Statistical significance of trend estimations for integrated water vapor time series obtained from GPS technique: a case study in Europe

Water vapor is an important medium for the transmission moisture and latent heat in the atmosphere. It is one of the most abundant and dominant greenhouse gases in the atmosphere, which is crucial for global warming. With higher temperatures, the specific humidity will also increase as predicted by the nonlinear Clausius-Clapeyron relationship, indicating a positive feedback loop. Hence, estimation of the trend of Integrated Water Vapor (IWV) in the atmosphere is of great importance for global warming research. However, previous studies have shown that the trends of IWV are usually rather small. Therefore, it is important to estimate the IWV trend and its associated uncertainty with a reasonable mathematical model for the homogenized time series from homogenously reprocessed GPS data sets. Since the 1990s, the Global Positioning System (GPS) has successfully been employed to retrieve IWV with a high temporal resolution, all-weather condition and with global coverage. In this work, we used the hourly GPS Zenith Total Delay (ZTD) time series for 1995.0-2017.0 at 21 European GPS stations derived from a homogeneous data reprocessing. For the conversion of ZTD to IWV, we employed the meteorological variables from ERA5, a state-of-the-art atmosphere reanalysis product newly released by the European Centre for Medium-Range Weather Forecasts (ECMWF). Then, we investigated the influence of noise model assumptions within the mathematical model on the uncertainties of IWV trend estimates. As expected, the results confirmed that the assumption of a white noise only model tends to underestimate the trend uncertainty. A first-order autoregressive process is the preferred mathematical model for a more realistic estimation of the IWV trend uncertainty.