Avaliação da aplicação dos dados do radar de abertura sintética Sentinel-1 para o mapeamento de inundação em áreas urbanas, periurbanas e rurais

Abstract

Floods are recurring events that cause natural disasters, affecting extensive areas and large numbers of people, resulting in significant human and economic losses. They occur in both rural and urban areas, with more pronounced impacts in the latter. In tropical countries, floods often result from intense rainfall, posing a challenge for obtaining images from sensors operating in the optical range due to the frequent presence of clouds. Synthetic Aperture Radars (SAR), especially Sentinel-1, emerge as a viable alternative, emitting electromagnetic waves in the microwave range that penetrate clouds, allowing monitoring under various atmospheric conditions. With the increase in natural disasters, remote sensing-based methodologies have become essential for mapping floods, offering a comprehensive view of these disasters. Synthetic Aperture Radars, in use for decades, are employed to map floods, with the Sentinel1A and 1B satellites being part of this approach. Despite some concerns about the suitability of Sentinel-1 for mapping urban areas, several studies have used techniques such as threshold identification and change detection to identify floods using radar images. In Brazil, the lack of in-depth studies on the use of Synthetic Aperture Radars for flood identification, especially in urban areas, represents a gap in remote sensing research. This study aims to fill this gap by investigating the application of Sentinel-1 imagery in the identification and mapping of floods, particularly in urban areas. The work also seeks to evaluate the effectiveness of the applied methodologies by comparing the results with validation maps provided by the international collaboration 'The International Charter Space and Major Disasters' and optical images obtained from the Sentinel-2 satellite. The present study showed that the application of images obtained from the Sentinel-1 satellite can be limited for urban floods in restricted areas, where the overall accuracy was around 0,08 in Rio Branco, Acre, compared to 0,95 in Beledweyne, Somalia, where the flooded urban areas were much more extensive.