Simultaneous monitoring of algal bloom and aquatic vegetation in eutrophic shallow lakes of the middle and lower yangtze river basin using deep learning

Abstract

In recent years, lake ecosystems have faced severe challenges, including the increasingly severe problem of lake eutrophication and the rapid degradation of aquatic vegetation (AV). These issues pose significant threats to water quality, biodiversity, and human health. Traditional methods struggle to accurately and simultaneously monitor algal bloom (AB) and AV due to their similar spectral characteristics. Furthermore, existing studies have predominantly focused on large lakes, leaving small and medium-sized lakes understudied. This study addresses these gaps by employing the UNet+ + deep learning model to simultaneously map AB and AV in 44 eutrophic shallow lakes (area>20 km²) in the middle and lower Yangtze River Basin (MLY). Utilizing Landsat remote sensing images from 2013 to 2023, the research provides a comprehensive analysis of the spatiotemporal dynamics of AB and AV. The results reveal significant spatiotemporal variations in AB and AV distribution. Additionally, the number of lakes experiencing algal bloom increased significantly (p < 0.01), and the area of aquatic vegetation exhibited a significant decrease. Unlike previous studies that focused on single categories or large lakes, this research addresses the complexity of "grass-algae coexistence" and provides a comprehensive understanding of their spatiotemporal dynamics. This research not only advances the application of deep learning in aquatic ecosystem monitoring, but also provides valuable insights for lake management and ecological restoration in the MLY region.