Mapeamento semiautomático de cultivos ilícitos de Cannabis Sativa no semiárido pernambucano mediante integração de imagens SPOT 5 HRG, dados geográficos auxiliares e conhecimento de campo

Abstract

The Cannabis sativa is the most demanded, illegal drug in the world, ahead of anfetamins, cocaine, and opiates. Cannabis planting and selling are prohibited in Brazil. The goal of this research was to develop an approach to detect illegal cultivations of Cannabis sativa in the semi-arid region of Pernambuco State, Brazil. The method included SPOT 5 satellite imageries, ancillary geographic data, field knowledge, fuzzy spatial modelling and object-oriented image classification. We analyzed field data gathered by police operations to destruct illegal plantations in June, 2007, November, 2007, May, 2008, and May, 2010; SPOT 5 HRG satellite images with overpasses from May, 2005 to December, 2010; and ancillary geographic data related to topography, vegetation, and land use. They were analyzed in a GIS data integration framework to prospect relevant spectral, spatial, and/or temporal patterns for definition of an operational model to indicate potential areas of occurrence of Cannabis plantations in the study area. The fuzzy model was important to produce a map indicating numerical possibilities of occurrence of Cannabis plantations in different regions of the study area. The object-oriented image classification allowed a semi-automatic detection of features compatible to Cannabis plantations in the SPOT images. Results showed that the approach proposed in this study is feasible for identifying potential areas of occurrence of Cannabis cultivation in semi-arid regions and at a regional scale. It also can be, in some extent, adapted to other regions of the country or even to other countries with similar challenge.