武汉市灰霾天气特征分析及基于支持向量机的能见度预报

利用武汉市2013年灰霾日气象数据和空气质量数据对灰霾天气特征及其影响因子进行了综合研究，获得了武汉市灰霾天气的主要影响因子，并使用支持向量机对灰霾日能见度进行了多因子综合预测。实验表明，支持向量机模型在短期预报中，±1 km、±2 km、±3 km误差范围内预报正确率分别达到733%、867%、967%，平均绝对误差在1 km内，实现了灰霾能见度高精度预报，优于多种预报模型。在第2、3天±3 km误差范围内的能见度预报准确率都达到90%，中长期预报能力较强，模型性能稳定

ANALYSIS OF HAZE CHARACTERISTIC AND VISIBILITY FORECAST BASED ON SUPPORT VECTOR MACHINE IN WUHAN CITY

With the daily meteorological data and air quality data of Wuhan city in 2013, the comprehensive research on the characteristics and influencing factors of haze weather is conducted, and the main meteorological influence factors for haze weather are gained. Support Vector Machine is proposed for haze days visibility forecasts based on multi factor. Research shows that, in the visibility forecast for short term, the accuracy of visibility prediction, based on support vector machine model, reached 733%, 867% and 967% within the error range of ± 1 km, ± 2 km and ±3 km, respectively. 〖JP2〗Owing to the MAE within 1km, it achieved high precision visibility forecast for haze days, and is superior to some other models. Besides, the accuracy of visibility forecasting during the first two and three days has reached 90% within the error range of ± 3 km, showing a strong capability and stable performance for long term forecasting