Abstract
Bio-energy plays an important role in the global energy mix, but especially in Brazil, where ethanol from sugarcane provides for more than 40% of the automotive fuel. Brazil, being the largest sugarcane producer and the second largest ethanol producer in the world, has seen a rapid expansion of sugarcane over the last decades, with Sao Paulo state contributing more than 60% to the nation's sugarcane production and experiencing a doubling in sugarcane acreage between 2003 and now. The direct and indirect land cover change induced by this expansion may lead to unforeseen and challenging socio-economic and biodiversity impacts elsewhere in the country. Together with partnering universities (mainly Utrecht University and UniCamp) we develop automatic land cover change monitoring algorithms operating on space-based remote sensing data and apply sustainability models to research the socio-economics effects of these changes. Since we have to apply these monitoring algorithms on large areas (state level and region level) and over multiple years, high requirements are put on processing power and data storage. Just last weeks we bought 20000 core hours on Cartesius via Frans Broos (TUDelft) to process over Sao Paulo state (roughly the size of the UK) over 13 years (2003-2015) for one optical sensor with a spatial resolution of 30mx30m and a potential temporal resolution of 16-days. On our own computer or our cluster this would take months, on Cartesius using 8 thin nodes it took 5 days. And we have the first results. In the foreseeable future we like to include more sensors and other states, which means we like to use Cartesius again. Since we already proved the success of our algorithms on Cartesius we are confident this project is feasible. LISA is not an option anymore since we started there, but were moved to Cartesius due to the large load on the bandwidth when reading in the remote sensing data stack. We still ask for a pilot project though because the expansion to other sensors and states brings other technical challenges we like to test on Cartesius. Thank you.
Netherlands