| dc.description.abstract | In previous research the variational Doppler Radar Analysis System (VDRAS), developed by National Centers for Atmospheric Sciences (NCAR), had been applied in real-time analysis of low-level wind and convergence for nowcasting of thunderstorms. By assimilating single or multiple-Doppler radar observations, VDRAS was also utilized to analyze and forecast the super-cells and squall lines. However, most of those experiments were conducted in a wide open region with flat surface. In this study, it was attempted for the first time to apply VDRAS in Taiwan and her vicinity, where the topography over the island set a complicated geographic environment, and the surrounding oceans limited any type of observations except meteorological radars. These two factors posed a challenging task to the success of the model forecast of meso-scale precipitation systems.
In order to test the performance of VDRAS in Taiwan area, a real case of Mei-Yu front, occurred on 14 June 2008 during Southwest Monsoon Expeiment/Terrain-Influenced Monsoon Rainfall Experiment (SoWMEX/TiMREX) IOP8, was chosen. The background field was prepared using observations from soundings and surface stations. Radar data collected by two Central Weather Bureau S-band radars, located at Cigu and Kenting respectively, were assimilated into VDRAS. Through the minimization of a cost function under the constraint of a numerical model (i.e., the 4DVAR method), an optimal analysis field was obtained for initialization, then followed by model forecasts. Furthermore, in order to investigate the influence on Quantitative Precipitation Forecast (QPF) by terrain effects, a series of forecasting experiments were designed in which the VDRAS analysis fields were carefully merged with Weather Research and Forecasting (WRF) model. The latter is known to have a better capability to resolve complex terrain.
Analysis fields obtained from assimilating radar data showed that VDRAS could produce dynamical and thermodynamic structures of a weather system rather reasonably. The model forecast, starting from the optimal analysis field, also agrees well with observations in terms of the moving direction of the main convective systems. The ETS score of two hours accumulated rainfall forecast at 6-mm threshold is 0.21. It was found that the QPF scores achieved by combining VDRAS and WRF could significantly exceed those of using WRF or VDRAS alone. Apparently this was attributed to the assimilation of radar data into VDRAS and the terrain-resolving capability of WRF.
| en_US |