基于MODIS的中国PM_(2.5)污染时空分布特征研究

Spatial and Temporal Distribution of PM_(2.5) Pollution in China Based on MODIS

近年来,中国各大城市雾霾天数有明显增长的趋势,雾霾天气的产生与细颗粒物PM_(2.5)有直接的联系。基于2013-2015年MODIS气溶胶光学厚度三级产品MOD08,考虑边界层高度、相对湿度、温度、风速、风向等气象因子,分季节构建BP(误差反向传播)神经网络模型,估算全国PM_(2.5)值,基于此值分析中国PM_(2.5)污染的时空分布。结果表明:(1)对气溶胶光学厚度(AOD)缺失值进行插值后夏季预测模型效果最好,R为0.840 1,春秋季预测模型效果较差,R分别为0.602 5、0.589 9。(2)PM_(2.5)空间分布差异性显著,秋季空间分布差异最显著,夏季降水丰富,空间分布差异显著性降低;高值主要出现在华北和西北小面积区域,低值出现在西部蔵区和海南。(3)高值的华北地区和低值的西部地区都有面积逐渐增加的趋势,高值逐渐扩散到西北小面积区域,低值扩散到中部地区。

In recent years, the haze days show a significant growth trend in those metropolises in China. The formation of haze days has a directly connection with the PM（2.5）. Based on the monthly mean AOD data by the moderate resolution imaging spectrometer（MODIS） from 2013 to 2015, considering the meteorological factors such as boundary layer height, relative hu-midity, temperature, wind speed, and wind direction, the BP（error-back propagation） neural network model in different season was built, and the value of PM（2.5） was estimated. Finally the temporal and spatial distribution of PM（2.5） pollution in China was analyzed. The results showed that after the aerosol optical thickness（AOD） missing values are interpolated, the summer fore-cast model has the best effect（R=0.840 1）, while the spring and autumn forecast model is not so good（R=0.602 5） and（R=0.589 9）. The spatial distribution of PM（2.5） was significant, especially the autumn, the spatial distribution in summer is lower due to the summer precipitation is rich, with high value mainly appeared in north China and northwest area, and the low value appeared in the west China and Hainan. The square of high value in north China and western of low value has a gradually in-creasing trend, the high value regin spreads to northwest area, and the low value regin spreads to the central area.