Geospacial Model for Identification of Areas With Fire Propagation Danger in the Serra da Canastra National Park, Brazil

Abstract

Despite the fire being part of the dynamics and evolution of the Brazilian Savanna (Cerrado), indiscriminate forest fires are a problem for Conservation Units longevity. The Serra da Canastra National Park (SCNP) is located in southwest of Minas Gerais, Brazil, and it has the fire as one of the main conflicting activity. This work had as aim to create a methodology for the generation of a potential danger model for fire propagation, which was applied in the SCNP. For that, geo-environmental variables associated with fire propagation were mapped: hypsometry, slope, wind surface effect, surface convexity index, vegetation index, drainage density and roads density. The spatial dependence among the areas of high fire recurrence and these geo-environmental variables was evaluated by Kolmogorov-Smirnov test, with weights associated with the variables. The modeling was generated in Geographic Information System (GIS) and employed Fuzzy probabilistic maps. The results showed that the Chapadão da Canastra and the Chapadão da Babilônia present a high potential danger of fire propagation. The danger of propagation in the flat plain between mountain slopes is medium, and in the buffer zone is lower. The model was validated, proving the method efficiency.