Métodos de detecção de mudança de uso e cobertura do solo para monitoramento ambiental

Abstract

Forest monitoring is currently indispensable for environmental planning. Through it, it is possible to detect and spatialize changes in land use and cover, becoming a tool for decision-making. In this context, the general objective of the study was to evaluate change detection methods (DM), which support the identification of changes in land use and cover in agricultural landscapes. The study was carried out in the Sarapuí river basin (SP, Brazil), using images from Landsat 5 and 8 satellites, in four different years. Three DM methods were evaluated, being them by normalized vegetation index (DMNDVI), by direct image difference (DMDID) and by change vector analysis (AVM), with and without radiometric and atmospheric corrections. Subsequently, a supervised classification of objects was performed and the respective DM thresholds were obtained. It was verified, using the Z test, that the methods differ statistically, except when the methods with and without correction are compared. Therefore, to define the most appropriate method, the methods were qualified by false positives and false negatives, and the DMNDVI presented better results. With the delimited areas of change, the intersection with the current use and land cover and with the priority areas for forest conservation in the study area was carried out. It was found that in the last thirty years, 22% of the total area of the basin suffered, at some point, changes in its use and soil cover, and in about 6.14% of its total area there was an increase in vegetation and in 15.97% loss of this coverage. In general, it is concluded that the DMNDVI method presents better results, helping in the decision-making process to identify areas for the payment of environmental services, that is, indicating areas where there was a gain in vegetation