Development of time-varying global gridded Ts-Tm model for precise GPS-PWV retrieval
Abstract. Water-vapor-weighted mean temperature, Tm, is the key variable to estimate mapping factor between GPS zenith wet delay (ZWD) and precipitable water vapor (PWV). In near real-time GPS-PWV retrieving, estimating Tm from surface air temperature Ts is a widely used method because of its high temporal resolution and a fair degree of accuracy. Based on the Tm estimates and the extracted Ts parameters at each reanalysis grid node, analyses of the relationship between Tm and Ts were performed without smoothing of data which will produce superior results than other similar studies. Analyses demonstrate that Ts–Tm relationship has significant spatial and temporal variations. Then static and time-varying global gridded Ts–Tm equations were established and evaluated by comparisons with radiosonde data at radiosonde 758 stations in the Integrated Global Radiosonde Archive (IGRA). Results show that our global gridded Ts–Tm equations have prominent advantages than other globally applied models. Large biases of Bevis equation or latitude-related linear model at considerable stations are removed in gridded Ts–Tm estimating models. Multiple statistical tests at 5 % significance levels show that time-varying global gridded model is superior to other Ts–Tm models at 83.64 % of all radiosonde stations, while no model is significantly better at 5.54 % of sites and others superior at only 10.82 % of sites. GPS-PWV retrievals using different Tm estimates were compared at a number of IGS stations. By application of time-varying global gridded Ts–Tm equations, the relative differences of GPS-PWVs at most sites are within 1 %. Such results are obviously superior to other Ts–Tm models. The differences between GPS-PWVs and radiosonde PWVs are influenced by other comprehensive factors instead of single Tm parameter. However evident improvements still exist at special site by using more precise Ts–Tm equations. PWV errors could decrease by more than 30 % during wetter seasons.