Resumen
Sensitivity analysis (SA) has long been recognized as part of best practices to assess if any particular model can be suitable to inform decisions, despite its uncertainties. SA is a commonly used approach for identifying important parameters that dominate model behavior. As such, SA address two elementary questions in the modeling exercise, namely, how sensitive is the model to changes in individual parameter values, and which parameters or associated processes have more influence on the results. In this paper we report on a local SA performed on a complex marine biogeochemical model that simulates oxygen, organic matter and nutrient cycles (N, P and Si) in the water column, and well as the dynamics of biological groups such as producers, consumers and decomposers. SA was performed using a ?one at a time? parameter perturbation method, and a color-code matrix was developed for result visualization. The outcome of this study was the identification of key parameters influencing model performance, a particularly helpful insight for the subsequent calibration exercise. Also, the color-code matrix methodology proved to be effective for a clear identification of the parameters with most impact on selected variables of the model.