Monitoring algal blooms in eutrophic inland lakes using OLCI: Overcoming cloud and sun-glint effects on low-quality...

Timely monitoring, detection, and quantification of algal blooms (ABs) are crucial for managing public health risks and understanding aquatic ecosystem dynamics. However, traditional algorithms often misclassify non-bloom waters (clouds, cloud shadows, and sun glint) as ABs. This study proposes a novel method (the NDVI-FBA_s algorithm) based on OLCI imagery to improve ABs extraction under complex observational conditions. Developed using 2754 samples, including four types of algal bloom and eight types of non-bloom water bodies, it integrates the normalized difference vegetation index (NDVI), which highlights the strong near-infrared reflectance of ABs, with the FBA_s index to capture red-band absorption features. Validation across diverse scenes demonstrated high performance: precision (1.0), recall (0.97), F1-score (0.98), and IoU (0.97), confirming the algorithm's effectiveness. The algorithm was applied to Chinese representative eutrophic lakes-Lake Dianchi, Lake Taihu, Lake Chaohu, Lake Xingkai, and Lake Hulun-over the 2016-2023 period. The findings indicate substantial seasonal and interannual variability in the occurrence of ABs. Peak bloom periods were observed from July to December in Lake Dianchi, May to November in Lake Taihu, and May to October in Lake Chaohu. In contrast, Lakes Xingkai and Hulun exhibited relatively low bloom frequency and extent. Moreover, the implementation of a national fishing ban in 2020 corresponded with a noticeable decline in ABs intensity in Lakes Dianchi, Taihu, and Chaohu, indicating a potential regulatory impact. Overall, the NDVI-FBA_s algorithm provides an effective and scalable approach for monitoring ABs in eutrophic freshwater systems, offering valuable insights into their spatiotemporal dynamics and supporting informed water quality management.