Abstract: Due to the frequent occurrence of natural disasters, complex changes in water vapor characteristics, and sparse sounding stations in Japan, it restricts the acquisition of high-precision water vapor, and lacks a high-precision Tm model in this area. In view of this article adopts the 2009 to 2016 Global Geodetic Observing System (GGOS) weighted average temperature (T_m) with the ERA-Interim 2 m T_s grid data, a new T_m model (JQTm model) suitable for the Japanese region that takes into account the seasonal variation of the T_m residual and the daily cycle variation is established. In addition, using the data of 13 sounding stations and 110 GGOS Atmosphere T_m grids in Japan in 2017, the accuracy of the newly established JQTm model in this paper is evaluated in Japan. The study found that compared with the GGOS Atmosphere T_m grid data, the bias and root mean square error (RMSE) of the JQTm model are 0.15 K and 1.92 K, respectively. The RMSE is 41.16% (1.33 K) and 44.41% (1.53 K) higher than the GPT2w-1 and GPT2w-5 models, respectively. Compared with the sounding data, the bias and RMSE of the JQTm model are –0.66 K and 2.14 K, respectively. The RMSE is 28.43% (0.85 K) and 29.61% (0.90 K) higher than the Bevis model, GPT2w-1 model, and GPT2w-5 model, respectively. The JQTm model can provide high-precision T_m values for the Japanese region and provide an important basis for studying atmospheric water vapor and extreme weather in this region.