|dc.description.abstract||An explicit simulation of Typhoon Morakot(2009) was studied. Three groups of numerical experiments were designed by varying microphysics parameterization schemes and vertical resolution, which were i) single-moment(WSM6_31) and double-moment(WDM6_31) schemes, ii) different vertical resolution(WDM6_21, WDM6_31, WDM6_mix45) and iii)different double-moment schemes(WDM6_31, Thompson_31, Morrison_31, Milbrandt and Yau_31). The study focused on the effects on the accumulated precipitation over mountainous areas, which was usually overestimated by model simulation. In addition to the comparison of radar reflectivity, typhoon tracks and intensity, precipitation maximum and patterns, a series of quantitatively statistics scores were evaluated between the observation data and model output.
In experiment i), WDM6_31 revealed a better performance on typhoon intensity and landfall time than WSM6_31. Besides, WDM6_31 also had better TSs and ETSs in 24-hour accumulated precipitation. In experiment ii), varying the vertical resolution had a small effect on typhoon precipitation maximum, intensity and landfall time simulation. No matter for 24-hour or 6-hour accumulated precipitation, WDM6_mix45 had the best performance among three simulations. Because of the finer vertical resolution in mid-to-upper layers, WDM6_mix45 avoided the unreasonable vertical velocity and radar reflectivity in upper layers. At the same time, WDM6_mix45 revealed the break down process of typhoon eye, which reduced typhoon intensity efficiently then lowered the overestimation of precipitation maximum in other simulations.
In experiment iii), Thompson_31, Morrison_31, Milbrandt and Yau_31 runs improved the positive bias for WDM6_31 run on radar reflectivity and precipitation maximum. On August 8th, Milbrandt and Yau_31 run had the smallest BIAS-1 value. While applying the statistics scores, the users should pay attention to the deficiencies of the scores, which could mislead the ability of model’s precipitation predicbility. On the vertical crosssection over terrain, WDM6_31 tended to produce more raindrops, Thompson_31 tended to produce more snows, whereas Milbrandt and Yau_31 tended to produce more ice flakes. It was the difference of original designs for every schemes that made the difference of precipitation simulations.