Aplicação de álgebra de mapas e análise multicritério no Google Earth Engine para mapear o risco de incêndios em Uruçuí-PI

Abstract

Forest fires have intensified in Brazil and constitute a major environmental issue in areas of agricultural expansion, such as the municipality of Uruçuí, in southern Piauí. In this context, this study aimed to identify fire-risk areas through the application of Multicriteria Analysis and Map Algebra in Google Earth Engine. Specifically, the study sought to select and weight environmental variables using the Analytic Hierarchy Process (AHP), integrate these variables into a continuous risk index, and validate its spatial coherence with hotspots detected by the VIIRS sensor in 2024. The model was structured based on the variables precipitation, land surface temperature, Normalized Difference Vegetation Index (NDVI), slope, altitude, and aspect, which were normalized and combined in a JavaScript environment within Google Earth Engine, resulting in a continuous index that was subsequently classified into quartiles. Validation included the generation of 366 daily maps and the spatiotemporal deduplication of the observed dataset, reducing it from 37,620 raw records to 1,417 deduplicated hotspots. The results showed that 77.56% of the hotspots were concentrated in the upper quartiles of the index, corresponding to the highest-risk ranges, with emphasis on the highest quartile, which accounted for 25.83% of the hotspots within 21.47% of the accumulated area, reaching an efficiency ratio of 1.20. The months of August, September, and October concentrated 76.2% of the occurrences, highlighting the seasonality of the phenomenon and the relevance of the monthly synthesis maps for the spatial interpretation of risk patterns. It is concluded that the model performed better as a tool for spatial risk ranking and for supporting the territorial prioritization of preventive actions than as a strict classifier of fire occurrence and non-occurrence. As a technical product, a script was developed in Google Earth Engine, with potential practical application in municipal environmental monitoring and management.