A comparison between the GNSS tomography technique and the WRF model in retrieving 3D wet refractivity field in Hong Kong
Abstract. Water vapor plays an important role in various scales of weather processes. However, there are limited means to monitor its 3-dimensional (3D) dynamical changes. The Numerical Weather Prediction (NWP) model and the Global Navigation Satellite System (GNSS) tomography technique are two of the limited means. Here, we conduct an interesting comparison between the GNSS tomography technique and the Weather Research and Forecasting (WRF) model (a representative of the NWP models) in retrieving Wet Refractivity (WR) in Hong Kong area during a rainy period and a rainless period. The GNSS tomography technique is used to retrieve WR from the GNSS slant wet delay. The WRF Data Assimilation (WRFDA) model is used to assimilate GNSS Zenith Tropospheric Delay (ZTD) to improve the background data. The WRF model is used to generate reanalysis data using the WRFDA output as the initial values. The radiosonde data are used to validate the WR derived from the GNSS tomography and the reanalysis data. The Root Mean Square (RMS) of the tomographic WR, the reanalysis WR that assimilate GNSS ZTD, and the reanalysis WR that without assimilating GNSS ZTD are 6.50 mm/km, 4.31 mm/km and 4.15 mm/km in the rainy period. The RMS becomes 7.02 mm/km, 7.26 mm/km and 6.35 mm/km in the rainless period. The lower accuracy in the rainless period is mainy due to the sharp variation of WR in the vertical direction. The results also show that assimilating GNSS ZTD into the WRFDA model only slightly improves the accuracy of the reanalysis WR and that the reanalysis WR is better than the tomographic WR in most cases. However, in a special experimental period when the water vapor is highly concentrated in the lower troposphere, the tomographic WR outperforms the reanalysis WR in the lower troposphere. When we assimilate the tomographic WR in the lower troposphere into the WRFDA model, the reanalysis WR is improved.