Model Error and parameter estimation in a simplified mesoscale prediction framework

Guillaume Vernieres, SAMSI, Josh Hacker, NCAR/RAL and Montserrat Fuentes, North Carolina State University

Abstract
A column (1D) model derived from the Weather Research and Forecast (WRF) numerical weather prediction (NWP) model is implemented for efficient experimentation with model physics. One of its purposes is to estimate a few parameters than can later be used in the 3D WRF in forecasting and nowcasting applications. We will present some preliminary results of parameter and model error estimation. The estimation procedure is based on observations of zonal and meridional wind speed, temperature, and water vapor mixing ratio. The covariances are approximated using an ensemble-based method, in which we have assumed that the model error can be represented by stochastic terms added to the zonal and meridional momentum equations as well as the parameter equations that are assumed to be slowly varying in time. The prior and posterior covariances are projected back to their geometrical interpretation to shed light on the physics of the assimilation and investigate the effect of assimilating different type of information. We will also analyze the posterior model error and optimized parameters arising from the assimilation of different synthetic, and possibly real, data.

Back to agenda