Résumé
The Amazon is the world's largest watershed and affects the South American and global climate. This region has seen an increase in the recurrence of extreme hydrological events (droughts and floods) in the recent decades. Located near Santarém and Óbidos, where records of maximum level were recently recorded, the Curuai floodplain is about 4,000 km² and is the subject of several studies. Remote sensing is the most appropriate tool for the study of this area due to its complexity and the reduced number of in situ measurement stations. The aim of this study was to estimate the flooded extent and volume from Sentinel-1 radar images for two hydrological years (2020-2021 and 2021-2022). For this purpose, an approach based on the classification by the multi-thresholds Otsu method was implemented. The results show that the flood extent varies from 31.58% to 54.85% of the study area and the variation of flooded volume is 19.69 km³. They also show that a larger area is flooded 100% of the time in the hydrological year 2021-2022 than in 2020-2021. The comparison with the literature reinforces facts such as the permanent presence of channels in the southeast and questions the classification by the multi-thresholds Otsu method. The limitations of this methodology open up perspectives for further study with regard to the comparison of several classification methods and the use of altimetry data for the calculation of the depth and, consequently, the flooded volume.