by Jeff Wise
If it’s remarkable to find a piece of MH370 with TV cameras rolling, imagine doing it twice.
Later that year Gibson was back on Ile Ste Marie, this time with a delegation of MH370 family members and a documentary crew. On the morning of December 8, the group split up and spent the day combing separate areas. The camera crew followed Blaine. Having driven along one stretch of shore on an ATV and found nothing, he turned around and was making his way back when he came upon a piece of debris at the edge of the wet sand. A wave had evidently deposited it within the few minutes since he had passed. “Appears to be Malaysia 370 interior cabin debris,” he declared.
And that wasn't all. Gibson found yet another piece under extraordinary circumstances. In a seaside village on mainland Madagascar he happened to spot a nine-year-old girl using a scrap of MH370 debris to fan a kitchen fire. “It was light and it was solid and it was part of the plane,” Gibson told The Guardian. “When I put the word out around the village, another guy turned up with another piece he had been using as a washing board for clothes.”
I found it implausible that one might happen upon someone fanning a fire with a flat object that just happened to be a piece of MH370. I became even more suspicious after I found that Gibson had given a different account of how he’d collected the piece in a posting to a closed Facebook group: “We combed the beach ourselves and found one piece of aircraft debris, a hexagonal rubber torque coupling seal… We showed pictures of debris to local people and asked if they had found any, and over the next two days different people brought us the remaining three pieces.”
No mention of the girl. Had he changed his story, or merely simplified it? And if the former, was he motivated simply by the desire to tell an interesting story, or by something darker?
Chapter 21
September 2016
As 2016 rolled on, the ships of the search flotilla continued to sail up and down the southern ocean, battling foul weather, seasickness, and monotony. There was a rare moment of excitement when their side scan sonar revealed the tell-tale oval-shaped debris field of a smashed vessel, but when the robot sub went down to take a closer look, it found a long-forgotten merchant ship.
Nevertheless, Australian officials continued to voice confidence that the seabed search would ultimately prove successful. On the second anniversary of the disappearance, ATSB head Martin Dolan told the Guardian that MH370 would be located before the end of the year. “It’s as likely on the last day [of the search] as on the first that the aircraft would be there,” Dolan said. “We’ve covered nearly three-quarters of the search area, and since we haven’t found the aircraft in those areas, that increases the likelihood that it’s in the areas we haven’t looked at yet.”
Martin was mistaken. Basic statistics will tell you that the further the search vessels moved away from the 7th arc, the lower the probability that they would encounter the wreckage on each successive pass. In the analysis that the DSTG had published the previous fall, it had calculated that there was effectively a zero percent probability that the plane hit the 7th arc north of 34° south longitude or south of 41° south longitude. And the ships had already scanned that segment of the arc out to a distance of 40 nautical miles in either direction.
In September I interviewed the DSTG’s Neil Gordon for Popular Mechanics and asked him why his team’s calculations hadn’t borne out. He told me that the failure of the seabed search meant that one of their initial assumptions must have been wrong. But which one? Gordon’s best guess had to do with an event that occured at 18:40, fifteen minutes after the SDU rebooted.
Amid its attempts to find the missing flight, Malaysia Airlines had tried to call MH370 on its satellite telephone. The call went through, but no one answered; it just rang and rang. The nature of the electronic interaction was such that no BTO value had been recorded, so no ping arc could be generated, and the signal wasn’t considered one of the seven canonical Inmarsat pings.
But a BFO value had been recorded, and it was telling. Its value was nearly a perfect match with the BFO values of the pings that followed in subsequent hours. Investigators had long interpreted this to mean that the direction of flight had been the same. That is, the plane had already turned south some time before 18:40. This assumption put a northernmost limit on where the plane could have flown.
In essence, the ATSB had staked their entire investigation on this interpretation of the 18:40 BFO value. But what if their assumption was wrong? It turned out that there's another way that that particular BFO value could have been generated. Because the Doppler precompensation algorithm doesn't take vertical motion into account, if the plane had been in a gradual descent—as, for instance, when coming in for landing—then it could have been heading east or north or west.
To be sure, it's hard to imagine why the plane would have been descending; generally when planes fly long distances they gradually climb, not descend. But if the plane hadn’t been cruising south at 18:40, then it might have been heading north, or flying in circles in the vicinity of the Andaman Sea, or who knows what. In any case, a later turn to the south would mean the plane could have crossed the 7th arc as much as 250 nautical miles further north than the DSTG had previously deemed possible.
In November of 2016, with just a sliver of the 46,000-square-mile search area still unscanned, the ATSB convened a meeting of its experts to grapple with all the different ways they could have gone wrong. The results of their discussion were published on December 20 in a report entitled MH370 – First Principles Review. The document confirmed what Gordon had told me, that the only way to make sense of the data was to assume that the plane must have wound up north of the previously defined search. The experts defined a new area, 10,000 square miles in size, that straddled the 7th arc between 32.5° and 36° south latitude. Given the signals received by Inmarsat, this was the only place the plane could possibly be. “Based on the analysis to date, completion of this area would exhaust all prospective areas for the presence of MH370,” the report affirmed.
On the day that the report came out, the ships were just finishing up their scan of the main search area. At the very end, they ducked into the newly defined area for a few weeks, then called it quits. The search was officially over. To hear the ATSB tell it, their calculations hadn’t been wrong; they’d just been unlucky. The plane must have taken a rather unlikely and surprising maneuver and wound up in the newly defined search area. But, due to circumstances out of their control, they would not be able to search there. The failure was not their fault. They could retire from the field with honor.
Chapter 22
November 2016
For the relatives of the missing passengers and crew, the end of the search was a devastating development. Back in 2014 the authorities had assured them that they would keep searching until their loved ones were found. Now they were cutting bait. “Stopping at this stage is nothing short of irresponsible,” scolded next-of-kin group Voice370.
As I read it, the failure of the seabed search amounted to a repudiation of the ATSB’s interpretation of the Inmarsat signals. If the plane had flown south, the seabed searchers should have found the plane’s wreckage. The fact that they hadn’t meant that Shah couldn’t have been the culprit.
But there had long been rumors of a smoking gun, a major piece of evidence that could tie Shah to the act. It had to do with Shah’s flight simulator. The captain was such a flying buff that he’d assembled an elaborate rig in his basement, complete with six monitors, two computers, a yoke, and rudder pedals. Back in June, 2014, the Guardian reported that investigators had found computer files on his flight simulator indicating that he “had plotted a flight path to a remote island far into the southern Indian Ocean where the search is now focused.” The following January, Byron Bailey wrote in the Australian that he, too, had received similar information, and that “my source … left me with the impression that the FBI were of the opinion that Zaharie was responsible for the crash.”
While these reports
were tantalizing, no one had ever presented any documentation to back them up. Then, in November, 2016, Irish journalist Mick Rooney released a portion of a secret Malaysian Police report that included raw data recovered from Shah’s home flight simulator. Of 671 save points that had been stored on various hard drives, six were of particular interest. The first four points appear to be snapshots from a continuous flight that takes off from Kuala Lumpur and climbs as it heads to the northwest, with the fourth point located over the Andaman Sea not far from where the real MH370 disappeared from military radar. The last two points are are located more than 3,000 miles away to the southeast. Curiously, at both of these final points the simulated plane’s fuel tanks are empty.
Assuming that these save points had been made in the course of a single flight up the Malacca Strait and then down into the Indian Ocean, it looked an awful lot like Shah had indeed practiced a suicide run. I decided to take a closer look. My father-in-law, Bob Masterson, happened to have a copy of the old flight-sim software that Shah had used, as well as a dusty Windows computer capable of running it. Once I had it set up in my attic I set about running the flight simulator, generating save points, then examining the data files that resulted. With practice I was able to get my files to match Shah’s quite closely.
What I found was that rather than a continuous autopilot flight up the Malacca Strait and then down into the southern Indian Ocean, the flight sim user had made a series of disconnected flights, changing his location between runs by manually moving an icon on a map display. In short, what I found didn’t look very much like a practice run for a flight to oblivion.
I paid particular attention to the last two save points, at which the plane is out of fuel and located over the remote stretch of the southern Indian Ocean. At the first point the plane is flying at 37,651 feet at close to 198 knots indicated airspeed, which is the speed recommended by the emergency checklist used when a plane has run out of fuel. At the second point, the plane is flying in much the same way, but the altitude had been manually decreased to 4,000 feet. In each case the plane is being held in a gentle descent of the sort one would hold in order to maximize the amount of time in the air.
To me, what this looked like was not a suicidal plunge, but an attempt to hand-fly a dual-engine-failure procedure. Why might Shah want to practice this? Maybe he wanted to see how he would perform in some famous real-life emergencies. On July 23, 1983, an Air Canada 767 en route from Montreal to Edmonton ran out of fuel at 41,000 feet due to an improperly calculated fuel load. In an amazing feat of airmanship, the pilots managed to glide the plane to a safe landing at a disused air base near Gimli, Manitoba. In the aviation world, this legendary feat has been memorialized as the “Gimli Glider.”
The parameters of the second save point struck me as strongly reminiscent of “The Miracle on the Hudson,” the 2009 incident in which a US Airways A320 hit a flock of geese that destroyed both its engines and then glided to a safe landing in the river.
The overall picture, then, looked ambiguous. Yes, Shah might have been practicing for a suicide flight. But he also might have been trying to become a safer pilot. Bear in mind, too, that he practiced engine-out procedures in a 777, he also flew a historical propeller transport, the DC-3. And three weeks later he played with a Boeing 737. This is hardly the behavior of a man with a monomaniacal obsession with his upcoming demise.
As we’ve already seen, Shah was a well-liked, balanced, and amiable person. The Royal Malaysian Police folder on Shah also included the results of a psychiatric evaluation which found that Shah had no psychological problems, family stress, money problems, or any other suggestion that he might be suicidal. “Zaharie is an experienced pilot and a competent and respected by peers,” the report states. “Information from friends and colleagues Zaharie show that he was a friendly, warm and jokes… we have not found, any changes in terms of psychological, social and behavioral patterns Zaharie Ahmad Shah before his flight.”
Certainly, pilots have crashed their planes into the ground and killed all their passengers. But in such cases there is invariably a history of stress and psychological dysfunction. Just a year after MH370 went missing, Germanwings pilot Andreas Lubitz crashed his Airbus into the French Alps. Lubitz had previously been grounded for psychiatric reasons and had torn up doctors’ notes excusing him for work on mental health grounds. Shortly before he killed himself, he performed web searches about deliberately crashing a plane.
Shah left behind no such clues. If he did commit mass murder-suicide, he must have been mentally unbalanced enough to contemplate one of the most heinous acts imaginable, and at the same time mentally composed enough to plan it without leaving behind any evidence—not a single Google search, not a single deleted email or stifled outburst—except, possibly, this single set of flight-sim data.
This level of self-control seems psychologically implausible to Katherine Ramsland, a professor of forensic psychology at DeSales University who has written 54 books, including Inside the Minds of Mass Murderers. “Usually, there’s something that people notice,” she says. “It’s one thing if you were single, but [Shah] had people around him, coworkers, family. They’d notice something different. He wouldn’t necessarily leak his decision or his planning, but there would be something different. To do something like this without anybody noticing anything would be really hard to do.”
Ramsland also doubts that Shah would have planned a suicide that involved flying the plane for seven hours after its diversion. “If somebody’s going to commit suicide, they’ll just go do it,” she says. Every other known instance of pilot suicide involves immediately flying the plane into the ground.
Then again, implausible isn’t impossible. “There will always be rare instances of things that surprise us,” she says. “Human beings are strange creatures.”
Chapter 23
April 2017
At this point, more than 20 pieces of debris had been found. And for us obsessives who still believed this case could be solved, this debris, more than anything else, was the thing to study.
While Gordon and his team at the DSTG were struggling to figure out where they’d gone wrong, David Griffin’s team at the CSIRO was attacking the mystery from another direction. Using data from NOAA’s Global Drifter Program, Griffin’s team had built a computer simulation of how debris might travel over time. They then ran the simulation many thousands of times to show all the different tracks the flaperon might have taken from various points along the 7th arc. Needless to say, this covered a huge swath of the ocean. Nevertheless, it was possible to calculate the zone from which the simulated flaperons would most likely reach Réunion.
As we’ve seen, an important component of this kind of modeling is windage. If an object floats high in the water, the wind will tend to push it along faster than if it were deeply immersed. To come up with a windage value for the flaperon, CSIRO scientists built six replicas of similar shape, size, and weight and set them adrift in the ocean off the coast of Tasmania. The results were then fed back into the model.
The outcome was puzzling. Most of the simulation runs ended with the flaperon floating far north of Réunion Island. Perhaps, Griffin’s team reasoned, the replicas weren’t accurate enough. To explore this possibility, the CSIRO team obtained a real Boeing 777 flaperon, then cut it down to mimic the damage experienced by the piece found on Réunion. When they put it in the water, voilà: Compared to the replicas, the real cut-down flaperon floated much more like the original. Once the new data was fed into the model, it showed that debris starting from within the high-probability search area was more likely to drift to Réunion Island. For the ATSB, this was welcome news. The drift modeling was pointing to the same stretch of ocean as their Inmarsat data analysis.
The CSIRO also used satellite radar altimetry data—super-accurate measurements of the ocean’s height over time—to get an even finer-grained look at ocean drift patterns. This analysis narrowed the hotspot of highest probability even further. “The
only thing that our recent work changes,” Griffin and his coauthors wrote in a paper released in April 2017, “is our confidence in the accuracy of the estimated location.”
There were a few problems with this picture, however. One was that, while the CSIRO’s refined model was excellent at predicting that the flaperon would turn up at the right place and time, it didn’t work so well for some of the other debris items. It couldn’t explain, for instance, how a piece of engine cowling had managed to float all the way to the southern tip of South Africa by December, 2015.
As with the flaperon, the CSIRO built a replica of the cowling fragment to test how the piece would have floated. They found that it floated so low in the water that it was virtually unaffected by the wind, and so moved more slowly across the ocean than a high-windage item like the flaperon. When they ran their simulation forward from 35° south, the cowling wound up nowhere near South Africa by the time the real-life object was collected. And some of the modeled low-windage debris washed up on the western coast of Australia where none had in fact been found.
The ATSB’s newly designated search zone had another major problem. It lay within their very first search area, and as such had already been partially scanned. In particular, the area around 35° south had been searched out to nearly 20 nautical miles in either direction. If the plane had crossed the 7th arc there and been plummeting near vertically at hundreds of miles per hour, its wreckage should already have been found.
So while the ATSB was making confident public pronouncements, the true situation was much more problematic and baffling. Investigators had developed multiple lines of inquiry into where the plane could have gone, but these lines did not converge. Choose any spot on the 7th arc, and there was a strong reason to believe that the plane hadn’t gone there. South of 39.5S was ruled out because the plane couldn’t fly that far. 36S to 39.5S was ruled out because it had already been searched. Debris drift modeling ruled out 34S to 36S. And a surface search shortly after the plane’s disappearance had long ruled out the area north of 34S.
