In the present study, a predictive technique incorporating driving forces was used to predict the atmospheric aerosol number concentration at the foot of Huangshan mountain which extractedthe driving force from the observation data by Slow Feature Analysis. To appraise its effectiveness, some prediction experiments were carried out using the hourly atmospheric aerosol number concentration in Huangshan. When the forecast step was 1, the correlation coefficient between the stationary model predictions and observation data was 0.6982; the correlation coefficent between the single external forcing model and observation data was 0.7390; the correlation coefficient between the double external forcing model and observation data was 0.7475. Adding external forcing can effectively improve the forecasting skills.%利用慢特征分析(Slow Feature Analysis, SFA)方法提取大气气溶胶时间序列的外强迫因子信息,并将此外强迫因子信息嵌入到预测模式中,建立一个包含提取外强迫因子信息的预测模式.利用该方法对2011年6月1日至2011年9月14日黄山山底的每小时大气气溶胶数浓度时间序列进行预测试验分析.结果表明,当提前预报一步时,平稳性模式的预测结果与实际观测数据的相关系数为0.6982,而单一外强迫模式的相关系数为0.7390,强迫模式的相关系数是0.7475,外强迫的加入可以有效的提高预测技巧.
展开▼
