The methodology is based on an a priori approach: this means that the results depend only on the structure of the model and on the set of variables which forms the data available for assimilation. The methodology is based on the computation of the sensitivities of the state variables to the model parameters, and enables one to analyze the role additional observations could play in constraining the model parameters. The methodology was applied to a one-dimensional (1D) primary production multinutrient model that describes the dynamics of pelagic ecosystem. The model describes nitrogen, phosphorus, and carbon cycles by means of a trophic chain constituted by two phytoplanktonic functional groups, one zooplanktonic pool and the detritus compartment. Nitrates, phosphates, and ammonia are considered as inorganic dissolved nutrients.

Results show that the sensitivities of the majority of the parameters are strongly correlated and, therefore, only 5 of 43 parameters of our model could be accurately calibrated, even if daily measurements of nutrients and chlorophyll a were available for 1 year and at three different depths. Most of the state variables show the highest sensitivity to parameters related to the water temperature, phytoplankton growth, and phytoplankton mortality. The analysis of this case study, which, in our opinion, is representative of oligotrophic mid-latitude environments, suggests that water shading coefficient, optimal temperature coefficients for small phytoplankton, optimal temperature coefficient for large phytoplankton, a grazing parameter, and a parameter that describes the influence of water temperature on biological and chemical kinetics could be simultaneously and efficiently calibrated.