Estimating Integrated Water Vapor Trends From VLBI, GPS, and Numerical Weather Models: Sensitivity to Tropospheric Parameterization

Abstract

In this study, we estimate integrated water vapor (IWV) trends from very long baseline interferometry (VLBI) and global navigation satellite systems (GNSS) data analysis, as well as from numerical weather models (NWMs). We study the impact of modeling and parameterization of the tropospheric delay from VLBI on IWV trends. We address the impact of the meteorological data source utilized to model the hydrostatic delay and the thermal deformation of antennas, as well as the mapping functions employed to project zenith delays to arbitrary directions. To do so, we derive a new mapping function, called Potsdam mapping functions based on NWM data and a new empirical model, GFZ-PT. GFZ-PT differs from previous realizations as it describes diurnal and subdiurnal in addition to long-wavelength variations, it provides harmonic functions of ray tracing-derived gradients, and it features robustly estimated rates. We find that alternating the mapping functions in VLBI data analysis yields no statistically significant differences in the IWV rates, whereas alternating the meteorological data source distorts the trends significantly. Moreover, we explore methods to extract IWV given a NWM. The rigorously estimated IWV rates from the different VLBI setups, GNSS, and ERA-Interim are intercompared, and a good agreement is found. We find a quite good agreement comparing ERA-Interim to VLBI and GNSS, separately, at the level of 75%.