Mapping evergreen forests in the Brazilian Amazon using MODIS and PALSAR 500-m mosaic imagery

In this study, we evaluate a methodology that uses dual-polarization L-band SAR 500-m mosaic PALSAR imagery to identify and map forests in the Brazilian Amazon and an algorithm that uses time-series MODIS imagery to map evergreen forest. IKONOS images were used to evaluate forest maps derived from PALSAR and MODIS imagery. A comparison between the PALSAR forest map and IKONOS forest maps shows that 91.4% of PALSAR-derived forest pixels had greater than 60% IKONOS-derived forest area. We also compared the PALSAR-derived forest map with the MODIS-derived evergreen forest map. Out of 1999 evergreen forest pixels in the MODIS evergreen forest map (the areas covered by the 11 IKONOS imagery), 1934 pixels were identified as forest by the PALSAR forest map, approximately 96.7% agreement. The results of this study highlight the potential of combining PALSAR and MODIS data for identifying and mapping evergreen forests in the Amazon.