*Davin Tes1, Tsuyoshi Kinouchi1
(1.Institute of Science Tokyo)
Keywords:Empirical orthogonal function, Gene expression programming , Spatiotemporal variability, Tonle Sap Lake, Water quality
The Tonle Sap Lake, a crucial ecosystem in the Mekong River Basin, experiences significant environmental changes due to modifications in the Mekong River and the lake system driven by human activities and climate change, but its comprehensive understanding is lacking. This study aims to develop a methodology for the detailed quantification of water quality and to comprehensively understand the factors influencing its spatiotemporal variability. Using gene expression programming (GEP), we established empirical regression models to estimate total suspended solids (TSS) and chlorophyll-a (Chl-a) from remotely sensed imagery, achieving high accuracy (R² = 0.979 for training and 0.996 for validation for TSS; R² = 0.872 and 0.943 for Chl-a). Empirical orthogonal function (EOF) analysis identified key spatial and temporal patterns, revealing seasonal Chl-a blooms along the southern shoreline during the dry season and high TSS concentrations in the northern lake center during the initial reverse flow period of dry years. These variations strongly correlate with water level anomalies and hydrological changes, highlighting the impact of hydrological regimes on water quality. Our findings suggest that modifications to the lake system, which alter its natural hydrological cycles, may contribute to water quality deterioration.