Improvement Of Stochastically Perturbed Parametrization Tendencies Scheme For Global Ensemble Forecasting Based On Bias Estimation

Stochastically Perturbed Parametrization Tendencies(SPPT)scheme is a traditional ensemble prediction model perturbation method,commonly used to describe the uncertainty brought by physical processes.It has been widely applied to the operational ensemble prediction systems of major numerical prediction centers,such as the CMA-GEPS of the Center for Earth System Modeling and Prediction of China Meteorological Administration,which has also introduced the SPPT scheme.However,the model inevitably has systematic bias.In order to reduce the impact of systematic bias on ensemble prediction,it is necessary to deeply analyze the systematic bias characteristics of CMA-GEPS,and improve the widely used SPPT scheme based on the systematic bias characteristics.Thus,the effects of random and systematic errors are considered simultaneously,providing a scientific basis for developing more comprehensive model perturbation method for global ensemble forecasting model.Based on the experimental data of ensemble forecasting run by the CMA-GEPS V1.1,this paper analyzes the characteristics of 21 ensemble members and the ensemble average potential temperature systematic bias,and estimates the ensemble forecast potential temperature systematic bias.The 10-day average of the control forecast bias before the forecast start time is taken as the ensemble forecast systematic bias.The estimated potential temperature systematic bias characteristics are analyzed.Finally,the empirical orthogonal function(EOF)method is used to extract the main characteristics of systematic bias and obtain the systematic bias tendency on the time integration step.In the integration process,the systematic bias correction method and the traditional SPPT scheme are combined to build a new model perturbation method(Bias correction of Bias Tenancy based on SPPT,SPPT-B)that combines systematic bias and random errors of ensemble forecasting.Several ensemble forecasting experiments are designed and conducted to explore the impact of SPPT-B method on global ensemble forecasting.The main conclusions are as follows:(1)The spatiotemporal distribution characteristics of the systematic bias for potential temperature in the CMA-GEPS mainly show that the systematic bias in the upper model layers is significantly larger than that in the middle and lower layers,and there is a trend of increasing with forecast lead times.Regardless of whether it is summer or winter,the distribution of systematic bias for potential temperature in the upper model layers(about 150 h Pa)is characterized by positive(negative)systematic bias for potential temperature in low latitude regions(middle and high latitude regions).The distribution of systematic bias in the middle and lower model layers show a characteristic of small(relatively large)systematic bias in low latitude regions(middle and high latitude regions),and the distribution characteristics of systematic bias for potential temperature of 21 ensemble members are basically consistent.(2)The analysis shows that the systematic bias of potential temperature in control prediction can better represent the systematic bias of potential temperature in ensemble prediction members.In this paper,the 10-average of the control forecast bias before the forecast start time is used to estimate the systematic bias of the ensemble prediction potential temperature,and the estimated systematic bias of the potential temperature is decomposed by EOF,and the first mode is used as the main part of the systematic bias of the ensemble forecast.The results show that the systematic bias of ensemble prediction represented by the first mode basically shows the characteristics of linear growth with forecast lead times,and the systematic bias in the upper troposphere is larger than that in the middle and lower troposphere.(3)The improvement effect of SPPT-B on ensemble forecasting is influenced by both the systematic bias correction method(INI-B)and the SPPT scheme(INI＿SPPT).SPPT-B has good advantages in improving ensemble forecasting skills.SPPT-B is better than INI-B and INI＿SPPT in improving the reliability and probability prediction skills of ensemble prediction systems.Only in the upper levels of the southern hemisphere in summer,tropical regions,and tropical regions in winter,INI-B is better than SPPT-B in the improvement of the consistency and CRPS.By comparing the effects of various experiments on improving the RMSE and Spread,SPPT-B has a better effect to improve the Spread of ensemble prediction than INI-B,and a better effect to reduce the RMSE of ensemble prediction than INI＿SPPT.