The Taking of MH370

by Jeff Wise

  Imagine that you have two ping rings, one created an hour after the other. The rings are basically concentric, with the second ring’s radius 300 nautical miles bigger than the first. Let’s say the plane started at some arbitrary point on the innermost ring. If that’s all we know, then the plane could have taken any of an infinite number of routes from the first to the second. It could have travelled radially directly outward at average speed of 300 knots. Or, if traveling at an average speed of 400 knots, it could have turned left or right at an angle.

  Assuming a certain speed allows you to draw a path between each successive set of arcs with a particular endpoint on the 7th arc. This technique has a somewhat surprising and important feature: due to the directional ambiguity of the BTO data, every route created in this way has a mirror-image route lying in the opposite direction. For instance, if you plot a route of gradually decreasing speed that curves off to the south, there will always exist an equally valid route that curves off to the north.

  Given the plane’s last known position and assuming a likely speed of about 500 miles per hour, experts reasoned that the plane’s most likely endpoint lay at either the northernmost or southernmost extremities of the 7th arc.

  Many journalists, including most of my fellow aviation analysts on CNN, thought the southern option more likely. After all, if the plane had gone north into the teeming heart of Asia, wouldn’t someone have picked it up on radar? The authorities agreed, and so a search was begun in the southern Indian Ocean under the auspices of the Australian Transport Safety Board (ATSB). Ships and planes from 14 countries scanned the ocean surface in what amounted to the largest search-and-rescue operation in history.

  But for me, there was a big problem with the southern option. If the plane went south, then everyone on board must have died, and my gut was telling me that whoever carried this out was too talented and motivated to quietly snuff themselves.

  With so much still unknown about the case, I found it easy to imagine all sorts of dark scenarios. Since no one had yet claimed responsibility, I wondered if the perpetrators were readying the plane in secret for a 9-11-style attack. Since most of the passengers were Chinese, it seemed to me that separatist Uighurs would be likely culprits.

  I wrote another article for Slate, positing that Malaysia’s announcement meant that “the plane ended its flight in a restive part of Central Asia.” This was quite different from what most other commentators were saying, and my editor was reluctant. I promised her that if I turned out to be wrong I would write a follow-up piece explaining my error and accepting responsibility.

  Chapter 4

  March 25, 2014

  As I was heading into CNN the third Tuesday after the disappearance, I received a text from a producer: the Malaysian prime minister, Najib Razak, was about to hold another press conference to discuss “developments.” My heart skipped a beat. Had they found the plane? I arrived at the Time Warner Center to find the studio abuzz. Maybe this long saga was about to draw to a close. But when Najib at last appeared at the podium, what he announced was something no one had anticipated. Yes, he declared, the fate of the plane had been determined. It had indeed gone south. But there were no bodies and no debris. Inmarsat, in concert with the UK’s Air Accidents Investigation Branch (AAIB), had made the determination through the power of mathematics alone.

  Based on their new analysis, Inmarsat and the AAIB have concluded that MH370 flew along the southern corridor, and that its last position was somewhere in the middle of the Indian Ocean west of Perth. This is a remote region, far from any possible landing sites. It is therefore with deep sadness and regret that I must inform you that, according to this new data, flight MH370 ended in the southern Indian Ocean.

  Never before in history had hundreds of people essentially been declared dead without any evidence but an undisclosed numerical analysis of undisclosed data. From within the hotel briefing room, onlookers could hear the wails of family members in a nearby room.

  Was it really possible to reach such a conclusion based on math alone? Many of the missing passengers’ family members were skeptical. They marched in protest to the Malaysian embassy in Beijing, where they hurled water bottles and battled with paramilitary soldiers in riot gear.

  To justify the conclusion, Malaysia’s acting transport minister held yet another press conference at which he released a report summarizing Inmarsat’s analysis.

  This time the metadata under discussion was not the Burst Timing Offset, but another parameter called the Burst Frequency Offset, or BFO. Just as the motion of a speeding train makes the pitch of its whistle go up or down—the Doppler effect—the relative motion of a satellite and an airplane shifts the frequency of radio signals transmitted between them. The BFO is a measure of this difference. Inmarsat was able to calculate what BFO values they would expect to see if the plane flew to the north versus what they would see if it flew to the south. They found that the observed values were much closer to the calculated southern values than to the calculated northern ones.

  The explanation made sense in principle, and the graphs and charts looked solidly scientific, but none of us in the green room understood exactly how these calculations worked. Nor, for that matter, did any engineers I knew. No one had ever before used BFO values to try to determine the location of a missing plane.

  Torie, my editor, asked if I was ready to write the apology piece I’d promised. I asked her to hang fire. So far the Malaysians had refused to release the complete Inmarsat data logs or to explain exactly how the scientists’ calculations had worked. I argued that until the details were released, it would be irresponsible to rubber-stamp their conclusions. I asked her to give me some time to try to verify what the prime minister had said.

  Fortunately, I had recently acquired a new group of friends who could help me do just that. In between going on CNN and writing for Slate, I’d started posting more technical MH370 material on my website, jeffwise.net. I’d launched the blog a few years before as a place to post short items about science and aviation, but had only attracted a small readership. Each post usually got only two or three comments. Once MH370 happened, it was a whole new game; each post might now get three or four hundred comments.

  Many of the new readers had serious engineering chops and contributed information I couldn’t have found any other way. Among them was Mike Exner, a veteran of the satellite communications industry who’d done pioneering work on the first GPS systems. Exner led me to a New Zealand space scientist named Duncan Steel who a few days earlier had begun posting details of the Inmarsat satellite’s orbital mechanics on a website of his own. Steel’s site and mine both grew into busy forums for technically savvy MH370 obsessives. If you found yourself wondering how a geosynchronous satellite responds to a shortage of hydrazine, or what a Boeing 777 Flight Management System does after it reaches its last entered waypoint, all you had to do was ask.

  Around the globe, a loose association of independent researchers went to work, mining the March 25 report for clues that we could use to reverse-engineer the full Inmarsat data set. As emails whizzed back and forth, someone came up with the idea of giving ourselves a name: the Independent Group. The puzzle was daunting, but the prospect of crowd-sourcing a solution to the world’s greatest aviation mystery was exhilarating.

  Chapter 5

  April 2014

  Malaysia’s revelation that the plane must have gone south effectively put the investigation in the hands of Australia, which by international treaty is responsible for search-and-rescue operations in that part of the Indian Ocean. After weighing the evidence, the ATSB decided to focus on a 23,000-square-mile area west of Perth. Ships and planes spotted many pieces of debris on the surface, provoking a frenzy of “Breaking News” banners, but none of it turned out to be from an airplane. Another week went by, and the ATSB shifted the search area 600 miles to the north. Later it would move the search area further south again. All this rejiggering was confusing, but it gave us CNN talking heads so
mething to chatter about on air.

  Mainly, though, we waited for the big news: the discovery of the first piece of confirmed debris. It seemed certain to happen any day now. Where one piece was found, others would be found nearby; by “drifting” the debris—that is, modeling the currents that had moved them around—investigators would be able to determine a likely area of impact. The wreckage would lie directly beneath the impact point.

  Once the wreckage was found, the mystery’s solution would be at hand. Amid the scattered debris, investigators would find the aircraft’s two black boxes. (Though they are called “black” their color is really international orange.) The first, the Flight Data Recorder, or FDR, contains thousands of parameters such as airspeed, altitude, and position. The other, the Cockpit Voice Recorder, or CVR, preserves flight-deck audio for the final two hours of flight. To make these crucial devices easier to locate, they are built with acoustic pingers that generate an ultrasonic signal when immersed in water.

  Once the approximate location of the plane’s resting place was identified, searchers would drag a kind of high-tech underwater microphone called a Towed Pinger Locator (TPL) back and forth listening for the acoustic pingers. The gear’s range is about a mile, so searchers would be able to sweep a relatively large area quickly. Once they detected the pinger, robot submarines would be dispatched to take pictures of the wreckage and bring the black boxes to the surface.

  A problem was looming, however. Pingers’ batteries are only designed to last for 30 days, and that deadline was fast approaching. If the batteries died before the black boxes were found, locating the wreckage would become vastly more difficult.

  In what amounted to a Hail Mary pass, the authorities decided to put the listening gear in the water without having yet found any floating debris. On April 4, an Australian naval ship deployed a TPL. Mirabile dictu: The very next day, the ship detected a faint signal. Hours later, another signal was detected 1.2 miles from the first. Then two more, a few miles away. It was a moment of enormous relief. After weeks of frustration, the answer to the riddle was at hand. The Malaysian minister of transport, Hishmmuddin Hussein, declared that he was “cautiously hopeful that there will be a positive development in the next few days, if not hours.” Search officials were upbeat, and so was the crowd in the CNN green room. Everyone was ready to bring this tragic saga to a definitive conclusion.

  The only Debbie Downer was me. I found this mind-boggling stroke of luck implausible, and pointed out that the pings detected were at the wrong frequency, and were located too far apart to have plausibly been generated by black boxes sitting stationary on the seabed. For the next two weeks, I was the odd man out on Don Lemon’s six-guest panel blocks, gleefully savaged by my fellow on-air experts.

  The Australians deployed an underwater robot called a Bluefin 21 to scan the seabed and find the source of the pings. They started in the area of maximum probability, so with every day that went by, the chance that they’d find anything dropped. After two weeks, the Bluefin 21 had searched the entire seabed within detection range of the pingers. There was nothing there.

  By the rules of TV news, the issue wasn’t settled until an official said it was. But nerves were wearing thin. One night, an underwater search veteran agreed with me that the so-called acoustic ping detections had to be false. Backstage after the show, he and another aviation analyst nearly came to blows. “You don’t know what you’re talking about! I’ve done extensive research!” the other analyst shouted. “There’s nothing else those pings could be!”

  Eventually, the pinger story petered out the way most stories do: the producers just stopped scheduling segments about it. A month later, a US Navy officer said publicly that the pings had not come from MH370 after all.

  The Australians called off the surface search. “It is highly unlikely at this stage that we will find any aircraft debris on the ocean surface,” said Prime Minister Tony Abbott. “By this stage… most material would have become waterlogged and sunk.”

  It had been 52 days since the plane had gone missing, Millions of dollars had been spent to absolutely no effect.

  For me, the acoustic pings offered an important lesson: the search authorities, and a large proportion of the media, did not have a reliable mastery of the story’s technical details. Until irrefutable evidence was in hand, every official pronouncement would have to be carefully fact-checked.

  Chapter 6

  May 2014

  My CNN bookings were winding down, but my obsession with the case kept growing. There was so much to chew on, with new leads emerging on multiple fronts. In the wake of the acoustic-pinger debacle, the authorities felt mounting pressure to release more information, especially from the impassioned and increasingly well-organized family members of the missing passengers. At last, on May 27, the Malaysians released the raw Inmarsat data.

  Working together, the scattered legion of amateur experts swan-dived into the 47-page trove. Much of the data turned out to be irrelevant, but all seven handshakes were depicted in great detail, and some other intriguing clues were revealed as well. One of the most surprising, and significant, was that the satellite communications system had not inadvertently been left on, as many of us had presumed, but had instead somehow become disconnected and then logged back on. This event had taken place at 18:25 Universal Time (UTC), just three minutes after the plane had disappeared from radar. Could this have happened accidentally, or was it further evidence of deliberate action by hijackers? Answering that question would be a priority in the months to come.

  Also highly significant were two brief sentences in the explanatory note that accompanied the data: “Inmarsat Classic Aero mobile terminals are designed to correct for aircraft Doppler effect on their transmit signals. The terminal type used on MH370 assumes a stationary satellite at a fixed orbital position.” At last, this told us why officials believed the plane had gone south instead of north.

  Here’s the short version:

  As it looks down from its orbit, an Inmarsat satellite has a line of sight to literally billions of radio-frequency devices, from cell-phones and walkie talkies to radio stations and radar dishes. In order to avoid getting swamped by all that babble, the airplane signal must land within the narrow band of the spectrum that’s been reserved for it. Very, very narrow: the frequency has to be accurate to within parts per billion.

  To find this window, satellite communications engineers grapple with a lot of sources of frequency error, one of them being the Doppler effect. To compensate for it, a piece of equipment within the airplane called the Satellite Data Unit (SDU) uses the aircraft’s position and speed to calculate the anticipated Doppler shift, then adds or subtracts this amount from the frequency at which it transmits to the satellite so the incoming signal hits at just the right frequency. Satellite companies routinely track and log the frequency so that they can fix any problems that might arise, and that’s why Inmarsat was able to turn over MH370’s BFO data to search officials.

  The SDU is located above the ceiling of the passenger cabin, right below the satellite antenna, which protrudes from the top of the airplane. Imagine an electronic version of an old-timey ham radio operator sitting underneath a radio tower. The SDU doesn’t generate information per se; it’s just providing the link between the aircraft and the satellite. You can think of it as somewhat analagous to a smartphone. When you turn on a phone, it connects to the cell network, but it doesn’t communicate with anyone until you send a text message, make a call, or activate an app.

  What the new report told us was that MH370’s SDU was programmed to assume that the satellite was orbiting over a fixed position at the equator. But in fact, 3F-1’s orbit had a slight wobble. Launched in 1996, it was intended to operate only for 13 years. As it aged, it ran low on the fuel that it required to stay precisely on location. The satellite’s wobble caused the plane’s electronics to incorrectly compensate for its own velocity and thereby left a trace of that motion hidden in the BFO value.

&
nbsp; Just after the plane disappeared from radar, the satellite’s wobble would have made a northbound plane’s transmission frequency too high. Then, after a few hours, the frequency would have fallen. Conversely, if the plane had flown south, the frequency would have been too low at first, and then risen. This latter pattern is precisely what Inmarsat scientists saw in their retrieved MH370 data.

  I spent a week playing with spreadsheets, running the numbers again and again until there was no doubt in my mind: the scientists had been right after all. The Inmarsat data did unambiguously show that the plane had flown south, not north, after it disappeared from primary radar.

  As promised, I wrote a piece for Slate explaining why I’d been wrong.

  Chapter 7

  June 2014

  Apart from precipitating my mortifying public climb-down, the release of the Inmarsat data was a thrilling development, dangling the possibility that with some elbow grease and a bit of mathematical savvy, we could crack the case. Since the BFO contained information about how the aircraft had been moving, it should allow us to figure out where on the 7th arc the plane had been at the time of the last ping. The wreckage of the plane should lie somewhere in the vicinity.

  Everyone set to work generating routes that matched the ping rings, then calculating how closely these routes matched the observed BFO data. Each time one of us came up with a route that seemed like a good fit, we would email it around to everyone else. I spent days working through the details of a route that ran through a waypoint southwest of the Cocos Islands and ended up around 38° south latitude. Was it the right one? It turned out to be impossible to tell from just the data at hand. As Mark Dickinson, Vice President of Satellite Operations at Inmarsat, later put it to me, “If you know the state of the aircraft, you can predict what the BFO is going to be, but if you only have the BFO number, it’s much harder to reverse engineer out all the components that make up that number.”