The effectiveness of ocean–colour data assimilation in providing robust biological–parameter estimates for basin–scale ecosystem models is investigated for a phytoplankton–zooplankton–nutrient model using North Atlantic satellite chlorophyll data. The model is forced by annual cycles of mixed–layer depth, day length, photosynthetically available radiation and a temperature–dependent phytoplankton maximum growth rate.
Although ocean–colour data are potentially limited in their ability to constrain model parameters because they provide information about the phytoplankton component only, this limitation is offset by the volume of data available covering the range of possible biogeochemical responses to similar and widely varying physical conditions. The results are improved by applying wintertime nutrient estimates based on in situ observations as an additional constraint.
Repeatability of parameter estimates obtained from independent samples is examined. Results obtained using regional and basin–wide sampling strategies for obtaining the optimization dataset are compared and the geographic applicability of the calibrated models is assessed.