Abstract:The global pressure and temperature model (GPT) is an empirical model providing pressure and temperature at any site in the vicinity of the earth’s surface. Aiming at mitigating the shortages of GPT, the GPT2w model is presented, which provides more meteorological data, with a higher horizontal resolution of 5° (GPT2w_5 model) and 1° (GPT2w_1 model). In order to test the accuracy of pressure and temperature of GPT, GPT2w_5 and GPT2w_1 model, the performance of the three models are compared and analyzed at selected sites, based on 1 year (2015.3 to 2016.2) high-precision sounding data from 29 sounding sites in China. The results show that all three models have obvious seasonality and sensitivity to latitude change; the GPT2w_1 model performs as well as the GPT2w_5 model and significantly better than the GPT model; but compared with the GPT2w_5 model, the GPT2w_1 model is more robust.