Development Of Local Ensemble Transform Kalman Filter Scheme And Application Research

The ensemble transform Kalman filter(ETKF)is an effective initial perturbation scheme for ensemble prediction,which has been widely used.However,The finite ensemble sample size,the same ensemble member setting in ensemble transform Kalman filter(ETKF)and the forecast model error may make the two remote state variables have higher spurious correlation,thus affecting the quality of the set disturbance.In order to effectively solve the problem of long distance spurious correlation,the localization idea is introduced to the ETKF scheme.Based on the GRAPES regional ensemble prediction system(GRAPES REPS),the local ETKF scheme(LETKF)is developed,and the key parameters such as the local scale of the LETKF scheme are determined.Through the case study of rainstorm and continuous batch test,the validity of LETKF scheme to eliminate long distance spurious disturbance is analyzed from the aspects of initial disturbance's correlation distribution,energy structure,evolution characteristics and ensemble prediction comprehensive inspection.The results show that LETKF localization can effectively eliminate the spurious correlation of distant ensemble disturbance,and further improve the quality of ensemble disturbance in GRAPES regional ensemble prediction.Meanwhile,Initial disturbance of the new scheme can be more reasonable to capture the fast-growing physical structure of analysis error,more accurately reappear the linear and nonlinear propagation and evolution characteristics of the forecast error in the numerical model.The main conclusions are as follows:(1)The LETKF localization initial perturbation scheme for GRAPES REPS is developed.The new LETKF localization scheme can effectively eliminate the long distance spurious disturbance information of the regional ensemble disturbance and improve the overall quality of the ensemble disturbance.(2)Under the conditions of current model horizontal resolution and fifteen ensemble members,the result of LETKF localization radius of 700 km is the best for eliminating the false distance and improving the quality of ensemble disturbance.The correlation distribution,energy structure and evolution characteristics of ensemble perturbation prove the validity of the localization scheme and the rationality of the corresponding disturbance structure.In addition,the introduction of low layer atmospheric information to increase the ensemble dispersion of the forecast,further optimizes the performance of the LETKF localization scheme.(3)Based on case analysis and batch test,the results show that the LETKF localization scheme works well on improving the quality of ensemble prediction,especially for forecast of the magnitude of rain,middle rain and rainstorm.Compared with GRAPES REPS,the overall quality of regional ensemble prediction of localization scheme has obvious advantages,especially for the improvement of temperature field.The development and propagation characteristics of ensemble disturbance and spread of different physical quantities and different levels are not consistent.Generally,the higher layer have a good quality in the troposphere and the lower layer are relatively poor.