| dc.description.abstract | Model for Prediction Across Scales-Atmosphere (MPAS-A) is the new generation global model developed by NCAR. However, the model prediction may not be improved without a good initial field. In order to improve the initial field, this research applies a hybrid system that combines two assimilation systems in which the background is from CWB_GFS, Grid-point statistical Interpolation (GSI) . The hybrid system assimilates the Global Positioning System Radio Occultation (GPSRO) data, including both bending angle and refractivity, in GFS to make a new analysis. The horizontal resolution in GSI-3DVar is T320 (approximately 37.5 km). The vertical resolution of both assimilation systems has 40 levels. Time window of assimilation in this research is 4.5 days, and the new analysis data are used as the initial field for MPAS-A. The horizontal resolution of MPAS-A is a variable resolution of 60-15 km. The higher resolution 15 km is centered is near Taiwan, the higher resolution domain covers East Asia and North Western Pacific.
In this research, we design three kinds of experiments. The first experiment assimilates GTS data, the second experiment assimilates bending angle data (BND) and the third experiment assimilates refractivity data (REF). The first case is Typhoon Soudelor (2015). Time of assimilation is from 0000 UTC 30 July to 1200 UTC 03 August, and the simulation duration is five days from the end of data assimilation with three different analyses. The second case is Typhoon Jelawat with the simulation of three different analyses for five days. At last, we analyze the difference be the initial fields and simulations for these two cases, and discuss the data impacts on the simulations. The tracks with assimilation of BND or REF in Soudelor are better than that with GTS. However, all the experiments over-predict the typhoon intensity. In Jelawat, however, the tracks with both BND and REF are worse than that with GTS. | en_US |