A comparative study of zenith tropospheric delay and precipitable water vapor estimates using scientific GPS processing software and web based automated PPP service

Abstract

The zenith tropospheric delay (ZTD) is a significant error source which affects the GPS signal propagation time. ZTD time series can directly or indirectly reflect the weather variations. In geodetic studies their computation is important since it improves position accuracy. Sophisticated software packages using network data of ground-based GPS receivers are used for this purpose. During the last years, various web based precise point positioning (PPP) services can provide position solutions. Additional products, such as ZTD estimates, can be derived both from PPP as well as from network solutions. Research institutes and laboratories like JPL provide similar services. This study’s objective is twofold. In a first instance we used the Bernese and GAMIT software packages and the PPP service via the JPL webpage to estimate ZTD values every hour for a period of one month. We selected a geographical area in Greece, where seasonal weather variations are frequent and GPS permanent station infrastructure is available. The estimated ZTD values derived from GPS processing for a single station were compared with the ZTD estimates derived directly from the Saastamoinen model using meteorological data from a co-located meteorological sensor as input to this model. The results show an rms agreement of about 45 mm. The second scope of this study is to compare precipitable water (PW) values between different processing schemes. For this purpose, we used the derived zenith wet delay estimates of each processing scheme and a global formula for the computation of the weighted mean temperature of the atmosphere for our area of study. The rms differences between the web PPP solution and the Bernese derived estimates were 0.35 mm. The rms of the differences between and the web PPP solution and the GAMIT derived estimates of PW were 1.67 mm. We conclude that this difference follows from the zenith hydrostatic delay component that needs to be properly modeled when high accuracy results are required.