![Sample our Computer Science journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5928/TOC-Subject-Sample-2021-Computer-Science.jpg"})
Abstract
To increase the monitoring potential of forest fires, an alert classification methodology using satellite-mapped hotspots has been established to help forest managers in the prioritization of which hotspot to be verified in the field, thus potentially improving the distribution of fire-fighting resources. A computer application was developed based on web-distributed geographical information technology whose main function is to interact automatically generated satellite hotspots and risk areas indicated in fire-susceptibility maps and classify them into five alert levels. The location of the hotspots is available continuously every 4 h, and a susceptibility map is produced daily through map algebra algorithm, which uses static (topography, vegetation and land use) and dynamic (weather) variables. Every process runs through automated geoprocessing routines. The methodology was tested during the dry period of 2007 in the Carajás National Forest, in the Brazilian Amazon, within an area of 400,000 ha. It is a critical area constantly threatened by fires caused by invasions and deforestation owing to intense agribusiness advances and mining activities in its surroundings. This situation results in observations of many hotspots inside the study area for the same day and almost the same time period, in places of extreme opposites, demanding complex rapid analysis and hindering the decision of the displacement of fire-fighting teams. Further, a major mining company operates within the National Forest area, maintaining actions of protection as part of its environmental mining license. Results are presented under three aspects: (i) the credibility of the daily susceptibility map (algorithm), which showed strong correlation between areas of greatest risks and the confirmed forest fires; (ii) the reliability of hotspots (alert levels), confirming 71% of fires; (iii) accuracy in the decision of which hotspot to be checked, which revealed the same number of verifications at different alert levels, 82% confirmed alert 5 hotspots (maximum) and only 50% from alert 1 (minimum), resulting in faster fire-fighting actions, minimizing burned areas and, in some cases, allowing fire control before its spreading. Therefore, the methodology demonstrated that GIS routines are able to determine the relationship between a reality-based, interpreted susceptibility map of the area and satellite-generated hotspots, highlighting the ones of highest hazard level through the alert classification, becoming an important tool to help decisions from the fire-control center, especially for high-risk regions. The methodology may be extrapolated to other forested areas.
Acknowledgments
We are grateful to Hiparc Geotecnologia Ltda. for permission to publish information about the methodology and the Forest Fire Alert System. Special thanks to engineers Paulo Bueno and Hamilton Carvalho from the Vale mining company, for their participation and support of the methodological tests; to meteorologists Adma Raia Silva and Ruibran Januario dos Reis from the weather company MGTempo; to environmental manager Laudicena C. Pereira from the Forest State Institute of Minas Gerais – IEF-MG – and researcher Luis Eduardo Maurano from the INPE, for their contributions to the technical discussions; and to GIS analyst Fabio Valério Xavier from Coffey Ltda. for help in digital processing. The first author thanks Brazil's Ministry of Education's Coordenação de Aperfeiçoamento de Pessoal de Nível Superior for his PhD grant. We are also grateful to the anonymous reviewers and the editor of IJGIS, who helped improve the article.
