Long-term spatiotemporal variability and drivers of chlorophyll-a in Haizhou Bay using Sentinel-3 OLCI and multi-source data

Global climate change is intensifying pressure on marine ecosystems, particularly coastal bays affected by natural variability and anthropogenic disturbances. Chlorophyll-a (Chl-a) is a key indicator of phytoplankton biomass and eutrophication status. This study investigated the spatiotemporal variability and potential drivers of Chl-a in Haizhou Bay during 2016-2025 by integrating Sentinel-3 Ocean and Land Colour Instrument (OLCI) imagery, in situ measurements, and multi-source ancillary data. Among the evaluated models, random forest showed the best overall performance and stability, outperforming support vector regression, extreme gradient boosting, and European Space Agency operational algorithms, with a coefficient of determination of 0.72 and a root mean square error of 0.66 in the logarithmic domain. The retrieved Chl-a fields were reconstructed into a spatiotemporally continuous 2016-2025 dataset for analyzing long-term dynamics and potential drivers. The principal contribution is the construction of this continuous dataset and its integration with environmental records to assess associations with mariculture activities, nearshore nutrient pressure, and episodic green-tide disturbances. Annual mean Chl-a initially increased, declined after 2019, and rebounded slightly in 2025, while remaining consistently higher nearshore than offshore. Chl-a peaked in May-June, and seasonal variability was associated with Pyropia cultivation, nutrient dynamics, and episodic green tides. Annual mean Chl-a was correlated with Pyropia cultivation area (r = 0.81, p  0.05). These findings provide a long-term evidence base for eutrophication control, mariculture management, and refined coastal management.