LAUSR

Tropical cyclone (TCs) prediction has improved in the last decade, but studying their structure and dynamics remains a challenging task due to limited in situ observations. The global navigation satellite system (GNSS) technique enables retrieving the state of the troposphere with high accuracy in all weather conditions. In this article, we use GNSS slant total delays (STDs) to verify the quality of the weather models during the passage of the TC Meranti, which was the strongest TC in 2016. STDs observed from 28 GNSS stations in Taiwan are compared with the STDs reconstructed from three meteorological data sets: the weather research and forecasting (WRF) model, the global forecast system (GFS), and the ERA-Interim (ERA) using a 2-D ray-tracing technique. Furthermore, two strategies were tested—including and excluding the delays due to liquid and ice water content. The results reveal a good consistency between the GNSS and the ray-traced STDs. The best agreement was found for the WRF, with a mean difference of −0.5 mm and a standard deviation of 29.4 mm when the hydrometeor delay was included. The mean hydrometeor contribution reached up to 2.8 mm for the WRF, while for the ERA and GFS, the contribution to the total delay was negligible. The recorded absolute percent differences were greater for the northern HENC station (mostly within 0.8%) than for the southern GOLI station (on average below 0.5%) due to a closer location to the passing Meranti. The most significant biases were seen in the ERA (around 7%) and the GFS (exceeding −3%) for the PLIM station, located in the central mountain range valley.