Deep learning architecture based on satellite remote sensing data for estimating ground-level NO2 across Beijing-Tianjin-Hebei Region

FAN Xuanshuo; WU Haibin; CHEN Xinbing; SONG Wei

doi:10.3969/j.issn.1673-6141.2023.02.009

Journals >Journal of Atmospheric and Environmental Optics >Volume 18 >Issue 2 >Page 181 > Article

Journal of Atmospheric and Environmental Optics
Vol. 18, Issue 2, 181 (2023)

Deep learning architecture based on satellite remote sensing data for estimating ground-level NO₂ across Beijing-Tianjin-Hebei Region

FAN Xuanshuo¹, WU Haibin^2,*, CHEN Xinbing², and SONG Wei²

Author Affiliations

¹Institute of Material Science and Information Technology, Anhui University, Hefei 230601, China

²School of Physics and Material Science, Anhui University, Hefei 230601, China

show less

DOI: 10.3969/j.issn.1673-6141.2023.02.009 Cite this Article

Xuanshuo FAN, Haibin WU, Xinbing CHEN, Wei SONG. Deep learning architecture based on satellite remote sensing data for estimating ground-level NO₂ across Beijing-Tianjin-Hebei Region[J]. Journal of Atmospheric and Environmental Optics, 2023, 18(2): 181 Copy Citation Text

show less

Fig. 1. Study area and distribution of ground-level monitor sites

Download full size | View in the Article

Fig. 2. DNN network

Download full size | View in the Article

Fig. 3. Model structure

Download full size | View in the Article

Fig. 4. Scatter plots of Sample-CV (a) and Site-CV (b)

Download full size | View in the Article

Fig. 5. Spatial distribution of NO₂. (a) September 5, 2019; (b) September 16, 2019; (c) October 18, 2019

Download full size | View in the Article

Data	Spatial resolution	Time resolution	Source
Monitoring sites NO₂/(μg·m^-3)	-	hour	CNEMC
NO₂ column densities/(mol·m^-2)	7.5 km $\times$ 3.5 km	day	S5P
Meteorological	0.25°	hour	ERA5
Elevation/m	90 m	-	SRTM3
Land use type	30 m	5 years	GlobelLan30
Population density /(people·km^-2)	30 arc-second	5 years	GPWv4

Table 1. Data information

View in the Article

Model	Sample-CV			Site-CV
Model	R²	RMSE	$M A E$	R²	$R M S E$	$M A E$
LR	0.68	9.42	7.37	0.55	11.52	9.05
GWR	0.73	8.97	6.46	0.71	9.71	7.20
SVR	0.75	8.39	6.33	0.72	9.35	6.69
GRNN	0.75	8.04	6.32	0.73	8.85	6.28
RF	0.80	7.74	5.39	0.71	9.70	6.84
DNN	0.80	7.72	5.31	0.74	8.95	6.01

Table 2. Comparison of the various models

Download Citation

Save the article for my favorites

Paper Information

微信扫一扫：分享

微信扫一扫：分享