Digital processing of a Landsat-TM time series for mapping and monitoring degraded areas caused by independent gold miners, Roraima state, Brazilian Amazon

摘要

A 12-year Landsat-Thematic Mapper (TM) time series (1987-1999) was used for mapping and monitoring evolution of degraded areas caused by independent miners ("garimpeiros") in the search for gold and diamond. Discrimination of target areas was achieved through third principal component images, due to their best enhancement of bare soil areas relative to the surrounding savanna vegetated terrain. A postclassification approach, based on image segmentation/region classification techniques, was used to map degraded areas. This procedure allowed to save time and to curtail inherent subjectivity commonly involved in visual interpretation, producing accurate land-cover change maps. According to these maps, degraded areas comprise 94.4 ha in 1987, 286.4 ha in 1991, and 404.4 ha in 1994. After "garimpeiros" abandoned the region sometime prior to 1994, vegetation soon started recovering mined areas as shown in images acquired in 1995, 1996, and 1999. According to these images, degraded areas decreased to 311.2, 283.7, and 246.2 ha, respectively. Data also indicate that vegetation regrowth is faster near the border of degraded areas, and becomes slower to the center of these areas, where damage process is more intense. Based on the satellite images-derived regrowth rates, it is possible to infer that degraded areas would not be entirely recovered by vegetation by the year 2019.