Assessment of the performance of GPS-PWV and rainfall event prediction by using precise products from different analysis centers

Precipitable water vapor (PWV) is a fundamental parameter in measuring atmospheric water vapor. By using the precise point positioning (PPP) technique, the PWV can be retrieved by Global Positioning System (GPS) satellites. Considering that the accuracy of PPP is highly dependent on the quality of the precise products provided by different analysis centers, this study thoroughly investigates the performance of GPS-ZTD (zenith total delay) and GPS-PWV derived from precise products of different analysis centers. The ZTD estimated by the GPS PPP is compared to IGS-ZTD (International Global Navigation Satellite Systems (GNSS) Service), and the results show that the correlation coefficient between them can reach approximately 0.96. In addition, to eliminate the influence of weighted mean temperature on GPS-PWV, three traditional methods of estimation of weighted mean temperature are evaluated. Results indicate that the bias of Braun model is smallest, which is only 1.3 K. Thus, the Braun model is adopted to estimate the GPS-PWV. The GPS-PWV based on the Braun model and IGS products is compared with RS-PWV (radiosonde) and ERA5-PWV (European Centre for Medium-Range Weather Forecasts Re-Analysis 5). The correlation coefficient between GPS-PWV and RS-PWV is approximately 0.97, and between GPS-PWV and ERA5-PWV is about 0.99. These results demonstrate that the GPS-PWV is reliable and stable. Then, the performance of GPS-PWV and rainfall event prediction based on GPS-PWV for different analysis centers are compared and analyzed. In terms of the GPS-PWV, the difference between these analysis centers is very small, the largest average bias is only 0.18 mm. However, for the rainfall event prediction, the maximum bias of successful prediction rate can reach approximately 11.43%. Thus, it can be concluded that the influence of precise products obtained from different analysis centers should be considered for rainfall event prediction based on GPS-PWV.