Positioning Error in GPS Measurements due to Atmospheric Excess Path Delay Estimated from Three-dimensional, Numerical Prediction Model Data

Abstract

We evaluate positioning errors due to atmospheric excess path delay. These are calculated using the 10-km spectral model data (the l0-km data) provided by the Japan Meteorological Agency (JMA), considering two GPS satellite constellations within about 2 hours. The first has high elevation satellites (case A) and the other has low elevation satellites (case B). We compare the errors with those calculated using a spherically symmetric atmosphere. The difference between both errors indicates the effect of azimuthal asymmetry of the atmosphere. In this calculation, wet, hydrostatic, and total delay cases are evaluated. The magnitude of positioning errors due to wet delay from the three-dimensional atmosphere generally shows baseline length dependency for both single and double differences. When the baseline ranges 500 km in the direction of themaximum horizontal gradient of water vapor, the horizontal positioning errors reach 40 cm in case A and 50 cm in case B, while the vertical one reaches 30 cm in case A and 15 cm in case B. The magnitude of the errors also increases as the difference in the altitude between the two sites increases. In addition, the directions of the horizontal error vectors heavily depend on the azimuth of the lowest elevation satellite. The horizontal errors due to the azimuthal asymmetry of atmospheric excess path delay are evaluated as follows; (1) errors due to wet delay range from 0.5 to 4 cm, (2) errors due to hydrostatic delay are up to 1 cm, and (3) errors due to total delay range from 0.5 to 5 cm. The maximum horizontal errors are detected on the baseline azimuths which are consistent with the maximum horizontal gradient of water vapor and/or temperature. For all cases the vertical errors are generally less than 1 cm. These results imply that realistic, three-dimensional atmospheric data is required to attain subcentimeter accuracy in space geodetic techniques considering the azimuthal asymmetry of the atmosphere. The use of numerical prediction data is the most realistic and efficient method for this purpose. In addition, we expect that this method will also contribute to improving the numerical prediction model in the meso scale using interaction with GPS meteorology.