% IMPORTANT: The following is UTF-8 encoded.  This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.

@ARTICLE{Yuan:1045966,
      author       = {Yuan, Yue and Shen, Fuzhen and Sheng, Chunyan and Zhang,
                      Zeming and Guo, Weihua and Zhu, Wengang and Zhu, Hui},
      title        = {{E}valuating the impact of anthropogenic drivers and
                      meteorological factors on air pollutants by explainable
                      machine learning in {S}handong {P}rovince, {C}hina},
      journal      = {Atmospheric pollution research},
      volume       = {16},
      number       = {12},
      issn         = {1309-1042},
      address      = {Amsterdam},
      publisher    = {Elsevier B.V.},
      reportid     = {FZJ-2025-03633},
      pages        = {102694 -},
      year         = {2025},
      abstract     = {Unexpected haze in North China Plain during the COVID-19
                      lockdown has been regarded as a natural window to explore
                      the meteorological impact on formatting PM2.5 pollution but
                      with limitations in quantifying weather elements’
                      contributions. In this study, daily data of six air
                      pollutants (including PM2.5, PM10, SO2, NO2, O3, and CO) and
                      six meteorological factors (including temperature, pressure,
                      relative humidity (RH), wind speed (WS), wind direction
                      (WD), and precipitation) from 2015 to 2020 across 16 capital
                      cities in Shandong province, China, was used to drive the
                      Machine Learning and the SHapley Additive exPlanation (SHAP)
                      models. By applying these models, contributions from
                      anthropogenic drivers to pollutant reductions and
                      contributions from meteorological factors to the haze event
                      were investigated. Results show that the COVID-19 lockdown
                      measures reduced concentrations of NO2, PM2.5, PM10, CO and
                      SO2 by −52.1 $\%,$ −40.0 $\%,$ −45.5 $\%,$ −29.4
                      $\%$ and −38.7 $\%$ respectively. On average, an 18.9 $\%$
                      increase in O3 was observed. PM2.5 pollution was mainly
                      driven by temperature with a SHAP value of 19.7 μg/m3,
                      followed by RH (5.8 μg/m3), precipitation (0.9 μg/m3), WD
                      (0.3 μg/m3), pressure (0.1 μg/m3) and WS (0.1 μg/m3)
                      during the haze period. Relative to the post-haze period,
                      high-pressure systems coupled with lower temperatures and
                      weakened surface winds hindered the dispersion of PM2.5
                      whilst higher RH was in favour of PM2.5 production during
                      the haze period. This study underscores the intricate
                      interplay between emissions, meteorological conditions, and
                      regulatory measures in air pollution, offering critical
                      insights into future air quality management strategies by
                      air pollution prediction.},
      cin          = {ICE-4},
      cid          = {I:(DE-Juel1)ICE-4-20101013},
      pnm          = {2112 - Climate Feedbacks (POF4-211)},
      pid          = {G:(DE-HGF)POF4-2112},
      typ          = {PUB:(DE-HGF)16},
      doi          = {10.1016/j.apr.2025.102694},
      url          = {https://juser.fz-juelich.de/record/1045966},
}