Satellite insights into diatom biomass dynamics and environmental drivers in the eastern China marginal seas

Diatoms are among the dominant phytoplankton groups in marginal seas and play a pivotal role in marine primary production and biogeochemical cycling. In this study, focusing on the eastern China marginal seas (ECMS), we developed a remote sensing model to retrieve diatom biomass ( C diat ) directly from remote sensing reflectance. Multiple linear regression and machine-learning methods were evaluated using in situ observations and satellite-in situ matchup datasets. Based on its balanced performance across the available evaluation datasets, the stepwise regression model was selected as the final regional model, achieving a coefficient of determination of 0.61 and a symmetric mean absolute percentage error of 48.3% in the satellite–in situ matchup validation. Application of this model to a 23–year record (2002–2025) of Aqua–MODIS observations revealed pronounced spatial and temporal patterns of C diat across the ECMS. Spatially, elevated C diat values were concentrated in coastal and estuarine waters and decreased markedly offshore. Strong seasonality was observed in coastal regions, characterized by summer maxima and winter minima. SHAP-based attribution analysis further identified iron, sea surface salinity, mixed layer depth, photosynthetically active radiation, silicon, and sea surface temperature as important environmental factors associated with nonlinear variations in C diat . River discharge was positively associated with C diat in estuarine waters, although these relationships weakened after removal of the seasonal cycle, indicating a substantial contribution from shared seasonal variability. These findings provide satellite-based insights into the multiscale variability of diatom biomass in the ECMS and improve our understanding of coastal ecosystem responses to environmental change.