Surface monitoring, vertical atmospheric column observation, and simulation using chemical transportation models are three dominant approaches for perception of fine particles with diameters less than 2.5 micrometers(...Surface monitoring, vertical atmospheric column observation, and simulation using chemical transportation models are three dominant approaches for perception of fine particles with diameters less than 2.5 micrometers(PM2.5) concentration. Here we explored an imagebased methodology with a deep learning approach and machine learning approach to extend the ability on PM2.5 perception. Using 6976 images combined with daily weather conditions and hourly time data in Shanghai(2016), trained by hourly surface monitoring concentrations, an end-to-end model consisting of convolutional neural network and gradient boosting machine(GBM) was constructed. The mean absolute error, the root-mean-square error and the R-squared for PM2.5 concentration estimation using our proposed method is 3.56, 10.02, and 0.85 respectively. The transferability analysis showed that networks trained in Shanghai, fine-tuned with only 10% of images in other locations, achieved performances similar to ones from trained on data from target locations themselves. The sensitivity of different regions in the image to PM2.5 concentration was also quantified through the analysis of feature importance in GBM. All the required inputs in this study are commonly available, which greatly improved the accessibility of PM2.5 concentration for placed and period with no surface observation. And this study makes an exploratory attempt on pollution monitoring using graph theory and deep learning approach.展开更多
将气象要素加入到基于气溶胶光学厚度(aerosol optical depth,AOD)的近地面大气颗粒物浓度估算是目前热门的技术手段之一。获取了江苏省南京市2014年3月—2019年2月期间的AOD,精细模式分数(fine-mode fraction,FMF)和PM_(2.5)质量浓度数...将气象要素加入到基于气溶胶光学厚度(aerosol optical depth,AOD)的近地面大气颗粒物浓度估算是目前热门的技术手段之一。获取了江苏省南京市2014年3月—2019年2月期间的AOD,精细模式分数(fine-mode fraction,FMF)和PM_(2.5)质量浓度数据,并结合天气研究和预报(weather research and forecast,WRF)模式得到的气象模拟数据,对南京市PM_(2.5)的质量浓度进行反演。结果表明,相比于AOD与PM_(2.5)进行相关性分析,通过FMF校正得到的精细气溶胶光学厚度AODf与PM_(2.5)的相关性分析能够取得更高的拟合系数,R2最高达到0.40。利用随机森林模型,引入含不同高度的气象因子对PM_(2.5)质量浓度建立反演模型,得到的拟合系数与各误差指标均优于仅含近地面气象因子的模型,表明PM_(2.5)质量浓度受到多因子共同作用的影响,能较好地为利用多源数据反演PM_(2.5)质量浓度提供依据和参考。展开更多
基金supported by the National Natural Science Foundation of China (Nos. 41822505 , 42061130213 )the Royal Society-Newton Advanced Fellowship (No. NAF\R1\201166)+1 种基金Beijing Nova Program (No. Z181100006218077)the TsinghuaUniversity Initiative Scientific Research Program。
文摘Surface monitoring, vertical atmospheric column observation, and simulation using chemical transportation models are three dominant approaches for perception of fine particles with diameters less than 2.5 micrometers(PM2.5) concentration. Here we explored an imagebased methodology with a deep learning approach and machine learning approach to extend the ability on PM2.5 perception. Using 6976 images combined with daily weather conditions and hourly time data in Shanghai(2016), trained by hourly surface monitoring concentrations, an end-to-end model consisting of convolutional neural network and gradient boosting machine(GBM) was constructed. The mean absolute error, the root-mean-square error and the R-squared for PM2.5 concentration estimation using our proposed method is 3.56, 10.02, and 0.85 respectively. The transferability analysis showed that networks trained in Shanghai, fine-tuned with only 10% of images in other locations, achieved performances similar to ones from trained on data from target locations themselves. The sensitivity of different regions in the image to PM2.5 concentration was also quantified through the analysis of feature importance in GBM. All the required inputs in this study are commonly available, which greatly improved the accessibility of PM2.5 concentration for placed and period with no surface observation. And this study makes an exploratory attempt on pollution monitoring using graph theory and deep learning approach.
文摘将气象要素加入到基于气溶胶光学厚度(aerosol optical depth,AOD)的近地面大气颗粒物浓度估算是目前热门的技术手段之一。获取了江苏省南京市2014年3月—2019年2月期间的AOD,精细模式分数(fine-mode fraction,FMF)和PM_(2.5)质量浓度数据,并结合天气研究和预报(weather research and forecast,WRF)模式得到的气象模拟数据,对南京市PM_(2.5)的质量浓度进行反演。结果表明,相比于AOD与PM_(2.5)进行相关性分析,通过FMF校正得到的精细气溶胶光学厚度AODf与PM_(2.5)的相关性分析能够取得更高的拟合系数,R2最高达到0.40。利用随机森林模型,引入含不同高度的气象因子对PM_(2.5)质量浓度建立反演模型,得到的拟合系数与各误差指标均优于仅含近地面气象因子的模型,表明PM_(2.5)质量浓度受到多因子共同作用的影响,能较好地为利用多源数据反演PM_(2.5)质量浓度提供依据和参考。
