Abstract: To provide scientific guidance for effectively preventing the impact of late frost disasters on local apple production, this study selected representative stations such as Wanrong, Baishui, Liquan, Shaanzhou, Fushan, and Xiongyue to establish and test a spatial-temporal model of apple flowering period. The model was used to reconstruct the apple flowering period in the main producing provinces of northern China from 1970 to 2019, and its spatial-temporal variation characteristics were analyzed. The climate risk of late frost during apple flowering period was analyzed based on meteorological indicators of late frost. The results showed:（1）The established apple initial flowering model could accurately simulate the changes in the initial flowering period of representative stations with independent data. The root mean square error of internal testing was 4.26 d, and the root mean square error of cross testing was 5.00 d.（2）The spatial-temporal model simulation of the initial flowering period of apples showed a significant trend of early flowering.（3）The frequency and number of days of occurrence of light frost during apple flowering were the highest, followed by medium frost, while the frequency and number of days of occurrence of heavy frost were relatively low.（4）From a spatial perspective, the frequency, number of days of occurrence, and climate risk index of apple late frost in the Bohai Bay area were relatively small, while the frequency, number of days of occurrence, and climate risk index of apple late frost in the northern part of the Loess Plateau were relatively large. Therefore, under the background of climate change, apples in the fruit areas of the Loess Plateau are greatly affected by late frost, and attention should be paid to key prevention measures.

Key words: apple, late frost freezing, initial flowering period, spatial-temporal model, climate risk