Research On GNSS Water Vapor Retrieval Method In China Based On Tm And ZHD Enhanced Models

Water vapor,as an important greenhouse gas in the troposphere,plays a vital role in global atmospheric radiation,energy balance and water cycle.Atmospheric weighted average temperature（T_m）and Zenith Hydrostatic Dealy（ZHD）are important factors affecting the calculation of atmospheric Precipitable Water Vapor（PWV）.In this paper,the construction of T_m model and ZHD vertical section model are respectively carried out by using multi-source data in China.The main work is as follows:1.In view of the problems that the existing T_m model only takes into account the influence of a single factor and has poor applicability in China,this paper uses the sounding station data in mainland China from 2015 to 2017 to analyze the relationship between T_m and multiple factors（surface temperature T_s,latitude,longitude and elevation）correlation.According to the characteristics of complex terrain and changeable climate in China,a multi-factor national unified T_m-C1 model and a regional T_m-C2 model（modeled according to the regions of the South,North,Northwest,and Qinghai-Tibet Plateau）that take into account the periodic residuals are constructed.Taking the T_m data of the sounding station in 2018 as a reference value,it is compared with the Bevis model and the GPT3 model.The results show that the T_m-C2 model has better applicability in China than the other models,and effectively reduces the influence of the seasonal,temporal and spatial distribution of T_m on the calculation of T_m,especially in the Qinghai-Tibet Plateau.2.The spatial and temporal distribution characteristics of the ZHD elevation scaling factor were analyzed,and the ZHD vertical profile model(ZHD_CH model)with a horizontal resolution of 0.25°×0.25°was constructed using the ERA-5 data.Taking the layered data of the sounding station and MERRA-2 as reference values,and compared with the GPT3 model with excellent performance.Experiments show that the vertical correction performance of the ZHD_CH model is improved by 0.21 cm（3.95%）and 0.47 cm（10.49%）,respectively,compared with the GPT3model.The user only needs to provide the annual accumulation day,the reference position information,the ZHD value at the reference position and the target height,and then the ZHD value at the target position can be obtained.3.GNSS water vapor retrieval is performed based on the ZHD ultrafast product(ZHD_OP)and the ZHD prediction product(ZHD_FC)provided by Vienna Mapping Functions 3（VMF3）of the Numerical Weather model.The Ts is provided by the weather station,and the T_m is calculated by the T_m-C2 model.The ZHD at the height of the VMF3 grid is corrected to the height of the GNSS station by the ZHD_CH vertical section model,and interpolate to the GNSS station with two horizontal interpolation methods（Bilinear Interpolation and Inverse Distance Weighted）to obtain the ZHD of the GNSS station,and then obtain the ZWD_OP and ZWD_FCfrom the ZTD provided by the GNSS,and finally obtain the ultrafast GNSS-PWV and real-time GNSS-PWV.Accuracy verification is carried out with the PWV provided by the sounding station as a reference value.The results show:（1）Bilinear interpolation method has better horizontal interpolation performance;（2）The ZHD without vertical correction had an average absolute bias of 12.43 mm and an average RMSE of 13.24 mm in GNSS water vapor retrieval.After vertical correction,the average absolute bias and average RMSE error of GNSS-PWV are1.05 mm（an increase of 91.55%）,2.49 mm（an increase of 81.19%）,respectively;（3）The water vapor inversion performance of ZHD_FC is close to that of ZHD_OP,indicating that the prediction products provided by VMF3 can achieve real-time GNSS water vapor prediction.