Using four dimensional variational assimilation to estimate model parameters: Sensitivity to the assimilation window length. WWRP/THORPEX Workshop on 4D-Var and Ensamble Kalman filter inter-comparisons

PULIDO M

Buenos Aires

Workshop; WWRP/THORPEX Workshop on 4D-Var and Ensamble Kalman filter inter-comparisons; 2008

WWRP/THORPEX

We present a technique to estimate model parameters in a nonlinear dynamical model of the middle atmosphere which is based on principles of four-dimensional variational data assimilation. A cost function that measures the mismatch between the model state and observations is defined. The control variables are the model parameters. The minimisation is performed using a conjugate gradient method, with the full adjoint model used to calculate the gradient of the cost function.The technique is used to estimate gravity wave drag. In this work the performance of the technique as a function of the assimilation window length and the ability of the technique to estimate nonlocal forcing effects are evaluated through twin experiments. We use the forward model to estimate model states where a prescribed gravity wave drag is imposed to theoretical and observational conditions. These model states are used as observations in the assimilation system in which the control variables are the components of the gravity wave drag field.