.

    What else would have been needed?

  42. @VictorI,

    I performed an analysis of the radar track about 11 months ago back in September 2014 in my paper “The Location of MH370”. Look at Section 5.3, and in particular see Table 5.1 “Radar Track Analysis”.

    This paper can be found at:

    In that paper I derived times and locations for post-ACARS events from 17:20 to 18:22 UTC. I determined that the radar data available at that time (which did not include the “Factual Information” which became available about 6 months later in March 2015) supported a strong case for a constant post-diversion (17:22 – 18:22) air speed of 500 +/- 3 knots. Your recent result of 498 knots (M0.84 at FL340) is consistent with my result from last year.

    A comparison of event timings is also of interest:

    Arrive IGARI: BLU (17:20:39) Victor I (17:20:31)
    Begin Diversion Turn: BLU (17:22:20) Victor I (17:22:52)
    End Diversion Turn: BLU (17:24:39) Victor I (?)
    Arrive Kota Bharu/GOLUD: BLU (17:36:38) Victor I ( 17:36:51)
    Arrive 6NM S Penang VPG: BLU (17:52:37) Victor I (17:52:35)

    The largest difference between our event times occurs for the start of the diversion turn, where the time in your paper is 32 seconds later than mine. The time differences for the other events are much smaller (2 – 13 seconds). My diversion turn time estimate only assumes (1) the ATSB radar map had accurate positions, (2) the ACARS data had accurate positions and times, and (3) the speed was steady from 17:07 to 17:22, at which time the plane turned and accelerated to a higher, constant speed. In addition, I used a 1 degree per second turn rate for all the turns, including the diversion turn. In my opinion the “impossibly sharp turn” is not real but rather just an artifact of time late errors in some of the recorded radar data near 17:23 UTC.

  43. @DrBobbyUlich: That is helpful to see that our results regarding true air speed are similar, especially in light of your not having the PSR timing details from the FI at your disposal. It all suggests that the conclusion of a flight at M0.84 and FL340 is reasonable.

    I started this effort without assuming a constant Mach number, attempting to derive the speeds based on the timestamps of the PSR data. The erratic and unreasonable speeds showed me that there was not sufficient accuracy in the data over small segment distances to obtain meaningful results, and I opted for a constant Mach number over a longer path.

    The sharp turn anomaly very well could have been created by missing radar data at the turn. I am still looking into this.

Leave a Reply

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

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