A comparative analysis of uncertainty modelling in GPS data analysis

Abstract

A thorough assessment and mathematical treatment of all relevant errors in GPS data processing and analysis are essential for the further use and interpretation of the processing results. In this study two mathematical approaches for the error handling are studied in a comparative way. A probabilistic approach is based on the construction of a fully populated variance-covariance matrix of zero difference phase observations by introducing the uncertainty measures of the respective influence parameters in terms of standard deviations. A deterministic approach interprets these uncertainty measures as error bands and uses formalisms from interval mathematics. Both approaches are applied to a simulated EUREF sub-network. The deterministic approach yields more realistic results, in particular with respect to the dependence of the uncertainty measures on the baseline length. © Springer-Verlag Berlin Heidelberg 2007.