Jing CHEN, Yuxiao CHEN, Hongqi LI, Jingzhuo WANG , Fajing CHEN
Numerical Weather Prediction Center, China Meteorological Administration, Beijing, China
ABSTRACT
Convective precipitation forecasts have been the hot topic of the meteorological research for many years. The high intensity and sudden occurrence of convective weather phenomena make them difficult to predict. The current method for predicting convective precipitation relies greatly on the Convective-scale ensemble model, which serves to produce probabilistic convective precipitation predictio and help improve the forecast quality.
A Convective-scale ensemble prediction system based on CMA Global/Regional Assimilation PrEdiction System (GRAPES) model, named the GRAPES Convective-scale ensemble prediction system (GRAPES-CEPS), was developed to promote the pre-summer convective precipitation forecast skill over south China .The GRAPES-CEPS ensemble calculates the initial condition perturbations using the ensemble transform Kalman filter (ETKF). Aside from the ICs perturbations, two stochastic schemes applied for each ensemble member: Stochastically Perturbed Parameterization Tendencies (SPPT) and Stochastically Perturbed Parameterizations (SPP) to describe the model uncertainty. In the two stochastic schemes, the random field which is described with first order Markov chain has a time-related characteristics and Gaussian distribution, and also has a continuous and smooth horizontal structure. GRAPES global ensemble system, which has been implemented in operational running since 28 Dec.2018, is used as Lateral boundary conditions to represent the uncertainty arose from lateral boundary conditions.
The traditional and neighborhood ensemble probability evaluation methods are applied to evaluate the skill of probabilistic convective precipitation forecasts over South China. The results shows skillful probabilistic forecasts of convective