I was reading (as you do) the Rochester Institute of Technology pages on Haiti geodata and disaster-risk analysis and noted a side-project: the Blue Tarp Detection Scheme, where Dr Dave Messinger used remote sensing and spectral analysis to detect and plot the locations of blue tarps. As anyone who has flown over a disaster zone or IDP camp knows plastic sheeting/tarpaulins/tarps have high contrast and are handed out during emergencies, so in many ways they make a find proxy for the presence of displaced populations, and possibly for the movement or change in these populations. I’d love to see more analysis of this as there are a number of factors that would affect a tarp count as a reliable indicator, including age, presence of temporary shelters before an emergency, the use of tarps for something other than shelter, and also the market that rapidly springs up around these products where they’re traded for cash and other goods. Nonetheless I’d expect that you could control for much of this, and generate some pretty interesting data. After a quick Google search I couldn’t find any papers that explored this use of remotes sensing which surprises me, I expect I’m just looking in the wrong place.
One reason that I’m so interested in this is that it’s pretty similar to a scheme I planned but never implemented for Darfur in 2005: at the UN Joint Logistics Center we were trying to both confirm changes to camp populations (which could be on the order of 80,000 in less than a month in some cases), and also try and understand how and when people were returning to their villages. We were also procuring and distributing hundreds of thousands of units of plastic sheeting (tarps) every season, accounting for something like 90% of the supply. With a virtual monopoly on distribution why not use a different colour each year, or label them somehow with symbols that could be detected remotely. With six-monthly remote sensing data collection and some relatively simple multi-spectral analysis, plus a bit of ground-truthing, you could expect to track how newly distributed orange tarps appear in the camp and then migrate out between camps or to villages, either through trade or by people returning (perhaps temporarily) to their homes.
Unfortunately this was yet another of those great ideas that never went anywhere: we never did anything about it, not even at the level of using the presence or change in tarps as a measure of camp population, or looking at seasonal and annual trends. There are lots of reasons it mightn’t have worked very well, and it also raises a few ethical issues particularly if we printed symbols on the tarps, not the least because we’d be potentially providing the government and other forces with a means to track movements.
I’ve gone on record before saying that in all the years I’ve been involved in the field of humanitarian information I struggle to think of more than three examples (outside of pure cartography) of where remote sensing and GIS analysis were used by humanitarian agencies to support operational – or even planning – decisions during a crisis. If we’d been more creative in 2005 there might just have been four examples.
Update (11:22 April 8 2011): Both John Crowley and Gisli Olafsson immediately suggested putting QR codes on the tarps and then using something like Walking Papers to process them. All good, but the logistics (and cost) of that printing (double sided too) is scaring me.
