FOREST DEGRADATOIN: HOW WATER STRESS, WILDFIRES, AND CLIMATIC FACTORS INFLUENCE SOUTHERN AMAZONIAN FORESTS

Remote sensing, satellite imagery, spectral response, Hyperion

Forest degradation can occur through several factors such as prolonged droughts, fires, wind storms, among others. Although these processes occur on a large scale in tropical forests and are in full intensification due to climate change, the use of surface reflectance for mapping forest degradation is still quite challenging. In this study we used multispectral data from the Landsat satellite and hyperspectral data from the EO-1 satellite to quantify forest degradation caused by fires in an area of experimental fires (2004 to 2010) on the southern of Amazonia. We used 18 Vegetation Indices (VI) being one specific for Landsat and 12 specific for EO-1, calculated for the burnt and control treatments. The most sensitive VIs of Landsat to changes in vegetation were the GRND and NDII, reaching values 87% and 48% lower than the unburned vegetation, respectively. From EO-1, PSRI was the most sensitive (94% higher than the control). Both the Landsat and EO-1 VIs detected the highest values at the end of the experiment (2010) and for the edge of the vegetation, which was evidenced by the drier and flammable vegetation detected by the increase in the PSRI and MSI VIs and the drop in NDVI and NDII. The use of hyperspectral VIs allowed the visualization of different processes occurring in the burned vegetation, such as the increase of anthocyanins and carotenoids and the decrease of photosynthesis and vegetation water. Our results show that hyperspectral VIs associatedwith carotenoid/chlorophyll changes are the key to detecting vegetation stress and degradation associated with forest fires in early years through satellite imagery.