Model for remote detection of areas naturally vulnerable to the occurrence of floods

Assessing the vulnerability of areas at risk to floods necessarily involves studying the morphometric characteristics of the relief and the drainage network and their contributions to the occurrence of the disaster. Thus, the aim of this study was to use multivariate discriminant analysis to define a function that can be applied in different locations to identify areas naturally subject to flooding. For this purpose, a cluster analysis was carried out for the hydrographic sub-basins of Alto Sapucaí, south of Minas Gerais, whose floods are recurrent according to civil defense. For the grouping of risk areas, a compaction coefficient of ۰.۷۵ and a form factor of ۰.۵۶ were obtained, both considered high risk for flooding. These results suggest that the cutoff value for the discriminant function less than ۳.۸۲ is indicative of an area naturally vulnerable to the occurrence of floods. These areas were corroborated with the mapping of municipalities classified in emergency situations by the civil defense. The accuracy value of the model obtained by comparison with the Civil Defense and Geological Survey of Brazil (CPRM) map was ۰.۷۶, indicating precision and robustness of the proposed model. The multivariate discriminant analysis enabled the identification of a mathematical function that can be applied in other locations, contributing to the rapid identification of places naturally prone to flooding. Thus, it can be an instrument for managing natural disasters, aiming to avoid the use and improper occupation of areas subject to recurrent floods, avoiding human and socioeconomic tragedies and serving as a tool for monitoring these areas

کلیدواژه ها:

Geoprocessing ، Hydrology ، Geomorphology ، Geographic Information System.