利用近邻因子提高二氧化氮遥感反演浓度的精度—基于随机森林算法

大气与环境光学学报 ›› 2023, Vol. 18 ›› Issue (3): 258-268.

利用近邻因子提高二氧化氮遥感反演浓度的精度—基于随机森林算法

符淼

广东外语外贸大学经济贸易学院

收稿日期:2022-01-11 修回日期:2022-02-28 出版日期:2023-05-28 发布日期:2023-05-28
通讯作者: E-mail: cnfm@163.com. E-mail:cnfm@163.com
作者简介:符淼 (1973- ), 广东雷州人, 博士, 教授, 硕士生导师, 主要从事环境经济学研究。E-mail: cnfm@163.com.
基金资助:
广东省自然科学基金自由申请项目;教育部人文社会科学研究规划基金项目

Improving the accuracy of NO2 concentrations derived from remote sensing using localized factors based on random forest algorithm

FU Miao

School of Economics and Trade, Guangdong University of Foreign Studies

Received:2022-01-11 Revised:2022-02-28 Published:2023-05-28 Online:2023-05-28
Contact: Miao Fu E-mail:cnfm@163.com
Supported by:
教育部人文社会科学研究规划基金项目 (17YJA790021), 广东省自然科学基金自由申请项目 (2017A030313439), 广州国际商贸中心研究基地专项资助 (JDZB202108)

摘要/Abstract

摘要： NO2是损害健康和破坏生态的主要大气污染物。本文基于NASA提供的Aura OMI遥感反演NO2浓度，利用采样点8 km内的经济、人口、路网和坡度数据，以及气象、植被和高程的点值数据，采用随机森林算法、地理加权回归 (GWR) 和多尺度GWR方法提高NO2浓度的预测精度。NASA原浓度R2为0.48，以上三种模型把交叉验证R2分别提高到0.74、0.71 和0.70，其中随机森林算法的精度最高，该算法的均方根误差 (RMSE) 和平均绝对误差 (MAE) 分别只有 6.4 μg/m3和4.98 μg/m3，且其速度远快于多尺度GWR，预测精度也高于大部分现有的同等范围研究。在浓度修正方面，局部化经济人口路网因子对预测精度提高的贡献至少为11.24%。此外，基于随机森林算法还给出全国县级城市 NO2浓度估计值的分布图。

关键词: 二氧化氮浓度, 近邻因子, 随机森林算法, 地理加权回归, 多尺度地理加权回归

Abstract: NO2 is a main air pollutant that damages human health and ecological environment. Based on NASA's NO2 concentrations retrieved from Aura OMI, the prediction accuracy of NO2 concentration is improved in this work using the random forest algorithm, the Geographic Weighted Regression (GWR) and the Multi-scale GWR model respectively. Localized data of economy, population, road network and slope within 8 km of the sampling point, as well as the point values of meteorology, vegetation and elevation are used as correction variables in the models. It is found that the three models increase the cross validation R2 of NASA's concentrations, from original 0.48 to 0.74, 0.71 and 0.70, respectively. Among the three models, the random forest algorithm is the most accurate one, with a low root mean square error (RMSE) of 6.4 μg/m3 and a low mean absolute error (MAE) of 4.98 μg/m3, and its speed is much faster than multi-scale GWR. In addition, the accuracy of random forest algorithm is also higher than that of most existing studies of similar extents. In terms of the concentration correction of NO2, it is found that the contribution of localized factors of economy, population and road network is at least 11.24%. In addition, based on the random forest algorithm, the distribution map of NO2 estimated concentration for county-level cities in China is also presented.

Key words: NO2 concentrations, localized factors, random forest algorithm, geographic weighted regression, multi-scale geographic weighted regression

中图分类号:

符淼. 利用近邻因子提高二氧化氮遥感反演浓度的精度—基于随机森林算法[J]. 大气与环境光学学报, 2023, 18(3): 258-268.

FU Miao. 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): 258-268.

参考文献 25

[1]	Li M, Zhang Q, Kurokawa J, et al. MIX: A mosaic Asian anthropogenic emission inventory for the MICS-Asia and the
	HTAP projects [J]. Atmospheric Chemistry & Physics, 2015, 15: 34813-34869.
[2]	Bechle M J, Millet D B, Marshall J D. National spatiotemporal exposure surface for NO2: Monthly scaling of a satellitederived
	land-use regression, 2000―2010 [J]. Environmental Science & Technology, 2015, 49(20): 12297-12305.
[3]	Chan K L, Khorsandi E, Liu S, et al. Estimation of surface NO2 concentrations over Germany from TROPOMI satellite
	observations using a machine learning method [J]. Remote Sensing, 2021, 13(5): 969.
[4]	Cooper M J, Martin R V, McLinden C A, et al. Inferring ground-level nitrogen dioxide concentrations at fine spatial
	resolution applied to the TROPOMI satellite instrument [J]. Environmental Research Letters, 2020, 15(10): 104013.
[5]	Knibbs L D, Hewson MG, Bechle M J, et al. A national satellite-based land-use regression model for air pollution exposure
	assessment in Australia [J]. Environmental Research, 2014, 135: 204-211.
[6]	Novotny E V, Bechle M J, Millet D B, et al. National satellite-based land-use regression: NO2 in the United States [J].
	Environmental Science & Technology, 2011, 45(10): 4407-4414.
[7]	Larkin A, Geddes J A, Martin R V, et al. Global land use regression model for nitrogen dioxide air pollution [J].
	Environmental Science & Technology, 2017, 51(12): 6957-6964.
[8]	Harper, A, Baker, P N, Xia, Y, et al. Development of spatiotemporal land use regression models for PM2.5 and NO2 in
	Chongqing, China, and exposure assessment for the CLIMB study [J]. Atmospheric Pollution Research, 2021, 12(7): 101096.
[9]	Shi Y Y, Li R D, Qiu J, et al. Spatial distribution simulation and underlying surface factors analysis of NO2 concentration
	based on land use regression [J]. Journal of Geo-Information Science, 2017, 19(1): 10-19.
	施媛媛, 李仁东, 邱娟, 等. 基于LUR的二氧化氮浓度空间分布模拟及其下垫面影响因素分析 [J]. 地球信息科学学报,
20	17, 19(1): 10-19.
[10]	Anand J S, Monks P S. Estimating daily surface NO2 concentrations from satellite data―A case study over Hong Kong using
	land use regression models [J]. Atmospheric Chemistry and Physics, 2017, 17(13): 8211-8230.
[11]	He Q, Kai Q, Cohen J B, et al. Spatially and temporally coherent reconstruction of tropospheric NO2 over China combining
	OMI and GOME-2B measurements [J]. Environmental Research Letters, 2020, 15(12): 125011.
[12]	Chen J, de Hoogh K, Gulliver J, et al. A comparison of linear regression, regularization, and machine learning algorithms
	to develop Europe-wide spatial models of fine particles and nitrogen dioxide [J]. Environment International, 2019, 130:
10	4934.
[13]	Wang Z, Uno I, Yumimoto K, et al. Impacts of COVID-19 lockdown, Spring Festival and meteorology on the NO2 variations
	in early 2020 over China based on in situ observations, satellite retrievals and model simulations [J]. Atmospheric
	Environment, 2021, 244: 117972.
[14]	Lamsal L N, Martin R V, van Donkelaar A, et al. Ground-level nitrogen dioxide concentrations inferred from the satelliteborne
	Ozone Monitoring Instrument [J]. Journal of Geophysical Research: Atmospheres, 2008, 113(D16): D16308.
[15]	Zheng K L, Huang Y, Yao X Y, et al. Correlation between PM2.5, NO2 and tourism activities, weather factors in Zhangjiajie
	City [J]. Journal of Atmospheric and Environmental Optics, 2020, 15(5): 347-356.
	郑凯莉, 黄毅, 姚小云, 等. 张家界市PM2.5、NO2与旅游活动及天气因素的相关性分析 [J]. 大气与环境光学学报, 2020,
15	(5): 347-356.
[16]	Breiman L. Random forests [J]. Machine Learning, 2001, 45(1): 5-32.
[17]	Zhai L, Li S, Zou B, et al. An improved geographically weighted regression model for PM2.5 concentration estimation in large
	areas [J]. Atmospheric Environment, 2018, 181: 145-154.
[18]	Meng X, Chen L, Cai J, et al. A land use regression model for estimating the NO2 concentration in Shanghai, China [J].
	Environmental Research, 2015, 137: 308-315.
[19]	Gilliland F, Avol E, Kinney P, et al. Air pollution exposure assessment for epidemiologic studies of pregnant women and
	children: Lessons learned from the Centers for Children's Environmental Health and Disease Prevention Research [J].
	Environmental Health Perspectives, 2005, 113(10): 1447-1454.
[20]	Health Effects Institute. Traffic-Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and
	Health Effects [R]. Boston: Health Effects Institute, 2010, Special Report 17.
[21]	Yang W B. An Extension of Geographically Weighted Regression with Flexible Bandwidths [D]. St Andrews: University of St.
	Andrews, 2014.
[22]	Zhao R, Zhang C X, Wu Y, et al. Analysis of spatio-temporal variations of tropospheric nitrogen dioxide in the North China
	plain based on EMI [J]. Journal of Atmospheric and Environmental Optics, 2021, 16(3): 186-196.
	赵冉, 张成歆, 吴跃, 等. 基于EMI观测华北平原对流层NO2的时空变化研究 [J]. 大气与环境光学学报, 2021, 16(3):
18	6-196.
[23]	Wang X H, Xu Y Z, Zhang C X, et al. Spatial-temporal variation of tropospheric NO2 concentration in Pearl River Delta
	based on EMI observations [J]. Journal of Atmospheric and Environmental Optics, 2021, 16(3): 197-206.
	王肖汉, 徐翼洲, 张成歆, 等. 基于EMI观测的珠三角地区对流层NO2柱浓度时空变化特征分析 [J]. 大气与环境光学学
	报, 2021, 16(3): 197-206.
[24]	Lee H J, Koutrakis P. Daily ambient NO2 concentration predictions using satellite ozone monitoring instrument NO2 data and
	land use regression [J]. Environmental Science & Technology, 2014, 48(4): 2305-2311.
[25]	Gu J B, Chen L F, Yu C, et al. Ground-level NO2 concentrations over China inferred from the satellite OMI and CMAQ
	model simulations [J]. Remote Sensing, 2017, 9(6): 519.