Processamento de Imagens dos Satélites Brasileiros CBERS-4 e CBERS-4A para Respostas Rápidas a Desastres

Satellite images can contribute to the identification and analysis of areas affected by natural disasters, through the use of Digital Image Processing (DIP) techniques that highlight areas of interest. The International Charter “Space and Major Disasters” is the world's main cooperation between space agencies for the free provision of emergency data and relies on Brazil in the process of responding to calls. Giving importance to emergency requests, this research proposes the use of Brazilian satellites to systematize DIP techniques as an aid to the management of disasters such as landslides and regional floods. We present two case studies: the landslides in Petrópolis (RJ), which occurred in 2022, and the regional floods in three provinces of Pakistan, in 2022. Data mining performed with the support of the Random Forest (RF) algorithm was adopted to extract the main attributes that were later combined in the RGB color space for quick identification of affected areas. The proposed color composition for Petrópolis was based on the selection of a principal component, vegetation index and hue component of the IHS color transformation, with a Global Accuracy (GA) of 81.82%, obtained from MUX/CBERS-4A data. The suggested composition for Pakistan was based on the selection of the NIR band and the first and second principal components, with an GA of 88.33%, using WFI/CBERS-4 images. All areas of interest could be successfully highlighted based on the suggested compositions.