Atmospheric modeling in GPS data analysis for high accuracy positioning

Abstract

In this paper we review the atmospheric modeling methods used in GPS data analysis. Due to the strong spatial inhomogeneity and temporal variability of atmospheric constituents, especially water vapor, accurate modeling of path delay in GPS signals is necessary for high-accuracy positioning (e.g., tectonics and sea-level change) and meteorological applications (climatology and weather forecasting). State-of-the-art path delay modeling consists primarily in parameter estimation. In this strategy, zenith path delays are estimated during the GPS data reduction. External correction is another common strategy, in which the wet path delay is measured by a remote sensing instrument (usually a microwave radiometer). However, the latter is not as generalized, and is rather used for specific field campaigns or local long term observations. Both strategies have led to quite similar coordinate accuracies (using daily GPS observations), at the level of 1–2 mm in the horizontal component and 5–10 mm in the vertical component. The external correction strategy is capable of achieving even higher accuracy under specific conditions. Recent models, including gradients in the parameter estimation strategy have only led to marginal improvement. A major limitation of both strategies seems to be the use of mapping functions for the hydrostatic path delay correction. With the parameter estimation strategy, this limitation applies also to wet path delay correction. The use of numerical weather prediction and analysis models, and/or spaceborne sounding instruments, is suggested for replacing mapping functions and possibly for performing directly the hydrostatic correction. New instruments, such as Raman lidars, might also be used for a more accurate external wet path delay correction in the presence of strong atmospheric inhomogeneity. Further work is still needed for achieving measurements of absolute water vapor distribution in the atmosphere for this purpose.