Abstract
Abstract. As an innovative use of Global Navigation Satellite System (GNSS), the GNSS water vapor tomography technique shows great potential in monitoring three-dimensional water vapor variation. Most of the previous studies employ the pixel-based method, i.e., dividing the troposphere space into finite voxels and considering water vapor in each voxel as constant. However, this method cannot reflect the variations in voxels and breaks the continuity of the troposphere. Moreover, in the pixel-based method, each voxel needs a parameter to represent the water vapor density, which means that huge numbers of parameters are needed to represent the water vapor field when the interested area is large and/or the expected resolution is high. In order to overcome the abovementioned problems, in this study, we propose an improved pixel-based water vapor tomography model, which uses layered optimal polynomial functions obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) by adaptive training for water vapor retrieval. Tomography experiments were carried out using the GNSS data collected from the Hong Kong Satellite Positioning Reference Station Network (SatRef) from 25 March to 25 April 2014 under different scenarios. The tomographic results are compared to the ECMWF data and validated by the radiosonde. Results show that the new model outperforms the traditional one by reducing the root-mean-square error (RMSE), and this improvement is more pronounced, at 5.88 % in voxels without the penetration of GNSS rays. The improved model also has advantages in more convenient expression.
Highlights
As the most active component in the troposphere, water vapor is one of the most difficult parameters to monitor and describe (Rocken et al, 1997)
Based on the surface pressure obtained from observed meteorological parameters, the zenith hydrostatic delay (ZHD) could be obtained by the Saastamoinen model, and ZWD was isolated from ZHD
The root-mean-square error (RMSE) and bias of slant water vapor (SWV) obtained from tomography results of two models are almost the same under different weather conditions, which indicates that the reconstructed SWVs of the improved model have similar accuracy to that of the traditional one
Summary
As the most active component in the troposphere, water vapor is one of the most difficult parameters to monitor and describe (Rocken et al, 1997). Since most of the GNSS tomography methods divided the troposphere of interest into finite voxels, and the water vapor density in each voxel is considered as constant, these methods with the above assumptions are defined as the pixel-based method This kind of method cannot retrieve the variations in voxels and breaks the continuous nature of the troposphere as well. The improved model uses the water vapor density obtained from the traditional model as the input value and outputs the fitting water vapor density by the layered optimal polynomial functions This new model has the advantage of reflecting the variations in voxels and keeping the continuity of water vapor in the troposphere
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
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
