Determination of zenith hydrostatic delay and its impact on GNSS-derived integrated water vapor

Surface pressure is a necessary meteorological variable for the accurate determination of integrated water vapor (IWV) using Global Navigation Satellite System (GNSS). The lack of pressure observations is a big issue for the conversion of historical GNSS observations, which is a relatively new area of GNSS applications in climatology. Hence the use of the surface pressure derived from either a blind model (e.g., Global Pressure and Temperature 2 wet, GPT2w) or a global atmospheric reanalysis (e.g., ERAInterim) becomes an important alternative solution. In this study, pressure derived from these two methods is compared against the pressure observed at 108 global GNSS stations at four epochs (00:00, 06:00, 12:00 and 18:00 UTC) each day for the period 2000-2013. Results show that a good accuracy is achieved from the GPT2w-derived pressure in the latitude band betwee n-30 and 30° and the average value of 6 h root-mean-square errors (RMSEs) across all the stations in this region is 2.5 hPa. Correspondingly, an error of 5.8mm and 0.9 kgm -2 in its resultant zenith hydrostatic delay (ZHD) and IWV is expected. However, for the stations located in the mid-latitude bands between-30 and-60° and between 30 and 60°, the mean value of the RMSEs is 7.3 hPa, and for the stations located in the high-latitude bands from-60 to-90° and from 60 to 90°, the mean value of the RMSEs is 9.9 hPa. The mean of the RMSEs of the ERAInterim-derived pressure across at the selected 100 stations is 0.9 hPa, which will lead to an equivalent error of 2.1mm and 0.3 kgm -2 in the ZHD and IWV, respectively, determined from this ERA-Interim-derived pressure. Results also show that the monthly IWV determined using pressure from ERAInterim has a good accuracy-with a relative error of better than 3% on a global scale; thus, the monthly IWV resulting from ERA-Interim-derived pressure has the potential to be used for climate studies, whilst the monthly IWV resulting from GPT2w-derived pressure has a relative e