Traditional GNSS space-based and ground-based tropospheric estimates are conducted separately, which limits the horizontal and vertical resolution of tropospheric products. This study evaluates the first step toward an integrated model: the development of an effective 3D ray-tracing algorithm for satellite-to-satellite (radio occultation) path reconstruction. Using a new approach for calculating gradients and interpolating refractivities, the authors develop a new algorithm for simulation of radio-occultation (RO) paths.
The accuracy of this algorithm is validated by simulating ten COSMIC-1 RO observations and comparing the results to the ROPP processing tool in terms of bending angle error. Significant differences in accuracy and calculation time are observed among settings.
Considering all collected results, our developed ray tracer proves to be a stable and fast method for simulating RO. The most crucial parameters for accuracy are the shooting threshold to the satellite and the vertical resolution of the model. Additionally, significant differences are found between ERA5 and COSMIC-1-based grid data sources. However, the highest-resolution refractivity models are suitable for achieving the highest simulation precision. Difficulties with signal convergence below 5 km altitude due to multipath effects are also identified.
In terms of calculation time, for the fastest settings, it takes slightly over 1 second per transmitter-receiver pair, resulting in 120-180 seconds per one RO event with 50 Hz and up to 2000 different excess phase values. Initial settings with the longest processing time take over 9 seconds.
For more information and detailed results, see the article:
Cegla, A., et al. GNSS signal ray-tracing algorithm for the simulation of satellite-to-satellite excess phase in the neutral atmosphere. J Geod 98, 42 (2024).
https://doi.org/10.1007/s00190-024-01847-0
Fig. 1. Flow chart of the ray tracing algorithm. “Eq. number’ refers to the equations described in the study. Green dots represent the positions of the ray points where interpolation and gradient calculations are performed, while red dots denote ray points located outside the voxel model
Fig. 2. Differences between simulated and observed bending angle. Dashed lines were processed with the wave optics option of ROPP, and solid lines with geometric optics options of ROPP. Red lines are ray tracing options with very high density of grid model and low threshold. Blue lines are values for the lower density of the grid model and relatively large threshold to satellite