Assessment of errors in Precipitable Water data derived from Global Navigation Satellite System observations. (July 2015)

• Record Type: Journal Article
• Authors: Hordyniec, Pawel; Bosy, Jaroslaw; Rohm, Witold
• Is Part Of: Journal of atmospheric and solar-terrestrial physics. Volume 129(2015:Jul.)
• Page Start: 69
• Page End: 77
• Publication Date: 2015-07
• Subjects: Atmospheric models -- GNSS meteorology -- Meteorological parameters -- Precipitable Water; Geophysics -- Periodicals; Atmospheric physics -- Periodicals; Géophysique -- Périodiques; Météorologie physique -- Périodiques; Electronic journals; 551.51
• DOI: 10.1016/j.jastp.2015.04.012
• Languages: English
• ISSNs: 1364-6826

Abstract

Abstract: Among the new remote sensing techniques, one of the most promising is a GNSS meteorology, which provides continuous remote monitoring of the troposphere water vapor in all weather conditions with high temporal and spatial resolution. The Continuously Operating Reference Station (CORS) network and available meteorological instrumentation and models were scrutinized (we based our analysis on ASG-EUPOS network in Poland) as a troposphere water vapor retrieval system. This paper shows rigorous mathematical derivation of Precipitable Water errors based on uncertainties propagation method using all available data source quality measures (meteorological sensors and models precisions, ZTD estimation error, interpolation discrepancies, and ZWD to PW conversion inaccuracies). We analyze both random and systematic errors introduced by indirect measurements and interpolation procedures, hence estimate the PW system integrity capabilities. The results for PW show that the systematic errors can be under half-millimeter level as long as pressure and temperature are measured at the observation site. In other case, i.e. no direct observations, numerical weather model fields (we used in this study Coupled Ocean Atmospheric Mesoscale Prediction System) serves as the most accurate source of data. Investigated empirical pressure and temperature models, such as GPT2, GPT, UNB3m and Berg introduced into WV retrieval system, combined bias and random errors exceeding PW standard level of accuracy (3 mm according to E-GVAP report). We also found that the pressure interpolation procedure is introducing over 0.5 hPa bias and 1 hPa standard deviation into the system (important in Zenith Total Delay reduction) and hence has negative impact on the WV estimation quality. Abstract : Highlights: We model Zenith Hydrostatic Delay and Precipitable Water with their uncertainties. We applied rigorous variance and bias propagation method. Direct observations provide unbiased data with 0.26 mm random error in PW. The major contribution to errors stems from interpolation and height corrections. NWP data are applicable if no collocated meteorological sensors are present.