Study On The Performance Of CESM-EnSRF Assimilation System And The Influence Of Different Ensemble Generating Methods On The Assimilation Performance

Ozone and methane are important atmospheric trace gases,which have an important impact on climate change.In addition,the FY3-E satellite is going to be launched which can achieve the profile of the global atmospheric compositions.In order to support the future application of this data in the climate model,it is necessary to construct an assimilation system which can quickly respond to observation data.This study has combined with the established CESM-EnSRF assimilation system and some modules in the system have been adjusted for simultaneously assimilating ozone and methane observations from the Atmospheric Infrared Detector(AIRS)into' the climate model.We have designed a ten times cycling assimilation test with the assimilation window of 6h.So the performance of the adjusted CESM-EnSRF assimilation system can be evaluated systematically.Furthermore the performance of the CESM-EnSRF assimilation system is greatly depend on the quality of the ensemble members cause of the estimation of the background error covariance is calculated by a set of ensemble members.The effects of ensemble member generation methods in the assimilation system have been studied.The performance of original Monte Carlo random perturbation method and time lag averaging method,growth model propagation method and conditional nonlinear optimal disturbance method used in the assimilation system have been compared.Ozone is selected as the target element in the experiment of this part,duo to its important climate effect.The results showed that:(1)The CESM-EnSRF assimilation system can improve the accuracy of background members by assimilating the ozone and methane satellite observation data from AIRS.The analysis fields are closer to the observation data than the background field.(2)After some adjustments,the CESM-EnSRF assimilation system can improve the forecast performance of ozone and methane in CESM model effectively,and the effects are mainly reflected in the stratosphere.However,at the top of the stratosphere,the assimilation performance to ozone is limited.In addition,the data cycling assimilation is more effective in improving the forecast accuracy of methane and ozone in the stratosphere.(3)The ensemble members generated by the Monte Carlo method have a problem of large dispersion;this condition is opposite in the Lag Average method.The probability distribution of Talagrand in the ensemble members generated by the Breeding Growth method is more reasonable then that by the Monte Carlo method.On this basis,the Conditional Nonlinear Optimal Perturbation method shows an improvement.(4)Four ensemble forecast methods all have a positive effect on the forecast of ozone in the model,the method used in the primary system has a better effect in the assimilation than the Lag Average Forecast method.In addition,the Breeding Growth method has a breeding cycle with the use of model,so the imbalance of the model cased by the Monte Carlo method has been adjusted;the assimilation performance shows an improvement.On this basis,the Conditional Nonlinear Optimal Perturbation method shows a small progress.(5)During the cycling assimilation period,four ensemble generating method all shows a progress in the improvement rate of assimilation in the stratosphere,and a small change in the troposphere.In addition,the advantage of the Conditional Nonlinear Optimal Perturbation method used in the ensemble assimilation system is mainly in the later period.