Rapid troposphere tomography using adaptive simultaneous iterative reconstruction technique

A near real-time troposphere tomography system (WATS-NRT) based on Global Positioning System (GPS) slant wet delays (SWD) was developed to reconstruct three-dimensional wet refractivity fields over a wide area. The NCEP Global Forecast System (GFS) short-range forecast products (GFS-FC) were taken as the tomography background to tackle the ill-condition issue, and the Adaptive Simultaneous Iterative Reconstruction Technique (ASIRT) was used for parameter estimation to avoid an inversion of a large sparse matrix. Experiments were set up for a 7-day period covering a severe storm event to evaluate the performance of the tomography system over an area with about 5000 tomographic nodes. Statistics of the computation time illustrate that WATS-NRT is efficient enough for providing 30 min interval tomography products in the studied area. Wet refractivity time series from the tomography product experience dramatic rises before the advent of heavy rain, which is not presented in the GFS-FC products. Comparisons also show general better agreements with the nearby radiosonde profiles in tomography products than GFS-FC, especially at the lower troposphere. In addition, average error RMS of SWD derived from tomography products is about 38 mm at an elevation of 15° and 9 mm in the zenith direction, compared to more than 70 and 16 mm for SWD from both GFS-FC and the reanalysis product ERA5, and compared to about 44 and 9 mm for ZWDINT method. These results suggest that WATS-NRT can be promising in some applications, such as monitoring extreme weather events, providing tropospheric delay estimates in geodetic data analysis and developing new mapping functions.