Modelagem espaço-temporal de incêndios urbanos em área aberta em Cuiabá, Mato Grosso

Abstract

The municipality of Cuiabá is subject to fire in its open area in the urban center and in rural areas. The fuel that allows such burnings is characterized in the urban center, by the undergrowth of the flowerbeds of avenues, parks, ecopoints, vacant lots, illegal waste deposits, represent fuels available for fires in open urban areas. In rural areas, the vegetation is more robust, consisting of larger grasses that form pastures, thin savanna, dense savanna and even transition forests. The highest occurrence of fires occurs during the dry season, especially in July and August, causing a very high volume of particulates to be released, visual pollution, harming the population's health and discomfort. This work, aiming to obtain a model to aid fire prognosis in an open urban and rural area of the municipality of Cuiabá, chose twelve variables (three geographic, three fire management, and six meteorological) to compose the membership equations required by Fuzzy logic principles also called Fuzzy Set Theory (TCN), using as information the fire occurrence data from the Public Security Secretariat of the State of Mato Grosso and the data obtained through the spectral index NBR (Normalized Fire Index) for detection of burn scars and evaluation of fire severity. The period corresponds to the dates 08/17/2013, 08/20/2014, 08/23/2015 and 08/28/2016, orbit and point 226/71 derived from the corrected reflectances for atmospheric effects and were obtained through the Science Center Platform Processing Architecture (ESPA) of the US Geological Survey (USGS) geological service; Development was mostly done in R language, with a demo panel in Shiny. All code is available in a public repository at https://github.com/demusis/fuzzyIncendio. The model was cross validated, showing a MAD of 16.88%. With these results, we conclude that the model obtained a credible result in relation to the results found, but an expansion of the calibrations with specialists, the model can be improved and serve as a tool to help managers, criminal experts and civil defense professionals regarding their public practices and policies.