基于支持向量机模式识别的大雾预报方法

Fog Forecast Method Based on SVM and Model Identification

选取1971-2000年11412月大雾发生前近地面层的气象要素（气温、降水、能见度、风向风速、相对湿度、云量等9个预报因子），将支持向量机（SVM）方法应用于大雾预报。采用支持向量机方法，应用径向基函数，建立了陕西公路站点大雾24h预报模型，并进行了大雾预报的模拟、训练，其寻优标准TS评分达到了理想的效果。

By means of Support Vector Machine （SVM） and nine meteorological element data near the surface layer before heavy fog occurring in December and November from 1971 to 2000 （air temperature, precipitation, visibility, wind speed, wind direction, relative humidity, and overall and low cloud cover）, a 24-hour heavy fog forecast model for the highway in Shaanxi Province is developed. The model, based on the Gauss kernel function, has put to trial use and gets satisfactory TS scores through simulating and training in fog forecasting.