5:15 PM - 7:15 PM
[AAS09-P05] Methodology for estimating Precipitable Water Vapor using GNSS and evaluation of its accuracy
Keywords:GNSS Meteorology, Precipitable Water Vapor, GEONET
In space geodetic technologies such as GNSS (Global Navigation Satellite System), microwave propagation delays due to water vapor in the atmosphere can be a source of error in precise geodetic and crustal movement observations. While attempting to eliminate this error source, it is possible to estimate the amount of water vapor over the receiving station as GNSS-PWV (Precipitable Water Vapor) by analyzing GNSS data, contributing to improve the accuracy for forecasting the occurrence of heavy rainfall.
The Total Atmospheric Delay (TAD) of the ground receiver station is first determined from the observation equation by the least squares method in estimating GNSS-PWV. The line-of-sight (satellite direction) atmospheric delay can then be expressed as the product of the TAD and an elevation angle-dependent mapping function. If the receiver stations are located close to each other, the mapping functions will have very similar values and the atmospheric delay parameters for each receiving station would be determined with a poor precision. Therefore, the TAD is estimated by solving the observation equation using an International GNSS Service (IGS) observation point as the coordinate reference point, which is more than 1000 km away from the receiver stations. Another method is to solve the equation by fixing the coordinates of the receiver stations with GSI’s F5 solution.
In this study, we compared the RS-PWV obtained from 14 JMA radiosonde (RS) observations in Japan with the GNSS-PWV of nearby GEONET (GNSS Earth Observation NETwork system) electronic reference points at monthly intervals from January to August, 2024. We also investigated the accuracy of GNSS-PWVs in different choices of coordinate reference points, and with and without fixing the coordinates of electronic reference points and considered the best analysis method in Japan. GAMIT software and JMA AMeDAS data were used to calculate PWV from GNSS observation data.
We found that the root mean square (r.m.s) of the errors between GNSS-PWV and RS-PWV values were about 3.00 mm range for coordinate reference points forming a baseline length of more than 1000 km from the electronic reference point, regardless of their combination. When Guam, Honolulu, Hong Kong and Ulaanbaatar were selected as coordinate reference points and the coordinates of the electronic reference points were fixed, the r.m.s was less than 3.00 mm, and the accuracy was maintained in the analysis every 15 minutes. As these four sites are located relatively far from Japan, the difference in elevation angle with the satellite would have affected the estimation of atmospheric delay parameters, and fixing the exact coordinates produced similar results. We also found that the accuracy of high PWV at electronic reference points in the north and east of the country declined slightly during the summer months. This suggests that GNSS-PWV shows spatiotemporally averaged values, which makes it more difficult to estimate particularly high PWV. However, there remains biases that can’t be explained by the difference in elevation between the electronic reference point and the RS parabolic point were found at each electronic reference point, suggesting that there would not be a local correspondence in calculating PWV from the TAD.
The Total Atmospheric Delay (TAD) of the ground receiver station is first determined from the observation equation by the least squares method in estimating GNSS-PWV. The line-of-sight (satellite direction) atmospheric delay can then be expressed as the product of the TAD and an elevation angle-dependent mapping function. If the receiver stations are located close to each other, the mapping functions will have very similar values and the atmospheric delay parameters for each receiving station would be determined with a poor precision. Therefore, the TAD is estimated by solving the observation equation using an International GNSS Service (IGS) observation point as the coordinate reference point, which is more than 1000 km away from the receiver stations. Another method is to solve the equation by fixing the coordinates of the receiver stations with GSI’s F5 solution.
In this study, we compared the RS-PWV obtained from 14 JMA radiosonde (RS) observations in Japan with the GNSS-PWV of nearby GEONET (GNSS Earth Observation NETwork system) electronic reference points at monthly intervals from January to August, 2024. We also investigated the accuracy of GNSS-PWVs in different choices of coordinate reference points, and with and without fixing the coordinates of electronic reference points and considered the best analysis method in Japan. GAMIT software and JMA AMeDAS data were used to calculate PWV from GNSS observation data.
We found that the root mean square (r.m.s) of the errors between GNSS-PWV and RS-PWV values were about 3.00 mm range for coordinate reference points forming a baseline length of more than 1000 km from the electronic reference point, regardless of their combination. When Guam, Honolulu, Hong Kong and Ulaanbaatar were selected as coordinate reference points and the coordinates of the electronic reference points were fixed, the r.m.s was less than 3.00 mm, and the accuracy was maintained in the analysis every 15 minutes. As these four sites are located relatively far from Japan, the difference in elevation angle with the satellite would have affected the estimation of atmospheric delay parameters, and fixing the exact coordinates produced similar results. We also found that the accuracy of high PWV at electronic reference points in the north and east of the country declined slightly during the summer months. This suggests that GNSS-PWV shows spatiotemporally averaged values, which makes it more difficult to estimate particularly high PWV. However, there remains biases that can’t be explained by the difference in elevation between the electronic reference point and the RS parabolic point were found at each electronic reference point, suggesting that there would not be a local correspondence in calculating PWV from the TAD.