Author(s): Matteo Redana; Yiming Lin; Xin Yi Chong; Tomas Maul; Karen Lee; Chris Gibbins
Linked Author(s): Xin Yi Chong, Matteo Redana
Keywords: Water segmentation GAM CNN ECOSTRESS reservoir
Abstract: Water temperature monitoring is crucial for informing water management strategies, especially in reservoirs and the river systems they regulate. The ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) is the highest resolution thermal sensor available, allowing assessment of Land Surface Temperatures (LST) in relatively small water bodies or rivers. A key first step in such assessment is the separation of wetted and non-wetted areas. Here we compare six methods for separating wetted areas and extracting ECOSTRESS pixels across the area of interest. Integrating multiple wavelength bands from diverse satellites improved the wetted area estimations. Two of the algorithms we tested outperformed the default ECOSTRESS water masking procedure allowing for a more accurate extraction of actual wetted pixels and estimation of water LST metrics. The integration of ECOSTRESS data and accurate estimation of up-to-date wetted areas provides the basis for an autonomous and near real-time framework for monitoring freshwater temperatures.
Year: 2025