Abstract: In August, 2022, the persistent extreme heat weather process in Sichuan Province was wide in scope, strong in intensity and long in duration. There were 74.8% sites in the province that broke the extreme value in the same period of nearly 70 years. Based on the hourly 2m temperature data of Sichuan in August 2022, the historical data of the same period, the 2m temperature forecast data of EC, CMA-GFS and CMA-MESO models, the characteristics and forecast error of this extreme heat weather process were analyzed. The results show that: (1) Affected by the terrain, the maximum temperature was high in the east and low in the west. (2) Affected by the urban heat island effect, the stations with the maximum temperature exceeding the historical extreme value mainly were concentrated in the west of the basin near the Longquan Mountains. (3) The average maximum temperature of the three representative stations was high, the anomaly was large, and the duration of high temperature was long. The accumulated precipitation in this period was obviously lower than the historical average value of the same period, and the number of precipitation days was less. (4) The accuracy of the three models for the minimum temperature forecast was significantly higher than that of the maximum temperature forecast, while the accuracy of the high temperature forecast of the optimal model was only 39.1%. (5) The forecasting advantages of EC model mainly lay in the low-altitude stations of the basin, while CMA-MESO model was better than EC model in some places of steep terrain. (6) For the maximum temperature forecast date, EC was the best for the basin and Panxi region, while CMA-GFS and CMA-MESO were the best for the West Sichuan Plateau.