A new zenith hydrostatic delay model for real-time retrievals of GNSS-PWV

Abstract. The quality of the zenith hydrostatic delay (ZHD) could significantly affect accuracy wet (ZWD) Global Navigation Satellite System (GNSS) signal, and from ZWD precipitable water vapor (PWV) can be obtained. ZHD is usually obtained a standard model – function surface pressure at GNSS station. When PWV retrieved stations that are not equipped with dedicated meteorological sensors for measurements, blind models, e.g., global temperature (GPT) commonly used to determine pressures these stations. Due limited accuracies GPT model-derived value also low accuracy, especially in mid- high-latitude regions. To address this issue, new model, named GZHD, was investigated real-time retrieval GNSS-PWV study. ratio total (ZTD) first calculated using sounding data 505 globally distributed radiosonde selected had over 5000 samples. It found temporal variation dominated by annual semiannual components, amplitude dependent upon geographical location Based on relationship between ZTD, developed back propagation artificial neural network (BP-ANN) method which took ZTD as an input variable. 20-year (2000–2019) 558 9-year (2006–2014) COSMIC-1 (Constellation Observing Meteorology, Ionosphere, Climate) data, were distributed, training samples model. GZHD evaluated two sets references: integrated ERA5 reanalysis data. performance compared GPT3, latest version. Results showed outperformed radiosonde-derived reference, measured root mean square error (RMSE) samples, GZHD-derived about 21 % better than GTP3-derived ones. ERA5-derived 30 GPT3-derived ZHD. In addition, derived 41 resulting evaluated, result indicated improved %.