A new method to transform satellite ocean color data into estimates of primary production and carbon fluxes is presented. A one-dimensional coupled physical-bio-geochemical model of the surface ocean is constrained to reproduce the seasonal evolution of surface chlorophyll concentration by adjusting the parameters of the 10-compartment trophic system using a variational technique. The method is applied to in situ surface chlorophyll data from station Papa, but averaged over the characteristic depth of ocean color measurements, and affected by errors compatible with those of satellite values. Using 35 measurements for the year 1976, it is found that five linear combinations of the trophic parameters can be adjusted. This adjustment is also valid for 1975 chlorophyll data. In general, the adjusted values of the trophic parameters are sensitive to their a priori values, but consistent results are found for the grazing rate (0.35 to 0.6 d-1), the phytoplankton mortality rate (very small), and the minimum concentration of zooplankton in winter (less than 0.16 mmolC m-3). Some carbon fluxes, namely photosynthetic carbon production in the euphotic layer (95 to 110 gC m-2 yr-1), its regeneration by grazing (60% of the latter), and the recycling efficiency of nitrogen (60%) seem to be robustly constrained, though primary production is apparently underestimated compared to the most recent ones. The export flux amounts to 35 to 40% of primary production, but its value depends on the particle sinking rate, which is not adjustable from chlorophyll data. This study suggests that simplified biological models, compared to the model used here, would be sufficient to achieve this task. ¿ American Geophysical Union 1996