Assimilating GNSS Tropospheric Products and Quantitative Evaluation of Their Contributions to Numerical Weather Prediction

Apart from the applications of navigation, positioning, and timing, the Global Navigation Satellite System (GNSS) plays an important role in improving the quality and reliability of numerical weather prediction (NWP) models. However, the difference and contribution of assimilating GNSS-derived zenith total delay (ZTD) and precipitable water vapor (PWV) to forecast results are less investigated, which becomes the focus of this study. A unified method of assimilating GNSS-derived ZTD/PWV is first proposed, and their difference and contribution to the forecasting performance of the Weather Research and Forecasting (WRF) model are quantitatively evaluated by focusing on the multiple meteorological parameters, such as precipitation, relative humidity (RH), temperature, and pressure. In addition, the effects of magnitude and seasonal characteristics of GNSS-derived ZTD/PWV on the WRF model are further analyzed during a case of severe convective weather. Central and eastern China was selected as the study area, and 287 meteorological stations, 452 GNSS/Met stations, and 11 radiosonde stations were selected over the whole year of 2018. Results indicate that the assimilation of GNSS-derived ZTD/PWV, particularly ZTD, enhances the forecast accuracy of different meteorological parameters, and the positive contribution degree increases as the magnitude of GNSS-derived ZTD/PWV increases. Compared with the traditional method, the root mean square error reductions of precipitation, RH, temperature, and pressure generated by a unified method of assimilating GNSS-derived ZTD/PWV are 31.9%/22.7%, 54.6%/44.0%, 44.7%/35.7%, and 37.1%/24.2%, respectively. These results show the feasibility and effectiveness of the proposed data assimilation method and verify the positive contribution of GNSS-derived tropospheric products in improving the performance of the WRF model, especially for severe convective event nowcasting.

I. Introduction

Numerical weather prediction (NWP) models, by solving mathematical and physical equations based on specific initial conditions, not only quantitatively and objectively predict future weather changes [1], but also serve as an indispensable foundation for contemporary meteorological forecasting and research [2]. In NWP models, the quality of the initial water vapor field represents the initial state of atmospheric water vapor distribution, which directly affects the forecast results of various weather elements, especially in the case of severe convective weather [3]. Therefore, improving the performance of the initial water vapor field is of great importance for NWP models [4].