基于深度学习的华东地区PM2.5浓度遥感反演

PM_2.5作为大气污染的主要来源,对人类身体健康有着极大的影响.本文提出基于深度学习模型的多要素联合PM_2.5反演方法,以PM_2.5浓度作为真值数据,引入Himawari气溶胶光学厚度(AOD)日数据产品与温度、相对湿度和气压等10个要素作为反演要素.为验证方法的有效性,采用华东地区2016～2018年的数据分季节开展实验,并与传统反演方法进行对比.结果表明,PM_2.5浓度与AOD、降水、风速、高植被覆盖指数呈正相关关系,与矮植被覆盖指数呈负相关关系,与温度、湿度、气压以及DEM的相关性随季节的变化而改变;基于深层神经网络(DNN)反演的PM_2.5精度高于传统的线性和非线性模型,各个季节R²均在0.5以上并且误差较小,其中秋季的反演效果最好R²为0.86,夏季为0.75,冬季为0.613,春季为0.566;模型的可视化结果显示,DNN模型的反演结果更接近地面监测站点插值的PM_2.5浓度分布,分辨率更高且更精确.

PM2.5 Inversion Using Remote Sensing Data in Eastern China Based on Deep Learning

PM_2.5, which is a major source of air pollution, has a considerable impact on human health. In this study, a multi-element joint PM_2.5 inversion method based on a deep learning model is proposed. With PM_2.5 concentration as the ground truth, 10 elements including the Himawari-AOD daily data products, temperature, relative humidity, and pressure, were introduced as inversion elements. To verify the effectiveness of the method, the experiment was carried out by season using remote sensing data in Eastern China during 2016-2018. The results demonstrate that PM_2.5 concentrations were positively correlated with AOD, precipitation, wind speed, and high vegetation cover index and negatively correlated with dwarf vegetation cover index. The correlation with temperature, humidity, pressure, and DEM changed with seasons. Comparative experiments indicated that the accuracy of PM_2.5 inversion based on the deep neural network is higher than that of traditional linear and nonlinear models. R² was above 0.5, and the error was small in each season. The R² value for autumn, which showed the best inversion, was 0.86, that for summer was 0.75, that for winter was 0.613, and that for spring was 0.566. The visualization of the model illustrates that the inversion result of the DNN model is closer to the PM_2.5 concentration distribution interpolated by the ground monitoring station, and the resolution is higher and more accurate.