Accurate pressure and temperature are indispensable in the GNSS water vapor retrievals. A refined regional empirical model, WHU_CPT, for estimating pressure, temperature and water vapor weighted mean temperature (Tm) with a horizontal resolution of 0.75° over China was developed. We found that the surface level reanalysis products are more reliable in reproducing surface temperature than pressure level products in the model construction. The model form was then determined individually for each meteorological variable by comparing performances of different models and by analyzing the Power Spectra Density of time series. Average RMS of pressure and temperature errors over China in WHU_CPT is about 4.07 hPa and 3.76 K, compared to 5.95 hPa and 5.92 K for GPT, 4.25 hPa and 5.14 K for GPT2, and 4.14 hPa and 5.14 K for GPT2w. Two Tm models were also developed in WHU_CPT. One is an empirical model, WHU_CPT (Tm1), which takes user location and time as inputs and the other one, WHU_CPT (Tm2), which additionally makes use of the measured temperature, is more suitable for users with surface temperature available. The mean RMS of Tm errors over China are about 4.45, 4.19, 3.81 and 2.97 K for the Bevis equation, GPT2w, WHU_CPT (Tm1) and WHU_CPT (Tm2).