SEMI-AUTOMATIC METHODOLOGY FOR DISASTER DETECTION IN IMAGES FROM BRAZILIAN EARTH OBSERVATION SATELLITES

• Grant number: 24/22293-4
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: May 01, 2025
• End date: April 30, 2028
• Field of knowledge: Physical Sciences and Mathematics - Geosciences - Physical Geography
• Principal Investigator: Thales Sehn Körting
• Grantee: Karine Bastos Leal
• Host Institution: Instituto Nacional de Pesquisas Espaciais (INPE). Ministério da Ciência, Tecnologia e Inovação (Brasil). São José dos Campos , SP, Brazil
• Associated research grant: 23/09118-6 - Content discovery in remote sensing image catalogs, AP.R

Abstract

This project aims to develop a methodology for the semi-automatic identification of coastal erosion, floods caused by sudden sea-level rise and increased precipitation, landslides, and wildfire-affected areas, using change detection with remote sensing images from the CBERS Program. Despite the availability of data from national satellites such as CBERS and AMAZONIA-1, there are significant gaps in their application to disaster studies in Brazil, particularly in addressing coastal disasters, which this project seeks to overcome. The proposed methodology combines machine learning and convolutional neural network (CNN), such as the use of CerraNetv3 and the Smart Mask Labelling (SML) module, to optimize semantic segmentation and identify affected areas. Multitemporal data will be integrated to detect landscape changes and alterations in vulnerable areas. These methodologies will enable faster and more accurate disaster analysis. Integration with the International Charter Space and Major Disasters is a key aspect of the project, which will facilitate the generation of cartographic products and reports to meet both national and international disaster management needs. The project also includes active participation in Charter meetings and collaboration with space agencies to ensure the technical compatibility of the developed products. By the end of the project, it is expected to deliver efficient algorithms, thematic maps, and high-precision technical reports, as well as disseminate the results through scientific articles in high-impact national and international journals. The project will contribute to modernizing disaster monitoring practices in Brazil and promote the use of national satellite data.