Estimating the Residual Tropospheric Delay for Airborne Differential Gps Positioning (a Summary)

When post-processing dual frequency carrier phase data, the residual tropospheric delay can easily be the largest remaining error source. This error can contribute a bias in height of several centimetres even if simultaneously recorded meteorological data are used. This is primarily due to the poor representation of the water vapour profile in the tropospheric delay models. In addition, a lack of real-time meteorological data would force the scaling of either surface values or standard atmosphere values; these are also unlikely to accurately represent the ambient atmosphere. To obtain the highest precision in kinematic GPS some advantage may be obtained by estimating this error source along with the position solution. The simple tests reported in this paper removed biases of several centimetres in height when estimating the residual tropospheric delay from GPS data recorded at an aircraft in flight. However, important limitations exist in the geometry of the satellite coverage which must be considered before the full reliability of the technique can be quantified. INTRODUCTION This paper provides a brief summary of our investigations into estimating the residual tropospheric propagation delay from GPS signals. This parameter is the remaining part of the tropospheric delay not predicted by empirical models. In post-processed dual frequency carrier phase data, it can easily be the largest remaining error source. Unlike most applications of the technique, we have used data recorded at an aircraft in flight. This idea was motivated by the fact that highly accurate aircraft positions are required for gravimetric, altimetric and photogrammetric surveying purposes. Increasingly, GPS is being used to provide the decimetre-level accuracy required for some of these techniques. This level of precision can be achieved using carrier phase observables, but we will show that unmodelled tropospheric effects could potentially contribute a bias of a similar magnitude. When processing GPS observations, a value for the tropospheric delay is predicted using empirical models which must be provided with meteorological values of the ambient