| Summary: | 博士 === 國立中央大學 === 大氣物理研究所 === 101 === A series of observation system simulation experiments (OSSEs) and real case study are conducted to investigate the application of the Doppler radar data assimilation technique for numerical model quantitative precipitation forecasts (QPF). A four-dimensional Variational Doppler Radar Analysis System (VDRAS) is adopted for all experiments. The first set of OSSEs demonstrates that when the background field contains the imperfect information predicted from a mesoscale model, the incorrect convective-scale perturbations in the background can result in spurious scattered precipitation. However, a smoothing procedure can be utilized to remove the fine structures from the primitive model output to avoid this over-prediction. Results from a second set of OSSEs indicate that the lack of low-elevation data due to beam blockage could significantly alter the retrieved low-level thermal and dynamical structures when different number of data assimilation cycles is applied. These impacts could lower the rainfall forecast capability of the model. The third set of OSSEs shows that, when the rainwater is assimilated over a long assimilation window, the nonlinearity embedded in the microphysical process could lead the minimization algorithm to a wrong direction, causing a further degradation of the rainfall prediction. However, using multiple short assimilation cycles produces better minimization and forecast results than those obtained with a single long cycle. A real case experiment based on data collected during Intensive Operation Period (IOP) #8 of the 2008 Southwest Monsoon Experiment (SoWMEX) is conducted to provide a verification of the conclusions obtained from OSSEs under a realistic framework. The microphysics scheme of VDRAS is extended from warm rain process to cold rain process. It is found that the retrieved water content would be underestimated if all radar reflectivities are assumed to be in the form of warm rain. This underestimation can be improved when the cold rain process is implemented into VDRAS. The VDRAS with ice physics can provide better rainfall forecast.
|
|---|
