| Typhoon is one of the most threatening weather systems. Typhoon Morakot (0908) caused extremely precipitation in the southeast coast of the Chinese mainland and Taiwan, which resulted in tremendous property losses and casualties. Therefore, it is very important to simulate Typhoon Morakot accurately, in order to study the formation mechanism of heavy precipitation for numerical weather prediction. In this paper, the track, intensity and precipitation of Typhoon Morakot are simulated using WRF model.First of all, the impact of physical parameterization schemes, including cumulus, microphysics and planetary boundary layer parameterization schemes on track, intensity and precipitation simulation has been studied. The results show that the Kain-Fritsch cumulus parameterization scheme performs best on the track simulation, and there is little difference among the intensities simulated by all the experiments. Cumulus convective parameterization schemes influence the spatial distribution of precipitation, while microphysics parameterization schemes affect the amount of precipitation by different distribution of precipitation particles. Through different near-surface friction and latent heat fluxes, planetary boundary layer parameterization schemes influence the simulation results, especially the typhoon axisymmetric structure.Meanwhile, the ensemble mean method is conducted to reduce the uncertainty of each experiment. The ensemble mean simulation shows that it can reduce the systematic errors, increase the reliability of the simulation and improve the simulation of extreme torrential precipitation. Considering the simulation of track and precipitation, WSM6_KF experiment is a good combination of cumulus convective and microphysics parameterization schemes.In order to select an appropriate coefficient of the spatial decorrelation scale, a single point assimilation experiment is conducted to study the characteristics of the statistical background error covariance matrix calculated by the NMC (National Meteorological Center) method. Five sets of data assimilation experiments using conventional observations (surface data alone, sounding data alone, both surface and sounding data, removing all the observations175km and360km away from the typhoon center) are conducted. Among all the experiments, the adjustment of the initial fields is the least by assimilating the sounding data alone, but with the integration of the model, the average track is slightly improved throughout the simulation period. In the early period of the simulation, the simulated track by assimilating both surface and sounding data is the closest to the observation, while in the late period, especially after landing the mainland, the simulated track by assimilating sounding data is the best one. Assimilating the closer observations to the typhoon center can improve the relative humidity and wind fields.Six experiments of assimilating radar data (RADAR1alone, RADAR3alone, RADAR4alone, and radial wind, reflectivity, and both radial wind and reflectivity data of the three radars) are designed. Results show that the average increments of the initial field are small by assimilating the radial wind data of the three radars, but the adjustments are still large over the study domain. Assimilating RADAR4shows the smallest adjustments to the initial fields due to the blockage of mountain areas. Compared to the control experiment, the simulated track is the best by assimilating both radial wind and reflectivity data, followed by assimilating RADAR3alone and the radial wind data of the three radars. The simulated typhoon intensities are not affected significantly in all the assimilation experiments. The amount of0-24h accumulated precipitation simulated by assimilating both radial wind and reflectivity of three radars is closer to the observation than that simulated by the control experiment. It also improves extreme precipitation centers for the simulation periods of24～48h and48～72h. |
PDF Full Text Request