State Key Laboratory of Remote Sensing Science, College of Global Change and Earth System Science, Beijing Normal University, Beijing, 100875, China.
Department of Atmospheric and Oceanic Science and ESSIC, University of Maryland, College Park, MD, 20740, USA.
Environ Pollut. 2023 Jun 15;327:121509. doi: 10.1016/j.envpol.2023.121509. Epub 2023 Mar 24.
Ground-level fine particulate matter (PM) and ozone (O) are air pollutants that can pose severe health risks. Surface PM and O concentrations can be monitored from satellites, but most retrieval methods retrieve PM or O separately and disregard the shared information between the two air pollutants, for example due to common emission sources. Using surface observations across China spanning 2014-2021, we found a strong relationship between PM and O with distinct spatiotemporal characteristics. Thus, in this study, we propose a new deep learning model called the Simultaneous Ozone and PM inversion deep neural Network (SOPiNet), which allows for daily real-time monitoring and full coverage of PM and O simultaneously at a spatial resolution of 5 km. SOPiNet employs the multi-head attention mechanism to better capture the temporal variations in PM and O based on previous days' conditions. Applying SOPiNet to MODIS data over China in 2022, using 2019-2021 to construct the network, we found that simultaneous retrievals of PM and O improved the performance compared with retrieving them independently: the temporal R increased from 0.66 to 0.72 for PM, and from 0.79 to 0.82 for O. The results suggest that near-real time satellite-based air quality monitoring can be improved by simultaneous retrieval of different but related pollutants. The codes of SOPiNet and its user guide are freely available online at https://github.com/RegiusQuant/ESIDLM.
地面细颗粒物 (PM) 和臭氧 (O) 是对健康危害极大的空气污染物。可以从卫星上监测到地面 PM 和 O 的浓度，但大多数检索方法分别检索 PM 或 O，而忽略了这两种空气污染物之间的共享信息，例如，由于它们有共同的排放源。本研究利用中国 2014-2021 年的地面观测资料，发现 PM 和 O 之间存在很强的关系，具有明显的时空特征。因此，我们提出了一种新的深度学习模型，称为臭氧和 PM 同时反演深度神经网络 (SOPiNet)，它可以每天实时监测，以 5 公里的空间分辨率同时对 PM 和 O 进行全覆盖。SOPiNet 采用多头注意力机制，根据前一天的条件，更好地捕捉 PM 和 O 的时间变化。将 SOPiNet 应用于 2022 年中国的 MODIS 数据，使用 2019-2021 年的数据构建网络，我们发现同时检索 PM 和 O 比分别检索它们的性能有所提高：PM 的时间 R 值从 0.66 增加到 0.72，O 的时间 R 值从 0.79 增加到 0.82。结果表明，通过同时检索不同但相关的污染物，可以改进基于卫星的近实时空气质量监测。SOPiNet 的代码及其用户指南可在 https://github.com/RegiusQuant/ESIDLM 上免费获取。
