COMPARISON OF GNSS PWV AND ERA5-DERIVED PWV BASED ON GNSS PWV IN HONG KONG, CHINA

Water vapor is the most abundant atmospheric gas, and it plays a vital role in the evolution of the Earth's climate. Precipitable water vapor (PWV) is a key factor in monitoring the climate and hydrological cycle. The use of GNSS to estimate PWV is a very effective method. This paper uses 17 satellite positioning reference stations in the Satellite Positioning Reference Station Network in Hong Kong, China, in 2017 to calculate the PWV and introduce the latest reanalysis data set of European Centre for Medium-Range Weather Forecasting (ECMWF) ERA5 into this study. The accuracy of ERA5-derived PWV was evaluated using the GNSS-derived PWV. In Hong Kong, the annual bias and RMSE values of GNSS-derived ZTD and ERA5-derived ZTD are 1.16cm and 1.78cm, while the annual RMSE values of GNSS-derived PWV and ERA5-derived PWV are 0.51cm and 0.57cm. The daily changes of GNSS PWV in 2017 are analyzed, and the results show that the ZTD effect of ERA5 reanalysis data derived in the small range area is not very ideal, but the accuracy of the PWV derived from ERA5 is better. * Corresponding author


INTRODUCTION
The troposphere is the lowest layer in the Earth's atmosphere, about 7 to 20 kilometers from the Earth's surface, where almost all climate change occurs. Water vapor in the troposphere plays a key role in global climate change and atmospheric processes (Holloway and Neelin, 2010;Torres et al., 2010;Wang et al., 2017). As an important component of the atmosphere, and as one of the most varied components in the troposphere, water vapor is highly involved in global water cycle and energy exchange (Zhang et al., 2019). In order to monitor and evaluate changes in the weather system, it is extremely important to accurately understand the distribution and changes of atmospheric water vapor. Ground to atmosphere top level cross sectional area The total amount of water vapor in the air column condenses into rain. The rainfall is usually called Precipitable water vapor (PWV). The ground-based GNSS water vapor estimation PWV technology provides a new means for tropospheric water vapor detection. Compared to traditional atmospheric water vapor detection, GNSS can be used in all weather conditions with low cost and high spatial and temporal resolution (Nilsson and Gradnarsky, 2006;Jin et al., 2007).
The concept of GNSS (Global Navigation Satellite System) Meteorology was first developed by Beavis and others. (1992) it is proposed that the use of ground-based GNSS-derived PWV has been widely used in meteorology. Especially for weather forecasting, there have been many scholars involved in the study (Fudeyasu et al., 2008;Benevides et al., 2015;Yao et al., 2017;Zhao et al., 2018 a). Monitoring extreme weather with GNSS PWV (Heffernan, 2013;Wang et al., 2018). Improvements to weather forecasting models (Gendt et al., 2004;Yang et al., 2013). In some small regions, these places have established a regional GNSS continuous operation reference station network, using these established GNSS continuous operation reference station network can be calculated based on the local PWV, which provides more ideas for studying the accuracy of GNSS PWV products in small areas.
In July 2017, the ECMWF released the fifth-generation reanalysis data set of ERA5 with a horizontal resolution of 31 km (Olauson, 2018). For the ERA5, the atmosphere is divided into 137 model levels from the Earth surface up to a height of 80 km and also interpolated to 37 pressure levels. Its temporal resolution reaches 1 hr (Albergel et al., 2018). Its data is available at https://cds.climate.copernicus.eu. The accuracy of a small range of ERA5-derived PWV can be evaluated using a small range of GNSS-derived PWV.

Data sources
Since the key parameter affecting GNSS-derived PWV is Zenith stratospheric delay (ZTD), the discussion of ZTD is added to the following study. This paper selects data from  In addition, get the data for the same time of ERA5 hourly estimates of the variables on the pressure levels' with a horizontal resolution of 0.25°×0.25°，covering pressure P(hPa), geopotential Z(m 2 •s 2 ) , temperature T(K) , Specific humidity q(kg•kg -1 ), relative humidity R(%), etc.

Processing Policies
Eighteen satellite positioning reference stations are processed using GAMIT 10.7 software and using the double-difference mode (Herring et al., 2010). The specific strategy is as follows: GNSS observation sampling rate of 30s, height cut-off angle of 10°, convective correction model using VMF1 model, 2h estimated once ZTD (B hm et al., 2006).
where k 1 =77.604K/hPa The partial vapor pressure e calculation formula is as (10).
The final formula for ZTD is: PWV can be derived by ERA5 reanalysis. First, the calculation formula of ZWD is derived, and the PWV derived from ERA5 is obtained by the formula (2).
where N ω = wet refractive index the N ω formula is as follows (Davis et al., 1985): where Z ω -1 = inverse compressibility level, which differs from 1 within 1% and which was set to 1 in this study. Take 3  The spring PWV has a distinctly high peak, indicating that the spring water vapor activity is intense, in line with the characteristics of the plum rain season in Hong Kong. The PWV is lower in winter, but the PWV changes significantly, and it is guessed that more precipitation time occurred in winter.
It can be observed from the figure that the GNSS-derived PWV has the highest coincidence with the ERA5-derived PWV in spring, and the PWV is different in other seasons, but the PWV trends are basically the same. It can also be seen from the figure that the difference between the GNSS-derived PWV and the ERA5-derived PWV is small, but since the estimated ZTD value is relatively inaccurate, the PWV accuracy of the two may be worse than ideal.