Identifying deforestation in Brazil using multiresolution satellite data

Abstract

The use of multiresolution satellite data to monitor deforestation on a continental/subcontinental scale is examined. MSS, local area coverage (LAC), global area coverage (GAC), and GOES data were applied to the study of deforestation in Rondonia, Brazil; the characteristics of these sensors are described. The probability thresholding and vegetation-index thresholding procedures used to process the data in order to differentiate between forest from nonforest are analyzed. The LAC, GAC, and GOES data are compared to MSS data. It is observed that GOES data is not useful for monitoring colonization projects due to excessive noise in the data; the GAC data is only applicable in large areas of contiguous forest clearing; the MSS data when available is applicable as a ground data reference source for differentiating cleared areas from primary; the LAC data are capable of delineating colonization clearings; and the probability thresholding procedure differentiates forest from nonforest more accurately than the vegetation-index procedure. The data reveal that the LAC data combined with the probability threshold procedure provide the best data-source/classification-procedure combination.