Comparisons of precipitable water vapor and zenith total delay derived from GNSS measurements, radiosondes, and ERA5 reanalysis data in the contiguous United States

Abstract

Precipitable water vapor (PWV) is an important parameter in studying hydrological cycles on a regional or global scale and monitoring local weather changes. Here, PWV and zenith total delay (ZTD) datasets were derived from Global Navigation Satellite System (GNSS) measurements at 430 stations in the contiguous United States from 2013 through 2020. These GNSS-derived ZTD and PWV datasets were assessed based on radiosonde (RS) data at 48 RS-GNSS collocated sites from 2013 through 2020 and ERA5 reanalysis products at 430 sites covering the same periods. Comparisons show that the GNSS-derived PWV dataset obtains overall accuracies in terms of root-mean-square (RMS) errors of 2.3 mm and 1.8 mm with respect to the RS measurements and ERA5 reanalysis products, respectively. Moreover, the GNSS-derived ZTD dataset achieves overall RMS errors of 1.4 cm and 1.0 cm in contrast to the RS data and ERA5 atmosphere products, respectively. In general, the quality of the GNSS-derived PWV dataset in the regions of interest mainly depends on the accuracy of ZTD estimations obtained from GNSS observations. Furthermore, mean values and seasonal variations of PWV time series derived from GNSS observations are in good agreement with those extracted from the ERA5 reanalysis products across the contiguous United States. The mean values of PWV time series in the southeastern regions vary from 20 mm to 45 mm and they are larger than those in the western regions, which is due to the topography distribution over the contiguous United States.