A weighted mean temperature model using principal component analysis for Greenland

The weighted mean temperature ( T m ) is an important parameter to convert the tropospheric zenith wet delay (ZWD) extracted from the global navigation satellite system (GNSS) signal into precipitable water vapor (PWV). The computation of T m requires vertical or ground meteorological parameters. However, most GNSS stations in Greenland lack in situ meteorological data, resulting in an accuracy degradation of the derived PWV. Using the most recent ERA5 reanalysis data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) from 1990 to 2018, we extract the principal components from a large amount of reanalysis data and then model T m using a data-driven principal component analysis (PCA) method. In comparison with classic periodic modeling approaches, our PCA model uses fewer parameters and considers temperature fluctuation with height. The proposed model is validated using observations from 11 radiosonde stations in Greenland from 2015 to 2019. The model’s bias and RMSE are − 0.110 and 4.447 K, respectively. The new model is also compared to the global pressure and temperature 3 (GPT3) and GTrop traditional grid models. The bias is reduced by 0.339 and 0.422 K, respectively, and the RMSE is reduced by 0.197 and 0.045 K, respectively.