Chlorophyll-a data assimilation using LSEIK improves coastal bloom representation in the Baltic Sea

Eutrophication is a major stressor on Baltic coastal ecosystems, significantly impacting phytoplankton bloom dynamics and biogeochemical variability. Accurate characterization of chlorophyll a concentration is crucial for understanding the timing, intensity, and spatial structure of phytoplankton blooms. However, in the Baltic region, observational data is sparse, and biogeochemical models often underestimate bloom intensity and fail to accurately describe seasonal evolution, particularly along the coast.

In this study, we investigate the impact of chlorophyll-a data assimilation (DA) on the simulation of phytoplankton blooms in the Baltic Sea. Satellite ocean-colour products and in situ observations are assimilated into a coupled physical–biogeochemical model using a Local Singular Evolutive Interpolated Kalman (LSEIK) filter. Both chlorophyll-only and combined SST and chlorophyll assimilation experiments are performed to assess their influence on bloom dynamics across different sub-basins. DA substantially improves the representation of phytoplankton bloom timing and magnitude at the basin scale. Relative to satellite-derived chlorophyll-a, DA reduces RMSD from 1.7 to 1.3~1.4 mg m⁻³ in spring and from 2.2 to to approximately 1.4 mg m⁻³ in summer for the chlorophyll-only and combined SST+chlorophyll assimilation experiments, respectively. Overall, the RMSD are reduced by 33~40% in DA runs over the full simulation period, indicating a significant improvement in the characterization of algal bloom intensity and large-scale spatial consistency. Comparisons with in situ observations shows regionally variable changes in correlation, indicating large differences between in-situ and satellite observations, while consistently showing a reduction in RMSD and an improvement in mean-state representation.

These results demonstrate that LSEIK-based chlorophyll-a data assimilation effectively constrains large-scale phytoplankton bloom dynamics in the Baltic Sea, improving the realism and spatial coherence of simulated chlorophyll fields. The findings highlight both the advantages and limitations of satellite-driven assimilation for representing coastal phytoplankton variability and provide insights for future developments in marine biogeochemical data assimilation.