Doppler radar observation is nowadays an important data source for mesoscale and microscale weather analysis and forecasting. In this paper, a regional 3D-Var system with the capability of assimilating Doppler radial velocities has been developed. The 3D-Var system is modified to include vertical velocity increments according to Richardson's equation, background cloud water and rainwater, and observation operator for Doppler radial velocity. The fifth-generation of NCAR-Pennsylvania State University mesoscale model MM5 is used for the simulation experiments. A heavy rainfall case study has been conducted. The case is a frontal rainfall event in Korean Peninsula. The Korea Jindo Doppler radar captured the most part of the rainfall process in South Korea. From the Doppler radial velocity assimilation experiments for this case study, the following conclusions are obtained:

a. The modified 3D-Var system which includes w increments and Doppler radial velocity assimilation is stable and robust. The wind and vertical velocity analysis after assimilating Doppler velocity data are sound and appropriate.

b. The rainfall forecast from 3D-Var analysis with Doppler velocity data is better than that without radar data, verified against the radial velocities and Korean AWS (Automatic Weather Station) rainfall with all the thresholds.

c. Generally speaking, 3D-Var with more 3-hr cycling obtained more improved rainfall forecast.

d. Compared with Korean operational forecast, 3D-Var with radar data assimilation performed better for heavy rainfall forecast.

In summary, the regional 3D-Var system showed positive impact of assimilating Doppler radial velocities on rainfall forecasts and its potential for operational application. More detailed sensitivity study, verification, diagnosis and comparison will be presented at the conference.