Catalogus

The present study attempts to meet this challenge. In a 1D framework at the DYFAMED station (NW Mediterranean Sea), near-monthly nitrate and chlorophyll profiles and daily surface chlorophyll concentrations are assimilated in a coupled dynamical–biological model using the tangent linear and adjoint models. Following sensitivity analyses that show that some parameters cannot be recovered from the data set used, assimilation of observed 1997 data is performed. The first inversion considered clearly shows that, in agreement with previous studies, (1) the data impose a C/Chl ratio that varies with depth (i.e. light) and (2) the "initial" conditions (e.g. winter nitrate profile) strongly constrain the annual biogeochemical fluxes. After assimilation of the 1997 data, the agreement between the data and the model is quantitatively improved in 1995 and 1996, which can be considered a good validation of the methodology. However, the order of magnitude of the biogeochemical fluxes, and especially of the particulate export and regenerated production, are not correctly recovered. An analysis of the simulations shows that this result is associated with a strong decrease in zooplankton concentrations. An additional constraint of maintaining acceptable levels of zooplankton is therefore added. The results are improved, but remain unsatisfactory. A final inversion, which takes into account the a priori estimates of the major annual fluxes, is then performed. This shows that there is no inconsistency between the NO₃ and chlorophyll data, the order of magnitude of the fluxes and the model. The work therefore demonstrates that recovering biogeochemical fluxes from available data of concentrations and stocks is not a straightforward exercise: the coverage and type of observations, and the nonlinearities of the biogeochemical model all contribute to this difficulty.