【摘要】 利用ECMWF预报场资料和地面观测资料开展支持向量机(SVM)、卡尔曼滤波(Kalman)两种统计方法对山岳型风景区气温集成预报方法的研究,并且结合自动气象观测站气温资料进行检验。结果表明:单一的SVM方法预报准确率较高且稳定,传统的Kalman方法预报效果不太理想,对长时效预报效果较差;加权集成后,3种集成方法预报稳定性较好、准确率比单一的SVM和Kalman预报方法均有所提高。从不同预报时效的检验结果看,各预报方法最高和最低气温预报准确率随着预报时效的增加均呈现降低的趋势。绝对误差分析表明,各预报方法最高气温的平均绝对误差均低于最低气温的平均绝对误差。更多还原

【Abstract】 Based on the ECM WF grid data and the observation data,the temperature forecast for mountainous scenic spot using SVM and Kalman methods was investigated. The results showthat the accuracy of the single SVM method is relatively high and stable,while that of the Kalman method is not sufficiently available especially for long-time forecast. The performance of integration solution of the two methods is better than that of the single one of them. M ore specifically,the forecasting accuracies of maximum and minimum air temperature decrease with the increasing prediction time for each method. The analysis also shows that the absolute error of forecasted maximum air temperature is smaller than that of minimum air temperature.更多还原