Google Massively Automates Tropical Deforestation Detection

Landcover change analysis has been an active area of research in the remote sensing com­mu­nity for many years. The idea is to make com­pu­ta­tional pro­to­cols and algo­rithms that take a couple of digital images col­lected by satel­lites or air­planes, turn them into land­cover maps, layer them on top of each other, and pick out the places where the land­cover type has changed. The best pro­to­cols are the most precise, the fastest, and which can chew on multiple images recorded under dif­ferent con­di­tions. One of the favourite appli­ca­tions of land­cover change analysis has been defor­esta­tion detec­tion. A par­tic­u­larly popular target for defor­esta­tion analysis is the tropical rain­forests, which are being chain­sawed down at rates which are almost as dif­fi­cult to com­pre­hend as it is to judge exactly how bad the effects of their removal will be on bio­log­ical diver­sity, plan­e­tary ecosystem func­tioning and climate stability.

Google has now gotten itself into the envi­ron­mental remote sensing game, but in a Google-​​esque way: mas­sively, ubiq­ui­tously, com­pu­ta­tion­ally inten­sively, plau­sibly benignly, and with probable long-​​term finan­cial benefits. They are now running a program to vacuum up satel­lite imagery and apply land­cover change detec­tion optomized for spotting defor­esta­tion, and for the time being targeted at the amazon basin. The public doesn’t cur­rently get access to the results, but pre­sum­ably that access will be rolled out once Google et al are con­fi­dent in the system. I have to hand it to Google: they are tech­ni­cally careful, but polit­i­cally aggres­sive. Amazon defor­esta­tion is (or should still be) a very polit­ical topic.

The par­tic­ular land­cover change algo­rithms they are using are appar­ently the direct product of Greg Asner’s group at Carnegie Institution for Science and Carlos Souza at Imazon. To signal my belief in the impor­tance of this project I’m not going to make a joke about Dr. Asner, as would normally be required by my back­ground in the Ustin Mafia. (AsnerLAB!)

From the Google Blog:

We decided to find out, by working with Greg and Carlos to re-​​implement their software online, on top of a pro­to­type platform we’ve built that gives them easy access to ter­abytes of satel­lite imagery and thou­sands of com­puters in our data centers.”

That’s an inter­esting comment in it’s own right. Landcover/​landuse change analysis algo­rithms pre­sum­ably require a rea­son­ably general-​​purpose com­puting envi­ron­ment for imple­men­ta­tion. The fact that they could be run “on top of a pro­to­type platform … that gives them easy access to … com­puters in our data centers” suggests that Google has created some kind of more-​​or-​​less general purpose abstrac­tion layer than can invoke their unprece­dented com­puting and data resource.

They back that comment up in the bullet points:

Ease of use and lower costs: An online platform that offers easy access to data, sci­en­tific algo­rithms and com­pu­ta­tion horse­power from any web browser can dra­mat­i­cally lower the cost and com­plexity for tropical nations to monitor their forests.”

Is Google sig­naling their devel­op­ment of a com­mer­ical super­com­puting cloud, à la Amazon S3? Based on the further marketing-​​speak in the bullets that follow that claim, I woud say absolutely yes. This is a test project and a demo for that business. You heard it here first, folks.

Mongobay points out that it’s not just tropical forests that are quietly dis­s­a­pearing, and Canada and some other devel­oped coun­tries don’t do any kind of good job in aggre­gating or pub­li­cally mapping their own enormous defor­esta­tion. I wonder: when will Google point its detec­tion program at British Columbia’s end­lessly exanding network of just-​​out-​​of-​​sight-​​of-​​the-​​highway clearcuts? And what facts and figures will become readily acces­sible when it does?


View Larger Map

Mongobay also infers that LIDAR might be involved in this par­tic­ular process of detecting land­cover change, but that wouldn’t be the case. Light Detection and Ranging is commonly used in char­ac­ter­izing forest canopy, but it’s still a plane-​​based imaging tech­nique, and as such not appro­priate for Google’s world-​​scale ambi­tions. We still don’t have a credible hyper­spec­tral satel­lite, and we’re nowhere close to having a LIDAR satel­lite that can shoot reflecting lasers at all places on the surface of the earth. Although if we did have a satel­lite that shot reflecting lasers at all places on the surface of the earth, I somehow wouldn’t be sur­prised if Google was responsible.

Which leads me to the point in the Google-​​related post where I confess my ner­vous­ness around GOOG taking on yet another service — envi­ron­mental change mapping — that should probably be handled by a demo­c­ra­t­i­cally directed, pub­li­cally account­able orga­ni­za­tion rather than a publically-​​traded for-​​profit cor­po­ra­tion. And this is the point in the post where I admit that they are taking on that function first and/​or well.

1 comment:

leave a comment