PPP Without Troposphere Estimation: Impact Assessment of Regional Versus Global Numerical Weather Models and Delay Parametrization

Abstract

Mapping functions based on global Numerical Weather Models (NWM) have been developed in recent years to model the tropospheric delay in space geodetic techniques such as the Global Navigation Satellite Systems (GNSS). However, the estimation of residual tropospheric delay is still a necessity when high accuracy is required. Additionally, correlation between the estimated tropospheric delay, the receiver clock offset and the station height component, prolongs the time required for the solution to converge and impacts directly the accuracy of the results. In this study, we applied tropospheric corrections from high resolution NWM in GPS processing, in an attempt to acquire rapid and accurate positioning results, waiving the need to estimate residual tropospheric delay. Although high resolution NWM have outperformed standard atmosphere parameters and global models, it is the first time they have been compared against NWM-derived corrections, such as the operational Vienna Mapping Function 1 (VMF1) parameters. The processing strategy employed utilizes different scenarios characterized by their (a) NWM temporal and spatial resolution (b) grid or site-specific domain and (c) delay parametrization. The results were assessed in terms of height components bias, convergence frequency and time as well as residuals of the GPS analysis. Results showed an overall scenarios agreement of about 20 cm for the height component. However, the site-specific domain and high resolution NWM scenarios outperformed the grid-based ones in most of the cases; centimeter compared to decimeter daily height time series bias along faster convergence time constituted their performance. The final height offset with respect to their ITRF14 values was almost three times larger for the grid-based scenarios compared to the site-specific ones. The iono-free least squares adjustment residuals analysis revealed similar patterns for all the scenarios while the estimated heights experienced a reduction on the days of heavy precipitation under most of the scenarios; for some of the stations the advantage of using direct ray-tracing became obvious during those days.