中文摘要：

为了方便准确地预测空气污染物浓度,基于气象因子聚类与多元回归的方法,以广东省顺德区为例开发了空气质量统计预报模型。预报模型能够较好地模拟出顺德区NO2、SO2、CO、PM10、PM2.5日均浓度和O3日最大8 h浓度水平和变化趋势,模型的模拟结果与实测值具有较高的相关性（相关系数R约为0.76）,标准化平均偏差为1.2%-13.4%,标准化平均误差为14.2%-30.3%,模型普遍略为高估各项污染物浓度水平。预报模型具有简单易行、节约人力物力、准确可靠等优点,适用于地级市及区县空气污染物的预报。

英文摘要：

To predict the concentration of air pollutants more accurately, an air quality statistical forecast model based on meteorological factors clustering and multiple regression was tested in Shunde District of Guangdong Province. The model could catch the concentration level and variation trend for NO2, 8O2, CO, PM10, PM2.5 and max O3-8h. The correlation between the simulation and observation was relatively high, with mean correlation coefficient about 0.76. The normalized mean bias （NMB） was 1.2%-13.4% and the normalized mean error （NME） was 14.2%-30.3%. Overall, the model overestimated the concentration slightly, With the advantage of saving labor and money, easy to operate and high accuracy, the forecast model is suitable for air pollutants concentration forecast in prefecture-level cities or counties and districts.