Applicability of methods for estimating GPS precipitable water in the Qinghai-Tibet Plateau

Atmospheric weighted mean temperature (Tm) and zenith hydrostatic delay (ZHD) are key parameters in determining the accuracy of GPS precipitable water (PW) estimations. Using radiosonde profiles with a high vertical resolution of 10–30m, we calculated Tm, ZHD and PW and investigated the applicability of the modeled Tm and ZHD, including the GPS PW obtained with them, in the Qinghai-Tibet Plateau (QTP). It is found that (1) the bias and RMS error of the Bevis Tm model in the QTP are less than 5% and 2%, respectively, with a diurnal variation that is small in the morning and large otherwise. (2) The ZHD models have an RMS error within 1% in the QTP, which is smaller than that in the plain of Central China, which is within 1.5%. The ZHD bias of the Saastamoinen model is less than 0.5% in the QTP, smaller than those of the Hopfield and Black models. The diurnal variations of the modeled ZHD bias and RMS error are complex in the QTP, varying with the ZHD model and station. (3) The relative RMS error of GPS PW increases with the station altitude, with a range of 12–66% in the QTP and within 8% in the plain of Central China. The diurnal variations of the GPS PW bias and RMS error in the QTP are similar to those of the modeled ZHD. (4) The accuracy of GPS PW is mainly affected by the ZHD model, and a good Tm can improve the accuracy of GPS PW in the QTP. ► The Bevis Tm model has a larger bias and RMS error in the QTP than in the plain of Central China. ► The RMS error of the ZHD model is smaller in the QTP than in the plain of Central China. ► The relative RMS error of GPS PW increases with the station altitude. ► The accuracy of GPS PW is mainly affected by the ZHD model, and a good Tm can improve the accuracy of GPS PW in the QTP.