【作者】 樊高峰； 张勇； 柳苗； 毛燕军；

【Author】 FAN Gao-feng1,2 ZHANG Yong2 LIU Miao1 MAO Yan-jun1 (1. Zhejiang Climate Center ,Hangzhou 310017; 2. National Climate Center, Beijing 100081)

【机构】 浙江省气候中心； 国家气候中心；

【摘要】 支持向量机(SVM)是基于统计学习理论的一种智能学习方法,可以用来解决样本空间的高度非线性的模式识别等问题。干旱是气候因子的高度复杂关系的一种异常反映。利用8月南方涛动指数,副高强度指数,极涡强度指数等15项因子,基于径向基核函数建立浙江秋季干旱预测模型,并应用交叉验证方式确定最优模型参数,进行秋季干旱预测,这与传统基于线性统计理论的预测方法有明显不同,得到较好效果。更多还原

【Abstract】 The support vector machine is an intelligent learning method based on the statistical learning theory, that may solve the problem of non-linear pattern recognition in sample space. Drought is reflected by climate factors’ high complexity and non-linearity. An autumn drought prediction SVM model is developed according to the 15 prediction factors, including the southern oscillation index, the subtropical high strength index, the polar vortex strength index, etc. cross-validation method is applied to find the best parameters of the SVM model. This kind of method differ much from general linear statistical theory, and is obviously superior in drought prediction.更多还原