Modis time series to assess pasture land

Land use conversion is a key factor in the mitigation of GHG emission. Maximum mitigation can be achieved when degraded pasture land is converted to biofuel crops. Remote sensing images, and in particular the MODIS time series data, have a great potential to asses degraded pasture land. This work has the objective to identify pasture land and its different levels of degradation in Mato Grosso do Sul state, Brazil. MODIS time series were used to obtain vegetation indices and fraction images. The wavelet technique was applied at various levels of decomposition to extract the input parameters in the WEKA J48 classifier. Pasture land was well distinguished from Cerrado. The distinction among different pasture land presented lower performance with best results for pasture with invasive plants followed by good pasture. Pasture land with bare soil patches and termite mounds were not distinguished from other classes of pasture.