关键词: 油菜, 花期预报模型, 有效积温法, 多元线性回归

Abstract: Based on the observational data of rape and meteorological data from 1995 to 2022 in Wuyuan, this study improved the rape flowering prediction model based on effective integrated temperature method. Starting with the date of rape budding and bolting, a multivariate linear regression approach was employed to enhance the prediction model. A meteorological factor model was established to predict the deviation in days between the simulated general flowering date based on effective integrated temperature method and the actual date, aiming to enhance the accuracy of rape flowering prediction model in Wuyuan. The improved and original models were compared and evaluated using simulation accuracy, root mean square error (RMSE), and relative error (RE). The results indicated that: (1)using 0°C as the threshold for effective integrated temperature and the average effective integrated temperature as the indicator, the preliminary prediction of the general flowering date of rape was performed, and the prediction accuracy improved as the flowering period approached. (2)Correlation analysis revealed that temperature was the primary meteorological factor influencing the rape general flowering period. A meteorological factor improvement model was established using mid-February average temperature, maximum temperature, and minimum temperature as independent variables and the deviation in days between the simulated general flowering date based on effective integrated temperature method and the actual date as the dependent variable. This model demonstrated statistical significance and passed the significance tests. (3)The evaluation of the prediction models before and after improvement showed that both methods yielded good prediction results. However, the meteorological factor improvement model produced superior simulation results, enhancing the accuracy of rape general flowering prediction model. The meteorological factor improvement prediction model, starting from the bolting, exhibited the highest accuracy in predicting the general flowering date of rape and can be effectively applied for rape flowering prediction.

Key words: Rape, Flowering prediction model, Effective integrated temperature method, Multiple linear regression