Detecção de Mudanças no Uso das Terras no Município de Pelotas (RS, Brasil), no período de 1985 a 2007, por meio de Processamento Digital de Imagens (original) (raw)

The human activity is the main agent of change in different landscapes, such as what occurs in the natural and rural areas. These changes may have different speeds, being perceived in small lapses of time, as days and months, or may be given over several years or centuries. So, the land use/cover change is one of the most important indicator of environmental conservation conditions and its economic status for a region, which has great association with the socioeconomics facts and phenomena, it is reflecting, as well, the public policies for regional or national developing. The purpose of this study was testing some methods of change detection, in Pelotas Municipality (RS, Brazil), from multitemporal orbital data, evaluating some techniques of digital image processing, such as: vegetation index (NDVI, EVI), tasseled cap transformation, principal components analysis and linear spectral mixture model. The orbital data were extracted from three sets of Landsat 5 TM images (WRS 221/82), acquired in 1985, 1995 and 2007, which they were coincided with the summertime, over the agricultural calendar of rice (the main agricultural activity of the municipality). The vegetation indexes were rather insensitive to detect changes among some classes of land use. Meanwhile, the best results were obtained using the linear spectral mixture model. It was also observed, during on the studied period, the urban growth and the decline in agricultural activity by family farming in the upper rural region, while the area occupied by rice plantations of agriculture business in the low lands, remained stable. Palavras-chave: land cover, agriculture, urban growth, rural-urban migration, multitemporal assessment. próximo artigo 3 9