Abstract: Based on daily minimum ground temperature data of 94 national meteorological stations in Inner Mongolia from 1971 to 2018,the last spring frost (ground temperature ≤0 ℃) dates of each station were calculated,and the spatiotemporal variations of the last frost were analyzed.It was found that the mean date of the last frost had a significant correlation with latitude and altitude,the lower the latitude and altitude,the earlier the last frost date.Interannual variation of the last frost date had a significant correlation with altitude,the higher the altitude,the larger the Interannual variation.Last frost dates of most stations had early trends that were related to altitude to some extent,the higher the altitude the more obvious the early trends.Last frost dates of most stations obvious abrupt changes of climate,abrupt changes of climate were related to latitude to some extent,the higher the latitude,the later the abrupt changes.On average,the last frost was non-damaging,light-damaging,medium-damaging and heavy damaging to crops at 49%,18%,16% and 17% of stations respectively.The number of non-damage level stations had a distinct increasing trend,the number of light-damage level stations had no obvious,the number of middle-damage level stations had a slight declining trend,and the number of heavy-damage level stations had adistinct declining trend.Empirical orthogonal function (EOF) analysis of last frost dates showed there were two main modes (of variation in last frost date)? The first EOF mode showed a whole anomaly structure type,with time coefficients that had a distinct declining trend and reflected climatic effects. The second EOF mode showed a north-south opposite structuretype with load values that were consistent with the latitudes and altitudes of the meteorological stations,and with time coefficients that had no obvious variation and reflected geographic effects.

Key words: Spring last frost;Minimum ground temperature;Linear trend;Abrupt changes of climate;Damage Levels;EOF analysis