Journal of Atmospheric and Environmental Optics, Volume. 18, Issue 3, 245(2023)
Estimation of PM2.5 concentration and analysis of influencing factors in China
Figures & Tables(11)
Fig. 1. Random forest model inversion accuracy. (a) Train dataset; (b) test dataset
Fig. 2. Spatial distribution of original and estimated PM2.5 concentration on August 20, 2018. (a) Original value; (b) estimate value
Fig. 3. Seasonal model inversion accuracy on test dataset. (a) Spring model; (b) summer model;(c) autumn model; (d) winter model
Fig. 4. Area model inversion accuracy on test dataset. (a) Eastern model; (b) central model; (c) western model
Fig. 6. Three-dimensional spatial effect diagram of influence factors on changes of PM2.5 daily concentration. (a) AOD and BLH;(b) LAT and BLH; (c) BLH and TMP; (d) TMP and RH
Table 1. Information table of influencing factors
View tableView in ArticleTable 1. Information table of influencing factors
Data type Data name Data code Units Data source Natural factors 10m u-component of wind U m·s-1 ERA5 Dataset
(https://climate.copernicus.eu)
10m v-component of wind V m·s-1 2m temperature TMP K Total precipitation PRE m Evaporation EVA m Relative humidity RH % Boundary layer height BLH m Elevation DEM m NASA SRTM Digital Elevation Dataset
(https://cmr.earthdata.nasa.gov)
Aspect ASPECT (°) Vegetation index NDVI ― MOD13A2 v6 Dataset
(https://lpdaac.usgs.gov)
Human factors Population POP ― Landscan Dataset
(https://landscan.ornl.gov/)
Nighttime lights NTL DN NPP VIIRS Dataset
(https://eogdata.mines.edu/products/vnl/)
Land use LUC ― MCD12Q1.006 Dataset
(https://lpdaac.usgs.gov)
Time and space factors Longitude LON (°) China Environmental Monitoring Station
(http://www.cnemc.cn/)
Latitude LAT (°) Day of year YOD day
Table 2. Seasonal model training parameter table
View tableView in ArticleTable 2. Seasonal model training parameter table
Model Main parameters n_estimators max_depth max_feature Spring model 119 36 12 Summer model 119 37 12 Autumn model 123 36 13 Winter model 124 33 13
Table 3. Area model training parameter table
View tableView in ArticleTable 3. Area model training parameter table
Model Main parameters n_estimators max_depth max_feature Eastern model 119 36 12 Central model 119 40 12 Western model 119 30 12
Table 4. Accuracy table of model inversion
View tableView in ArticleTable 4. Accuracy table of model inversion
Kind Model Train dataset Test dataset R2 ERMS/(µg·m-3) R2 ERMS/(µg·m-3) Overall China model 0.99 3.97 0.91 10.13 Area Eastern model 0.99 3.48 0.91 9.41 Central model 0.99 3.54 0.93 9.07 Western model 0.98 4.85 0.86 12.71 Season Spring model 0.98 4.56 0.85 11.42 Summer model 0.97 2.57 0.78 7.15 Autumn model 0.99 3.16 0.91 8.29 Winter model 0.99 5.15 0.91 13.67
Table 5. Model precision comparison table
View tableView in ArticleTable 5. Model precision comparison table
Model Train dataset Test dataset R2 ERMS/(µg·m-3) R2 ERMS/(µg·m-3) Multiple linear regression 0.24 29.73 0.23 29.72 Extreme gradient boosting 0.78 15.75 0.75 16.47 Random forest 0.99 3.99 0.91 10.13
Tools
Get Citation
Copy Citation Text
Yuan CAO, Mingyan GONG, Fei SHEN, Jinji MA, Guang YANG, Xiwen LIN. Estimation of PM2.5 concentration and analysis of influencing factors in China[J]. Journal of Atmospheric and Environmental Optics, 2023, 18(3): 245
Paper Information
Category:
Received: Nov. 15, 2021
Accepted: --
Published Online: Jun. 29, 2023
The Author Email: MA Jinji (jinjima@ahnu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20