The Assimilation of Phytoplankton Functional Types for Operational Forecasting in the Northwest European Shelf

Abstract

This paper proposes the use of assimilation of phytoplankton functional types (PFTs) surface chlorophyll for operational forecasting of biogeochemistry on the North-West European (NWE) Shelf. We explicitly compare the 5-day forecasting skill of three runs of a physical-biogeochemical model: (a) a free reference run, (b) a run with daily data assimilation (DA) of total surface chlorophyll (ChlTot), and (c) a run with daily PFTs DA. We show that small total chlorophyll model bias hides comparatively large biases in PFTs chlorophyll, which ChlTot DA fails to correct. This is because the ChlTot DA splits the assimilated total chlorophyll into PFTs by preserving their simulated ratios, rather than taking account of the observed PFT concentrations. Unlike ChlTot DA, PFTs DA substantially improves model representation of PFTs chlorophyll. During forecasting the DA reanalysis skill in representing PFTs chlorophyll degrades toward the free run skill; however, PFTs DA outperforms free run within the whole 5-day forecasting period. We validated our results with in situ data, and we demonstrated that (in both DA cases) the DA substantially improves the model representation of CO 2 fugacity (PFTs DA more than ChlTot DA). ChlTot DA has a positive impact on the representation of silicate, while the PFTs DA seems to have a negative impact. The impact of DA on nitrate and phosphate is not significant. The implications of using a univariate assimilation method, which preserves the phytoplankton stochiometry, and the impact of model biases on the nonassimilated variables are discussed.