SVM方法在热带气旋风雨及温度预报中的应用  

Application of SVM method in weather forecasting during tropical cyclone process

作　　者：汤驰[1] 冯文[2] 官满员 施思[1] 吴文娟[1] 吴佳妮[1] 

机构地区：[1]海口市气象局,海口570100 [2]海南省气象局,海口570100

出　　处：《南京信息工程大学学报（自然科学版）》2015年第5期458-462,共5页Journal of Nanjing University of Information Science & Technology（Natural Science Edition）

基　　金：国家自然科学基金(40821092)

摘　　要：热带气旋的强弱和移动路径会直接影响到周围大气中气压、温度、露点等气象要素的变化.为更好地了解热带气旋对海口市的影响,通过收集影响海口市热带气旋关键因子,建立热带气旋风雨影响预报因子库,基于SVM方法对热带气旋在过程降水量、最大风速和平均温度进行趋势预报.结果表明,该方法对影响海口市热带气旋的过程降水量、最大风速和平均温度都有较好的预测效果,但对于超过15 m/s的最大风速和200mm以上降水量级上存在一定的偏差,这可能与SVM模式中预报因子库中关键因子不全及模式的择中原理使结果趋于平均化相关.The strength and movement path of tropical cyclones will directly affect the changes of the ambient atmosphere in aspects of air pressure,temperature,dew point and other weather elements. To better understand the impact of tropical cyclones on climate of Haikou city,this paper collects the key factors of tropical cyclones affecting Haikou city,thus establishes the impact predictor pool,and forecasts the trends in precipitation,maximum wind speed and the average temperature based on the Support Vector Machines（ SVM） method. The results show that this method generally works well in trend prediction of the three weather elements during tropical cyclone process affecting the Haikou city. But the method show some deviation in forecast accuracy for wind speed higher than 15 m / s and rainfall greater than 200 mm,which may be due to the lack of some key factors in predictor pool and the averaging process in principle of SVM model.

关 键 词：支持向量机 热带气旋 预报 

分 类 号：P49[天文地球—大气科学及气象学]