WATER TRANSPARENCY TRENDS IN AMAZON BASIN FROM LONG-TERM LANDSAT DATA

Water transparency is an important water quality measurement as a proxy to the amount of suspended and dissolved constituents in the water. In the Amazon basin, the water bodies are mainly controlled by annual flooding pulse, which drives the suspended sediments and phytoplankton dynamics in white, organic matter in black waters, and phytoplankton dynamics in clear water. However, anthropogenic impacts, such as massive deforestation and unrestricted mining, may change water transparency patterns. This study aims to developed a Mixture Density Networks Secchi Disk Depth (Zsd) based on Landsat long-term data and evaluate its trends. In-situ data and Landsat surface reflectance product allowed the algorithm validation. Then, after applying the algorithm to the 1985-2022 Landsat time series, this study computed time Zsd trends at key Amazon Basin locations based on water types. Results showed positive and significant changes for Madeira, Amazon Rivers, and Monte Alegre Lake.