On the consistency and variability of GNSS-estimated tropospheric gradients

The tropospheric wet delay is an important error source in precise GNSS positioning and is routinely modeled through the estimation of zenith wet delay (ZWD) and horizontal tropospheric delay gradients. While GNSS ZWD has been successfully used in climate studies and operational numerical weather prediction (NWP), the meteorological exploitation of tropospheric gradients remains limited, partly due to challenges in their interpretation, consistency, and sensitivity to processing strategies. The gradients reflect horizontal asymmetries in the neutral atmosphere and can, in principle, be inferred from ZWD differences between nearby GNSS stations, assuming a suitable vertical scaling of refractivity gradients. In this study, we investigate the consistency and variability of single-station GNSS-estimated tropospheric gradients using dense station pairs in southern Sweden from the SWEPOS GNSS network. We compare single-station gradients estimated directly from GNSS processing with inter-station horizontal gradients derived from ZWD differences. The two types of gradients are linked using water vapor scale heights derived from ERA5 atmospheric profiles, together with the assumption that the refractivity gradient scales with the amount of water vapor. Using one year of data, we assess the impact of different processing configurations and evaluate the temporal and spatial variability of GNSS tropospheric gradients. Our results show that, on a per-station basis, ZWD estimates are generally stable under commonly adopted processing options, whereas gradient estimates are, as expected, significantly more sensitive to processing settings, such as elevation cut-off angles and temporal constraints. Furthermore, a high degree of correlation between single-station gradients and inter-station horizontal gradients is found for station pairs with separations of less than about 25 km. We therefore propose that inter-station gradients can be used as a reference for tuning GNSS gradient estimation strategies, ensuring consistency in gradient magnitude. These findings highlight both the potential and the challenges of GNSS-estimated gradient products and provide guidance for their application in atmospheric monitoring.