Stochastic Perturbation On Individual Parameterization Scheme Tendencies In WRF Modeling System

With the development of the numerical forecasting model,most countries and regions have applied it as the primary method of daily operations.And weather model has become one of the main means for small-scale and severe weather.Due to the chaos of the atmosphere,as well as and inevitable observational errors,make numerical models deviate from atmospheric truths.To reduce this kind of bias,applying the ensemble prediction method for the probabilistic forecast is research hotspots in the scientific community.The Stochastically Perturbed Parameterization Tendencies scheme(SPPT)is one of the wildly used ensemble approaches with excellent performance in Weather Research and Forecasting model(WRF).But the scheme has the disadvantage of being unable to estimate the errors produced by different parameterization processes individually,and there is no better consideration of the interaction between errors and different parameterization processes.This study addresses this issue by proposing three individual perturbation schemes,including cumulus,planetary boundary layers,and radiation parameterization process.The performance of those schemes are evaluated against conventional multi-physical methods,multi-parameter,SPPT,stochastic kinetic energy backscattering approaches(SKEB).A correlation-preserved method is designed and applied to the combination scheme,which perturbs model parameterization scheme independently.The contribution of perturbation on different parameterization processes was evaluated,And the optimal value of perturbation parameters was discussed.The main conclusions of this study are listed as follows:1)Designed a Stochastically Perturbed Cumulus Parameterization Tendency(SPCPT)scheme,it can reduce the convective precipitation,significantly decreasing the forecast bias of rainfall.When using empirical parameters,the scheme provides a significant improvement in the probability forecasting performance of rainfall.Compared to the multi-cumulus parametrization scheme,SPPT scheme,and SKEB scheme,it further reduced precipitation RMSE by 3.6%,7.5%,and 9%.The scheme also boosts the variables spread of the upper atmosphere.2)Stochastically Perturbed PBL Parameterization Tendency scheme is more sensitive to gridpoint variance.The desired value of this parameter is determined by 30 sets of tests using random selection.And the appropriate value of this parameter is 0.50.Two case study has found that suitable decorrelation scales are conducive to the spatial distribution of precipitation,larger decorrelation time can increase spread,and large grid variance can effectively reduce system error.The scheme has a greater response to gridpoint variance than the SPPT.Long-term simulations have found that when using optimization parameters,the scheme can represent a more reasonable error pattern,with improvement in forecasting near-ground variables and precipitation.3)Based on the theory that there is a partial correlation in different physical tendencies.A perturbation method that can retain correlation is designed and applied to both longwave and shortwave radiation tendencies.This method improves the spread of over 37% of upper air temperature,which also can reduce warm bias on 2m temperature significantly.4)Stochastic Perturbations for Independent Parameterization Tendencies(SPIPT)scheme estimate uncertainty rising from cumulus,planetary boundary layer,and radiation.It incorporates the advantages of these three methods and able to adjust perturbation amplitude for different parameterization tendencies.Using empirical parameters in SPIPT,the spread of upper air temperature and wind gains by up to 166% and 143%,respectively.It is also reducing the error of around 4% in ground variables and significantly improving the precipitation GSS.This method reduces the temperature wind and precipitation errors further than the SPPT scheme by about 1.5%,3.8%,and 7.6%.5)The cumulus parameterization has more than 55% contribution to the SPIPT scheme.The interaction between parameterization processes ignored by SPPT schemes has more than30% contribution in temperature forecasting while SPIPT scheme is applied.The perturbation schemes proposed in this study have increased ensemble system performance mainly due to taking that interation into consideration.