Abstract
Abstract. Troposphere tomography, using multi-constellation observations from global navigation satellite systems (GNSSs), has become a novel approach for the three-dimensional (3-D) reconstruction of water vapour fields. An analysis of the integration of four GNSSs (BeiDou, GPS, GLONASS, and Galileo) observations is presented to investigate the impact of station density and single- and multi-constellation GNSS observations on troposphere tomography. Additionally, the optimal horizontal resolution of the research area is determined in Hong Kong considering both the number of voxels divided, and the coverage rate of discretized voxels penetrated by satellite signals. The results show that densification of the GNSS network plays a more important role than using multi-constellation GNSS observations in improving the retrieval of 3-D atmospheric water vapour profiles. The root mean square of slant wet delay (SWD) residuals derived from the single-GNSS observations decreased by 16 % when the data from the other four stations are added. Furthermore, additional experiments have been carried out to analyse the contributions of different combined GNSS data to the reconstructed results, and the comparisons show some interesting results: (1) the number of iterations used in determining the weighting matrices of different equations in tomography modelling can be decreased when considering multi-constellation GNSS observations and (2) the reconstructed quality of 3-D atmospheric water vapour using multi-constellation GNSS data can be improved by about 11 % when compared to the SWD estimated with precise point positioning, but this was not as high as expected.
Highlights
For some years, global navigation satellite systems (GNSSs)-based tropospheric tomography has been regarded as one of the most promising techniques to reconstruct the temporal–spatial variation of atmospheric water vapour (Flores et al, 2000; Crespi et al, 2008)
For GNSS troposphere tomography, the horizontal resolution of voxels is first determined according to the number of voxels and the coverage rate of GNSS stations located in the bottom layer
A comparative experiment using single- and multi-constellation GNSS data derived from different numbers of stations revealed that increasing the station density improved the quality of tomographic results, with the root mean square (RMS) accuracy of slant wet delay (SWD) residuals increasing by about 16 % when compared to the result of using multi-constellation GNSS troposphere tomography
Summary
GNSS-based tropospheric tomography has been regarded as one of the most promising techniques to reconstruct the temporal–spatial variation of atmospheric water vapour (Flores et al, 2000; Crespi et al, 2008). By discretising the area of interest into finite voxels, the water vapour information in divided voxels can be reconstructed under the assumption that the unknown estimated parameters are constant during a given period (Radon, 1917; Flores et al, 2000) This technique has been proven by some feasibility studies with GPS-only observations (Troller et al, 2002; Bender and Raabe, 2007; Chen and Liu, 2014) as well as the simulated multi-constellation GNSS observations (Crespi et al, 2008; Bender et al, 2011b; Wang et al, 2014; Benevides et al, 2015c, 2017).
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call