Earth Planets Space, Vol. 52 (No. 10), pp. 685-690, 2000LETTER
Yoshinori Shoji1, Hajime Nakamura1, Kazumasa Aonashi1, Akinori Ichiki1, Hiromu Seko1, and Members of GPS/MET Japan Summer Campaign 1997 in Tsukuba
1Meteorological Research Institute, 1-1 Nagamine, Tsukuba, Ibaraki 305-0052, Japan
(Received January 7, 2000; Revised September 9, 2000; Accepted September 9, 2000)
Abstract: Simultaneous GPS and water vapor radiometer (WVR) observations were carried out at Tsukuba, Japan from May 1 to June 30, 1998. The precise point positioning method of the GIPSY/OASIS-II software package (GIPSY) was used to retrieve precipitable water vapor (GPS_PWV) from GPS data, which was then compared with precipitable water vapor observed by WVR (WVR_PWV). They agreed quite well with the root mean square difference of less than 1.5 mm. However, periodic variations were found in the difference between GPS_PWV and WVR_PWV (dPWV). It was also found that semi-diurnal or diurnal components of these variations had a positive correlation with site displacement due to Ocean Tidal Loading (OTL). Two months of dPWV data were decomposed by the period of a component of OTL, and then composite time series data with a period equal to that of the component were made. This process was performed for K1, O1, M2, and S1 components of OTL. In each component, a periodic variation in dPWV appeared which was similar to those of the simulated GPS_PWV errors from OTL effects calculated with 'GOTIC' (Sato and Hanada, 1984), a program for the computation of OTL effect. Inclusion of OTL effects into GIPSY analysis reduced dPWV. In the M2 component, the amplitude of the dPWV was reduced by about 80%. This suggests that the OTL components calculated by the GOTIC succeeded in simulating the actual site displacement by OTL effects in Japan. On the other hand, in K1 components, the amplitude of dPWV without OTL in GIPSY is 1.5 times larger than the simulated GPS_PWV error, with considerable error remaining even in the case of GIPSY analysis with OTL. The error may be due to multi-path effect, temperature dependency on conversion from Zenith Wet Delay to PWV, or instrument dependency of WVR on temperature. Analysis utilizing much longer data periods than the present two months is required to overcome these difficulties.