Unraveling the Mystery of MH370
Part 1 of 3
The following article is the first of my ‘World Mystery Series’.
I. Introduction
Planes do crash, but they don’t vanish from the sky. The disappearance of Malaysia Airlines (MAS) flight MH370 ten years ago is one of the greatest mysteries in aviation history.
From the very first day 8 March 2014, I found myself strongly driven by inexplicable motivation to tackle the mystery. Looking back, I spent roughly four years (2014-2018) investigating the case almost incessantly, delving ever deeper into the case. Throughout those years, I normally spent one hour or so everyday reviewing news reports, collecting evidence, performing data analysis, studying research works and exchanges between aviation experts in various blogs. Even if I haven’t solved the mystery, I believe I’ve at least greatly advanced my knowledge in aviation.
This journal article is a recap of my journey of investigation. The following sections present a brief overview of essential topics covered in the process. While salient points are highlighted, technical details are kept to a minimum. Readers should note that one might need to spend weeks, if not months, of intensive research to have a firm grasp on each individual topic.
II. Ground Zero
It is most crucial to follow closely and correctly the sequence of events that happened since the take-off of MH370 (registered 9M-MRO) from Kuala Lumpur (KL) Airport at 16:41UTC on 7 March 2014. Basically, everything seemed to be normal right until the plane was about to enter the Vietnam FIR (figure 1). It was at that moment that the final communication with KL ATC was made (at 17:19:30UTC), as KL ATC was going to handover to Ho Chi Minh (HCM) ATC. Just one minute later, the plane reached the waypoint IGARI. Then another minute later (17:21UTC) the plane went electronically dark. Here one should first get familiar with some basics of aircraft communications with the ground including the transponder, ADS-B and ACARS. What ‘electronically dark’ means that the aforementioned communication systems stopped. As such, the plane would disappear from the radar used by the ATC. The sudden shut down of these communication systems and disappearance from radar could simply mean that the plane has crashed into the sea. However, the whole story changed when the plane was then found to be captured by the radar to be heading back towards the Malay Peninsula. One should however note that there is a tiny gap here. MH370 was not continuously tracked by the radar to be turning back. It was deduced that the plane picked up by the radar must be MH370 because there were no other commercial planes near the turn back location at that time. It is an extremely important point and I’ll come back to this later. The finding that the plane has turned back could imply that it had experienced serious mechanical problem and was badly in need of emergency landing. But the fact that it didn’t attempt landing would then imply that either the pilots were incapacitated, or it was a deliberate flight maneuver by a rogue pilot to evade detection. The latter makes sense as it is not difficult to shut down these communication systems from the cockpit. Furthermore, the shut down of these systems right near the FIR boundary during the handover between two ATCs was considered to be a highly strategic action to delay possible response by the ATCs. Now the big question is: Who did it? We will discuss this later.
III. Radar Tracking
In the previous section we discussed the turn back of MH370 near waypoint IGARI. Even with its transponder and ACARS shut off, the plane should still be captured by radar as a moving object. But the detection depends on its flight altitude. The plane would be captured by civilian or secondary radar (hereafter SSR) if it flies high. If the plane flies at very low altitude, it might escape from the SSR radarscope, but could be captured by military or primary radar (hereafter PSR) which is normally set to detect low level invasion. Indeed, the plane was known to have been tracked by at least three radars, namely the SSR at Kota Bharu Airport (which captured the flight towards and across Malay Peninsula), the SSR at Butterworth Airport (which tracked the plane’s passage south of Penang and subsequent northwest turn up the Malacca Strait towards Pulau Perak), and PSR at Pulau Perak (which tracked the plane across the Malacca Strait on airway N571 till passing waypoint MEKAR). In fact, there is vast radar coverage over the Malay Peninsula and Malacca Strait (figure 2). Yet most radar data were not released. One intriguing fact was that the plane was found to undergo sophisticated maneuvers which included drastic and frequent altitude changes (figure 3). Rather than attempting for emergency landing, it seemed more likely the plane was navigating via waypoints while evading detection.
One crucial piece of information regarding the radar tracking was the so-called ‘Lido radar image’, which was shown to the victims’ families (i.e. next-of-kin or NOK) in Lido Hotel, Beijing on 21 March 2014 (figure 4a, b). The image showed significant anomalies in the radar plots which defied explanation. Researcher Ron Belt has conducted a detailed analysis [1] of the radar data. Furthermore, it’s noteworthy that there were discrepancies on the flight routes tracked by the military radar between those presented by Malaysia Air Force Chief Rodzali Daud in March 2014 (he later denied it!), and those of the subsequent official versions (figure 5). These issues are nontrivial. We will come back to them in Part 3 of the article.
IV. Satellite Detection
One major breakthrough came when it was found that the turn-back plane later responded to satellite signals, i.e. it had ‘electronic handshakes’ or ‘pings’ with Inmarsat satellite 3F-1 orbiting above the equator in Indian Ocean for 6 more hours after the last radar signal. One should note that even with its transponder turned off, the aircraft still has a SATCOM system or a satellite data unit (hereafter SDU) in the ACARS which communicates with the satellite. This SDU cannot be easily turned off/on in the cockpit without highly sophisticated procedures, nor is it under the pilots’ work routine (normal pilots should even have no idea what it is!). For the MAS, the Inmarsat service which it subscribed to provided two types of data. The first was Burst Time Offset (BTO) which could determine the aircraft’s distance away from the satellite. The second was Burst Frequency Offset (BFO) which could be used to deduce the relative motion of the aircraft w.r.t. the satellite, much like the Doppler shift effects. With the BTO data, the Inmarsat staff were able to determine the precise distance of the aircraft from the satellite at each ‘ping’ (figure 6a, b). They also developed a novel mathematical technique to determine the relative motion of the aircraft w.r.t. the satellite with the BFO data, thereby deducing its flight direction. With the BTO analysis, the plane was determined to have flown along either a northern or southern corridor. As the BFO analysis matched with the southern corridor (figure 7), the aircraft was deduced to have ended up near the ‘7th Arc’ in the southern Indian Ocean (SIO) west of Australia. Technical details of the analysis will not be covered here. Interested readers may refer to the seminal research paper “Bayesian Methods in the Search for MH370” prepared by DSTG [2]. The DSTG study demonstrated that the BTO data matched well with a high, straight and fast flight towards the south over the SIO. The paper showed a probability density function (pdf) of the plane’s endpoint location which has a maximum value near 38S 88E. The pdf drops to very low values north of about 35S. This ‘heat map’ showing the probability of the plane’s endpoint formed the basis of the subsequent search zone in the SIO.
The Inmarsat data was widely regarded as the ‘gold standard’ for the search of MH370. Without it, there would have been absolutely no clue to find the plane. However, there is a very intriguing detail in the data. Note from figure 6b that there was a log-on request at 18:25UTC and 00:19UTC. While the 00:19UTC log-on request was believed to correspond to fuel exhaustion and subsequent activation of the auxiliary power unit (APU), the 18:25UTC request was very unusual as we do not expect the SDU to ‘reboot’ this way. Without this reboot, the plane could have gone completely dark forever, without even satellite response. There is no reasonable explanation for why the pilot/hijacker would disable the ACARS (thereby the SATCOM) with sophisticated actions (if he knows how to do so), and then go through the same sophisticated process to turn it back on. The strangest fact was that the SDU reboot occurred just shortly (roughly 10 mins) after the plane had gone out of the final radarscope. It looked like a highly unusual serendipity that we were able to continually track the missing plane after the final major turn (FMT) into the SIO. Aviation journalist Jeff Wise (JW) noticed this anomalous event, and made a compelling case that the Inmarsat data might have been spoofed or tempered with. But why? It’s a mind-boggling question and we will revisit this in the later section.
V. The Search
Initial search [map]
Shortly after the incident on 8 March 2014, flight search missions were launched from Gulf of Thailand to Vietnam and Cambodia. Following the identification of a debris field by a Cathay Pacific flight off the Vietnam coast, search was conducted in the South China Sea (SCS) by Vietnam’s Search & Rescue Control Centre. The debris were later identified to be not related to the aircraft. On 9 March, SASTIND, a civilian agency under the Chinese Ministry of Industry and Information Technology reported three large objects at 6.7N 105.63E triggering further search there by Vietnamese aircrafts and Chinese ships. But nothing aircraft related could be found. On 10 March, Malaysia extended the search to Malacca Strait following military radar detection of the turn back, but without any finding.
SIO search [map]
Based on the Inmarsat data discussed in the previous section and ocean drift from possible impact location near the 7th Arc, a 42-days aerial surface search [video], coordinated by Australian Maritime Safety Authority (AMSA), was conducted starting on 18 March 2014. Australian Transport Safety Bureau (ATSB) used underwater microphone ‘Towed Pinger Locator’ to detect the black boxes, and deployed underwater robot ‘Bluefin 21’ to scan the seabed. But nothing was found.
Interestingly, between 16 and 26 March, multiple large objects (maximum length >20m) near 44S 90E were spotted by satellites of Australia, China (Gaofen-1) and France (Pleiades 1A, 1B) and Airbus Terra SAR-X analysed by MRSA of Malaysia [report]. However, that far south location was apparently outside the known fuel range of MH370. Hence search was not conducted in that region despite the high significance of those debris finding. Instead, the search areas were shifted to the north along the 7th Arc. In October 2014, autonomous underwater vehicle (AUV) was launched from the Australian-contracted survey ship M/V Fugro Discovery, with search areas [map] mainly north of the Broken Ridge. Efforts were again proved futile. The SIO search basically terminated by the summer of 2016. But in early 2018, search interest suddenly revived with the U.S. registered company Ocean Infinity offering a no-find-no-pay search [map]. What a generous offer! Might there be other motives behind?
Before moving to the next section, I’d like to pose two questions for readers: 1) Given the DSTG benchmark study, what’s the rationale to keep shifting the search areas northward? In particular, what pilot input should be made to have the plane ending up in that northern location? 2) What were those large objects near 44S detected by multiple satellites? Could they be regular flotsam & jetsam? Try to check the world’s shipping routes and assess the probability. Researcher and mathematician Brock McEwen offered a critical review of the ATSB search decisions [3].
Or the conclusion of the searches is that: The plane was never found in the SIO because it’s not there?
VI. Hypothetical Scenarios
The conclusion that MH370 crashed into the remote SIO was solely based on Inmarsat data together with novel and sophisticated mathematical methods. Science didn’t answer the question why it did so. To address this question, different scenarios have to be considered. They can be categorised into either motive-based or non-motive-based. Motive-based scenarios include bombing or missile shoot down, pilot suicide, rogue pilot and hijacking. Non-motive-based scenarios include accidents such as mechanical or electrical failure, onboard fire, hypoxia, etc. I conducted a historical survey of aviation disasters and found some case examples for each scenario. Then I compared each case with MH370, but could find little similarities. The case of MH370 was truly unique and unprecedented. The comparison is briefly summarised in a table (see presentation slide 45).
The problem with the mechanical failure or fire scenario was that it could not explain why there was (i) no distress signal or mayday, (ii) no attempt for landing but the sophisticated maneuvers (discussed in Section III), and most prominently, (iii) why the plane could remain in air for so many hours. The bombing/shoot down scenario could explain (i) but not (ii) and (iii).
It’s noteworthy that the SIO search area initially delineated by ATSB was strongly based on the hypothesis of a ‘ghost flight’ (i.e. a totally incapacitated pilot due to depressurization leading to hypoxia) which was proposed by DSTG and Independent Group (IG). In this case, the plane would continue its flight in autopilot mode (absolute or magnetic heading) until it runs out of fuel. Interestingly, such ghost flight scenario had historical precedent – Helios 522. This scenario makes sense if one solely interprets the Inmarsat data after the FMT. However, it remains difficult to explain how the transition from active pilot input (as evidenced before the FMT) to ghost flight occurred. Perhaps noting this problem, the official narrative gradually shifted towards the pilot suicide scenario. There had been precedent cases of pilot suicide (e.g. German Wing 9525). But had there been any pilot taking a long flight to commit suicide? Well, there was indeed a curious case of USAF pilot Craig Button which was suspect of a long suicide during a training mission in 1997. However, there was absolutely zero case of long suicide following a convoluted and evasive route with so many passengers! Nobody commits suicide without reason. Had the captain of MH370 shown any incentive or signs of committing suicide before that fateful night? I shall conclude with a definitive answer – NO. This topic will be discussed in Section XI. Let’s say we have eliminated all the above scenarios, what’s left is the ‘rogue pilot’ scenario. Could MH370 have been hijacked? It’s very unlikely to happen with the stringent security measures in commercial aircrafts (e.g. the cockpit is always locked) after the 9/11 attack. Besides, there was no Squawk Code 7500 (for hijacking) sent from the plane. Let’s say a hijacker was smart enough to break into the cockpit, incapacitated both the captain and the first officer, then turned off the transponder immediately disabling the Squawk. Why then did the hijacker take such a long flight to nowhere until fuel exhaustion? Perhaps a reference case was Ethiopian flight 961. But there were still stark differences between MH370 and ET961. Now we are left with only two possibilities – that the rogue pilot was the captain himself, or that a highly sophisticated electronic remote hijack with the use Boeing-Honeywell Uninterruptable Autopilot (BHUAP) [4] was performed. We shall leave this to later sections. For a summary of the likelihood of different scenarios, see presentation slide 91.
VII. Debris Discovery
The conclusion that MH370 crashed into the SIO was purely based on the Inmarsat data. Yet nothing was found in the search areas near the 7th Arc. So it remained a matter of faith to trust the Inmarsat analysis. However, a major discovery came in July 2015 when a flaperon was found on the shore of Reunion Island (figure 8) which was identified with 9M-MRO (i.e. MH370) by its serial number. More debris were later found in Mozambique, Madagascar, South Africa as well as other places (see presentation slides 49-50, 55-65 for details). Most debris were identified with different parts of the B777 aircraft with varying levels of certainty. Some followers of the MH370 mystery might have cast doubt on the veracity of the debris. Now let’s just presume the debris to be genuinely from the doomed aircraft. A few common characteristics can be identified in the debris discovery: 1) They were mainly found ashore in the southwestern part of Indian Ocean; 2) They were relatively few in number, 3) Most were external parts of an aircraft; 4) The majority were from the R.H. side of an aircraft (figure 9), 5) There were different degrees of bio fouling. Point 1 can simply be explained by the debris being driven by the ocean current to reach their destinations (figure 10). More on this will be discussed in Section IX. Points 2-4 are significant as they shed light on the nature of the crash, while point 5 may hint on the location of crash. The forensic analyses of debris will be presented in the next section.
VIII. Forensics of Debris
Impact analysis
One prominent feature immediately identified from the recovered flaperon in Reunion Island was the intact (damaged) leading (trailing) edge (figure 11). What kind of impact would result in this damage pattern? One conclusion is that it entered the water in a near horizontal position (figure 12) which is typical of a ditching attempt. In this way the trailing edge would be eroded by water. If the plane crashed into the sea in a steep dive, like what one would expect from either pilot suicide (high speed nosedive) or ghost flight followed by engine flame out (steep spiral dive), the front edge of the flaperon would have been severely crushed. Though the trailing edge damage of the flaperon could alternatively be explained by fluttering on flight, it’s not convincing to me.
A member of the CAPTIO team (a research group on MH370), aerospace engineer Argiris Kamoulakos, conducted the study [5] of ditching using dynamic numerical simulations based on the methods of Finite Elements to simulate structures and Smoothed Particle Hydrodynamics (SPH) to simulate fluid flow. He concluded that it was a ditching that could have resulted in the damage of the flaperon's trailing edge, while an in-flight loss of the flaperon would most likely result in other types of damage than those visible on the flaperon. The results were consistent with the findings of the structural study by the General Delegation to Armament (DGA) laboratory, where the flaperon had been transmitted because of the French judicial investigation. The DGA study was incorporated into the final investigation report published by the Malaysian authorities in July 2018. These experts give strong credence to the hypothesis of a ditching during the last moments of the missing aircraft.
Interestingly, damage photos of the Captain Sully case (US Airways Flight 1549 ditching on the Hudson River in 2009) showed similar trailing edge damages on the wing (figure 13) and even stripped off engine cowling (see the ‘RR’ piece in S. Africa). Nevertheless, the two cases do have subtle differences. The Sully case was a controlled ditching to minimize the impact force (thereby maximize the floating time of the plane) for survival of passengers, whereas the overall debris forensics of this case pointed towards a hard ditching, targeting to sink and disappear the plane as fast as possible. Moreover, the debris from the vertical stabiliser resembled the damage found in Asiana Flight 214 that crash landed at San Francisco International Airport in July 2013 (figure 14). This damage could be caused by flying debris from the front when the plane hit the water horizontally.
There was more evidence of ditching, according to the analysis by aviation disaster investigator Larry Vance. He examined closely the outboard flap discovered in Tanzania, and found compression fracture in the end plate caused by spanwise crushing force that crush the flap and flaperon together when the right-wing tip dug into water possibly during ditching (try to use your imagination). He also found ‘V-shaped’ black smudge witness mark on the outside of the seal pan endplate probably caused by crushing by the outboard end of the flaperon due to severe spanwise force. Furthermore, there was indication of damage caused by the support track and carriage assembly when they were pulled out through the entry hole, which is possible when the extended flap was being pulled backward through the water during ditching. Readers may refer to presentation slides 79-84 for detailed illustrations.
Marine biology examination
One extremely interesting aspect of the flaperon recovered in Reunion is that it was attached by vast number of goose barnacles (figure 15). Barnacles often attach to floating objects in the ocean, growing in size with time. The size of the barnacles and their species type provide important clues on the time they spent drifting over the ocean, and the areas they drifted past related to the sea surface temperature.
The barnacle attached to the Reunion flaperon belong to the species Lepas anatifera. Figure 16 shows the global distribution of Lepas anatifera and Lepas australis (Schiffer & Herbig 2015). One can see that Lepas anatifera is normally found in tropical to sub-tropical warm waters above 18C. Notably, the species is not recorded south of 33°S (Hinojosa et al., 2006). One can see that the warm Agulhas Current transports organisms from the Indian Ocean to the southern tip of Africa. In fact, L. anatifera is reported in a diversified assemblage of goose barnacles from the Cape Town peninsula (Whitehead, Biccard & Griffiths, 2011), at the westernmost waning of the Agulhas Current. No wonder the ‘RR’ debris discovered in Mossel Bay, S. Africa was initially found attached by barnacles (see presentation slide 60). Note the warm ocean current (red arrows) in the Indian Ocean responsible for dispersal of L. anatifera originated from the tropical ocean north of 15S (figure 17). This is remarkable as the flow of the current was in contrary with the ocean drift model results used in the ATSB official report to support the drift of the flaperon from the 7th Arc (to be discussed in the next section).
In a detailed French study in 2016, the age of the barnacle colony on the flaperon was estimated at 476 days (or 15-16 months) based on the size (36 mm) of the biggest specimen (MH370 Safety Investigation Report, Appendix 2.6A) [6]. As the growth of the specimen stopped on the day of discovery of the flaperon (29/7/2015), the initial colonisation dates back to 10/4/2014, or 33 days after the plane’s disappearance. Using oxygen isotopes in the study, results indicated that in the months preceding the sample, the flaperon had drifted progressively from very warm waters (28.5 ± 1°C) to cooler waters (25.4 ± 1°C) present near the Reunion Island (MH370 Safety Investigation Report, Appendix 2.6B) [6]. This indicated the L. anatifera on the flaperon should have originated from tropical waters.
Fascinated by these findings, I visited the Scripps Institute of Oceanography in La Jolla, CA in the summer of 2016 to study the marine biology in detail. I was amazed to find that the L. anatifera attached to the Reunion flaperon were likely coastal species which usually populate the atolls surrounded by warm waters. Further study of the bathymetry of the Indian Ocean showed these atolls only exist in three regions – Cocos Is., Maldives and Chagos Archipelago. All these regions are located far away from the 7th Arc, where no L. anatifera could be expected because of much colder water.
IX. Ocean Drift Modeling
We can expect that the debris discovered were brought to their destination by ocean currents. In the SIO, one can see a large anti-clockwise gyre. Intuitively, if MH370 did crash near the 7th Arc, one can imagine the debris might follow this gyre and drift to the southwestern part of the SIO. In fact, there is a variety of ocean drift models which can simulate the drift of the debris. There are two types of such models – forward drift and reverse drift. In support of the ATSB search, David Griffin et al. of Commonwealth Scientific and Industrial Research Organisation (CSIRO) simulated the path taken by the flaperon across the Indian Ocean, and concluded the July 2015 arrival date of the flaperon at Reunion is consistent with impact occurring between latitudes 40S and 30.5S on the 7th Arc. But one must be cautious that ‘consistent’ doesn’t mean proof. First, there was a problem with the timing of debris discovery at Reunion (flaperon) and S Africa (RR piece) if they drifted from the 7th Arc. Suppose both debris were soon discovered once they came ashore, the flaperon seemed to reach Reunion too late while the RR piece reached S Africa too early. There is also another paradox here. For impact in a more southern location, one would expect many debris would have drifted to the coast of Western Australia (WA). On the other hand, a northern location would be in conflict with the DSTG study (which assumed autopilot input), as well as the BFO analysis discussed in Section IV. Out of intense curiosity, I personally went beach combing along the WA coast all the way from Geraldton to Cape Leeuwin in 2017, obviously without any findings. I believe that if there were aircraft debris drifted to the WA coast, they would certainly be picked up by the myriad Aussies on the beaches.
Noting the inadequacies of the CSIRO work, I started studying the results of other ocean drift models which simulated the drift of the flaperon and other debris. These include the reverse drift model of MeteoFrance and GEOMAR of Germany; the forward drift model Adrift and the oceanographic simulations of SANDER+PARTNER. Moreover, data from the global drifter program and the NOAA buoy drift study were also examined. In choosing the right drift modeling result, there are two important factors to consider: 1) the windage factor (i.e. how far the debris emerges above the water surface to ‘catch’ the wind), especially for the flaperon; and 2) Stokes drift (i.e. nonlinear net drift near the ocean surface which results from surface waves). The French who kept and examined the flaperon concluded that it drifted entre deux eaux (immersed under water), which could imply low windage and weak Stokes drift. The totality of drift modeling results showed that the search zones near the 7thArc could hardly reconcile with the debris findings. In particular, the reverse drift models generally indicate a much more northern origin location. Moreover, there seemed to be a ‘hot spot’ in the tropical SIO near 5-15S 70-80E. The area is located to the east of Diego Garcia. Wouldn’t this result match quite well with the marine biology examination in the previous section?
X. Ocean Acoustic Studies
When there is a significant impact on the ocean surface, acoustic gravity waves will be generated which propagate to distant locations. These acoustic waves could be picked by hydrophones which normally detect underwater earthquakes and nuclear tests. If an aircraft crashes into the ocean and sinks to the ocean floor, similar acoustic waves can be expected. Interestingly, on the day of the disappearance of MH370, hydroacoustic station off Cape Leeuwin (HA01) operated by Comprehensive Nuclear-Test-Ban Treaty Organisation (CTBTO) and another station off Rottnest Island near Perth operated by Australia’s Integrated Marine Observing System (IMOS) detected the sound of possible impact of an aircraft on the ocean. The sound signal, recorded at 01:30UTC on 8 March 2014, was discovered by a team led by Alec Duncan, an underwater acoustics specialist at Curtin University’s Centre for Marine Science and Technology in Perth, Australia. The signal was believed to have originated somewhere along a strip extending to the northwest of the Indian Ocean which is clearly out of the range of the ATSB search along the 7th Arc.
A study by Cardiff University, UK in 2017 found a weak signal recorded by HA01 at 00:50:00UTC with origin centered near 43S 94E and event (E1) source between 00:25UTC and 00:31UTC. The same station recorded at 01:34:40UTC another signal with origin centered near 23S 96E and event (E2) source between 01:11UTC and 01:16UTC. At the time of that study, it was believed that E1 could not be related to MH370, as the event time was only a few minutes after the final satellite handshake (at 00:19UTC) and the source location was too far for the plane to reach from the 7th Arc. When I read this study, I immediately recalled the large objects near 44S detected by multiple satellites (discussed in Section V). In a subsequent study, the same group of researchers found that the sea floor elasticity may greatly boost up the traveling speed of acoustic gravity waves, thus the locations of impact for E1 and E2 might both be much further from those pinned down in the the previous study (figure 18). In addition to HA01, there was another hydrophone station (HA08s) at Diego Garcia which recorded acoustic signals on the same day. These signals suggest an impact location in the northern part of the SIO. However, the signals were distorted by noise which was believed to be caused by a military exercise in the area. What’s more suspicious is that there were 25 minutes of data from HA08s missing, which was not explained by CTBTO. What happened, and what was indeed recorded in that 25 minutes?
(to be continued in Part 2)
References
[1] Ron Belt, “Analysis of Malaysian Radar Data”, 29 May 2014.
[2] Samuel Davey et al., “Bayesian Methods in the Search for MH370”, Defence Science and Technology Group. Springer Open (2016).
[3] Brock McEwen, “Time to Investigate the Investigators”, 16 January 2015.
[5] Argiris Kamoulakos, “The end of flight MH370: a ditching study of the flaperon hitting the surface of the sea”, 13 March 2020.
[6] The Malaysian ICAO Annex 13 Safety Investigation Team for MH370 (2018), “MH370 Safety Investigation Report”, 2 July 2018.
Recommended Book
Larry Vance, MH370: Mystery Solved (Kindle edition, 2018).