Simulation of first flowering date for apple and risk assessment of late frost in main producing areas of northern China

An accurate model of apple was established to simulate the first flowering date in the main apple producing areas of northern China. Taking the Yining, Aksu, Wanrong, Baishui, Liquan, Shanzhou district, Fushan, and Xiongyue in Xinjiang, Loess Plateau, and Bohai Bay, as examples, a systematic investigation was made to evaluate the climate risk for the late frost of apple in flowering stage. The phenological and meteorological observation data was collected during the same period from the representative stations in each area. The parameters of thermal time model were revised to establish the first flowering date model of apple in different stations. Using internal and cross validation methods, the applicability of thermal time model was evaluated when simulating the first flowering date of apple. Then, the model was used to reconstruct the first flowering date of apple from 1962 to 2019, in order to explore the variation trend for the first flowering date of apple and the climate risk of late frost to apple flower in each fruit region. The frequencies and days were determined in various grades of late frost, as well as the climate risk, combined with the meteorological index of late frost and the last day of its occurrence. The results showed that: 1) The models for the first flowering date of apple in Xinjiang, Loess Plateau, and Bohai Bay were able to accurately simulate the changes of the first flowering date in the representative stations in each region. The data was independent, where the root mean standard errors (RMSE) of internal validation in each region were 2.5-4.8, 2.4-5.0, and 2.3-3.0 d, and the R2 values were 0.80-0.81, 0.66-0.89, and 0.74-0.90. In the cross validation, the values of RMSE were 5.0-6.6, 3.4-6.0, and 3.6-4.1 d, and the values of R2 were 0.41-0.73, 0.54-0.92, and 0.88, respectively. 2) From 1962-2019 (except that Fushan was from 1980 to 2019), the first flowering date of apple showed a significant trend of advance, where the trend coefficients was -1.84 - -1.04, -3.09 - -2.62, and -2.37 - -1.88 d/10 a, respectively. 3) The occurrence frequency and days of light frost were the highest, followed by the medium frost, while, those of heavy frost were lowest. There were relatively large climate risk indexes of apple late frost in the fruit region of Xinjiang and the Loess Plateau. Specifically, the maximum of 2.38 was in Yining, and the second value of 1.81 was in Wanrong, with its light and medium frost occurring once about every 2 and 4 years, respectively, while, the annual average numbers of days of late frost were 0.90 and 0.31 d, respectively. Followed by Baishui, the light and medium frost occurring once every 3 and 14 years, respectively, while the annual average numbers of days of late frost were 0.62 and 0.14 d, respectively. Nevertheless, the climate risk index of apple in the late frost was 0 at Fushan in Bohai Bay. The apples in the fruit area of Xinjiang and Loess Plateau were more affected by the late frost, whereas, that in the Bohai Bay was basically not affected by the late frost. The findings can provide a scientific guidance to effectively prevent the impact of late frost disaster on local apple production.