Constructing time-series submerged aquatic vegetation by integrating process-based modeling and satellite images

Abstract

Submerged aquatic vegetation (SAV) plays a critical role in lake ecosystem health. However, quantifying the spatiotemporal patterns of SAV biomass remains challenging due to limited time-series data. To address this challenge, we integrated a process-based SAV dynamic model with a satellite-based SAV biomass estimation model to construct a time-series SAV dataset for Lake Zhanbei, a sub-lake within China's largest freshwater lake, Lake Poyang. The integrated model effectively captured SAV biomass dynamics, with model performance of R²=0.60 and RMSE=0.24 kg/m² compared to measured data. Results showed that SAV was more abundant near floodplain areas. A significant decline of SAV biomass was observed from 0.76 kg/m² (2021) to 0.19 kg/m² (2022), primarily due to a drop in the annual average water level from 14.1 m (2021) to 13.4 m (2022) caused by extreme drought. Water level was the most sensitive driver of SAV biomass, while temperature also had a notable impact under optimal water levels. Our scenario simulations revealed that global warming could enhance SAV growth, while nutrients had minimal effects. Compared with in-situ measurements from previous publications, the integrated model offers a cost-effective and high-resolution approach to study SAV dynamics, with potential applications in other lakes.