论文部分内容阅读
为了提前1个月预报出水稻稻曲病发生的气象条件适宜程度,根据中长期预报原理,采用因子膨化滑动相关普查、空间拓扑和最优相关技术,筛选出对综合稻曲病指数影响最显著的预报因子,分别构建基于气象要素、海温因子、大气环流指数的预报模型,并对3种模型的预报结果采用算术平均、加权平均和多元回归方法进行集成。结果表明,建立的3种模型均通过了显著性检验,预报效果较为理想,经过集成后提高了单个模型的拟合精度和独立样本试报的准确性,其中多元回归集成的效果更好。因此,建立的稻曲病预报模型可投入业务使用,预报结果将为稻曲病防治工作提供较为充足的时间。
In order to predict the suitable meteorological conditions of rice false smut disease one month in advance, according to the medium and long-term forecasting principle, the factors influencing the comprehensive smut index were screened by using the factor popularization sliding correlation survey, the spatial topology and the optimal correlation technique , The forecasting models based on meteorological elements, SST, and atmospheric circulation index are respectively constructed, and the forecasting results of the three models are integrated by arithmetic mean, weighted average and multiple regression methods. The results show that all the three models have passed the significance test, and the forecasting result is ideal. After being integrated, the fitting accuracy of individual models and the accuracy of independent samples are improved, and the multiple regression integration is better. Therefore, the established model of the false smut prediction can be put into operation, and the forecast result will provide sufficient time for the prevention and control of false smut.