SIO and Reverse Drift

If you follow @MikeChillit in whole or part on Twitter, you probably know I’ve tinkered around with drift patterns in the Southern Indian Ocean off and on. I’ve done that mostly as a reaction to the efforts of others who appear to be deliberately conducting analysis in a way that concludes Australia has been searching in the “correct location” all this time, and the reason it hasn’t found anything is because there isn’t anything to find.

Admittedly, the search for MH370 has been complicated. We don’t have much information about what may have happened, let alone what actually happened. But that should not have paralyzed the effort to find the plane. Nearly everyone agrees the “Seventh Arc” metric developed by Inmarsat from BTO exchanges between the plane and its 3-F1 satellite is probably reasonably useful. It marks a mostly north-to-south path over the South Indian Ocean south of Java that runs some 4,000 km before reaching a spot near -40° S, where most also agree the plane could not have flown beyond.

But instead of setting up to methodically search the entire 4,000 km arc that passes north, west, and a bit south of Western Australia, ATSB has persisted in searching only the most distant 1,000 km: leaving three-quarters of the Seventh Arc unsearched. Why? ATSB has argued at times it is because “all of those who know anything about anything say that’s the only place on the Arc MH370 could be”. But those people “who know everything there is to know” never stand up and agree. They remain nameless, faceless ghosts: figments of Australia’s now badly broken image and imagination. In essence, Australia has stated time and time again that if the plane is farther north, it needs to get up and drift down to the search area so Fugro can find it. This doesn’t just border on the absurd, it is certifiable.

So, after trying hard to avoid investing the personal time and effort into putting together a proper picture of drift patterns in SIO, I have now reversed course and have done just that. It is / was a huge task. A lot of drift data is involved, but it also requires another major consideration for which I have yet to find much data: wind speed and direction across SIO between the equator and -40° S. I have estimates for the northwest portion of the area that appear to be quite adequate, but I still have nothing for the southeast (west of Australia), or for the southwest along the South African coast. So, the graphics I present below will change when I obtain that information. I know in general terms what it will look like. Those who hope it will show that drift moves from SE Australia to the Mascarene Island areas will be disappointed. It just doesn’t. Winds along the Western Australian coast flow mostly to the northeast and north. They mostly begin over the Circumpolar Current south of -40° S where a large chunk breaks off and curves up the WA coastline. A fair amount of water does the same thing, but quickly wraps back to the SW and recycles.

How Drift Moves West of Australia

At a minimum, those Western Australia northbound winds move debris toward Java, not the Mascarenes. Those winds begin falling apart between Geraldton and Exmouth where they catch westbound winds from Timor Sea. A small amount of debris re-curves to the SW, and some curves up to the NW and the Mascarenes, but it is nothing like the drift already in that portion of the Indian Ocean.

It is indeed possible for a few things to drift from SW Australia to Pemba Island in the NW, and beyond, but it is not common. It is not the default location drift moves toward in that part of the ocean. So, finding a flaperon on Reunion Island COULD be a result of drift from the Perth area in relatively rare circumstances, but no one would expect an entire tranche of debris from the Perth area to end up all the way up to Madagascar and eventually East Africa. That kind of debris relocation / accumulation can only come from somewhere west of its most southerly location: currently southern Madagascar, Reunion Island, and Rodrigues Island. Moreover, the debris that has not been stopped at those Island locations has drifted up around the north end of Madagascar, then south via the Mozambique Channel, and / or further north yet to Pemba Island and beyond.

That is simply how it works out there and no one is going to put the remains of MH370 SW of Perth with bad drift modelling and convoluted claims. The plane is wherever it is, and that is certainly not where Australia has camped out for the better part of three long years. Everyone is tired of that horrible close-to-the-vest claim. If anyone with credentials really thinks the plane is where the search is, they need to stand up and take questions. They need to stop hiding behind the press officer in Canberra.

The following chart was constructed from nearly 600,000 NOAA drift records. That is one-quarter of the 2.4 million records for the 35 year period between the early 1980s and early 2016. That is about 4 records per drifter per day for as long as each drifter was functional: several years in most instances. Only the midnight UTC record was retained and used for analysis purposes.

The dataset is formally known as “envids spatial temporal” data. Each record has twelve data elements. Those of primary interest for drift analysis purposes were “ve” (velocity east), “vn” (velocity north), and “speed”. They were used to obtain ‘heading’ or ‘bearing’. GPS coordinates, also in each record, were used to verify sampled outcomes. All ‘resultant’ data were dropped into a 1° by 1° matrix for final analysis. Summary metrics were developed for drift direction and other factors. The effects of small local currents are necessarily included in the analysis at the 1° by 1° level, and can be seen in vector fluctuations. Those effects could be further enhanced by reducing matrix size to 0.5° by 0.5°, but that level of precision does not appear to be necessary. Indeed, the analysis does not leave ambiguity about how drift works in the South Indian Ocean.

As the analysis progressed, currents alone tend to move fairly uniformly from the northeast (Timor Sea area) to the southwest (tip of South Africa). There is plenty of evidence to support that proposition, not the least of which is that it is known that a large volume of water pushes into the South Indian Ocean from the Pacific Ocean via Timor Sea and Banda Sea; it has to go somewhere. It tends to move toward South Africa and from there into the South Atlantic and some into the Circumpolar Current.

If one were to attempt to reverse drift at that first level of analysis (water current flow), it would point to Sumatra and Java as the likely origin of MH370 debris. But that would completely ignore the significant contributions from wind, which tends to move at about 14 knots to the northwest.

When wind and current vectors were combined, the result is as you see it below. Important to note it is consistent with most independent drift modeling that has emerged in the past year or so from Geomar and others. NOAA’s Rick Lumpkin also did an informal reverse drift model for the author, and it was entirely consistent with these results.


This is the reverse drift graphic result from about 600,000 drifter records over a period of about 35 years. It includes some of the effects of wind, but will be refined as more precise wind data become available. The chart is intended to make it possible to follow the vectors to find the likely starting point. For example, from Reunion Island the likely starting point on the Seventh Arc is a short distance northeast of Zenith Plateau. Numerous analyses lately have concluded that the area between Batavia Seamount and Zenith Plateau appears to be the most likely start point for Reunion Island debris. There is no known way to pinpoint the origin more precisely.

180 Days Prior to Reunion

This informal chart was prepared by NOAA’s Dr. Rick Lumpkin at the author’s request in February 2016. It’s purpose was to use NOAA drifter data to approximate the location of MH370 drift six months after the crash. The reverse drift path is clearly to the northeast of Reunion Island, and appears headed directly for Sunda Strait.

The graphic directly above prepared by Rick Lumpkin does not include wind vector data. If it did, it would move directly toward Timor Sea and Exmouth, Australia.

It is the author’s understanding that yet another drift model is due to be released soon by a Western Australia professor who has prepared several analyses. Almost all, so far, have been at odds with those prepared by Geomar and others.

Comments are closed.