最小二乘支持向量机在云量预报中的应用

Application of least squares support vector machine in prediction of cloud cover

基于2003—2006年逐年1、8月WRF区域数值预报产品和单站观测资料,采用最小二乘支持向量机回归方法,结合选取合适的参数和核函数,分别按月通过不同长度样本序列建立了台北和厦门站总云量和低云量短期释用预报模型,利用2007年1、8月样本资料对模型进行了预报和检验,并与神经网络方法进行了对比。结果表明:最小二乘支持向量机回归方法的预报效果要好于神经网络方法;两站不同长度样本的总云量和低云量预报模型,预报效果较好,其预报准确率不会因为训练样本的减少而降低。可见,最小二乘支持向量机回归在云量等气象要素释用预报方面,具有较好的应用前景。

Based on WRF model products and station-observed data of January and August from 2003 to 2006,the short-term forecast models of every month by the Least Squares Support Vector Machines（LSSVM）method for total cloud cover and low cloud cover at the Taibei and Xiamen station are established by selecting the appropriate parameter and kernel function,according to different sample length.The models are tested using the data of 2007,and compared by BP-ANN.Results show that the forecasting effect of LSSVM is better than BP-ANN,the forecasting effect of models corresponding to different sample amount is acceptable,and decrease of training data number doesn＇t reduce the forecast accuracy of LSSVM method.The result also shows that LSSVM is an effective method for total cloud cover and low cloud cover prediction,and it is a hopeful method for other meteorological element forecasting.