Significance of discrepancies in the processing of GPS data with different algorithms

Summary GPS phase measurements can be processed by using a number of different algorithms. These include models based on (interferometric) single differences, double differences and triple differences, as well as an algorithm using the direct observable, i.e. the pure phase measurement. Theoretically, subject to including appropriate a priori covariance matrices, which express the geometrical configuration (correlation) between the different nominal observables, the results should be identical. Conversely, neglecting these covariances could lead to different numerical answers. Identical GPS data sets are processed with software packages based on different algorithms, and the resulting discrepancies established, in order to asses whether or not they are significant. Similarly, differential ionospheric and tropospheric corrections are evaluated and their significance determined for baselines of different lengths.