The Prelimiary Improvements Of Cumulus Convective Parameterization Schemes In GRAPES_Meso Model

Some effective improvements of cumulus convective parameterization schemes are proposed mostly based on common numerical weather prediction models, such as WRF, MM5 and MSM. Several methods are put forwarded to improve the cumulus parameterization schemes in GRAPES_Meso (Global/Regional Assimilation and PrEdiction Mesocale) in this paper. Firstly numerical compared tests were conducted to evaluate Kain-Fritcsh eta (hereinafter KFeta) scheme and Betts-Miller-Janjic (hereinafter BMJ) scheme. Combining with the results of previous research, three modifications to the Kain-Fritcsh convective parameterization are implemented into the GRAPES_Meso.The numerical experiments were carried out to evaluate the improved schemes. There have some main concluions:(1) There are no obvious difference between KFeta and BMJ in winter, but there are some differences in summer.The statitic character of meteorological fields show that KFeta performances lightly better than BMJ, but it doesn’t as good as BMJ in the middle atmospheric layer. The results of T-Score verification indicate that KFeta scheme has a higher score than BMJ scheme, especially in moderate rain to severe rain over southern China.(2) In this paper, three kinds of improvement of Kain-Fritcsh cumulus parameterization schemes are put forwarded:1) the temperature perturbation as trigger mechanism is decomposed into vertical and horizontal temperature perturbation associated with water vapor(hereinafter KF1); 2) the water vapor advection is directly added into the original temperature perturbation in trigger function(hereinafter KF2); 3) both two modifications(KF1 and KF2) are combined as an whole in trigger function to take into account the water vapor(hereinafter KF3).The improved scheme can better reflect the actual obervation than the old KF when simulates either heavy rains or typhoons, and KF3 scheme performances better than the other improved scheme, especially simulating the precipitation and the typhoon path.(3) The main conclusions are as follows:the fake precipitation is less in KF1 than the original scheme in the case of rainstorms; the center of the heavy rainfall is better simulated in KF2; the range and intensity of precipitation are better improved by KF3. In the typhoon cases, the simulated intensity is the best in KF1 than the others; while the simulated track error is the smallest in KF2; simulated of these two parts all improved in KF3.Besides, the T-Score verification and RMSE indicate that KF2 and KF3 scheme are better than KF1.