A Hyper-Dense GNSS Receiver Network for Monitoring Time and Spatial Variations of Precipitable Water Vapor (PWV)

We employ the GNSS meteorology technique to measure precipitable water vapor (PWV) from the propagation delay of GNSS signal in the atmosphere. We installed a hyper-dense GNSS network using 15 receivers with a horizontal spacing of 1-2 km in Uji, Japan (Uji network). We also obtained precipitation with a rain gauge at a nearby operational weather station and rain cloud distribution by an X-band radar. We selected 40 days from April 2011 to March 2013, when considerable precipitation was detected. Difference in PWV within 10 km was 3-10 mm during a heavy rain. We found PWV increased 10-20 minutes before a passage of a rain cloud. The maximum value of PWV correlated well with the amount of precipitation on the ground. The variance of PWV between the GNSS sites was enhanced during a heavy rain. For a future practical hyper-dense GNSS network system with many receivers, we consider to use inexpensive single frequency (SF) receivers. Because SF receiver cannot eliminate the ionospheric delay by itself, we interpolate the delay referring the delay measured by the nearby dual frequency (DF) receivers. We investigated ionospheric delay by the Uji network, taking advantages of Quasi-Zenith Satellite System (QZSS) that gives signals at high elevation angles. During a travelling ionospheric disturbance (TID), a wavy structure with a horizontal scale of several tens km was recognized. The ionospheric delay was compensated by a linear and quadratic interpolation, then the resulting error of PWV compared with DF solution was about 1.50 mm in RMS. For a real-time estimation of PWV, we used real-time satellite clock information corrected by GEONET. Difference of PWV between the real-time analysis and the post processing with the final orbit was 0.7 mm in RMS. We estimated an overall error of PWV with a dense SF-receiver network on a real-time basis was 1.7 mm in RMS.