Estimation of GPS Site Height by using Dynamic Hydrostatic and Wet Mapping Functions Based on Radiosonde Refractivity Profile

GPS site height time series of GEONET (GPS Earth Observation Network which consists of about 1000 sites) of GSI (Geographical Survey Institute) in Japan show seasonal variations and several days variations of several ten mm. If some of the variations come from atmosphere, the one of the most plausible error sources in the atmosphere is "climatic" mapping function generally used in GPS analyses. We thus studied characteristics of site height time series of GEONET and "dynamic" mapping function (DMF) based on radiosonde profiles. We estimated GPS site heights by using the DMF and compared those with estimates by using climatic mapping function. Radiosonde data of 20 sites observed by JMA (Japan Meteorological Agency) and GEONET site coordinate time series based on GSI's routine analysis from 1997 to 2002 were used. After correction of offsets of height caused by earthquakes, periodic variations of the height time series were computed by fitting sinusoidal harmonic function to the time series. We analyzed four GEONET sites near radiosonde sites for two years in 2001 and 2002 by using Bernese GPS software. The reference site locates in the central Japan and mean distance of other three sites from the site is about 1000 km. DMF was computed by ray tracing, where isotropic distribution of refractivity in spherical layers was assumed. Because in-situ surface meteorological observation is not available in GEONET sites, we interpolated surface meteorological data observed at about 150 sites to the GEONET sites. The hydrostatic DMF was used to compute slant hydrostatic delays and the wet DMF and its derivative were used to estimate hourly zenith wet delay with daily coordinates. The estimates were compared with those by using Niell's mapping function (NMF) [1996]. The annual component of the height time-series of GSI's routine analysis have mean amplitude of about 10 mm and show extremum around day of year (DOY) 30 and 210. The hydrostatic DMF has significant annual variations from 0.2% to 0.5% of mapping factor, and it is about two times larger than those of NMF in the northern Japan while the mapping factors of DMF and NMF show almost similar in the southern Japan. Those in wet DMF are about 1% at the four sites. Mapping factors of hydrostatic and wet DMF show maxima around the end of January and the middle of January, respectively. The difference of two height time series of NMF and DMF shows seasonal variations of about 2-3 mm, where the magnitude is about two times larger than the difference of two height time series by using NMF and NMF but ignoring annual variation of its hydrostatic term. Several day variations of about 3 mm are also seen in the difference of two height time series of NMF and DMF. We have confirmed that most of the seasonal trend in the difference came from hydrostatic term of DMF, and that several day variations in the difference are contributed by both hydrostatic and wet DMFs, where the former is larger in winter season and the latter is larger in summer season, respectively. The facts suggest that hydrostatic DMF is important to remove systematic annual variations which can not be explained climatic mapping function. However, the magnitude of the seasonal variations in the difference of the two height time-series of DMF and NMF are too small to explain those in height time series of about 20mm.