Poster Session P1.30 Simultaneous Retrieval of Microphysical Parameters and Atmospheric State Variables with Radar Data and Ensemble Kalman Filter Method

Monday, 1 August 2005
Regency Ballroom (Omni Shoreham Hotel Washington D.C.)
Mingjing Tong, SOM/CAPS, University of Oklahoma, Norman, OK; and M. Xue

Handout (532.9 kB)

In our recent observing system simulation (OSS) experiments with ensemble Kalman filter method (Tong and Xue 2005, MWR), the wind, thermodynamic and microphysical fields of a supercell thunderstorm are retrieved very accurately from simulated single-Doppler radar radial velocity and reflectivity data. Forecasts starting from the analyses remain very good for at least 2 hours. However, when real data are used, both analysis and forecast will be affected by model errors. For the storm scale, microphysics parameterization is an important source of error, which can affect the strength, structure and evolution of the convective systems.

Past studies have shown that a reasonable agreement between simulated and observed radar reflectivity structures within thunderstorms is obtained only after tuning numerous coefficients in the model microphysics. This study examines the impact of the errors in some of these coefficients or parameters on the retrieved model state and tries to determine the ability of the EnKF method in correcting these errors. This is done by retrieving simultaneously selected microphysical parameters and the model state.

The ARPS model that our EnKF system is based on includes a three-class-ice microphysics scheme. The parameters that we are dealing with include the intercept parameters of rain, snow and hail drop size distributions and the densities of snow and hail. OSS experiments are performed in which individual parameters are retrieved separately or in combinations. The retrievals of individual parameters are quite successful. Not only can the parameter values be retrieved accurately through the assimilation cycles, the retrieved model state can be as good as and occasionally even better than the case of no parameter error. The retrieval is found to be sensitive to the variance of perturbations added to the parameters. Results of simultaneously retrieving all five parameters will also be presented.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner