Abstract: In recent decades, extreme climate occurred frequently on the Loess Plateau. It is thus particularly important to study and predict the occurrence of extreme climate. Available researches on extreme climate mainly focus on the changing characteristics of the event itself, but ignore the correlation between average temperature and its changing trend. We used linear trend estimation, Mann-Kendall test, sliding t test and Pearson correlation analy-sis to study the variation trend of extreme temperature and its correlation with average temperature on the Loess Plateau based on the daily maximum temperature, minimum temperature and average temperature data of 79 meteo-rological stations from 1986 to 2019. The results showed that the extreme warmth index in the Loess Plateau region showed a significant upward trend, the extreme cold index showed a significant downward trend, and the frequency of extreme high temperature events increased. Most of the extreme temperature indices had abrupt changes in the middle and late 1990s and in 2012, and the extreme temperature showed a downward trend from 1998 to 2012, which better responded to the phenomenon of global warming hiatus. The increasing trend of mean temperature in the gully region, the rocky mountain region and the valley plain region was more obvious than that in other regions. The stations with large trend of extreme temperature index almost all occurred in the region with large increases of mean temperature. The increases of average temperature in small increments increased the frequency of extreme high temperature event, with the change range of extreme low temperature and its rate being greater than the extreme high temperature. Climate warming on extreme temperature index had a different effect, small changes in the average temperature in the Loess Plateau made the climate transfer towards the direction of more frequent heat.

Key words: extreme climate, extreme temperature index, heat wave, frost damage, Loess Plateau