Integrating NWP products into the analysis of GPS observables

The processing of the Global Positioning System (GPS) observables is regularly performed using some "a priori" standard meteorological values. These can be very unrealistic if only few observations are available since the actual atmospheric conditions over the geodetic network are neglected. This seems to be the case of randomly moving GPS receivers far from continuously observing reference stations, as geodetic coordinates and tropospheric parameters cannot be accurately estimated. More realistic initial values for the troposphere may be obtained from a Numerical Weather Prediction (NWP) model. Thus, the impact of the ingestion of NWP forecasts into the GPS data processing is analyzed for short time series in terms of the geodetic vertical coordinate. The non-hydrostatic MM5 model with boundary and initial conditions given by the ECMWF analysis is used to compute estimates of the zenith wet and dry delays. These modeling simulations are then ingested into the GPS data analysis of the GIPSY software package. First, this approach is applied to GPS data gathered at the UK permanent stations of the COST-716 first benchmark campaign where it is possible to estimate a stable 24 h reference solution. Here, an improvement of about 60% in the vertical coordinate bias is found. Still, the absolute accuracy cannot expected to be better than 1 dm due to the remaining mismodeling of the troposphere. Then, the modeling approach is applied to a kinematic analysis of GPS buoy data. In this case also, a qualitative improvement is noticeable as demonstrated by a comparison with independent satellite radar altimeter data. (C) 2002 Published by Elsevier Science Ltd.