Use of Landsat Tm Fraction Images to Quantify the Optical and Sar Data Relationships for Land Cover Discrimination in the Brazilian Amazonia

This paper presents a comparison between L-band HH JERS-1 SAR and Landsat TM data for land cover assessment in an Amazonian test site. For this, a study area located in the south of Rondônia State, western Brazilian Amazonia, was selected. The Landsat TM data acquired on August 1 1997 were converted to vegetation, soil, and shade fraction images, using a Linear Spectral Mixing Model. These fraction images are produced to enhance different characteristics of land cover, expressed as different mixtures of a few number of image components. The JERS-1 data acquired on August 08 1997 for this investigation were processed to standard level 2.1, i.e., 16 bits ground range, 3 looks and a pixel spacing of 12.5 m. To compare JERS-1 and TM fraction images, a regression analysis technique was used. The multiple regression analysis showed coefficient of determination (R) of 0.86 between SAR and the fraction images. The visual inspection of JERS-1 SAR image showed the high contrast between forest (medium gray) and cleared areas (dark gray). There is a direct relationship between SAR and shade fraction images and an inverse relationship between SAR and soil fraction images. Based on these relationships, it can be pointed out the utility of SAR data as an alternative data source for land cover classification in the Amazonian region.