Mapping the water cover of a protected area in the Amazon using Sentinel 1A data

Abstract. A large diversity of terrestrial cover is observed in the Amazon rainforest, which includes large areas prone to floods. These areas are globally important as they are humid. This research analyses the potential of SAR Sentinel 1 images as identification tools for hydric cover in conservation areas of the Amazon Biome in the state of Maranhão during the rainy season in the Amazonian region. Via Google Earth Engine (GEE), a collection of scenes from the mission Sentinel 1A was collected, referring to the comprehended period, January and June 2024. These scenes were submitted to mathematical bands in a GEE processing tool, which extracted the smallest pixel value observed each month. Through identification work on-site, the reference areas and the sampled areas were inspected before validation. A cut-off limit was defined to what is classified as water and as non-water after comparing to the targeted references the statistical samples observed on-site, and, to conclude, the records of pluviometric precipitation were evaluated for the analyzed area. The results show that the average backscatter values indicative of water coverage is approximately between −24dB and −21dB and are, therefore, compatible with the samples found in the Brazilian Amazon. The chosen cut-off was −20dB, which achieved hitting a rate of 76,6%, allowing us to observe that the months with the biggest recorded precipitation (March and April) also correspond to the months with the largest extension of hydric cover in the analyzed area.