CONICET | Buscador de Institutos y Recursos Humanos

Oceanic and atmospheric global numerical models represent explicitly thelarge-scale dynamics while the smaller-scale processes are not resolved so thattheir effects in the large-scale dynamics are included through subgrid-scaleparameterizations. These parameterizations represent small-scale effects as afunction of the resolved variables. In this work, data assimilation principles areused not only to estimate the parameters of subgrid-scale parameterizationsbut also to uncover the functional dependencies of subgrid-scale processes asa function of large-scale variables. Two data assimilation methods based onthe ensemble transform Kalman filter (ETKF) are evaluated in the two-scaleLorenz ?96 system scenario. The first method is an online estimation that usesthe ETKF with an augmented space state composed of the model large-scalevariables and a set of unknown global parameters from the parameterization.The second method is an offline estimation that uses the ETKF to estimate anaugmented space state composed of the large-scale variables and by a spacedependent model error term. Then a polynomial regression is used to fit theestimated model error as a function of the large-scale model variables in orderto develop a parameterization of small-scale dynamics. The online estimationshows a good performance when the parameter-state relationship is assumedto be quadratic polynomial function. The offline estimation captures bettersome of the highly nonlinear functional dependencies found in the subgrid-scaleprocesses. The nonlinear and nonlocal dependence found in an experiment withshear-generated small-scale dynamics is also recovered by the offline estimationmethod. Therefore, the combination of these two methods could be a useful toolfor the estimation of the functional form of subgrid-scale parameterizations.

enviar mensaje