Integrated assessment of future land use in Brazil under increasing demand for bioenergy

Environmental impacts of a future increase in demand for bioenergydepend on the magnitude, location and pattern of the direct and indirectland use change of energy cropland expansion. Here we aim at 1)projecting the spatiotemporal pattern of sugar cane expansion and theeffect on other land uses in Brazil towards 2030, and 2) assessing theuncertainty herein. For the spatio-temporal projection, four modelcomponents are used: 1) an initial land use map that shows the initialamount and location of sugar cane and all other relevant land useclasses in the system, 2) an economic model to project the quantity ofchange of all land uses, 3) a spatially explicit land use model thatdetermines the location of change of all land uses, and 4) variousanalysis to determine the impacts of these changes on water,socio-economics, and biodiversity. All four model components are sourcesof uncertainty, which is quantified by defining error models for allcomponents and their inputs and propagating these errors through thechain of components. No recent accurate land use map is available forBrazil, so municipal census data and the global land cover map GlobCoverare combined to create the initial land use map. The census data aredisaggregated stochastically using GlobCover as a probability surface,to obtain a stochastic land use raster map for 2006. Since bioenergy isa global market, the quantity of change in sugar cane in Brazil dependson dynamics in both Brazil itself and other parts of the world.Therefore, a computable general equilibrium (CGE) model, MAGNET, is runto produce a time series of the relative change of all land uses givenan increased future demand for bioenergy. A sensitivity analysis findsthe upper and lower boundaries hereof, to define this component's errormodel. An initial selection of drivers of location for each land useclass is extracted from literature. Using a Bayesian data assimilationtechnique and census data from 2007 to 2012 as observational data, themodel is identified, meaning that the final selection and optimalrelative importance of the drivers of location are determined. The dataassimilation technique takes into account uncertainty in theobservational data and yields a stochastic representation of theidentified model. Using all stochastic inputs, this land use changemodel is run to find at which locations the future land use changesoccur and to quantify the associated uncertainty. The results indicatethat in the initial land use map especially the shape of sugar cane andother land use patches are uncertain, not so much the location. From theeconomic model we can derive that dynamics in the livestock sector playa major role in the land use development of Brazil, the effect of thisuncertainty on the model output is large. If the intensity of thelivestock sector is not increased future projections show a large lossof natural vegetation. Impacts on water are not that large, except whenirrigation is applied on the expanded cropland.