In this study, two Weather Research and Forecasting model (WRF) experiments with gray-zone (GZ9) and convection-permitting (CP3) resolution are conducted for summer from 2009 to 2019. The surface air temperature (T2m) and precipitation are evaluated against in-situ observations and the Global Satellite Mapping of Precipitation (GSMaP) dataset. The results show that both experiments successfully capture the spatial pattern and daily variation of T2m and precipitation, though cold bias for temperature and dry bias for precipitation exist especially over the regions south of 35°N. Compared to GZ9, CP3 reduces the cold and dry bias over the southern TP. In addition, analysis of the diurnal variation of precipitation shows that both experiments simulate the advanced occurrence time of maximum precipitation over the eastern TP but postpone that over the central and western TP. Both experiments simulate a bimodal structure of the diurnal cycle of precipitation amount (PA). Further investigation reveals that GZ9 has more low-level clouds and prevents shortwave radiation from reaching the surface during daytime, leading to lower maximum surface air temperature (Tmax) in GZ9, while CP3 has more low-level clouds over the southeastern TP and preventing the outgoing longwave radiation and compensating the heat loss during nighttime, resulting in higher minimum surface air temperature (Tmin) in CP3. Besides, more water vapor over the southeastern TP results in more precipitation and thus reduced dry bias in CP3 over the southeastern TP.