Abstract: Abstract: Extreme climate events pose significant risks to global food production. With changes in climate and underlying surface conditions, the spatial-temporal characteristics of extreme temperatures and their impact on regional crop yields are often overlooked. Based on monthly temperature data from national meteorological stations from 1951 to 2020, trend lines and wavelet period methods were used to explore the characteristics of extreme temperature changes under LUCC conversion. And applying a stochastic frontier model, the impact on regional crop output is quantitatively analyzed. The results indicate that during the study period, the regional extreme temperature index showed a fluctuating upward trend in time series, with obvious periodic characteristics, mainly around 30 years, and gradually showing long-term characteristics of 50 years or more, while the spatial distribution showed heterogeneity; The maximum value of the regional extreme temperature index all appeared after 2000, and the regional temperature increased significantly, and the extreme weather phenomenon became increasingly prominent with the time series, which had a general impact on the regional agricultural output, especially the maximum temperature minimum (TXn) index increased by 1%, and the corresponding output decreased by 0.57% (p=0.05). Our research findings will provide information on adaptation strategies to mitigate future extreme climate impacts.