Comparison of Envisat ASAR deforestation monitoring in Amazônia with Landsat-TM and ALOS PALSAR images

Abstract. Envisat ASAR image mode (IMS) and wide-swath (WS) scenes were studied for their usability for monitoring deforestation in the Amazon Rainforest. High-resolution (30m) IMS of the dry season (August) of two subsequent years, 2004 and 2005, showed unambiguous changes in the land-cover with a strong decrease in backscatter for deforested areas. In addition, SAR shadows of trees bordering the deforested areas, as well as strong backscattering from trees that act as corner reflectors on the opposite side of the deforestation field were clearly visible. 100m-resolution WS data from dry and wet seasons were also visually analyzed in comparison with Landsat TM and ALOS PALSAR images. Dry season WS images showed good agreement, but wet season WS images showed little agreement between low backscatter and deforestation areas by visual interpretation. Because of the 100m-resolution as well as the weak vegetation penetration of C-band, single WS scenes do not seem reliable for detecting deforestation if the area is not completely cleared. We suppose that using the high revisit capability of Envisat ASAR would however improve its capability. A multi-sensor approach, combining optical, L- and C-band SAR data, could help to detect different stages of deforestation and would definitely improve the all-weather capability of an operational deforestation monitoring system