[6-2]

Hybrid Data Assimilation Methods on a Modified Shallow Water Model

Michael Goodliff (University of Reading), Javier Amezcua (University of Reading) and Peter Jan van Leeuwen (University of Reading)

Hybrid data assimilation schemes are becoming more widely used in Numerical Weather Prediction (NWP). These methods combine ideas from successful schemes such as 4DVAR, the ensemble transform Kalman filter (ETKF) and Particle Filters. The motivation behind hybrid schemes is to combine benefits of different methods into the framework of others, theoretically benefiting from both methods. Although some of these hybrid schemes are being used operationally now, several basic questions on the reasons behind their performance are still open.

We systematically compare the performance of Ensemble Transform Kalman Filter (ETKF), Ensemble Transform Kalman Smoother (ETKS), Sequential Importance Resampling (SIR), 4DENVAR, 4DENVAR-QVA and a Particle Filter with a 4DENVAR proposal density on a 1D shallow water model which allows for convection and rain. Using the analysis root mean square error (RMSE) as a metric, these schemes have been compared considering observation regimes and different observation densities.