Abstract:In order to reduce the limitations and bias of the temperature and humidity profiles extrapolation scheme of the GRAPES(Global/Regional Assimilation and Prediction System) 3-DVar systems for the upper-atmosphere background,this paper proposes to use climatic data to reconstruct upper-atmosphere vertical profiles of temperature and humidity.First,a series of simulation experiments were conducted by a one-dimensional variational data assimilation system to analyze the bias of the current extrapolation program and its effects on inversion results.Then,vertical profiles were constructed using upper-atmosphere climatic data,and assimilation deviation was also discussed.Finally,assimilation experiments were performed and analyzed through the GRAPES global analysis and forecast system.The one-dimensional variational system experiments showed that there was deviation between the structure of the current upper background field of temperature and humidity profiles extrapolation scheme and the actual conditions of the upper atmosphere.Its inversion results suggested a greater negative impact of the stratosphere and troposphere.The application of climatic data to correct the upper atmospheric background field(temperature and humidity profiles) could reduce the deviation by over 50%,which further proved the feasibility of using upper climatic data to replace the upper background field produced by the temperature and humidity profiles interpolation scheme in the current GRAPES model assimilation system.The results from the GRAPES global 3DVar assimilation experiment indicated that using climatic data to replace the temperature and humidity profile structure of the upper atmosphere can significantly improve the inversion results of the stratosphere and upper troposphere.The upper atmosphere climate profile corrections for the background field can improve the analysis results of the GRAPES-3DVar system,reboot the upper atmosphere satellite channel and optimize the assimilation and forecast result of the GRAPES model.The final results showed that the maximum of the upper background temperature and humidity profiles extrapolation scheme error in the one-dimensional experiments could reach tens of degrees or more around the 1 hPa level,not only affecting the stratosphere but also having an impact on the troposphere.Using climate data to correct the GRAPES upper temperature and humidity data,the temperature deviations can be reduced by over 50%,which fully proves the feasibility of replacing the interpolation assimilation system in the current GRAPES model with the top-level climate data scheme.The global GRAPES-3DVar assimilation test results showed that the new scheme not only significantly improves the quality of the analysis of the stratosphere,but also the upper troposphere.This paper has preliminarily improved the upper atmospheric temperature and moisture profiles construction scheme for the GRAPES model.However,there is still room for optimizing the GRAPES-3DVar model upper-atmosphere assimilation.Channel selection schemes for different kinds of data and error correction techniques need to be further studied.The data used in the assimilation cycle experiment were conventional observations and microwave thermometer data(AMSU-A) from NOAA-15/16.The NCEP reanalysis data were used as testing data.The control group(CTRL) used the original upper-atmosphere background field produced by the temperature and humidity profile extrapolation scheme;the experiment group(TEST) used the climate profile correction scheme.During the experiments,the control group and the experimental group both used the same data and satellite channel selection method.The radiative transfer model was RTTOV7 in the GRAPES system.