Improving the weighted mean temperature model: A case study using nine year (2007-2015) radiosonde and COSMIC data in Hong Kong

The important role of a humidity conversion factor in mapping the zenith wet delay (ZWD) onto precipitable water vapour (PWV) makes it a vital parameter in Global Navigation Satellite System (GNSS) meteorology. The accuracy of the humidity conversion factor depends mainly on the quality of the atmospheric weighted mean temperature (T-m) calculation. Several effective models have been developed to calculate T-m with parameters such as the station's location, date of measurement and surface temperature. However, the drawbacks of these models, including low accuracy in some areas and complex model functions, cannot be ignored because they prevent further application in atmosphere analysis. Therefore, a new discrete integral formula for highly accurate calculation of T-m is proposed. Moreover, the linear relationships between time, surface temperature (T-s), height (H) and T-m are well studied based on radiosonde and Constellation Observation System for Meteorology Ionosphere and Climate (COSMIC) historical data (2007-2015). A new regional T-m model was investigated on the basis of parameters time, T-s and H in Hong Kong, China. The evaluation results indicate that the improvement of T-m derived from the new integral is > 4.6 K based on radiosonde data. In addition, the statistical results of T-m with different sampling based on radiosonde data and a COSMIC "wetPrf" profile show that the improvement of T-m derived from the new integral are > 2.10 and 2.75 K, respectively. Finally, wetPrf profiles collocated with Hong Kong and the 45,004 radiosonde station collected from January 1 to October 31, 2016, are used to evaluate the accuracy of the new T-m model. Compared with the Bevis model, the statistical results indicate that the T-m accuracy is improved by 34.75%. Therefore, the model for T-s and time is suggested to estimate the T-m in Hong Kong.