• Journal of Atmospheric and Environmental Optics
  • Vol. 18, Issue 3, 258 (2023)
FU Miao*
Author Affiliations
  • School of Economics and Trade, Guangdong University of Foreign Studies, Guangzhou 510006, China
  • show less
    DOI: 10.3969/j.issn.1673-6141.2023.03.007 Cite this Article
    Miao FU. Improving the accuracy of NO2 concentrations derived from remote sensing using localized factors based on random forest algorithm[J]. Journal of Atmospheric and Environmental Optics, 2023, 18(3): 258 Copy Citation Text show less
    The comparison of the distributions of NO2 ground-level observed and satellite-derived concentrations. (a) Ground-level observed NO2 concentrations; (b) NASA satellite-derived NO2 concentrations
    Fig. 1. The comparison of the distributions of NO2 ground-level observed and satellite-derived concentrations. (a) Ground-level observed NO2 concentrations; (b) NASA satellite-derived NO2 concentrations
    The histograms of the cross validation concentrations from the three algorithms.(a) GWR_NO2; (b) MGWR_NO2; (c) RF_NO2
    Fig. 2. The histograms of the cross validation concentrations from the three algorithms.(a) GWR_NO2; (b) MGWR_NO2; (c) RF_NO2
    Cross validation results of the GWR, MGWR and random forest. (a) GWR; (b) MGWR;(c) random forest; (d) random forest 2
    Fig. 3. Cross validation results of the GWR, MGWR and random forest. (a) GWR; (b) MGWR;(c) random forest; (d) random forest 2
    The comparison of the predicted concentrations from the four approaches for Tibet. (a) NASA_NO2; (b) GWR_NO2;(c) MGWR_NO2; (d) RF_NO2
    Fig. 4. The comparison of the predicted concentrations from the four approaches for Tibet. (a) NASA_NO2; (b) GWR_NO2;(c) MGWR_NO2; (d) RF_NO2
    County-level NO2 concentrations predicted by the random forest algorithm. (a) Nationwide; (b) North China Plain;(c) Yangtze River Delta; (d) Guangdong Province; (e) Urumqi, Xinjiang
    Fig. 5. County-level NO2 concentrations predicted by the random forest algorithm. (a) Nationwide; (b) North China Plain;(c) Yangtze River Delta; (d) Guangdong Province; (e) Urumqi, Xinjiang
    VariableCountMeanStdMin25%50%75%Max
    Obs_NO2149431.85012.5045.43522.32931.42140.20576.919
    NASA_NO2149422.49014.8620.6009.60019.55034.40064.800
    GWR_NO2149431.08710.9887.86022.89130.00338.15180.722
    MGWR_NO2149431.85312.022-0.54422.82230.78940.03573.671
    RF_NO2149431.96210.4798.51524.03530.95538.97564.301
    Table 1. Statistical description of observed,NASA and the cross validation concentrations
    ModelCV rCV R2SlopeRMSEMAEMAPEReg R2
    NASA0.69370.48130.824614.380911.86550.4066NA
    GWR0.84260.71000.74056.78955.20210.19040.7880
    MGWR0.83700.70050.80477.01565.30530.20020.9400
    Random forest0.85820.73650.71926.42234.98500.1953NA
    Random forest 20.84130.70780.70656.79865.22830.1994NA
    Table 2. The comparison of cross validation results of the models
    Miao FU. Improving the accuracy of NO2 concentrations derived from remote sensing using localized factors based on random forest algorithm[J]. Journal of Atmospheric and Environmental Optics, 2023, 18(3): 258
    Download Citation