Why We Need Data Withheld by Malaysia

Radar data shown to MH370 families but never released.

It’s now almost three years since the disappearance of MH370. After an exhaustive underwater search in the Southern Indian Ocean (SIO) of around 120,000 square kilometers, we are still unable to locate the fuselage of the aircraft. The main evidence we have to locate the wreckage are the communication signals between the aircraft and a ground station in Perth, relayed by an Inmarsat satellite in a geosynchronous orbit above the Earth. From these signals, assuming they have not been manipulated, we do know that MH370 turned south and ended its flight along an arc in the SIO we now call the “7th arc”. We have also recovered debris from the crash that has washed up on the shores of several countries in Eastern Africa. Although the satellite data and the timing and location of the recovered debris support the theory that aircraft terminated in the SIO, they do not provide enough precision to locate the aircraft.

In order to estimate where MH370 lies in the SIO, a number of us, including the official investigative team and we in the MH370 Independent Group (IG), have used mathematical models to reconstruct paths that replicate how MH370 might have been flown. These models incorporate a number of assumptions, some of which we can be fairly certain are correct, and others that we are less certain about. For instance, assumptions about the performance of the aircraft, including minimum and maximum speeds, fuel consumption, and autopilot behavior, are all fairly well known. Other assumptions about the “complexity” of the path are less known. There are some that maintain that the simplest of reconstructed paths, i.e., paths with the fewest numbers of changes in speed, direction, and altitude, are the most likely. In fact, the current underwater search area is derived from this assumption. But with the failure to find the wreckage, it is time to re-evaluate this assumption. After all, the aircraft did not follow a random path. Rather, the aircraft likely was actively flown by the pilot (or pilots) at least until it made the final turn towards to the SIO. If the diversion was intentional, then there was a reason for the pilot choosing to fly along the particular path. As a result, models that reconstruct the path using a series of random events are likely to fail because they do not account for the motivation and intentions of the pilot.

If we are going to consider complicated paths that might include turns, holding patterns, changes in speed, and descents, the search area grows to an unmanageable size. We therefore have to introduce other assumptions and/or constraints to limit the size to something searchable.

Fellow IG member Richard Godfrey and I recently wrote a paper in which we predict a crash site for MH370 using certain data found on Captain Zaharie Shah’s home simulator. The simulator data include six position coordinates from a flight the Captain created using Microsoft Flight Simulator. The positions first show the aircraft on the runway at Kuala Lumpur International Airport, and then progressively airborne in the Malacca Strait, the Andaman Sea, and the SIO. The simulation ends with the aircraft running out of fuel in the SIO. Recognizing that the positions in the Andaman Sea and the SIO align with an ice runway at McMurdo Station, Antarctica, Richard and I hypothesize that the Captain used this same final waypoint when he programmed the flight computers of MH370. By combining this final waypoint with the available satellite data, we predict a terminus on the 7th arc well north of the current search. This theory also predicts that the aircraft entered a holding pattern near Car Nicobar Airport in the Andaman Sea before turning towards the SIO.

The failure of the underwater search and the modeling work of Richard and me shows the importance of using all the available evidence in defining possible crash sites. However, much of the evidence has never been made public. For instance, the data obtained from the Captain’s computer was from a secret Royal Malaysia Police (RMP) report that was never officially released. We are only aware of it because the report was leaked to French media organizations.

Australian reporter Marnie O’Neill recently asked fellow IG member Don Thompson and me to make a list of what important evidence was being withheld, and she used this as the basis of a story for news.com.au. Here are some examples of how Malaysia is showing that it is not fully committed to the finding MH370:

Inadequate Response to Disappearance

  • After the disappearance, there were only two attempts to reach MH370 using SATCOM voice (18:40 and 23:14 UTC).
  • There was no attempted military air intercepts as MH370 turned back and flew across the Malaysian peninsula despite plane detection by military radar in real time.
  • There was a delay of four hours after the disappearance before search and rescue (SAR) efforts began.
  • There was a delay of four days (March 12, 2014) before search shifted from South China Sea to Indian Ocean, despite having radar data.

Denied, Omitted, or Ignored Data

  1. Radar captures of MH370 in the Malacca Strait were shown to the victims’ families in Beijing on March 21, 2014, but radar captures between 18:02 and 18:22 UTC were never shared with the ATSB. (See figure above.)
  2. The partial data set of raw radar data made available to the ATSB was never shared publicly.
  3. The ATSB report released in June 2014 includes statements about a radar capture of MH370 at 19:12 UTC in the Andaman Sea. Later, the ATSB acknowledged the data to be from Singapore radar, and considering the distance from Singapore, likely was from an aircraft with radar capability operating in the Malacca Strait or Andaman Sea. No mention of this data is included in the Factual Information released in March 2015, yet if this data exists, it would place the terminus in the SIO much further north than where MH370 was searched.
  4. The existence of telephone records indicating a connection of the First Officer’s cell phone to a tower on Penang Island was first denied by Malaysia and not included in the Factual Information report released on March 2015. The secret RMP report has detailed information about this connection.
  5. The simulator data recovered from the Captain’s computer suggest a simulated flight with points in the Andaman Sea and the SIO. Malaysia first denied the existence of this data and did not include the data in the Factual Information report released in March 2015. The secret RMP report included some information about the simulator data, but the details about how the data was extracted and analyzed are unknown.
  6. The secret RMP report documents WeChat activity on the Captain’s cell phone while MH370 was lined up on the runway, only one minute before takeoff. The details of this activity are not presented in the RMP report. No mention of this data was included in the Factual Information report released on March 2015 despite its extreme relevance.
  7. Malaysian authorities have shown no timeliness in retrieving possible MH370 debris recovered from the shores of Eastern Africa.

As Don said in the article, “My own ‘hot button’ is that military long-range air defence surveillance data from assets operated at seven, possibly even eight, sites across four nations is absent from the data set available to ATSB.”  Don explains that “[t]hose [sites], all within range of the flight path MH370 is believed to have taken, are located at Lhokseumawe, Sabang/Pulau We and Sibolga in Indonesia; Car Nicobar and Port Blair in the Indian Andaman Islands; Khok Muang and Phuket in Thailand; and Western Hill, Penang, Malaysia. Any one of them, or all collectively, could provide the vital clues to the plane’s whereabouts.”

If we have any hope of re-starting the search for MH370, we need all the available data so that we can properly constrain the models we use to predict the terminus. The time for Malaysia to release all it has is long overdue.

91 Responses to “Why We Need Data Withheld by Malaysia”

  1. Kenyon says:

    @Victor, thank you for the new blog site. I second your call for more data and increased transparency from Malaysia.

    Additionally the French need to release their official forensic analysis of the right Flaperon. My latest Flaperon Failure Analysis can be found at the following link. I would be thrilled to toss it in the waste basket of ‘nice try’ should the French or others present the official analysis.


  2. ALSM says:

    Thank you for setting up this blog, Victor. Nicely done. RPI site also looks nice. I agree with all you stated in your opening post.

    I respect the need for some confidentiality on much of the personal material in the RMP leaked documents, but Malaysia should have made public ALL the technical and circumstantial information that is key to understanding the likely path, possible equipment maintenance problems and possible motivations. In particular, all the details about the 2 cockpit initiated cell phone calls at 16:39 and 17:52 should have been in the FI. The “10 second radar data” provided to the ATSB should also be made public, along with an explanation for why the 18:02-18:22 data was withheld even from ATSB.

  3. Victor Iannello says:

    @ALSM: Thank you, Mike. I wish we also knew more about the WeChat activity that occurred while lined up on the runway and one minute from takeoff.

    @Kenyon: Thank you for your comments, Tom. I hope your flaperon analysis gets more looks.

  4. Trip says:

    I appreciate your work around China’s misdirection. I was wondering if there was such a data base for other countries response. I always think of Sherlock Holmes and the case of barking dog. What was curious was that the dog didn’t bark when expected. Three categories of information: verifiable truth, unknown veracity and verifiably false. In addition to the missing pieces of information listed above where else did Malaysia attempt to mislead? What did Russia, the US, Thailand or Indonesia say? Was anyone transparent in their willingness to help solve this?

  5. gecina says:

    Thank you for doing this, Victor. Just a question on this part from Item 5:

    “… The secret RMP report included some information about the simulator data, but the details about how the data was extracted and analyzed are unknown.”

    In this link from Jeff’s blog, the software used to examine Shah’s hard drives is listed, as well as some of the methodology:


    “Exhibit 6 TOTAL
    MK 20 Inspection date and time on the computer via the CMOS settings.
    MK 22, MK23, MK24, MK25, MK26
    • Forensic Tool Kit (FTK) 4.2
    • FTK Imager 3.1
    • FTK Registry Viewer 1.6
    • Internet Evidence Finder 6.1
    • EnCase V6.18”

    “Each disk has the image using FTK Imager to ensure data integrity
    which will be analyzed in the hard drive. Before an image is used, the test data integrity verification is carried out to ensure that the image is the same Hash Value and not contaminated.”

  6. Henrik Rydberg says:

    Thank you, Victor, for opening up this channel, and for your creative, detailed and advanced efforts to solve this mystery. I imagine this site will soon be filled with people that have truly missed the opportunity to collaborate with their peers on this level.

  7. Richard says:

    Excellent Web Site and first Post!

    I completely agree we need full disclosure and total transparency on the part of the authorities.

    We also need open minds to consider that the final major turn south may have occurred much later than the first satcom call at 18:39 UTC and even up to 19:36 UTC. The Inmarsat satellite data at 18:39 UTC can mean a turn southwards but can also mean a descent as shown in our joint paper that you reference.

    Extrapolating back from the Inmarsat satellite data between 19:41 UTC and 00:11 UTC, places the final major turn close to Car Nicobar airport and the so called N.W. Point that the Malaysian Preliminary Report and ATSB Report reference without explanation.

    This in turn places the MH370 end point much further north, than the ATSB search area.

  8. Victor Iannello says:

    @Trip: I don’t know of any database that chronicles the response by country. Although Malaysia seems to be the biggest offender of withholding information, other countries also could have more information to share.

    @gecina: We know what tools were used, but we don’t know, for instance, why the MS Flight files were not recovered in whole. We only have fragments of each file.

    @Henrik Rydberg: Thank you for your comment, and I hope you come back often.

    @Richard: As you know, we are in violent agreement. There is no reason for the “NW Point” to be shrouded in secrecy, especially since the ATSB has already disclosed radar capability in that area. Either the plane was captured by radar in the Andaman Sea or it wasn’t.

  9. Sabine Lechtenfeld/littlefoot says:

    Victor, thanks for giving us a new platform for discussing mh370. I hope it will inspire informed and rational discussions.
    I concur wholeheartedly with all major points of your first post. Without getting the withheld data, further progress seems to be difficult.
    Richard (Godfrey) and Victor:I find your paper which evaluates the possibility that the plane came down much further North on the 7th arc for various reasons, very interesting. Reconciliating the existing data with the deliberate actions of an abductor/perp who had a plan and knew what he was doing, seems to be a move into the right direction, even if the exact motivations for crashing the plane into the SIO instead of opting for a more spectacular and overtly political action, are still unclear to us.

  10. Victor Iannello says:

    @Sabine: Thanks for your comments. If a radar target really was captured in the Andaman Sea, it not only invalidates the underwater search area, but also raises serious questions about the motivation of the pilot. At a very least, officials should explain the meaning of the NW radar point that the ATSB presented in its June 2014 report, later acknowledged to based on “clearance” from Singapore radar, and then never discussed again.

  11. Victor Iannello says:

    On Jeff Wise’s blog, @TBill asked me, “Can you elaborate briefly on your comments (above on this blog) about ATSB not getting the radar data all the way out to 1822? Believe the FI says MH370 went to MEKAR+10 so ATSB would have known that much? Are we saying ATSB/IG knew generally about it from the Lido presentation but were not given the raw data which might be important?”

    It is my understanding that the ATSB was unaware of the radar data in the Malacca Strait shown to the families at the Lido Hotel in Beijing, and the ATSB and the DSTG never used it in their analyses. The data that Malaysia supplied to the ATSB had the next-to-last capture at 18:02 and then a final capture at 18:22. The missing data between those two times that is shown in the Lido Hotel image establishes that MH370 joined airway N571 at VAMPI. To me, this is an important clue as to how the plane was flown that should not be ignored. In fact, in the FI from March 2015, the following text is included in Figure 1.1B: “The primary target (military radar) appeared to track west-northwest direction joining RNAV Route N571 at waypoint VAMPI then to 10Nm north MEKAR”. This seems consistent with the Lido Hotel image.

    If the radar captures shown in the Lido Hotel image are valid, why were they not publicly released, and why were they withheld from the ATSB? If the radar captures shown in the image are invalid, why was an intercept of airway N571 at VAMPI referenced in the FI?

  12. Sabine Lechtenfeld/littlefoot says:

    @Victor, if the NW radar capture of alleged Singaporean origin over the Andaman Sea reported by the ATSB in June 2014 isn’t just a figment of someone’s imagination, it raises indeed several serious questions concerning the motives of the pilot for straying so far into a Northwestern direction before turning the plane around. And,as you say, it would invalidate the underwater search area. But it also raises serious questions concerning the decisions of the ATSB who apparently ignored this radar capture when they defined the underwater search area, although they themselves had reported the radar point. Why was it never mentioned again and why wasn’t it used for defining the search area? There may be valid reasons, but we need to know.

  13. Victor Iannello says:


    To recap, the following statement was in the ATSB report from June 2014: “The aircraft passed close to a NW point at 1912”.

    When Niels Tas asked the ATSB about this statement, he received the following reply: “The NW point at 1912 was an assumed theoretical location at 8° 35.719’N, 92° 35.145’E initially chosen to provide clearance from the known radar sources (mainly Singapore). A line from IGREX to the 1912 point was used as an upper bound for the airplane performance work after loss of radar contact (the min flight distance would be turning south right after loss of radar). This point did not affect the Doppler analysis, just the fuel burn, which affected the range measurements. Analysis had included using the upper bound (IGREX/1912 point) and the lower bound (direct from the 1822 point)…”

    I tried getting a clarification from the ATSB, but they only told me that the information provided to Niels was the same information provided to the DSTG. I suspect the source of the information is a close ally of Singapore. On the basis of the information, the DSTG opted to not include the NW point in their analysis that was released in December 2015.

    Perhaps the DSTG chose to ignore the NW point because it was ambiguous, and because it was not consistent with their interpretation of the BFO at 18:40, which for level flight would imply the plane was already flying south. At the time that the previous ATSB report was written in June 2014, the BFO at 18:40 was still being ignored. That could explain why the NW point was first considered in the ATSB report and then later ignored in the DSTG report. That would also mean that the possibility that the NW point was a true radar capture and there was a descent at 18:40 was not considered by the DSTG.

  14. Sid says:

    Thank you for setting up the site. The questions being asked should not trouble the authorities unless there is something to be hidden.

    The official gap between 18:02 and 18:22 is inconsistent with the usual primary radar tracking concept where the time series of radar returns is associated with a track. Transponder data is not used (and in this case there is none). How can one confidently associate the 18:22 point with the 18:02 point without knowledge of the intervening track?

  15. Victor Iannello says:

    @Sid: Welcome, Sid. I hope you comment here often. And in your first comment, you raise an excellent point.

  16. DennisW says:

    Just a hello for now.

    To Victor’s point above, data is so very important. While the DSTG Book did not change the course of the search, the glimpse of the previous flight paths plus ACARS data had a definite impact on our analytics. Had this data been in the public domain earlier, I really believe we (the collective we) would have had a different view of what the ISAT data, BFO in particular, could actually tell us.

    I can think of no valid reason why the data from previous flights of 9M-MRO has been withheld. Likewise the radar data from the accident flight.

  17. Victor Iannello says:

    @DrBobbyUlich: Thank you for your comment. Political rants are allowed only if they are relevant to finding MH370. One of my goals in creating this blog was to encourage more technical discussion. I hope this forum also allows more technical interaction between IG members and important contributors like you. After Duncan Steel suspended activity on his blog, I think there has been a void in this regard.

    I agree with everything you said about the behavior of the autopilot while in LNAV mode with a lateral offset and reaching an EOR. I look forward to hearing back after you incorporate an accurate wind model.

  18. Victor Iannello says:

    @DennisW: Welcome to the blog. I agree with all you said.

  19. Vector-1 says:

    Great work on the blog and the details in the article.

    The big gap in the radar could well be due to the unreliability of primary radar.

    Clearly everybody believes the missing data should be released. However I am not sure if you the realise the radar data may well be included within the RMP report, including the missing radar data levels between 17:39 and 18:25.

    There are levels given there of 44,700ft and elsewhere three separate levels thousands of metres apart at exactly the same time. However if these really are the readings, they are clearly impossible and a decision could well have been made to omit them from the FI report, otherwise it would cast doubt on just about everything else in the report.

    If these are the recorded levels, they should still be published and leave it to others to find a logical explanation. The 45,000ft was mentioned by the New YorkTimes as early as 14 March and presumably the police would have access to the radar data in May 2014 when preparing their report on the flight crew and would not be particularly interested in the significance of the levels and just reported the figures they had.

    On the assumptions it was a deliberate act it is clear there are two reasons it would have been flown to the SIO and not crashed at the first opportunity like other suicide by plane examples or used in a spectacular event. The first possibility was to create the world’s greatest aviation mystery, the second was to ensure there could be no blame attributed to anybody to shield the family of the guilty party.

    One question: what is there to show the Malaysians were monitoring the radar in real time after 17:21?

    I have not seen anything other than evidence they seemingly extracted the route from radar recordings and altered the search location accordingly in the following days as more information became available.

  20. Victor Iannello says:

    @Vector-1: Welcome, and thank you for your comment.

    The radar data probably was included in the full RMP report as an image of the DVD cover was shown. However, I don’t believe that the radar data was leaked to French media, and so it remains unavailable to the public.

    In the RMP report, the radar data related to the cell phone connection near Penang showed an unrealistically high altitude of FL447, as you say, and high speeds up to 578 knots. The high altitude is surely due to miscalibration of the radar head, and I suspect the high speeds are due to intermittent and/or inaccurate position information. The poor quality of the data could explain why Malaysia is reluctant to release the raw data and also why the DSTG chose to heavily filter the raw data.

    Also in the RMP report Folder Appendix (1), Appendix J-1, there is a timeline of the crew’s activities. On p127/227 in the PDF, there a figure with an element labeled “FARIQ AB HAMID DETECTED AT BBFARLIM2 LBS ONLY” with a handwritten note that says “radar bukit bendera (army) 35100 feet”. This confirms that the radar head was at Western Hill, Penang. But the fact that the altitude is 35100 feet rather than FL447 suggests that the Malaysians at some point might have adjusted the radar data to reflect the miscalibration of the radar head at the time that MH370 captures were made.

    We don’t know whether or not the Malaysian were monitoring the radar in real time. They have made conflicting statements about this. Hishammuddin Hussein said they detected the plane in real time but regarded it as friendly and so did not respond with an intercept. But the delayed and misplaced SAR actions were initially explained by the failure to detect the plane in real time. The Malaysians seem to chose whatever explanation is convenient at the time.

  21. DennisW says:


    “Hishammuddin Hussein said they detected the plane in real time but regarded it as friendly and so did not respond with an intercept. But the delayed and misplaced SAR actions were initially explained by the failure to detect the plane in real time.”

    My personal take on it is that the Malays knew full well the plane was diverted Westward, whether by radar or other means. The SAR activities in the time immediately following the diversion were bizarre by any measure. Bizarre in that these activities were tightly coordinated by high level Malay officials, and not the people who would normally be running an operation of that type. It would be analogous to Barack Obama getting involved in a SAR mission in the US. Even ICAO was critical of the Malay response, which as I understand it, is quite unusual for that organization.

    In any case, I think the Malays were happy to be searching in the South China Sea because they knew the plane was not there. Along comes Inmarsat to spoil the narrative. Oops! The Malays had to make a diving catch, and embarrassed themselves in the process. It is true, as you say, that the search in the South China Sea was either the result of not having the radar data in real time (or near real time) or suppressing it. I tend to believe the latter.

    Of course, none of this speculation is helpful relative to finding the aircraft. It is unlikely we will ever know what actually took place during and shortly after the diversion unless a whistle blower comes forward.

  22. Victor Iannello says:

    @DennisW: The 1MDB scandal demonstrates that not all that Malaysian officials try to keep hidden remains hidden. Time will tell.

  23. DrB says:


    Thanks for setting up this blog. I do hope all political rants will be excluded.

    With the assumption of a single FMT, I have been able to find only one candidate route – a Constant True Heading (CTH) at Holding speed passing near IGOGU and ANOKO.

    With only one candidate route identified, two quandaries immediately arise:
    a) Why or how would the flight crew use CTH?
    b) Why or how did Holding speed result if there was no racetrack pattern flown?

    I believe both these unusual and unexpected results could be unintended consequences of an unsuccessful attempt to loiter in the vicinity of WITT in Banda Aceh.

    As discussed previously, I have recently evaluated the possibility of a route that passes ~10 NM to the right side of ANOKO. Here is a link to a figure showing this route near the FMT:


    It is possible to create this candidate route with no flight crew inputs to the FMS after ~ 18:26 UTC. This is much earlier than I had expected, and it allows up to 15 minutes for incapacitation to occur before the first unanswered phone call at 18:40.

    The following is a notional sequence of events that would, I believe, produce this resulting route:

    1) There are only three pilot entries to the FMS needed after 9M-MRO began following N571 near Pulau Perak.

    2) At that time the speed was ~ LRC and the altitude was ~ FL360 (= even for westward travel and possibly as low as FL340). Those were set soon after the time of the initial diversion at 17:21. The high speed across the military radar track indicates LRC was used.

    3) After reaching MEKAR on N571, suppose the Active Route 1 Legs page contained the following waypoints: NILAM / IGOGU / ANOKO. The first two waypoints are just ahead on N571. ANOKO may have been added as the first pilot action, replacing LAGOG and the remainder of the N571 waypoints after IGOGU. Of course, we don’t know if the full N571 airway was selected or the first few waypoints were individually entered. For the scenario I describe here, it makes no difference which was actually done. In either case, any waypoints after IGOGU were replaced with ANOKO.

    4) ANOKO might have been selected because it is the first point in the ANOKO2C Standard Terminal Arrival Route (STAR) to WITT. It seems reasonable to me to use ANOKO as a fix for a loiter there while technical problems are being addressed, because that would allow rapid entry to the STAR for a possible emergency landing at WITT if needed. If the aircraft flew up the Strait of Malacca rather than landing at Penang, perhaps because of an unusual technical fault, there would be no apparent reason to continue on N571 past IGOGU, but there could be a reason to turn off N571 to ANOKO and loiter there.

    5) The second pilot action is to add a Holding pattern to the Active Route 1 Legs page with ANOKO as the fix.

    6) These first two steps were completed prior to ~ 18:25. Now the Active Route 1 Legs page contains NILAM / IGOGU / ANOKO / Holding (at ANOKO).

    7) The third and last pilot action needed is to initiate a 10 NM lateral offset to the right of the current track from NILAM to IGOGU (N571). This was executed at ~ 18:25:40, just after passing NILAM. Perhaps the intent of the offset maneuver (a SLOP) was to avoid the possibility of a mid-air collision after the power restoration, performed several minutes prior, did not restore TCAS and radio communications.

    8) Now 9M-MRO is programmed to fly 10 NM to the right of ALL the remaining waypoints now in the active list: IGOGU and ANOKO.

    9) When 9M-MRO is approximately “abeam” of IGOGU (which is now ~ 10 NM on the left side of the aircraft), the FMS begins the FMT and turns toward the point 10 NM to the right side of ANOKO. The FMT begins at ~ 18:38:35 and it is completed at ~ 18:40:28.

    10) At 18:39:56 the first satellite phone call is received but not answered. Note that the FMT is actually completed during the phone call.

    11) At approximately 18:41:40 the FMC automatically begins to reduce speed to the current Best Holding value of ~ 257 KIAS (at the current weight of 207 MT and at FL360) so that it can enter the upcoming Holding pattern at the appropriate (reduced) speed. No pilot action is needed for this to occur; it is done automatically by the FMC. The VNAV now controls the speed to follow the Best Holding KIAS as the weight changes (i.e., it will slow down as fuel is consumed and the weight is reduced so as to achieve maximum endurance).

    12) At about the same time, which is 2 minutes before reaching the ANOKO + OFFSET position, an END OF OFFSET message appears on the CDU informing the pilot that the offset must be removed in order for the subsequent Holding pattern to be flown. This is because the Holding pattern will be cancelled by the FMC if the route offset is not equal to zero when the Holding fix (ANOKO) is reached.

    13) There is no pilot action at that time to zero the lateral offset, and at ~ 18:43:40 an END OF OFFSET error occurs, cancelling the Holding pattern (but not changing the speed, which previously began to follow Best Holding using VNAV).

    14) Immediately following, a second error, “END OF ROUTE”, occurs because there are no entries in the Active Route 1 Legs page after the (now cancelled) Holding pattern.

    15) When an END OF ROUTE (EOR) error [as well as a ROUTE DISCONTINUITY (RD) error] occurs, the FMC changes LNAV to a default lateral navigation method which maintains constant heading. There is some debate about whether this heading is constant in true bearing or in magnetic bearing, but there seems to be agreement that a constant true heading (CTH) is possible, either by design in the FMC or by having the NORM/TRUE switch set to the TRUE position. I can say that I have been unable to find a candidate constant magnetic heading route. The only candidate route seems to be at constant true heading, and this, in fact, is what the best PC-based simulators do for both EOR and RD errors. I am not aware of any extended, documented simulations being done on a high-level B777 simulator to determine which heading mode is actually utilized, but this kind of test would settle the issue.

    16) The true bearing between IGOGU and ANOKO is exactly 180.0 degrees south. This would also be the true course followed by 9M-MRO up to ANOKO, even with a lateral offset.

    17) Once the EOR error occurs, the heading is maintained at the value it had at the time of the error, which is slightly LARGER than 180 degrees true. This is because at that time there was a slight crosswind from the right side (west), causing the aircraft to have a heading of ~ 181 degrees in order to fly the track at exactly 180.0 degrees true.

    18) After CTH navigation ensues, the aircraft begins by following the due south track fairly closely – until the wind changes.

    19) Soon after the EOR error, the wind shifts from westerly to easterly and strengthens. This begins pushing the aircraft to the west side of a due south track, since it is no longer following a “straight” track but simply keeping its nose pointed in the same direction.

    20) As the crosswinds change along the route, the aircraft drifts laterally according to the strength and direction of the crosswinds.

    21) The easterly winds peak near 25 knots before subsiding at about 18S latitude, where the winds switch back to the west. Strong westerly crosswinds up to ~ 60 knots then produce a path that changes course by up to ~ 4 degrees when in the SIO.

    22) The TAS slows from ~ 455 knots to ~422 knots as the fuel is consumed and the aircraft weight declines, with second-engine fuel exhaustion occurring at ~ 00:17:29.

    The constant heading navigation methods are the most difficult to model using a computer, and they are also less accurate. Fortunately, fairly accurate global wind data from satellites are available every three hours for multiple altitudes, and this will allow modeling of the lateral drift and ground speed variations. Initial results using a 2-D wind model are encouraging in that a good match can be obtained with BTO/BFO/Wind/PDA errors. It remains to be demonstrated that a full 4-D wind model also provides an acceptable fit. This work is in progress.

  24. M Pat says:

    Absolutely delighted with your initiative in setting up this forum for debate.
    Firstly it gives hope that following the end of the subsea search, the appetite to pursue this mystery is not dwindling. We can’t let it fizzle out.
    Secondly, having followed this in depth since Day 1 on TMF/PPRuNe/DS/JW/Reddit, I think you are the ideal person to provide an environment for open, collaborative, evidence based technical discussion. We are very lucky.

  25. Victor Iannello says:

    @M Pat: Welcome and thank you. Perhaps we can collectively make some progress. Please comment often.

  26. Sabine Lechtenfeld/littlefoot says:

    @Dennis, I think your idea has merit that the very bizarre and totally inadequate reactions of the Malaysians during the time immediately after the plane was reported missing by the Vietnamese ATC, might indicate that they knew full well the plane had been abducted and who was the perp. I always had a hard time to believe in a string of one totally incompetent reaction after another which effectively delayed all SAR activities for hours. But if this was not just mere incompetence it takes indeed involvement and coordination from higher authorities. I wouldn’t exclude this possibility.Interestingly the Malaysian authorities contradicted themselves several times in the immediate aftermath of the plane’s vanishing. As Victor said, they give it the spin which suits them best at any given moment. Unfortunately the Malaysians have never been challenged by the ATSB for their many shortcomings and failures to report important informations. The Australians seemed to treat the Malaysian authorities with kid gloves – presumably for geopolitical reasons. The same can be said for other nations as well, though.

  27. Mick Rooney says:

    @Victor, the DVD-R cover image included in folder 5 of the RMP report is labelled “DCA RADAR DATA”. Without knowing what the DVD-R contains, I remain sceptical that it contains anything more than SSR data gathered from ATC when the transponder was operational.

  28. Victor Iannello says:

    @Mick Rooney: I think it’s possible that the radar data on the DVD was only the civilian SSR data, as you suspect. But it may include both the PSR and SSR from civil radar that was included in the March 2015 FI. Or it may include the radar data, both civilian and military, up until Pulau Perak, that was included graphically in the RMP report. Whatever is on the DVD, it’s more than we have now.

  29. Trip says:

    QZ8501 Air Asia crash in December 2014. Has anyone researched the fact that Air Asia’s majority shareholders are Malaysian and they are headquartered in KL? Any possible link to mh370 and mh17?

  30. TBill says:

    Thank you for setting up the new blog Victor.
    2nd paragraph you say aircraft performance including autopilot is understood, but I was thinking it is still not exactly clear what happens after a waypoint discontinuity (True or Magnetic heading?). I’d add that to a list of things we’d like/need to know.

  31. Paul Smithson says:

    Nice to see this new initiative, Victor.

    Having looked very closely at the radar data in FI, LIDO and the positions included in the RMP (radar captures close to Penang and the phone LBS) I have reached the conclusion that:-

    1) The diversion towards Penang and then WNW up the Malacca strait definitely happened. I have therefore formally abandoned my earlier “head home” theory that involved a turn-back for Kuala Lumpur from IGARI. Other than the “right angle turn”, some minor slant range distortions and the lacuna in the straits, there is nothing particularly odd about radar trace as we know it.

    2) A turnback to the left provides a good fit in timing and position with the first primary radar captures coming back towards Kota Bharu.
    This looks to me like:-
    a) Initial turnback is a 180 (turnback on to reciprocal heading) on to 239/240T from the IGARI-BITOD heading of 059.
    b) Course adjusted a little to point at KB
    c) After KB course adjusted again to point just south of Penang.
    d) Possibly further slight heading changes crossing the Malay peninsular to produce a “bow” shape.
    e) Once we “smooth” small errors in radar position/time, the flight up to Penang appears to have been performed at constant speed of a little over 500kts GS ( ~490TAS). As you have pointed out, this is close to M0.84 (a little under, I think?).

    3) The course after Penang appears to me to be a straight line.
    a) Using the LBS position and the adjacent radar positions (adjusted for slant range/altitude) and the LIDO slide, the aircraft appears to have executed a smooth turn south of Penang on to a constant heading (track over ground) very close to 290T
    b) The positions on LIDO slide for which we have times are those few positions that did NOT line up with the vast majority of others (presumed lateral error). This is what allows the times for those positions to be legible – as opposed to the others that are overlaid on each other. I have assumed that the “true” position of these timed positions lies on a common track with the vast majority of primary radar captures.
    c) I make the speed over ground of the LIDO track 506-508kts over ground – again with tailwind of 13-15kts. This is nearly identical to speed between turnback and Penang. There is no indication of speed change after Penang, at least up to the position assumed to represent 181446 [1814?6]
    d) The position of the 182212 “final radar capture” is inferred, not known. The positions typically attributed to it using continuation at same speed/heading correspond closely with the end of DSTG’s smoothed radar trace.
    e) At FL340, the 182212 inferred position should still have been well above radar horizon of Western Hill, although it is getting very close to maximum nominal range of Western Hill radar.

    From the above, I conclude that:

    1) Turnback performed initially on to a reciprocal heading
    2) Course adjusted to fly over KB
    3) Course adjusted to fly towards Penang (just south of)
    4) Turn south of Penang on to heading (track) 290T
    5) Constant speed throughout – all the way from Turnback through the LIDO slide trace
    6) Course followed does not require the aircraft to be flying a waypoint route – it could just as easily have been flying “headings” (with rather few changes up to Penang and no change after Penang).
    7) 182212 position is inferred and may not be reliable
    8) I agree with DSTG that 182212 is “too close” to the series of BTOs 182527-182815. If aircraft remained on same heading of 290T (as I provisionally assume) then it must have slowed down a little. This remains true even if we ignore 182212 and extrapolate from the positions up to 1814. I haven’t yet done my calculations to estimate speed reduction amount/timing to obtain consistency with the first ping rings.

  32. Please can you advise how backtracking the debris found via known ocean currents effects thinking on search areas. Thank you.

  33. Warren Platts says:

    Hi Victor,

    Well done on the new blog site. One thing about the Lido radar image that’s bugged me and as you’re probably well aware is that “295R 200nm from Butterworth AB” tag. The actual position of that bearing and range from Butterworth lies approximately 72 nm due east from the radar track at that time (18:22). An error so blatant, it makes you wonder about the rest of the diagram.

    Then there that one offset point a little before the circle. I once calculated it would take like a 70 degree bank to make that point if it is accurate. Well beyond the overbank protection IOW, although I’ve since learned the pilot can override the overbank protection if he or she wanted to. Probably it’s just an error–radar bearings are not nearly as accurate as radar ranges, but then there were those rumors of the plane juking like a fighter jet at times…

    As for the missing data, the Lido figure had that big circle drawn around it, as if to draw attention to it, suggesting that maybe it WAS discussed at the Lido conference (presumably in Chinese?). Is there a reliable transcript in existence of the entire conference? The gap could have been some sort of radar shadow–or possibly a result of being below the radar?

    Re: the latter possibility, maybe the outlier point right before the gap is related. My understanding is that if you want to dive a plane rapidly, rather than simply pushing the nose over and thus resulting in strong negative g forces, it’s better to sharply bank over to the right (say) followed by another bank to get back on track. So the gap and outlier point could be related, I suppose, if it were the case that extraordinary control inputs were taking place….

  34. Victor Iannello says:

    @TBill: Yes, you are right. We are not sure which roll mode occurs after an end-of-route, but we do know the behavior for each roll mode.

    @Paul Smithson: Thank you for commenting. The Lido Hotel radar image suggests that MH370 joined airway N571 at VAMPI.

  35. Victor Iannello says:

    @Michael Hilton: Thank you for joining the discussion, Michael.

    There are people here like Richard Godfrey that are much more capable of answering this question than me because they have performed detailed studies using the database of drifter trajectories. The simple answer is that areas along the 7th arc further north than the last underwater search area seem to better explain the lack of debris on the shores of Western Australia and the timing of the arrival of the flaperon at La Reunion in July 2015.

  36. Victor Iannello says:

    @Warren Platts: Welcome, Warren. Please continue to comment.

    Your points are all valid, and demonstrate the need for Malaysia to release the raw radar data. We are forced to guess about things that might be known to Malaysia.

  37. TBill says:

    Here are some of the AutoPilot unknowns:
    > Behavior after discontinuity?
    > Available waypoints in MH370 FMC data base
    > MH370 Magnetic correction tables (what date?)

  38. ALSM says:


    Re the magnetic correction table, we know the following from a pvt Feb 2016 email I received fm an ATSB official:

    “I have confirmed my assumptions from before. The FT ADIRU on the Boeing 777 has a fixed value mag map lookup table used for determination of magnetic heading or magnetic track values. The FT ADIRU provides inertial navigation in a true earth reference frame. The latitude, longitude, heading, track and ground speed are all determined relative to true north. Based on the latitude and longitude values a local magnetic variation value is selected from a lookup table in software. This local magnetic variation value is applied to true track to determine magnetic track, and applied to true heading to determine magnetic heading.

    My understanding is that the FT ADIRU on the MH370 aircraft had a 2005 Mag Map installed. The value for local magnetic variation is not adjusted for the annual rate of change.”

  39. TBill says:

    Thank you ALSM I was able to find 2005 mag version for FS9/FSX. Believe that is more curvature in the SIO compared to today. I also have 2017.

  40. Mick Gilbert says:

    G’day Victor, thank you for establishing this forum. Now that the underwater search has been “suspended” I fear that MH370 will slip rapidly from the public’s mind if not for efforts such as this.
    With regards to the famous Lido Hotel slide, I have long suspected that the Royal Thai Air Force radar station at Hat Yai was the source of those radar plots. I have a couple of reasons for suspecting Hat Yai.
    First, the discontinuity in the middle of those plots aligns quite closely with terrain masking from a mountain range to the west of Hat Yai. The Hat Yai radar station is located atop a 300 metre high hill some 10 kilometres south of the Songkhla international airport. The discontinuity in the radar plots occurs roughly between the 241 and 255 radials from Hat Yai. If you plot those radials as bearings on a topographical map you find that they very neatly bracket a section of high ridgeline that forms part of the Nakawan Range some 25 km to the west of Hat Yai. That section of ridgeline, which rises sharply some 140m above the surrounding terrain and then falls away almost as sharply, would mask an airplane flying around or below 35,000 feet.
    Second, the termination of the Malacca Strait radar plots just past MEKAR also coincides with another section of high terrain on roughly the 267 radial from Hat Yai.
    Third, on 19 March 2014, Air Vice Marshal Montol Suchookorn of the Royal Thai Air Force announced that the RTAF had provided radar data that related to MH370 to the Malaysian authorities. Two days later at the Lido Hotel, Bejing, we see new radar data on a hastily prepared slide that is markedly different from the stylised representation of the airplane’s track from KL to IGARI and back across the Peninsula; I don’t think that was happenstance.
    As for the issue of “no attempted military air intercepts”, well that’s just a Boy’s Own Annual fantasy. Most air forces in that region (Singapore may be the exception) do not maintain a quick reaction alert capability (a) because the geo-political environment doesn’t require it and (b) because it is exorbitantly expensive and hard on crews. QRA requires a pair of fuelled and typically armed fighters with flight crews, ground crews, air traffic control and air combat/defence (ground or airborne radar) crews available 24/7. And on top of all that you generally need an air-to-air refuelling capability ready to go if you expect to be able to recover your fighters after they’ve executed an expedited intercept requiring supersonic flight on afterburner. If you don’t have an AAR capability then the effective range of your QRA becomes very limited.

  41. DennisW says:

    @Mick Gilbert

    Well, that is certainly an interesting piece of information. Of course, the 295 radial in the graphic is even less accurate relative to Hat Yai than it is to Penang Hill. I think the radial label is the primary reason most people assumed Penang Hill.

  42. Mick Gilbert says:

    @DennisW It interesting, perhaps instructive, to plot a bearing of 115 (reverse of 295 radial) from the last radar plot just past MEKAR; you’ll find it runs through Ipoh, home of the Royal Malaysian Air Force School. I’ve often wondered if the Air Force School was given the job of preparing that slide.

  43. DennisW says:

    @Mick Gilbert

    It certainly does. That is interesting.

  44. Victor Iannello says:

    @Mick Gilbert: Thank you for your comments about Thai radar. When you refer to the Hat Yai installation, I assume you mean the radar station near Khok Muang at coordinates 6.844,100.420.

    The IG at one point considered the effect of terrain masking for an aircraft at 35,000 and 37,000 ft, geometric altitude. Bob Hall used STK software to create several plots for the radar heads in the region based on line-of-sight obstruction. Here is one that considers the radar installations in Malaysia and Thailand for a target at 35,000 ft. The point labeled T3 is Khok Muang, and the point labeled M1 is Western Hill (Penang Island). Unless there were changes in altitude, terrain masking from T3 does not explain the hole in the radar data. Nor did T3 have the capability to capture the target at 18:22 UTC.

  45. DennisW says:


    Nice graphic. The radar head at point T2 is interesting relative to coverage. Why was that radar not considered? What/where is the physical installation for point T2?

  46. Victor Iannello says:

    @DennisW: T2 is an installation at Ko Samui Island at 9.492,99.987. T1 is also interesting. It is an installation in Phuket at 7.881,98.316, and should have seen MH370 past 18:22.

    Duncan Steel compiled the list of radar installations some time ago. You can find the locations and elevations here. You can see the actual installations using Google Earth.

  47. DennisW says:


    Thx. I do have the earlier Steel table. Was not sure if the STK graphic followed the same labeling convention. The Phuket installation had a very good shot at the aircraft. My assumption has always been that it was inactive at the time. T2 would appear to have much more appeal than T1 relative to the range and the “hole”.

  48. Victor Iannello says:

    @DennisW: If the STK image is accurate, MH370 seems to be out of range of T2 after turning past Penang Island. It looks like it would have needed to be much higher than 35,000 ft for the hole in data to be explained by terrain masking from T2. I have another plot at 37,000 ft and the results are not much different.

  49. DennisW says:


    Sure. If the STK graphic is taken completely literally what you say is clear. However, I always interpret such graphics/analytics as an approximation to reality. Modeling RF propagation with precision is simply not possible. I tend to focus on general features rather than absolute boundaries unless the line of sight horizon is involved. Even then, diffraction and atmospheric lensing effects can muddle the results a bit.

    So far, T2 is the only reasonable explanation for the hole that I am aware of. The time of flight and distance from the cell phone registration to loss of radar contact does not seem to support an altitude change explanation.

  50. Victor Iannello says:

    @DennisW: Good points. Another question is why Western Hill missed captures after 18:02.

  51. Don Thompson says:

    @Richard: “Extrapolating back from the Inmarsat satellite data between 19:41 UTC and 00:11 UTC” leads to the most significant reappraisal of the elements used to re-construct 9M-MRO’s route south. Exploring circumstances that the final southbound ‘leg’ of 9M-MRO’s flight occured much later than previously considered, i.e., later than 18:39UTC, transforms the outcome on the 7th arc.

    @Victor: ““The NW point at 1912 was an assumed theoretical location at 8° 35.719’N, 92° 35.145’E initially chosen to provide clearance from the known radar sources (mainly Singapore).” This is, indeed, an curious reply. The options for ‘mainly Singapore’ resources include:

    a) RSAF fixed long range surveillance site on Bukit Gombak, Singpore, but that is far out of range from the ’10N’ locality;

    b) RSAF’s G550 CAEW surveillance aircraft, if on patrol;

    c) An RSAF Fokker-50 MPA, if engaged on a Malacca Straits Patrol/EiS patrol;

    d) another source relayed through Singapore’s Information Fusion Centre.

    @Vector-1: The radar issue is less ‘unreliability’ rather it is Malaysia’s failure to provide any detail concerning what was recorded by their systems, and their neighbours’ systems. No detail on the sites involved, the capabiities of those sites, and so on. However, that information is available, spread about the public domain, and I have been collating it with the aim of making reliable interpretation of what has been published. No doubt, more questions will ensue.

    @Paul Smithson: Glad to read of your revised conclusion!

    @Mick Gilbert: Victor has responded that previous analysis of the radar horizon from the Hat Yai (Khok Muang) RTADS-III site would provide good coverage of the Straits of Malacca, I concur with that and now have access to a modelling tool which leads to the same conclusion. It’s worth noting that Thailand has an agreement with DCA Malaysia to share SSR data to ensure service continuity for its FIR surveillance and air traffic management service, the Khok Muang RTADS-III site also hosts the SSR antenna feeding the civil ATM network.

    It’s my intention to complete an revised and expanded analysis including all relevant Malaysian radar capabilities and the long range, air defence surveillance, systems deployed around the Straits of Malacca (Indonesia, Thailand, Singapore, and India/Andaman & Nicobar Is). I haven’t got a timescale for that but I’ll refocus efforts.

  52. Don Thompson says:

    @Dennis, all.

    Consider the ‘Beijing Lido’ track to be a composite from two sources: a) Bukit Puteri, and b) Western Hill.

    The systems at both these sites employ phased array antennas which focus ‘pencil beams’ to scan in the vertical while the antenna rotates for 360º azimuth coverage. The vertical scan requires high & low power beams to cover the detection envelope: lower power for higher elevation, shorter range; higher power for lower elevation, further range. What we see in the ‘Beijing Lido’ are the targets detected by the high power, low elevation, extended range, element of the system.

    Further, 7th & 8th March 2014 radiosonde data from weather stations at Phuket, Kota Bharu, Penang, and Medan all show conditions for atmospheric ducting.

  53. Paul Smithson says:

    Another “speculative hypothesis” for the lacuna that I have proposed is that this represents the area where radars in Malaysia (Western Hill) and Indonesia (Lhokseumawe and Medan) oppose each other. Perhaps the WH hill radar suppresses radar reflections in that region because of interference from the “opposing” radars?

  54. TBill says:

    The radar data is very interesting.

    Why are so many proposed MH370 flight paths including some IG paths going so close to the Sabang radar after 18:40 FMT? It would seem the PIC would try to go further around the Sabang coverage, and Indonesia basically said no apparent MH370 siting in their airspace.

    The McMurdo path on the Z FS9 simulator does a great job of going around Indonesia air space, and the Iannello/Godfrey McMurdo path from VOCX at least disguises MH370 from a Sabang view point, and perhaps might have been modified to include an offset or altitude adjustment.

  55. DennisW says:

    @Don T

    The elephant in the room is the hole in the radar track which the Bukit Puteri and Western Hill composite does not seem to adequately explain.

  56. Don Thompson says:

    @DennisW wrote: “the hole in the radar track which the Bukit Puteri and Western Hill composite does not seem to adequately explain”

    It does if one considers the long range element of the detection envelope is 180-190nm to 250nm out from the antenna.

    Bukit Puteri to 18:07:16, track intersect with ‘circle’ = 250nm
    Western Hill to 18:22:12, target beyond MEKAR = 250nm
    Note that the common central point for radials to the 18:07:16 and 18:07:06 targets is Bukit Puteri.

  57. DennisW says:

    @Don T.

    Do you have a reference for the long range detection envelop? Frankly, I struggle with your 180-190nm number unless there is a range gate involved. If Western Hill is range gated beginning at the 180-190nm distance, that would explain the hole.

  58. Mick Gilbert says:

    @Victor @Don T. Gents, thank you for your responses.
    Victor, yes, I’m refering to the Royal Thai Air Force radar station at the coordinates quoted (the radar antenna is about 50 metres north of north-east of those coordinates).
    Bob Hall’s graphic appears to show no coverage of the flight path at 35,000 feet by Hat Yai (Khok Muang) which leads me to suspect that the radar mast head height was not modelled at around 315 metres ASL or the intervening terrain has been modelled at its actual height ASL rather than its apparent height to the radar. Either way Bob’s graphic appears to have the shape of the masking for Hat Yai correct but it understates the range of the radar coverage. It is worth noting that if you expand the coverage by 50-odd kilometres you get a reasonable match to both the discontinuity and the termination point.
    On a separate note, there seems to be an assumption that all military radars in that region were operational at the time MH370 was in the air, early morning hours on a Saturday. Informal discussions with an experienced air force air combat officer who has had some exchange experience in the region suggests otherwise. He is of the opinion that most of the RTAF, RMAF and TNI-AU surveillance radar units would have been non-operational on the weekend. Obviously the status of military surveillance assets is a touchy subject (revelations here in Australia that our Jindalee OTH radar is not operated on a 24/7 basis generated some public debate) so getting to the bottom of that may prove problematic. The only military radar station that can be confirmed as operating on the morning of 8 March 2014 is the RTAF station at Hat Yai (Khok Muang) because of its feed into the civilian system. Which brings me back to suspecting the RTAF radar station at Hat Yai (Khok Muang) as the source of the Malacca Strait radar plots.

  59. Victor Iannello says:

    @Mick Gilbert: I think you are confusing the T2 and T3 ranges in the figure. As @DennisW said, it is T2 (Ko Samui Island) which qualitatively shows the terrain masking corresponding to the hole. The T3 (Hat Yai) radar is shown as covering much of the path in the Malacca Strait, but coverage does not extend to the 18:22 point. I would be surprised if Bob did not correctly use STK, as that is his expertise.

    Perhaps Don Thompson can provide independent verification of radar range from the radar sites using the other tool he references.

  60. Mick Gilbert says:

    @Victor. You’re quite right, Victor, I was looking at the T2 coverage (a salutary lesson in not attempting this sort of thing on the bus!).
    I’d be interested in hearing Don’s assessment of the coverage from Hat Yai between the 240 and 255 radials.

  61. Don Thompson says:

    Refer to (1) for radar horizons from Khok Muang (assuming 25ft antenna height above ground alt of the site).

    @DennisW I’m not suggesting a simple range gate, rather the combination of gating and the ‘fine’ beam forming of the phased array antenna. Without the beam forming these ‘3D’ radars can’t discern target altitude. See Christian Wolff’s useful material at (2), specifically case 1, on that page, where he expands on the long range surveillance radar design. I’m suggesting that what is depicted by the ‘Beijing Lido’ track is a result of range gating and processing only certain beam elevations.

    (1) http://imgur.com/feAbx5Q
    (2) http://www.radartutorial.eu/06.antennas/Long-%20and%20Short-Range.en.html

  62. Victor Iannello says:

    @Don Thompson: The radar range image you shared for Khok Muang is very similar to what Bob Hall calculated using STK, and verifies that result.

  63. PeterC says:

    @Don Thompson:
    I know it is a bit of an ask, but, can you please generate radar range images for ALL the PSR’s in the area from FL200 to FL400 in 5k-feet steps similar to the one you shared for Khok Muang above, or link us the resource you used to make that plot so we can do it ourselves ?

  64. DrB says:


    For anyone who may find it useful, here is an EXCEL file (36 MB) containing all the relevant MH370 GDAS weather data:


    This file has worksheets covering 1500 hours UTC on 2014.03.07 to 0300 hours on 2014.03.08 in 3 hour increments. The first worksheet in the file contains explanatory notes, and it has a data viewer so that any available parameter may be displayed.

  65. Victor Iannello says:

    @DrBobbyUlich: Thank you, Bobby. One of these days I’ll incorporate a full 4D interpolation for wind and temperature into my models.

  66. DennisW says:


    Yes, I actually spent 13 years of my life working as a SLAR engineer at Goodyear Aerospace in Litchfield Park, AZ. I have a pretty good grasp of the system aspects of radar.

    Are suggesting the hole is formed by beam width considerations in the elevation direction at a range of between 120km (nominal start of hole) and 160km (nominal end of hole)? You might want to try a “back of the envelop” calculation on that one.

  67. Don Thompson says:


    The two segments of track depicted in the image are:
    a) 18:00 to 18:07, approx 170nm to 250nm from Bukit Puteri (south of Kota Bharu),
    b) VAMPI to 18:22, approx 180nm to 259nm from Western Hill (Penang Island)

    I don’t regard the gap in the track as a ’hole’, that is, targets missing from a single site’s tracking. I consider it as a gap resulting from the limit of coverage from two radar sites.

    I’m suggesting that each system recorded targets of interest filtered by range. If the detection envelope was limited in such a way, the full vertical sweep would be unnecessary. I understand that it’s possible to do this, I don’t have the domain specifics of how it’s achieved, hence relied on advice from others more knowledgeable.

    My working assumption, at the moment, is that Malaysia derived it’s tracking from the TUDM/RMAF ADS sites at Bukit Puteri and Western Hill plus the air traffic management PSR sites at Butterworth AB and Kota Bharu.

  68. Don Thompson says:


    The plot from Bob Hall is consistent with the tool I have used. If you want to try an independent analysis, consider QGIS with an available plug-in.

  69. DennisW says:


    I like your concept. No quarrel with that at all. To suggest that an aircraft at 30,000 feet at a range of 120km can fall out of the beam width while an aircraft at 30,000 feet at a range of 160km will be seen is simply not workable from a simple geometry and beam width basis. If it was done at all, a range gate is the only method I know of. However, I can’t think of any good reason to establish range gates at those particular ranges. Maybe.

    Altitude determination is typically done by dithering the beam vertically, and looking at the peak of the reflected energy. It is quite different than simple detection (there vs not there).

  70. Don Thompson says:


    Litchfield Park?

    We may, unknowingly, have passed each other along N Litchfield. I once did a stint at Luke.

  71. DennisW says:

    @Don T

    Many of my neighbors flew out of Luke. Played a lot of bridge with Captain (now Colonel ret.) Steve Dwelle who was solo pilot for the T-Birds and his foxy wife. Good times. I liked the area a lot, and both my kids were born there.

    Believe it or not SR-71’s could and did land at the tiny Litchfield Park Airport (in the middle of the night) and stuck there noses through a hole in the wall so we could install a side-looker in them. I never actually got to see the aircraft.

  72. Mick Gilbert says:

    @Don T. Thank you for the latest radar coverage modelling for Hat Yai (Khok Muang), Don.
    I’m sure that you have noted the very sharp drop in coverage between the 238 and 241 radials; that aligns quite nicely with the “thinning” of the radar plots leading up to the start of the gap on the Lido Hotel slide.
    It is interesting that the plots subsequent to the resumption of contact near VAMPI are comprised of two (if not three) clusters. The first of those clusters sits roughly between the 255 and 258 radials, roughly coincidental with a brief up tick in radar coverage before it falls away rapidly. The anomaly, however, is the absence of returns associated with the not dissimilar up tick in coverage around the 248 radial.
    Having now looked at a higher resolution image of Bob Hall’s graphic I note that coverage from Ko Samui Island appears to extend sharply at roughly the same place that the Hat Yai coverage falls away. As I understand it, Samui would have been operating at the time because, like Hat Yai, it provides coverage for civilian area control.
    Given the vagaries of primary radar and the limitations of modelling, I suspect that the Lido Hotel slide was constructed largely from Hat Yai’s radar data with perhaps some data from Ko Samui Island representing the final cluster of returns to just past MEKAR. As I noted previously the Malacca Strait radar data emerged on a (hastily prepared?) slide that differed markedly from the slide based on Malaysian civilian radar only two days after the Thais announced that they had provided military radar data to the Malaysian authorities. Further, unlike almost every other military radar in the region, we know that the military radars at Hat Yai (Khok Muang) and Ko Samui Island were operational given their dual role as civilian AAC assets. Now, all of the foregoing may be a series of coincidences but I’m not inclined to think so.

  73. Don Thompson says:


    It is the case that Khok Muang’s coverage could be interpreted as providing the first segment, 18:00UTC to 18:07UTC, of the track over Str of Malacca. However, the gap and the second segment are not consistent with that interpretation.

    If Khok Muang’s range limitation, terrain obstruction considered, caused the loss of targets at 18:07UTC and failure to acquire the target across the ‘gap’, there is nothing in the range analysis, computed by two separate tools, that suggests Khok Muang might have later re-acquired the target of interest in the vicinity of VAMPI.

    Using GE I constructed a 3D model of the Khok Muang radar horizon and a path depicting a track at 35,000ft: the track breaks through (above to below) the horizon near 18:07UTC point/entry to gap. That’s a third analysis.

    Further note, it’s the SSRs co-located at Phuket and Khok Muang that feed the civil ATM systems and, from which, data is shared with DCA/MY.

    NB: I purposely refer to the RTADS-III sites as Phuket, Ko Samui and ‘Khok Muang’, the reference to Hat Yai has often been exploited in other forums to misrepresent the antenna location as Hat Yai airfield.

  74. Mick Gilbert says:

    @Don T.

    Thanks for the further analysis, Don. With a number of analyses pointing to Khok Muang as a possible/likely source of the 1800 – 1807 UTC segment of the track, doesn’t it make sense to consider it as one part of a composite? Could Ko Samui account for the 1813 – 1822 UTC part of said composite?

  75. Victor Iannello says:

    From the Bangkok Post. Note that the statement implies that there were no captures by Thai radar after MH370 passed Butterworth. That said, we know that it is hard to trust any single official statement, although official statements from the Thais tend to be more accurate than from the Malaysians.

    The Royal Thai Air Force detected a radar signal of a plane turning and flying to the north of Malaysia’s capital Kuala Lumpur before disappearing early on March 8, around the same time Malaysian Airlines flight MH370 was reported missing.

    Air force chief Prajin Juntong said the radar signal was received from a plane around midnight of March 8, when the missing jet left Kuala Lumpur and headed northeast towards Vietnam.

    “The air force’s radar stations in Surat Thani, Phuket and Songkhla’s Hat Yai received some signals that need to be verified by both the Aeronautical Radio of Thailand Co and Malaysian radars to check whether they are the same,” ACM Prajin said.

    “A plane made a U-turn before flying to the north of Kuala Lumpur and Butterworth army base in Malaysia before vanishing,’’ the air force chief said.

    ACM Prajin said it was possible the plane could have headed to the Straits of Malacca.

    He said the plane’s direction was consistent with Malaysian Prime Minister Najib Razak’s comments the plane made a U-turn and flew over an area of the southern border of Thailand and then headed northwest to the Strait of Malacca.

    He said he would send the radar information to Malaysia to verify whether it was consistent with information they have about flight MH370.

    He also said Thailand was assessing the possibility that its radar stations lost the plane’s signal because it had already flown out of radar range. Thailand would also have to find out whether the signals picked up by other radar stations were from the same plane.

    When asked why it took so long to release the information, air force spokesman Air Vice Marshal Montol Suchookorn said, ‘’Because we did not pay any attention to it. The Royal Thai Air Force only looks after any threats against our country, so anything that did not look like a threat to us, we simply look at it without taking actions.’’

  76. Mick Gilbert says:


    Yes, Victor, the reporting was all over the shop back then. As early as 18 March 2014, ACM Prachin Chantong (also referred to as Prajin Juntong) was saying that RTAF radar had “… detected a passenger plane taking off from Kuala Lumpur and did a U-turn, diverting to Butterworth, Penang Island before heading into direction of the Strait of Malacca.” The next day he confirmed that Surat Thani (Ko Samui) had tracked MH370. (http://www.thaivisa.com/forum/topic/712004-thai-air-force-radar-may-have-picked-up-mh370/)

    When you look at Bob’s coverage map, Surat Thani would have lost MH370 pretty much over Buterworth, however, Khok Muang would have been able to track it from there up to the point it became masked by terrain at 1807 UTC. For me, the only question remaining about the radar plots up the Malacca Strait is whether Surat Thani reacquired MH370 at 1813 UTC and tracked it until 1822 UTC. If so, that would provide a composite composed entirely of RTAF radar data that was handed to the Malaysian authorities just two days before it was seen publicly. That is a much simpler answer to the Lido Hotel slide conundrum than some of the extraordinarily complex technical solutions that have been posited (and yes, I’m also saying that I’m not smart enough to understand most of the latter).

  77. Don Thompson says:


    “doesn’t it make sense to consider it as one part of a composite [.. with..] Ko Samui”

    The idea does warrant consideration, but it doesn’t fit. By modeling Ko Samui’s capability using the same three tools, it is evident that at the range from the antenna, the radar horizon is too high to have captured the targets depicted between VAMPI and the 18:22 point.

    For the RTADS-III network, only Phuket remains to be considered: it would have unrestricted line-of-sight to all the depicted target locations in the area of interest. I did contemplate possible obstructions, nothing fitted.

    Indonesia’s surveillance sites don’t fit. That only leaves Malaysia’s.

  78. Don Thompson says:


    Surat Thani is not a radar head location, I understand it is the location of the sector operations centre for the three remote radar heads comprising the RTADS-III network, i.e., Phuket, Khok Muang, and Ko Samui.

  79. Mick Gilbert says:


    Thanks for the information on Ko Samui’s coverage. And yes, you’re right, despite the airport on the island being referred to as Surat Thani Ko Samui, Surat Thani itself is on the mainland; the naming conventions in that part of the world can be a bit confusing.

    Despite being one of the best sited radar heads in the region (sitting atop a 600m high hill on an island with no terrain obstructions for 60-70 km) I can see the range/horizon problem for Ko Samui that you’ve described. At 37,500 feet (~FL350) the target at VAMPI would have been within the instrumented range of Ko Samui’s AN/FPS-130 (ARSR-4) but just on or below the horizon and at MEKAR it would have been just beyond instrumented range and just below the horizon. I could make a case for ducting but that would be a rather inelegant “solution”. I’m still intrigued though by the alignment between Ko Samui and intervening terrain and the gaps in the VAMPI to MEKAR plots at 1815:30-1816:20 UTC, 1818-1819 UTC and 1820:30-1821:12 UTC. What does your QGIS coverage map for Ko Samui look like between the 216 and 231 radials?

    The two problems I have in trying to assign the VAMPI to MEKAR plots to the RMAF radar station at Western Hill, Penang are:
    (a) the RAT-31DL has an antenna rotation period of 10 seconds which should yield 6 plots per minute; the enhanced images of the slide show the plot rate is much more likely to have been 5 per minute which points to a source radar with a 12 second rotation period like one of the Thai AN/FPS-130s; and
    (b) the pattern of plots doesn’t look like what you’d expect to see for a military radar tracking a target that was bigger than a B-52 with a relatively constant speed, altitude and heading that was well inside the radar’s instrumented range and over water well above the horizon. The plots do, however, look like what you might expect to see for a target at maximum range, low on the horizon traversing the very edge of the coverage.

    I am struck by the fact that a lot of very clever people have been struggling with the Lido Hotel slide conundrum for some time with no clear and agreed solution in sight. Might it be time to reframe the problem? With no obvious solution for the data and assumptions, what assumptions might be changed to fit the problem to a solution? Target altitude (~FL350) , vertical speed (0 fpm) and track (~290°) are key assumptions that we seemed to have settled on early in the piece that have constrained possible solutions. Do you think that there’s anything to be gained by re-examining them?

  80. Mick Gilbert says:


    Can I ask what ATSB report you are referring to when you say “The ATSB report released in June 2014 includes statements about a radar capture of MH370 at 19:12 UTC in the Andaman Sea.”?

  81. Victor Iannello says:

    @Mick Gilbert: The report was released on June 26, 2014, with the title “MH370–Definition of Underwater Search Area”. In an upcoming post, I’ll have more to say about what we can glean from the 1912 radar data, including the following statement from the report: “The aircraft passed close to a NW point at 1912”.

  82. Don Thompson says:


    Indeed, the AN/FPS-130X antenna rotation speed is 6rpm, as it is for the AMS/Selex RAT-31DL and AMS/Marconi Martello S743D of the RMAF (I may have previously misquoted 10rpm)

    1) The time stamps that can be discerned from the Lido image are consistent with a 10 second scanning period.

    2) The track is depicted as two separate segments, both segments begin with ‘dense’ target clustering which then fades as the target progresses north west. My interpretation is that the targets were initially detected at a range where adequate performance was possible and the targets were moving away from the antenna toward the edge of detection range.

    3) The radar horizons, as modelled without consideration for atmospheric effects, preclude Khok Muang or Ko Samui as the source. If the radar horizon for each of those two sites is extrapolated further, to allow for refraction/ducting in the atmosphere, sustaining the gap becomes a ‘problem’.

    4) The artefacts evident in the depiction of track describing the IGARI turn, and the 02:07:06 and 02:07:16 timestamps at hte end of the first segment, provided some discrimination for my hypothesis of Bukit Puteri and Western Hill providing the ‘military radar’ data.

    However, the fundamental concern is the consistency of reporting. Two weeks after the loss of 9M-MRO, the Malaysian Air Force Chief of Air Operations discussed the information represented on this image. No subsequent publication references this track and it wasn’t shared with ATSB. To support DSTO’s analysis, Malaysia provided radar data up to the start of this track, and provided a location for a target, described as isolated, at 18:22. Without the intervening data, how can the integrity of the 18:22 target be assured, how can the assumptions for the track to MEKAR along N571 be validated? Resolute clarification is necessary.

    As I set out in 2014 and further discussed here, all the long range air defence surveiilance assets, their locations, and capabilities are known, there is no justification for secrecy.

  83. DennisW says:

    @Don T

    “The track is depicted as two separate segments, both segments begin with ‘dense’ target clustering which then fades as the target progresses north west. My interpretation is that the targets were initially detected at a range where adequate performance was possible and the targets were moving away from the antenna toward the edge of detection range.”

    The gap remains a major issue. It is not adequately explained with respect to Western Hill which should have easily seen the aircraft all the way through the gap. Vertical beam width cannot account for the gap based on very simple geometrical considerations. The only explanation which aligns with physics is low altitude flight through the gap which might create problems of its own with respect to average speed.

    An equipment failure or power outage and subsequent repair would work, but would seem to be remote. It is also possible that Bukit Puteri was tracking the aircraft and Western Hill was dormant and alerted when the aircraft was heading out of the range of Bukit Puteri.

  84. PeterC says:

    I have downloaded QGIS and having had a quick look, it is obviously a very powerful tool. I have found a decent simple tutiorial to get started. But having never used a GIS program before, it is obviously going to take me a long time to learn the “in-and-outs”.

  85. MH says:

    This hole/gap in the radar coverage can not just be an one-off occurrence. If it happened for MH370 then it must be happening for every flight taking that path. if it is not happening for other flights then wonder why it occurred???

  86. BarryC says:

    I believe that some of the issues observed with the radar track towards VAMPI and MEKAR are related to variable anomalous propagation (ducting), possibly affecting both the X and Y planes of the transmitted/reflected pulse. Usually this type of ducting is noted when the effective range is extended “over the horizon”, but can happen the other way when an inversion layer does the opposite and a target normally expected to be within range is missing, i.e. the transmitted pulse was sent into space; hence no return. Another likely cause is multi-path refraction where phase cancellation can effectively cause the loss of the return signal – more likely over a calm sea, or the path taken exceeds the receiver gate time, and is lost as the next TX pulse is sent.

    Any of the above could have been happening, and the erratic plotting of some of the returns can also be explained the same way.

    For those trying to make sense of this, a good primer for this type of radar ducting/multi-path (anomalous propagation) can be found at:-


  87. Brock McEwen says:

    Hi Victor; thank you for opening this forum – much appreciated.

    In the spirit of advancing the science without rancor or bias, my first question is strictly technical:

    The BTO arcs are derived from a “distance from plane to satellite”, which is itself derived from taking the signal’s transit time, and multiplying it by the speed of light. The signal transit time is the BTO value itself, with two adjustments:

    1) all BTO values are recorded with a large bias term removed – so this must be added back.
    2) logon requests (special class of BTO) have embedded in them an extra processing time – so this must be removed

    While the add-back for 1) has been supplied to the nearest 10 microseconds, the deduction for 2) – 4,600 microseconds – appears rounded to the nearest 100 microseconds.

    My question is simple: what is the potential rounding error embedded in the 4,600 value – and to what does this incremental rounding error translate with respect to the calculated position of any arc generated by a logon request?

    I consider it pertinent to this forum because the “7th” is one of these logon arcs. I bring it to this forum because various attempts to engage the IG on this specific issue in the past have not produced a meeting of the minds.

    Thanks in advance to whoever can help improve my understanding.

  88. DennisW says:


    Probably not.

  89. BarryC says:


    “Probably not.”

    The question of why these variable type radar returns hadn’t been observed in prior flights was raised. The answer is those flights were tracked using SSR where the transponder responded on a much lower frequency (1090 MHz) with the aircraft derived positional data.

    So, why not?

  90. Don Thompson says:

    AP (anomalous propagation) effects will have influenced the PSR performance on the night of 7-8th March. Archived radiosonde data logged by weather stations at Kota Bharu, Penang, Medan, and Phuket all exhibit temperature inversions at altitude.

    During the course of that night the DCA (and other ATM services) will, routinely, have tracked civil traffic using SSR. The SSR systems provide a similar range of coverage as the military Long Range Air Defence Surveillance PSRs, approximately 250nm. While its assumed 9M-MRO’s transponder did not resume service after 17:21UTC, surveillance of all other traffic will have been routinely recorded. That surveillance data provided a means to calibrate the altitude discriminated and recorded by the LR-ADS PSRs.

    A function of RMAF’s Sentry C² system is correlation between the LR-ADS PSRs, SSR and filed flight plans. It’s reasonable to assume both sets of data, PSR and SSR, were available.

    We don’t know if a calibration process was undertaken. No instance where radar data has been exploited; ‘Lido’, Factual Information, RMP Folders, is accompanied by evidence that altitude of an assumed track for 9M-MRO has been rigourously correlated using the available sources.

  91. Victor Iannello says:

    Comments here are closed. Please continue the discussion under the new post.