Interpolating atmospheric water vapor delay by incorporating terrain elevation information

Abstract

In radio signal-based observing systems, such as Global Positioning System (GPS) and Interferometric Synthetic Aperture Radar (InSAR), the water vapor in the atmosphere will cause delays during the signal transmission. Such delays vary significantly with terrain elevation. In the case when atmospheric delays are to be eliminated from the measured raw signals, spatial interpolators may be needed. By taking advantage of available terrain elevation information during spatial interpolation process, the accuracy of the atmospheric delay mapping can be considerably improved. This paper first reviews three elevation-dependent water vapor interpolation models, i.e., the Best Linear Unbiased Estimator in combination with the water vapor Height Scaling Model (BLUE + HSM), the Best Linear Unbiased Estimator coupled with the Elevation-dependent Covariance Model (BLUE + ECM), and the Simple Kriging with varying local means based on the Baby semi-empirical model (SKlm + Baby for short). A revision to the SKlm + Baby model is then presented, where the Onn water vapor delay model is adopted to substitute the inaccurate Baby semi-empirical model (SKlm + Onn for short). Experiments with the zenith wet delays obtained through the GPS observations from the Southern California Integrated GPS Network (SCIGN) demonstrate that the SKlm + Onn model outperforms the other three. The RMS of SKlm + Onn is only 0.55 cm, while those of BLUE + HSM, BLUE + ECM and SKlm + Baby amount to 1.11, 1.49 and 0.77 cm, respectively. The proposed SKlm + Onn model therefore represents an improvement of 29–63% over the other known models.