A Multi-Source Methodology for Bathymetric Mapping: Integrating In-Situ Measurements with Sentinel-2 Spectral Indices in Mexican Dams

Abstract. This study proposes a multi-source methodology for monitoring bathymetry in continental water bodies in central Mexico by integrating in-situ and satellite-based techniques. A 3D-printed Unmanned Surface Vehicle (USV), equipped with echo sounders and GPS, was used to collect high-resolution depth data from five dams: Cointzio and Queréndaro (Michoacán) and Mata, Soledad, and Esperanza (Guanajuato). This data was correlated with Sentinel-2 imagery accessed through the Microsoft Planetary Computer, which provided multispectral, spatio-temporal data across different seasons. To assess water body delineation accuracy, several water indices were compared, including the Normalized Difference Water Index (NDWI), Automated Water Extraction Index with shadows (AWEI_sh), without shadows (AWEI_nsh), Modified NDWI (MNDWI), Sentinel Multi-Band Water Index (SMBWI), and Sentinel-2 Water Index (SWI), along with the Scene Classification Layer (SCL). SWI consistently yielded the most reliable contours. Although the SCL layer struggled in areas with dense aquatic vegetation, misclassifying water surfaces, it proved useful when combined with SWI. This integration produced the most accurate results for most dams— except in Queréndaro, where dense hyacinth cover in parts of the water body and irrigation agriculture in the surroundings impedes the correct detection of water body boundaries. A strong correlation between USV data and satellite-derived contours confirms that combining in-situ and remote sensing sources offers a robust and precise framework for bathymetric mapping in inland waters of Mexico.