Construction Of Tm Model And PWV Inversion By Integrating FY-4A And ERA5 Dat

The water vapor plays a leading role in the atmospheric greenhouse effect and the earth’s water cycle,its change can indirectly affect the vertical stability of the atmosphere and the formation and evolution of weather system.Therefore,high-precision water vapor information is very important for modern meteorological operations and climatology research.Precipitable Water Vapor(PWV)is often used to characterize atmospheric water vapor content,and is a key parameter in climate research and extreme weather early warning.Compared with the traditional water vapor detection technology,PWV inversion using the Global Navigation Satellite System(GNSS)has the advantages of all-weather,high spatial and temporal resolution and low cost,and can achieve a large range of high precision water vapor monitoring.This study takes GNSS inversion of PWV as the main line.For the existing atmospheric weighted mean temperature(T_m)model parameters,modeling data sources to be optimized and the model construction only relies on a single sounding site or single grid point data,Based on fusion data(FY-4A GIIRS data and ERA5 reanalysis data),a regional T_mmodel(FY-ETm model)over China is established.In view of the problem that it is easy to lack measured meteorological parameters when inverse PWV with GNSS,the accuracy and applicability of PWV inversion using GPT3 model combined with different T_mmodels in various regions of China land were analyzed,and the accuracy of PWV inversion using FY-ETm model was verified by PWV obtained from GNSS stations as reference values.The existing water vapor fusion model and PWV vertical correction model depend on specific reanalysis data,and the selection of data type and the determination of model expression need further study.An empirical grid model based on Askne model and spherical harmonic function fitting PWV deviation is proposed.The results show that:(1)The FY-ETm model shows high accuracy.With the 84 sounding stations as the reference value,the average annual Bias and RMS of FY-ETm model are-0.02K and 5.79K,respectively,which are 3.62K(Bias),0.8K(RMS),2.54K(Bias)and 0.63K(RMS)higher than those of Bevis and GPT3 models.With the 120 evenly distributed reanalysis grid data of ERA5 as the reference value,the average annual Bias and RMS of FY-ETm model are 0.01 K and 3.32K,respectively,which are 0.97K(Bias),0.13K(RMS),2.94K(Bias)and 1.71K(RMS)higher than those of Bevis and GPT3 models.Compared with GPT3 model,FY-ETm model also shows obvious accuracy improvement in western and northern China.(2)PWV inversion methods by the T_mcalculated by GPT3,Bevis and improved Bevis combined with the temperature and pressure parameters calculated by the GPT3 model,three methods based on the GPT3 model to calculate PWV were recorded as G/PWV,GB/PWV and GBG/PWV.With the 50 sounding stations as the reference value,their average annual Bias were 0.06mm,0.17mm and 0.42mm,respectively,and the average annual RMS were 4.47mm,4.48mm and 4.54mm.In the Chinese land area,the stability of PWV retrieval by the three methods increases with the increase of latitude.The PWV results based on GPT3 model inversion can reach millimeter-level accuracy.In the case that the measured meteorological parameters at the station cannot be obtained in the process of GNSS inversion of PWV,the GPT3 model can be used for PWV inversion.With the PWV obtained from GNSS station as the reference value,the accuracy of PWV inversion from FY-ETm model is similar to that from GNSS station,Bias ranges from-0.5 mm to 0.5 mm.(3)Profiles of 20 radiosonde stations in Qinghai Tibet Plateau,China,along with the latest publicly available C-PWVC2 model are used to validate the local performance.The average annual Bias and RMS of the ASV-PWV and C-PWVC2 models were-0.44 mm,3.44mm,and-1.36mm,2.51mm,respectively.The accuracy of the ASV-PWV model in spring and winter was better than that in summer and autumn,and the accuracy of the C-PWVC2model had no obvious seasonal change.Using PWV data of 7381 ERA5 grid points as reference values,the average annual Bias and RMS of the ASV-PWV model were-0.73 mm and 4.28mm,respectively.The variation range of Bias was-4.5～1.5mm,and the main variation range was-2.5～0.5mm,accounting for 90.59%.The variation range of RMS was1～18 mm,and the main variation range was 1～6 mm,accounting for 78.3%.In general,the Bias value of the ASV-PWV model showed a trend of first increasing and then decreasing with the increase of the pressure value at the grid dot,and the RMS value increased with the increase of the pressure value at the grid dot.