IMIT   21220
INSTITUTO DE MODELADO E INNOVACION TECNOLOGICA
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Off-line and online estimation of subgrid parameters using ensemble-based data assimilation
Autor/es:
PULIDO M
Lugar:
Paris
Reunión:
Workshop; Data Assimilation for Detection and Atribution of Climate Change; 2014
Institución organizadora:
CNRS-ANR
Resumen:
Oceanic and atmospheric global numerical models represent directly the large-scale dynamics while the smaller-scale dynamic features are not resolved in the model so that their effects in the large-scale dynamics are included through subgrid parameterizations. These parameterizations model small-scale effects as a function of the resolved variables, thus they contain a set of unknown parameters.  In this work, the estimation of parameters of subgrid parameterizations that model the small-scale variables in the Lorenz'96 system is evaluated using two techniques based on the local ensemble transform Kalman filter. An on-line estimation approach uses the local ensemble transform Kalman filter with an augmented space state composed by  the model variables and the set of unknown parameters. The other technique is an off-line estimation approach that first uses the local ensemble transform Kalman filter to estimate an augmented space state composed by  the model state and the ``missing'' forcing produced by the unresolved variables to the resolved variables. In a second offline stage, the set of unknown parameters are optimized using the estimated forcing in each assimilation cycle. We show that there is a time lag when estimating the model error which assumes a constantpersistent model for the forcing. The online quadratic estimation shows an excelent performance.  Neither the online nor the offline estimation appears to give an accurate overview ofthe stochastic characteristics of the small-scale subgrid effects.