Jeff Wise

Last week, the Joint Investigation Team conducting a criminal investigation into the downing of idH17 issued their preliminary findings. Here’s what I think are the main takeaways.

— The findings strongly endorse the work of “open source intelligence” pioneer Eliot Higgins and his group, Bellingcat. In the immediate aftermath of the shoot-down, it was accepted by nearly every pundit and journalist that the missile had been fired accidentally by poorly trained militiamen who had somehow gotten their hands on an SA-11 Buk launcher and had a acquired a target without bothering to first identify it. But by painstaking work and great resourcefulness, the Bellingcat team was able to piece together an extremely convincing timeline, by which the launcher was brought across the border from a specific Russian military unit, was transported under the direction of the GRU (Russian military intelligence), shot down MH17, and was sent back across the border that night. As I’ve written previously, the timeline described by Bellingcat does not fit with the hapless-militiaman scenario very well. As the New York Times reported, “It is unlikely that anyone not connected with the Russian military would have been able to deploy an SA-11 missile launcher from Russia into a neighboring country.”

— While still admiting the possibility that the Buk crew acted on its own, the report shifts the emphasis to the once-unthinkable: that the missile launch was ordered by higher-ups:

…an investigation is conducted into the chain of command. Who gave the order to bring the BUK-TELAR into Ukraine and who gave the order to shoot down flight MH17? Did the crew decide for themselves or did they execute a command from their superiors? This is important when determining the offences committed by the alleged perpetrators.

As the New York Times put it, the JIT has signaled that it intends “to build an open-and-shut case against individual suspects and to diagram the chain of command behind the order to deploy and launch.”

One can just about imagine a wet-behind-the-ears lieutenant, newly trained and sitting nervously in the cab of his Buk TELAR, messing up and accidentally firing a missile at an unidentified target. But it is harder to imagine an experienced senior officer mistakenly giving the order. Indeed, the higher one goes up the chain of command, the less likely that the decision was made without explicit or implicit endorsement by an immediate superior. The implication, then, is that the order to shoot down MH17, if it did come from anywhere, came from the very top.

Continue reading Implications of the JIT’s MH17 report

Last week we discussed what we know about the first hour of MH370’s disappearance, based on primary radar data and the first Inmarsat BTO value. Today I’d like to talk about the BFO data and what it can tell us about MH370’s fate.

As longtime readers of this blog well know, the Burst Frequency Offset (BFO) is a type of metadata that measures how different the frequency of an Inmarsat signal is from its expected value. It is an important value to a communications satellite operator like Inmarsat because if the value gets too large, the system will be operating outside its approved frequency limit. One cause of such a change would be if a satellite begins wandering in its orbit, which indeed was the case with MH370. The fact that the Satellite Data Unit (SDU) aboard MH370 did not properly compensate for drift in the Inmarsat satellite overhead is the reason the BFO data contains a signal indicating what the plane was doing.

While each of the BTO values recording during the seven “pings” tells us fairly precisely how far the plane was from the satellite at that time, the BFO data points taken individually do not tell us much about the plane was doing. Taken together, however, they indicate three things:

1. After the SDU logged back on with Inmarsat at 18:25, the plane took a generally southern course. If we didn’t have the BFO data, we wouldn’t know, from the BTO data alone, whether the plane followed a path to the north or to the south (see above.)
2. The plane had turned south by 18:40. The BFO value at the time of the first incoming sat phone call at 18:40 indicates that the plane was traveling south.
3. At 0:19:37 the plane was in a rapid and accelerating decent.

However, as I’ve previously described, if all of these things were true, then the plane would have been found by now. So at least one of them must be false. In the course of my interview with him, Neil Gordon said that the ATSB is firmly convinced that #3 is true, and that as a result he suspects that #2 is not. Specifically, he points out that if the plane were in a descent at 18:40, it could produce the BFO values observed. Thus it is possible that the plane did not perform a “final major turn” prior to 18:40 but instead loitered in the vicinity of the Andaman Islands or western Sumatra before turning and flying into the southern ocean. If this were the case, it would result in the plane turning up to the northeast of the current search area. An example of such a route has been described by Victor Iannello at the Duncan Steel website.

It is worth nothing that such a scenario was explicitly rejected as unlikely by the Australian government when they decided to spend approximately $150 million to search 120,000 square kilometers of seabed. The reason is that it was deemed unlikely that the plane would just happen, by chance to be descending at the right time and at the right rate to look like a southward flight. For my part, I also find it hard to imagine why whoever took the plane would fly it at high speed through Malaysian airspace, then linger for perhaps as much as an hour without contacting anybody at the airline, at ATC, or in the Malysian government (because, indeed, none of these were contacted) and then continuing on once more at high speed in a flight to oblivion.

Well, is there any other alternative? Yes, and it is one that, though historically unpopular, is becoming imore urgent as the plane’s absence from the search area becomes increasingly clear: the BFO data is unreliable. That is to say, someone deliberately altered it.

There are various ways that we can imagine this happening, but the only one that stands up to scrutiny is that someone on board the plane altered a variable in the Satellite Data Unit or tampered with the navigation information fed back to the SDU from the E/E bay. Indeed, we know that the SDU was tampered with: it was turned off, then logged back on with Inmarsat, something that does not happen in the course of normal aircraft operation. It has been speculated that this depowering and repowering occurred as the result of action to disable and re-enable some other piece of equipment, but no one has every come up with a very compelling story as to what that piece of equipment might be. Given the evident problems with the BFO data in our possession, I feel we must consider the possibility that the intended object of the action was the SDU itself.

When I say BFO tampering has been “historically unpopular,” what I mean is that almost everyone who considers themselves a serious MH370 researcher has from the beginning assumed that the BFO data was generated by a normally functioning, untampered-with SDU, and this has limited the scenarios that have been considered acceptable. For a long time I imagined that search officials might know of a reason why tampering could not have occurred, but I no longer believe this is the case. When I questioned Inmarsat whether it was possible that the BFO data could have been spoofed, one of their team said “all Inmarsat can do is work with the data and information and the various testings that we’ve been doing.” And when I raised the issue with Neil Gordon, he said, “All I’ve done is process the data as given to me to produce this distribution.” So it seems that the possibility of BFO spoofing has not been seriously contemplated by search officials.

If we allow ourselves to grapple with the possibility that the BFO data was deliberately tampered with, we quickly find ourselves confronting a radically different set of assumptions about the fate of the plane and the motives of those who took it. These assumptions eliminate some of the problems that we have previously faced in trying to make sense of the MH370 mystery, but introduce new ones, as I’ll explore in upcoming posts.

One minute after MH370’s flight crew said “Good Night” to Malaysia air traffic controls, and five seconds after the plane passed waypoint IGARI at 1720:31 UTC, the plane’s Mode S signal disappeared from air traffic control screens. As it reached the border of the Ho Chi Minh Flight Information Region (FIR) approximately 50 seconds after that, the plane made an abrupt 180 degree turn. The radius of this turn was so small, and the ground speed so low, that it appears to have been effected via a semi-aerobatic maneuver called a “chandelle.” Similar to a “box canyon turn,” this involves climbing under power while also banking steeply. The maneuver offered WWI pilots a way to reverse their direction of flight quickly in a dogfight.

Chandelles are not a normal part of commercial 777 operation. They would not be used by pilots responding to in-flight fire.

The fact that such an aggressive maneuver was flown suggests that whoever was at the controls was highly motivated to change their direction of flight. Specifically, instead of going east, they wanted to go west.

At the completion of the left-hand U-turn the plane found itself back in Malaysia-controlled airspace close to the Thai border. It flew at high speed (likely having increased engine thrust and dived from the top of its chandelle climb) toward Kota Bharu and then along the zig-zaggy border between peninsular Malaysia and Thailand (briefly passing through the outer fringe of Thai airspace) before making a right-hand turn south of Penang. We know this “based mostly on the analysis of primary radar recordings from the civilian ATC radars at the Kuala Lumpur (KUL) Area Control Centre (ACC) and at Kota Bahru on the east coast of Malaysia; plus (apparently) the air defense radars operated by the RMAF south of Kota Bahru at Jerteh, and on Penang Island off the west coast,” according to AIN Online.

At 18:02, while over the small island of Pulau Perak, the plane disappeared from primary radar, presumable because it had exceeded the range of the radar at Penang, which at that point lay 83 nautical miles directly behind the plane. Then, at 18:22:12, another blip was recorded, 160 miles to the northwest.

The most-asked question about the 18:22 blip is: why did the plane disappear then? But a more pressing question is: why did it reappear? If the plane was already too faint to be discerned by Penang when it was at Pulau Perak, then how on earth could it have been detected when it was three times further away?

One possibility is that it was picked up not by Malaysian radar, but by the Thai radar installation at Phuket. An AFP report from March 2014 quoted Thailand’s Air Marshal Monthon Suchookorn as saying that Thai radar detected the plane “swinging north and disappearing over the Andaman Sea,” although “the signal was sporadic.”

At 18:22, the plane was approximately 150 miles from Phuket. This is well beyond the range at which Penang had ceased being able to detect the plane. What’s more, when the plane had passed VAMPI it had been only about 120 miles from Phuket. If it hadn’t seen the plane when it was at VAMPI, how was it able to detect it when it was 30 miles further? And why just for a momentary blip?

I don’t believe that, as some have suggested, the plane climbed, was detected, and then dived again. As Victor Iannello has earlier pointed out, the plane was flying at around 500 knots, which is very fast, and suggests a high level of motivation to be somewhere else, not bleeding off speed through needless altitude changes.

I propose that what happened at 18:22 was that the plane was turning. Entering into a right bank, the plane would turn its wings temporarily toward the Phuket radar station, temporarily presenting a larger cross section. Then,  when the plane leveled its wings to straighten out, the cross section would shrink, potentially causing the plane to disappear. Continue reading How MH370 Got Away

Last month Robyn Ironside, the National Aviation Writer at the News Corp Australia Network, published what struck me as an extremely important article in the Daily Telegraph about the work of scientist Patrick De Deckker, who had obtained a sample of a Lepas anatifera barnacle from the French judicial authorities and conducted an analysis to determine the temperature of the water in which the barnacle grew. A snippet:

The same 2.5 centimetre barnacle was used by both French and Australian examiners — but different techniques applied. “For my analysis, I used a laser to create little holes of 20 microns, over the length of the barnacles. In all we did 1500 analyses,” said Professor De Deckker.

Intrigued, I reached out to Ironside, asking if she could tell me more about De Deckker’s work. She very graciously did just that, and shared this extremely interesting nugget, a verbatim quote from De Deckker:

The start of the growth was around 24 degrees (Celsius) and then for quite some time, it ranged between 20 and 18 degrees (Celsius). And then it went up again to around 25 degrees.

This is surprising. The graphic above shows the water temperature in July 2005, which I take to be a rough proxy for the water temperature in March 2014. (I would be extremely grateful if someone could extract granular sea-surface temperature maps for March 2014 to July 2015 from NASA or NOAA databases available online.) It shows that the waters in the seabed search area are about 12-14 degrees Celsius. To find 24 degree water would mean trekking 1000 miles north, above the Tropic of Capricorn.

It has long been known that Lepas anatifera do not grow in waters below about 18 degrees Celsius, and that in order to begin colonizing the flaperon (if it began its journey in the search zone) would have had to first drift northwards and wait for warmer months and warmer latitudes. What’s peculiar is that this particular Lepas would have to have waited a good while beyond that, until the flaperon arrived in water six degrees above its minimum. As I’ve written before, Lepas naupali are common in the open sea and in general are eager colonizers of whatever they can glue their heads to.

Peculiarity number two is that after this period of initial growth the flaperon then found its way into significantly colder water, where most of its total growth took place. What’s weird is that every drift model I’ve ever seen shows currents going through warm water before arriving at Réunion. Where the heck could it have gone to find 18-20 degree water? And how did it then get back to the 25 degree waters of Réunion Island, where it finished its growth?

I’m frankly baffled, and am appealing to readers to ponder historical surface temperature data and drift models to help figure out what kind of journey this plucky Lepas might have found itself on.

 

Today I’d like to discuss some of the implications of what DSTG scientist Neil Gordon said in the course of the interview I published yesterday.

In particular, I’d like to look at what he told me about the ATSB’s interpretation of the 0:19:37 BFO value. Essentially, Gordon assures us that the experts have looked at what the manufacturers know about how these boxes work, and the only interpretation they can come up with is that the BFO value was the result of a very steep rate of descent–specifically, 5,000 fpm at 00:19:29 and then 12,000 to 20,000 fpm at 00:19:37. This is got a gentle deterioration; it’s accelerating at about 1/2 g, so that in another 8 seconds, at that rate, the descent will be at 19,000 to 35,000 fpm, that is to say going straight down at 187 to 345 knots. Remember that the plane had already been losing speed and altitude for five to fifteen minutes before the second engine even flamed out, and losing more altitude in the subsequent two minutes before the 00:19:29 ping was logged. Thus, both velocity and acceleration point to a situation in which the plane will be hitting the surface in short order. Bearing in mind that the plane would be in a spiral dive if unpiloted, I can’t see how it could have traveled more than 5 nm from the last ping, let alone 15nm, let alone 40 nm. It would have hit soon, and it would have hit hard.

One possible explanation would be the idea that the plane was in a phugoid: plunging quickly, then rising again, then plunging again. But as I wrote in a previous post, simulator runs by Mike Exner suggest that these extreme rates of descent are characteristic of the later stages of an unpiloted post-flameout plunge, when phugoid effects are overwhelmed. Thus, if the ATSB is correct in interpreting the final BFO value as a very steep plunge–as Gordon assures us they must–then the plane should be well within 15 nautical miles of the seventh arc.

The chart above (based on the invaluable work of Richard Cole) shows a band of seabed, marked in red, defined by an outer border that is 15 nm beyond the 7th arc and an inner border that is 15 within the 7th arc. As you can see, this band has almost entirely been searched out to the 99% confidence level as defined in Figure 2 of my previous post (located at the intersection of the 7th arc and 94.85 degrees east). All that remains is a rectangle approximately 17 km wide and 150 km long, for a total area of 2,550 sq km.

According to Figure 3 in that same post, the DST calculates that the probability that the plane crossed the seventh arc northeastward of 96.75 degrees east longitude is effectively zero. To search to this longitude would require covering another 3,700 or so sq km. Thus, to cover all the seabed that MH370 could plausibly have reached, if the ATSB’s BTO and BFO analysis is correct, would require another 6,250 sq km of seabed scanning, which is more or less what the ATSB has been planning to search anyway. Unfortunately, the search at present is not taking place in either of these remaining areas.

As I see it, there are four possibilities at this juncture:

1. Both the BFO and the BTO analysis are correct, and the plane is lying somewhere in the remaining 6,250 sq km described above.
2. The BTO analysis is correct, but the BFO analysis is wrong. In this case, the plane was not necessarily descending with great rapidity, and instead might have been held in a glide, and is most likely in “Area 1” shown above.
3. The BTO analysis is incorrect, and the BFO analysis is correct. The plane was indeed descending very rapidly during the last ping, but the plane was further to the northeast somewhere in “Area 2.”
4. Both BTO and BFO analysis are incorrect. The plane could be just about anywhere.

I happen to believe that the DSTG knows what it is doing, and that 2 through 4 are not the case. On the other hand, the unsearched areas remaining are at the far fringes of likelihood, and so don’t feel that #1 is a high-probability option, either. No doubt some will argue that the plane might have been overlooked within the area already searched, despite assurances from officials that if it was there they would have seen it.

Frankly, we’re running out of compelling options.

When Australia’s Transport Safety Board (ATSB) was tasked with finding missing Malaysia Airlines flight 370, it tapped another arm of the government, the Defense Science & Technology Group (DSTG), to tell it where to look. There a team led by Dr. Neil Gordon devised a mathematical approach based on Bayesian analysis to weigh all the possible routes that the Boeing 777-200ER could have flown, given the seven Inmarsat “pings,” the plane’s fuel load, environmental conditions, and the different settings available on the autopilot. From this they derived a probabilistic “heat map” of where the plane might have wound up (Fig 1, above). The results showed that the jet most likely flew fast and straight, at high altitude, before running out of fuel and crashing. It was this analysis that allowed the ASTB to define the search area currently being scoured for traces of seabed wreckage. Yet, with less than 10 percent of the area left to be searched and not a trace found, it now appears they looked in the wrong place. Earlier this summer, the three nations responsible for the investigation—Malaysia, China, and Australia—jointly announced that they would not be extending the search after the last portion is completed this fall. Last month Dr. Gordon went on record for the first time to explain what might have gone wrong and where the next place to look for the plane should be. His answers formed the basis of an article for Popular Mechanics; for the readers of this blog I present a less filtered version of what Dr Gordon had to say.

One of the crucial decisions you had to make was how to treat the 18:22 radar return. In your report, you wrote, “The final reported position from radar was very at very long range from the sensor and there was a long time delay between it and the penultimate radar report. The report is at long range and it is likely to have rather poor accuracy because of the angular errors translates into large location errors at that range.” Are you confident that that radar return is not anomalous, it actually comes from the plane?

You’ve got to understand what our job in this investigation is. Our job is to take the data as presented to us by the accident investigators and project a trajectory from that.

Was there any explanation or speculation on why a plane would be detected at that point but not before or after?

I guess it was that they’ve just got snapshots off the radar screen. I’m speculating here but I would imagine they’ve recorded a video of the screen but they don’t necessarily have a digital backup of the measurements.

Continue reading Exclusive Interview with Top MH370 Search Mathematician Neil Gordon

In last month’s New York magazine article about Zaharie Ahmad Shah’s flight simulator, I cautioned against treating the recovered data as a smoking gun:

…it’s not entirely clear that the recovered flight-simulator data is conclusive. The differences between the simulated and actual flights are significant, most notably in the final direction in which they were heading. It’s possible that their overall similarities are coincidental — that Zaharie didn’t intend his simulator flight as a practice run but had merely decided to fly someplace unusual.

What I failed to question was the report’s assumption that the six points all belonged to a single flight path. On closer examination that assumption seems ill supported. Rather, it seems more likely that the six points were recorded in the course of  two or possibly three separate flights. They were interpreted as comprising a single flight only because together they resembled what investigators were hoping to find.

The first four points do appear to show a snapshots from a continuous flight, one that takes off from Kuala Lumpur and climbing as it heads to the northwest. Between each point the fuel remaining decreases by a plausible amount. Each point is separated from the next by a distance of 70 to 360 nautical miles. At the fourth point, the plane is at cruise speed and altitude, heading southwest in a turn to the left. Its direction of flight is toward southern India.

The fifth and sixth points do not fit into the pattern of the first four. For one thing, they are located more than 3,000 miles away to the southeast. This is six or seven hours’ flying time. Curiously, at both points the fuel tanks are empty. Based on its fuel load during the first four points, the plane could have flown for 10 hours or more from the fourth point before running out of fuel.

The fifth and sixth points are close together—just 3.6 nautical miles apart—but so radically different in altitude that it is questionable whether they were generated by the same flight. To go directly from one to the other would require a dive so steep that it would risk tearing the aircraft apart.

The picture becomes even more curious when we examine the plane’s vertical speed at these two points: in each case, it is climbing, despite having no engine power.

The ATSB has speculated that in real life MH370 ran out of fuel shortly before 0:19 on March 8, and thereafter entered into a series of uncontrolled porpoising dives-and-climbs called phugoids. In essence, a plane that is not held steady by a pilot or autopilot, its nose might dip, causing it to speed up. The added speed willl cause the nose to rise, and the plane to climb, which will bleed off speed; as the plane slows, its nose will fall, and the cycle will continue.

Could a phugoid cause a plane to climb—663 feet per minute at point 5, and 2029 feet per minute at point 6? The answer seems to be yes for the fifth point and no for the sixth. Reader Gysbreght conducted an analysis of 777 flight-simulator data published by Mike Exner, in which an airliner was allowed to descend out of control from cruise altitude in the manner that the ATSB believes MH370 did.

A diagram produced by Gysbreght is shown at top. The pink line shows the plane’s altitude, starting at 35,000 feet; the blue line shows its rate of climb. Worth noting is the fact that the phugoid oscillation does indeed cause the plane to exhibit a small positive rate of climb soon at first. But by the time the plane reaches 4000 feet — the altitude of the sixth point — the oscillation has effectively ceased and the plane is in a very steep dive.

Gysbreght concludes:

As expected for a phugoid, the average rate of descent is about 2500 fpm, and it oscillates around that value by +/- 2500 fpm initially. The phugoid is apparently dampened and the amplitude reduces rapidly. I was slightly surprised that it reaches positive climb values at all. Therefore I think that 2000 fpm climb is not the result of phugoid motion.

Not only is the plane climbing briskly at the sixth point, but it is doing so at a very low airspeed—just above stall speed, in fact. If the pilot were flying level at this speed without engine power and pulled back on the controls, he would not climb at 2000 feet per minute; he would stall and plummet. In order to generate these values, the plane must have been put into a dive to gain speed, then pulled up into a vigorous “zoom climb.” Within seconds after point six, the simulated flight’s speed would have bled off to below stall speed and entered into an uncontrollable plunge.

Perhaps this is why Zaharie chose to record this particular point: it would have been an interesting challenge to try to recover from such a plunge at low altitude.

What he was doing at points 5 and 6, evidently, was testing the 777 flight envelope. This might seem like a reckless practice, but I think the opposite is the case. From time to time, airline pilots do find themselves in unexpected and dangerous conditions. For instance, as Gysbreght has noted, “On 7 october 2008 VH-QPA, an A330-303, operating flight QF72 from Singapore to Perth, experienced an In-flight Upset west of Learmonth, West Australia. The upset was caused by a freak combination of an instrumentation failure and an error in the flight control software, which resulted in an uncommanded pitch-down. The vertical acceleration changed in 1.8 seconds from +1 g to -0.8 g.” It would be better to experience a situation like this for the first time in a flight simulator in one’s basement, rather than in midair with a load of passengers and crew.

What Zaharie clearly was not trying to do was to fly to McMurdo Station in Antarctica, as some have speculated.

For one thing, while a 777 is fully capable of flying from Kuala Lumpur to Antarctica, it was not carrying enough at point 1 to make the trip. And if one were trying to reach a distant location, one would not do so by running one’s tanks dry and then performing unpowered zoom climbs.

The misinterpretation of the flight simulator data offers a couple of cautionary lessons. The first is that we have to be careful not to let a favored theory color our interpretation of the data. The investigators believed that MH370 flew up the Malacca Strait and wound up in the southern Indian Ocean, and they believed that Zaharie was most likely the culprit; therefore, when they found data points on his hard drive that could be lumped together to form such a route, that’s what they perceived.

A second lesson is that we cannot uncritically accept the analysis made by officials or by self-described experts. Science operates on openness. If someone offers an analysis, but refuses to share the underlying data, we should instinctively view their claims with suspicion.

Last month, I published an article in New York magazine about a secret Malaysian police report which included details of a simulated flight into the southern Indian Ocean. As Victor Iannello revealed in a comment earlier today, that information came from French journalist Florence de Changy, who had come into possession of the full police report but only shared a portion of it with me.

I have not seen the full report, but would very much like to, because I would like to form my own judgement of what they mean, and I think everyone who is interested in trying to figure out what happened to the missing plane, including the next of kin, are entitled to the same. Some people who have read the full reports have suggested that they give the impression that the recovered simulator files do not in context seem all that incriminating. Other people who have seen the full report have told me that the report contains material that makes it hard to doubt that Zaharie is the culprit. Of course, it’s impossible to rely on someone else’s say-so. We need to see the full report.

The reason I am writing this post now is that earlier today Florence published an article in Le Monde in which she describes having the full report as well as another, 65-page secret document on the same topic. Meanwhile, another French newspaper, Liberation, has also published an article indicating that they, too, have a copy of the report. And private correspondence between myself and a producer at the television network “France 2” indicates that he has as well.

Meanwhile, I know that independent investigators here in the US have the documents as well.

At this point, the secret documents are not very secret. Someone within the investigation has been leaking them like crazy, obviously with the intention that their contents reach the public. My understanding is that this source has placed no restrictions on their use. So journalists and independent investigators who have copies of these documents need to do their duty and release them — somehow, anyhow. Some people that I’ve begged and implored to do so have said that they fear legal ramifiations. Well, if it’s illegal for you to have these documents, then you’ve already broken the law. Use Wikileaks or another similar service to unburden yourself.

Free the data!

UPDATE 8/14/16: Apparently Blaine Alan Gibson has the document, too, according to a rant he post on Facebook. He reveals that the entire set of documents is 1,000 pages long.