Evaluate And Refine The Model To Inverse Tropospheric Water Vapor With Ground-based GNSS

Tropospheric zenith total delay(ZTD)and weighted mean temperature(Tm)is a key parameter to inverse water vapor from ground-based Global Navigation Satellite System(GNSS).Nowadays,the GPS2 w model can provide a good effort on the meteorological parameter all over the world.But the applicability of the model needs to be researched further in China where is with a vast territory,complex terrain and diverse climate.In order to build a model that fits China best,this paper will study the applicability of the Tm values calculated from GPT2 w model with the grid resolution of 1°and 5°over China.They are defined as G1-Tm and G5-Tm respectively.At the same time,the meteorological parameter calculated from GPT2 w are combined with the zenith hydrostatic delay(ZHD)model and the zenith wet delay(ZWD)model.They are used to calculate the ZTD values of GPT2 w with grid resolution of 1°and 5°respectively.At the same time,the applicability of these two ZTD models are to be evaluate in China.In addition,based on the meteorological parameter provided by sounding stations in China,Bevis model,regional Bevis-L model,E-Tm model,H-Tm?model,H-Tm?model are built with different elevation in China respectively.Then compare the applicability of these four new models with Bevis model and G1-Tm model.The research details and experimental results of this paper based on ground-based GNSS are the following:1.The data provided by sounding stations from 2013 to 2015 over China are used to calculate the Tm values by integral method.The calculated values are treated as the true value.They are used to evaluate the applicability of G1-Tm and G5-Tm over China from all aspects.The evaluate results show the followings:(1)The average bias,MAE and RMS of G1-Tm in a year is-1.32 K,3.38 K and 4.34 K respectively which is better than G5-Tm;(2)The average MAE and RMS in a season appears seasonal differences obviously from temporal distribution;(3)The average model of bias in a year decreases with the increase of the longitude while the average MAE and RMS in a year increase with the increase of the latitude from space distribution.But when the elevation is lower than 3000 m,the applicability of G1-Tm is better than G5-Tm significantly.On the other hands,when the elevation is higher than 3000 m,G5-Tm is better.2.This paper regards the ZTD in 2015 collected from 185 GNSS stations in the Crustal Movement Observation Network of China(CMONOC)as the true values.Then evaluate the applicability of G1-ZTD model and G5-ZTD model in China in detailed.The evaluate results are the followings:(1)The average bias,MAE and RMS of G1-ZTD in a year is 1.38 cm,2.75 cm and 2.88 cm respectively which is better than G1-ZTD model.(2)The average RMS of a season exist seasonal difference obviously in each of the two model in the eye of temporal distribution.(3)In the eye of spatial distribution,the average RMS of a year exist the trend that the RMS decrease with the increase of the latitude.At the same time,the RMS decrease with the increase of the elevation obviously very much.3.China is divided into several parts by the elevation of sounding stations.Then Bevis model,regional Bevis-L model,E-Tm model,H-Tm?model,H-Tm?model are built in China respectively.Statistic and analyze the accuracy distribution of the G1-Tm model and the five new models mentioned above in different elevation and different season.The statistical results shows the followings:(1)The whole accuracy of the regional Bevis-L model built by surface temperature is better than G1-Tm model in each elevation area.(2)For nonmeteorological data model,G1-Tm model has a better applicability in the elevation of 1000 m area;in the elevation from 1000 m to 2000 m area,the H-Tm?model and H-Tm?model consist seasonal complementary with G1-Tm model;the applicability of H-Tm?model and H-Tm?model is better in the area where elevation is larger than 2000 m.