State Key Joint Laboratory of ESPC, School of the Environment, Tsinghua University, Beijing 100084, China; State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China.
State Key Joint Laboratory of ESPC, School of the Environment, Tsinghua University, Beijing 100084, China; State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China.
J Environ Sci (China). 2020 Dec;98:85-93. doi: 10.1016/j.jes.2020.04.042. Epub 2020 Jun 14.
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 (PM) concentration. Here we explored an image-based methodology with a deep learning approach and machine learning approach to extend the ability on PM 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 PM 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 PM 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 PM 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.
基于图像的深度学习和机器学习方法扩展细颗粒物（PM）感知能力研究
表面监测、垂直大气柱观测和化学输送模型模拟是感知直径小于 2.5 微米（PM）浓度的细颗粒物的三种主要方法。在这里，我们探索了一种基于图像的方法，使用深度学习和机器学习方法来扩展对 PM 的感知能力。我们使用了 6976 张图像，结合了上海（2016 年）的每日天气条件和每小时时间数据，通过每小时的地面监测浓度进行训练，构建了一个由卷积神经网络和梯度提升机（GBM）组成的端到端模型。使用我们提出的方法对 PM 浓度进行估计的平均绝对误差、均方根误差和 R-squared 分别为 3.56、10.02 和 0.85。迁移能力分析表明，在上海训练的网络，只需在其他位置微调 10%的图像，就能达到与在目标位置训练的数据相似的性能。通过在 GBM 中分析特征重要性，还定量分析了图像中不同区域对 PM 浓度的敏感性。本研究中所需的所有输入都是常见的，这极大地提高了对没有地面观测的地点和时间的 PM 浓度的可及性。本研究还尝试使用图论和深度学习方法进行污染监测。
