Hybrid precipitable water vapor fusion model in China

Abstract

High-precision precipitable water vapor (PWV) can be obtained using the global navigation satellite system (GNSS). However, the spatial resolution of the GNSS-derived PWV is insufficient, particularly for areas without sufficient GNSS stations. Although other techniques, such as radiosonde sounding, can also be used to obtain PWV, generating PWV images with high precision and high spatial resolution remains difficult. Therefore, this research focuses on the generation of such PWV images. A hybrid PWV fusion (HPF) model is proposed, which combines the polynomial fitting and spherical harmonic functions using the GNSS-/radiosonde-derived PWV at specific stations and European Centre for Medium-range Weather Forecasting (ECMWF)-derived PWV at grid-based points. The initial HPF model value is calculated based on the global pressure and temperature model 2 wet (GPT2w) model. Additionally, the large PWV difference and small PWV residual are estimated using the polynomial fitting and spherical harmonic functions, respectively. An optimized method of determining the weightings for the different PWV sources is also applied in resolving the established HPF model. PWV derived from 254 GNSS stations of the Crustal Movement Observation Network of China (CMONOC), 78 radiosonde stations of the Integrated Global Radiosonde Archive Version 2 (IGRA2) data set, and 961 grid points of the ECMWF ERA-Interim data sets over China are used to perform the experiment. Numerical results verify the accuracy and reliability of the proposed HPF model with an average root mean square (RMS) of approximately 1.75 mm for the entire 2014 in China, and under 3 mm in southeast China and in the summer in the selected four regions. This result indicates the good performance of the proposed HPF model and meets the requirement of weather nowcasting.