Close Window

In this study, statistical multitemporal analysis was applied to evaluate the capability of reflectance, vegetation indices {lsqb}normalized difference vegetation index (NDVI) and soil adjusted vegetation index (SAVI){rsqb}, normalized difference infrared indices (NDII5 and NDII7), and fraction images, derived from spectral mixture analysis (SMA), to distinguish intact forest from four classes of degraded forests: nonmechanized logging, managed logging, conventional logging, and logged and burned. For this purpose, a robust time series dataset of Landsat Thematic Mapper 5/Enhanced Thematic Mapper (TM/ETM{plus}) images was used in conjunction with forest inventory transects and data on disturbance history. The study area is located near two important sawmill centers-Sinop and Claudia, in Mato Grosso State-in the southern Brazilian Amazon. Most of the remote sensing measures tested to distinguish intact forest from degraded forests showed statistically significant changes. Fraction images, particularly green vegetation (GV) and nonphotosynthetic vegetation (NPV), were the most effective means tested for identifying conventional logging and logged and burned forest in the region. The GV change, detected from intact forest to conventional logging and logged and burned forest classes, persists no more than 1 yr, but the NPV change is still significantly different for up to 2 yr. In the second and third years following a degradation event, a significant regeneration change signal was observed in reflectance and fraction images, which can be useful for identifying these types of forest disturbances in areas where optical satellite images cannot be acquired every year

Close Window