Least travel time ray tracer version 2 (LTT v2) adapted to the grid geometry of the OpenIFS atmospheric model

Abstract

Electromagnetic signals commonly used in geodetic applications, such as the Global Navigation Satellite System (GNSS), undergo bending and delay in the neutral gas atmosphere of the Earth. The least travel time (LTT) concept is one of the approaches to model signal slant delays via a ray tracing (RT) procedure. In this study, we developed an LTT-based RT algorithm (LTT v2), where the three-dimensional refractivity field of the atmosphere is based on the atmospheric model data. This representation is complete in a sense that the domain of the RT conforms to the native grid geometry of the atmospheric model. In principle, the LTT-based RT algorithm is seen as an extension of an atmospheric model for signal delay evaluation. The atmospheric states are generated using a global numerical weather prediction model, the Open Integrated Forecast System of the European Centre for Medium-Range Weather Forecasts. In the LTT v2 model, some physical and numerical approximations are improved compared to the original implementation, called “LTT v1”. We compare the slant delays products of the two models. Additionally, a comparable modelling setup is created with the state-of-the-art VieVS Ray Tracer (RADIATE). The skill of slant delay estimation is assessed using metrics that are indicative of the quality of GNSS products derived using the GROOPS (Gravity Recovery Object Oriented Programming System) orbit solver software toolkit of the Graz University of Technology. The metrics used are GNSS orbit midnight discontinuities (MDs) and residuals of ground station precise positioning with respect to the IGS14 reference. Employment of slant delay products of the LTT RT algorithm in GNSS processing shows similar performance with v1 and v2. The GNSS orbit MDs are reduced by around 3 % when using the LTT v2 model, while root-mean-square residuals of ground station precise positioning are 5 % lower with LTT v1. The consistency of both metrics is improved slightly using LTT v2, as seen by the metrics' standard deviation values. Intercomparison with RADIATE indicates significantly better performance of LTT v2, which we attribute entirely to the much larger amount and lossless utilization of weather model data as input to LTT v2 versus RADIATE.