Determination of GPS Tropospheric Delays by Utilizing Ground-based GPS Network Measurements and its Applications for Weather Forecasts and Precise Point Positioning

摘要

The delay occurs while GPS signals pass through theionosphere and troposphere and it decreases the accuracyof GPS positioning. Therefore, it must be eliminated forGPS precise point positioning (PPP). Through studies ontropospheric delay measurements of the GPS signal, it isnow possible to apply the technology of GPS networkmeasurements to weather forecasts. Recently, the KoreaMeteorological Administration (KMA) has introducedGPS precipitable water vapor (PWV) to numericalweather predictions. KMA have been using upper-airobservation systems to observe atmospheric water vapor,such as radiosonde and radiometer. However, theseobservation networks weren't dense enough to analyze therapid change in under-synoptic scale atmosphere. Thenetwork of GPS is dense enough in Korea, and it can bean alternative of upper air observation method.To calculate GPS PWV, we need air pressure andtemperature measurements at GPS station. About 100GPS permanent stations are being operated in Korea.However, only 10 GPS permanent stations have their ownweather sensors connected to the GPS receiver. Toovercome this limitation, interpolation of pressures andtemperatures from nearby automatic weather station(AWS) is needed for the GPS station without ameteorological sensor. In this study, we comparedKriging and reverse sea level correction (RSLC) to verifywhich the suitable AWS interpolation method is.Eventually, RSLC showed more accurate result. The rootmean square error (RMSE) of RSLC was 0.2 hPa and1.2°C , on the other hand , the RMSEs of Kriging were 1.3hPa and 1.8 °C. We calculated GPS PWV by usingGIPSY 5.0 and evaluated the PWV accuracy bycomparing with PWVs from radiosonde and radiometer.The RMSE of GPS PWV was about 3 mm as comparedwith radiosonde or radiometer. We calculated GPS PWVsduring severe weather periods and analyzed thecharacteristics of atmospheric water vapor changes.We suggested a new tropospheric delay correctionmodel based on zenith total delay (ZTD) from GIPSYprocessing result. The model provides tropospheric delaycorrection to GPS users into grid map structure. To makethe tropospheric delay grid map, we calculated ZTDs ofeach GPS site in Korean Peninsula. Then, we interpolatedthe ZTDs by using inverse distance weighting (IDW)algorithm to calculate the values of grid points. Thetropospheric delay correction model consists of zenithhydrostatic delay (ZHD), zenith wet delay (ZWD) andZTD. As the result of tropospheric delay model accuracyanalysis, the RMSE between grid map data and GPS sitedata was 0.7 mm in ZHD, 7.6 mm in ZWD and 8.5 mm inZTD. Finally, we applied the model to single frequencyrelative positioning algorithm and analyzed thepositioning accuracy enhancement. As the result,positioning accuracy was improved up to 36% in case ofrelative positioning of Suwon (SUWN) and Mokpo(MKPO), that the baseline distance was about 297km.