Comparison and Validation of Spatial Modeling of Fire Risks Considering Different Prediction Methods

Abstract:

The environmental problems of fires change the dynamics of the planet modifying and destroying their cycles and ecosystems. The human being is responsible for almost all the fires, but he is also protagonist of prevention initiatives. Thus, it becomes necessary to plan actions to combat these environmental damages. Since the geographic location is a important attribute, this research aims to support prevention and control of fires generating and validating maps with prediction risk of fire models applied in João Pessoa city. The data were modeled, processed, handled and analyzed in ArcGIS software v10.0 and Matlab, as well as the generation and overlay of thematic maps using multicriteria analysis, weighting the variables and fuzzy logic. In the next step, it was made the data validation considering the real data and the results demonstrated that the templates generated with the aid of fuzzy logic showed the coefficient of determination above 85%. The rainfall variable was the factor that contributed significantly to the models having greater reliability. This variable was not used and not specifically recommended in other methods compared in this study. The factors that contributed to the high degree of vulnerability risk of fires: high slope, vegetation, areas of high concentration of people, subnormal agglomerations and regions within the influence of the road network and hydrography. Finally, this study aimed to contribute to the decision making of the environment, social security and defense managers quickly and accurately using a few variables and low cost.

Keywords:
Fires; Geoprocessing; Risk maps; Validation