|dc.description.abstract||The purpose of this study is to explore the feasibility of assimilating atmospheric state variables observed as well as derived from Doppler radar data to improve the prediction of a thunderstorm development. In the methodology, the complete algorithm is composed of three steps, namely (1) wind retrieval, (2) thermodynamic retrieval, and (3) moisture adjustment. The performance of the proposed method is investigated by six experiments, and all of them are conducted under the framework of OSSE (Observation System Simulation Experiment). From these experiments, it is attempted to address the following issues: (1) verification of proposed method (2) the impact of the assimilation frequency and assimilation time interval, (3) influence of assimilating with/without sounding data, and (4) influence of assimilating Doppler radar data and retrieval pressure field only. In (4), the vertical momentum equation is utilized to obtain the absolute potential temperature fields.
The results demonstrate that the accuracy of storm rainfall prediction can be improved after assimilating observed/retrieved atmosphere state variables from Doppler radar data. However, the impact on rainfall prediction from the assimilation frequency and time interval is insignificant. By contrast, the model forecast can be improved substantially if the data from a sounding released within the analysis domain is available. Furthermore, after the first assimilation with sounding, if one conducts a second assimilation even without the information from sounding observations, additional improvements can be achieved. The application of this method to real case study would be a natural extension of this study in the future.