Quantifying the impact of environmental loading on Zenith Tropospheric Delays in Europe

We assess the impact of the environmental loading on the Zenith Total Delay (ZTD) time series estimated within the second re-processing campaign of the EUREF Permanent GNSS Network (EPN). In particular, we used one solution provided by the ASI (Agenzia Spaziale Italiana Centro di Geodesia Spaziale, Italy), and two solutions provided by the GOP (Geodetic Observatory Pecny, Czech Republic) EPN analysis centers, along with the combined products. We find that ZTD time series derived within individual solutions are characterized by pure autoregressive noise, which is reduced during the combination in favor of white noise. This reduces the error of ZTD trends and is of great importance for numerous applications, as climate analyses, where trend is taken into account. Combination procedure does not however affect spatio-temporal patterns of ZTD residuals (linear trend and seasonal signals removed beforehand). We observe no impact of non-tidal oceanic loading on the ZTD residuals. Then, we compute ZTD differences from the two GOP solutions, which only differ by unmodelled non-tidal atmospheric loading. We prove that there is a similarity between the ZTD differences and non-tidal atmospheric loading which is strongly demonstrated in terms of unusual loading events, as significant inter-annual signals or large seasonal peaks. As these similarities account for 54% of ZTD differences, this indicates that unmodelled non-tidal atmospheric loading effect contributes to the ZTD residuals interpreted as a noise, affecting errors of trends. Therefore, we recommend that the non-tidal atmospheric loading is included at the observation level, once high-significance of ZTD parameters is required.