Abstract: Based on the hourly temperature data of national multi-source fusion products and national ground meteorological observation stations from 00:00 on January 1, 2018 to 23:00 on August 31, 2020 (Beijing time), the applicability of fusion temperature products in Yunnan is evaluated according to the indicators such as average error, root mean square error, correlation coefficient and time error. The results show that the average error of hourly temperature of fusion products was between −1 ℃~1 ℃, the root mean square error was less than 1.8 ℃, and the correlation coefficient was above 0.95. For the extreme temperature test, it was found that the extreme temperature errors were larger in northwest Yunnan. The maximum temperature had a large error from January to February. The error between the maximum temperature occurrence time of the fusion product and the actual maximum temperature occurrence time was 0 hours, accounting for 70% of the total samples. In the time inconsistent samples, the time error within 2 hours accounted for 86%, and the fusion products accounted for more in advance than the actual ones. The minimum temperature had the largest error from May to June, and the time error of 0 hours accounted for 73% of the total samples. In the time inconsistent samples, the time error within 2 hours accounted for 41%, more than 12 hours accounted for 40%, and the time delay of extreme value was relatively high. On the whole, the fusion temperature product has good applicability in Yunnan, and the inversion effect of the maximum temperature is better than that of the minimum temperature. Due to the influence of precipitation, terrain, altitude and other factors, the fusion temperature product has a certain systematic error in Yunnan.