Data assimilation and a pelagic ecosystem model: parameterization using time series observations

Variational adjoint assimilation of time series observations is used to estimate the optimal parameters of a nitrogen-budget, upper ocean, mixed-layer ecosystem model. Observations collected at the Bermuda Atlantic Time-Series Study (BATS) site are taken as an example of a time series. A twin experiment using simulated data of the same type and frequency as the BATS observations first demonstrates the adequacy of the observations to estimate the model parameters and model the ecosystem annual cycle. This experiment further shows that some of the model parameters cannot be estimated independently. This conclusion leads to a simplification of the model and a redefinition of its parameters. Based upon the success of the twin experiment to estimate all model parameters, an attempt to assimilate actual observations from BATS was undertaken. The assimilation of real data leads to the conclusion that, even though the frequency and type of observations is adequate to estimate the model parameters, the considered model is not appropriate for the annual cycle of the BATS ecosystem. (C) 1998 Elsevier Science B.V. All rights reserved.