Collection Of Mesoscale Heavy Precipitation Forecasting Techniques For The Study

China is located in the famous East-Asian monsoon region. There are lots of disasterscaused by mesosacle systems. How to predict the time, the location and the intensity about thesever weather is always the difficulty in the NWP research field. Ensemble prediction is a newstochastic dynamic forecasting technology developed in recent years. During the past decademajor NWP centers started to run ensemble forecast systems with their global models (Molteniet al.1996, Toth et al.1997, Houtekamer et al.1996). It has been demonstrated that suchensemble forecasts can be significantly advantageous to a single control forecast. Since theensemble workshop in 1994, encouraged by the success of global medium-range ensembleforecast, investigations have been made in examining the short-range ensemble forecast withmesoscale models. Mesoscale ensemble forecasting which aim for improving the forecasting ofthe sever weather has been become the top topic in the research field of NWP. A series ofresearches have been demonstrated that mesoscale ensemble forecast based on thesub-synoptic-scale numerical models could provide useful information for short-range forecastissues such as sever storms, cyclogenesis, precipitation.Recently, issues of how to generate theinitial perturbation fitting for the limited area were considered, as well as considerations ofmodel errors. Besides these, the boundary condition-related uncertainty needs more carefulconsideration.In order to search the appropriate ensemble perturbation method fitting for the convectiveinstability, to further understand the predictability of mesoscale precipitation and convectiveevents in the East-Asian monsoon region. A mesoscale ensemble prediction system using anon-hydrostatic mesoscale NWP model (GRAPES-Meso) is developed for the motivations ofimprovements of the heavy rain forecasts in summer season in China. This system, calledGRAPES Mesoscale Ensemble Prediction System(GRAPES-MEPS). Firstly, in order to studythe model related uncertainties of GRAPES-MEPS, convective factors in cumulusparameterization, multi-physical parameterizations and physical tendency have been considered.Secondly, small perturbations made by Breeding of the Growing Mode (BGM) have beenconsidered to account for the initial uncertainties of GRAPES-MEPS. Besides, a new ensembleapproach—— 'Hybrid Ensemble' has been applied to take into account the effects on the modelresolution. Some mesoscale convective events were preliminarily chosen for the case studies,such as flash flooding, squall line and torrential rainfall. A primary experiment and verificationresults show that GRAPES-MEPS is feasible and could capture some characteristic ofmesoscale system. The major work and remarks of this study are summarized as follow:(1)Fully considered about the uncertainties of the short-range forecast, a mesoscaleensemble forecaset system(GRAPES-MEPS) has been set up, which mainly consists of threeparts :initial perturbation ,model perturbation and ensemble verification.(2) To study the model uncertainties of the GRAPES-MEPS, a flash flooding case (10 July2004 in Beijng) was particularly chosen for the study of a 36h mesoscale ensemble forecastfirstly. Compared with 4 different experiments, we have found that each scheme has differentresult, even in the same scheme, members differ from each other. Totally, mesoscale ensemblesystem composed of multi-physical parameterizations perform better than of singleparameterization.(3) It is useful to add appropriate initial information when construct an ensemble forecastsystem. In other words, An effective ensemble forecast system should contain the uncertaintiesnot merely from the model itself, but also from the initial conditions.(4)To study the initial uncertainties of the GRAPES-MEPS, the Breeding of GrowingMode (BGM) method was adopted with a 12h time window cycle. Some members could reflectthe structure of the errors and capture some information of initial uncertainties. The ensembleverification results indicate that the spread among members has been increased and the forecastreliability of mesoscale ensemble forecasting has been improved. The results are preliminarybut encouraging, it could need more research and experiments to further improve theGRAPES-MEPS.(5)Although a small quantity of heavy rain events have been chosen for theGRAPES-MEPS experiments and their verifications, the results indicate that theGRAPES-MEPS has ability to capture the mesoscale system characters. Futhermore, there is areasonable spread among members and the ensemble mean could improve the location andintensity of the heavy rain forecasts.