Europe-wide air pollution modeling from 2000 to 2019 using geographically weighted regression

Youchen Shen; Kees de Hoogh; Oliver Schmitz; Nicholas Clinton; Karin Tuxen-Bettman; Jørgen Brandt; Jesper H. Christensen; Lise M. Frohn; Camilla Geels; Derek Karssenberg; Roel Vermeulen; Gerard Hoek

doi:10.1016/j.envint.2022.107485

Europe-wide air pollution modeling from 2000 to 2019 using geographically weighted regression

Youchen Shen^*, Kees de Hoogh, Oliver Schmitz, Nicholas Clinton, Karin Tuxen-Bettman, Jørgen Brandt, Jesper H. Christensen, Lise M. Frohn, Camilla Geels, Derek Karssenberg, Roel Vermeulen, Gerard Hoek

^*Corresponding author for this work

Research output: Contribution to journal › Article › Academic › peer-review

2 Citations (Scopus)

3 Downloads (Pure)

Abstract

Previous European land-use regression (LUR) models assumed fixed linear relationships between air pollution concentrations and predictors such as traffic and land use. We evaluated whether including spatially-varying relationships could improve European LUR models by using geographically weighted regression (GWR) and random forest (RF). We built separate LUR models for each year from 2000 to 2019 for NO₂, O₃, PM_2.5 and PM₁₀ using annual average monitoring observations across Europe. Potential predictors included satellite retrievals, chemical transport model estimates and land-use variables. Supervised linear regression (SLR) was used to select predictors, and then GWR estimated the potentially spatially-varying coefficients. We developed multi-year models using geographically and temporally weighted regression (GTWR). Five-fold cross-validation per year showed that GWR and GTWR explained similar spatial variations in annual average concentrations (average R² = NO₂: 0.66; O₃: 0.58; PM₁₀: 0.62; PM_2.5: 0.77), which are better than SLR (average R² = NO₂: 0.61; O₃: 0.46; PM₁₀: 0.51; PM_2.5: 0.75) and RF (average R² = NO₂: 0.64; O₃: 0.53; PM₁₀: 0.56; PM_2.5: 0.67). The GTWR predictions and a previously-used method of back-extrapolating 2010 model predictions using CTM were overall highly correlated (R² > 0.8) for all pollutants. Including spatially-varying relationships using GWR modestly improved European air pollution annual LUR models, allowing time-varying exposure-health risk models.

Original language	English
Article number	107485
Journal	Environment International
Volume	168
DOIs	https://doi.org/10.1016/j.envint.2022.107485
Publication status	Published - Oct 2022

Keywords

Geographically and temporally weighted regression
Land-use regression
Random forest
Spatially varying coefficient
Spatiotemporal variation

Access to Document

10.1016/j.envint.2022.107485Licence: CC BY

1-s2.0-S0160412022004123-mainFinal published version, 9.55 MBLicence: CC BY

Cite this

@article{78da83c455d24aa9afd65d189f3e2279,

title = "Europe-wide air pollution modeling from 2000 to 2019 using geographically weighted regression",

abstract = "Previous European land-use regression (LUR) models assumed fixed linear relationships between air pollution concentrations and predictors such as traffic and land use. We evaluated whether including spatially-varying relationships could improve European LUR models by using geographically weighted regression (GWR) and random forest (RF). We built separate LUR models for each year from 2000 to 2019 for NO2, O3, PM2.5 and PM10 using annual average monitoring observations across Europe. Potential predictors included satellite retrievals, chemical transport model estimates and land-use variables. Supervised linear regression (SLR) was used to select predictors, and then GWR estimated the potentially spatially-varying coefficients. We developed multi-year models using geographically and temporally weighted regression (GTWR). Five-fold cross-validation per year showed that GWR and GTWR explained similar spatial variations in annual average concentrations (average R2 = NO2: 0.66; O3: 0.58; PM10: 0.62; PM2.5: 0.77), which are better than SLR (average R2 = NO2: 0.61; O3: 0.46; PM10: 0.51; PM2.5: 0.75) and RF (average R2 = NO2: 0.64; O3: 0.53; PM10: 0.56; PM2.5: 0.67). The GTWR predictions and a previously-used method of back-extrapolating 2010 model predictions using CTM were overall highly correlated (R2 > 0.8) for all pollutants. Including spatially-varying relationships using GWR modestly improved European air pollution annual LUR models, allowing time-varying exposure-health risk models.",

keywords = "Geographically and temporally weighted regression, Land-use regression, Random forest, Spatially varying coefficient, Spatiotemporal variation",

author = "Youchen Shen and {de Hoogh}, Kees and Oliver Schmitz and Nicholas Clinton and Karin Tuxen-Bettman and J{\o}rgen Brandt and Christensen, {Jesper H.} and Frohn, {Lise M.} and Camilla Geels and Derek Karssenberg and Roel Vermeulen and Gerard Hoek",

note = "Funding Information: This work was supported by EXPANSE and EXPOSOME-NL projects. The EXPANSE project is funded by the European Union{\textquoteright}s Horizon 2020 research and innovation programme under grant agreement No 874627. The content of this article is not officially endorsed by the European Union. The EXPOSOME-NL project is funded through the Gravitation programme of the Dutch Ministry of Education, Culture, and Science and the Netherlands Organization for Scientific Research (NWO grant number 024.004.017). The authors declare no competing financial interest. Funding Information: This work was supported by EXPANSE and EXPOSOME-NL projects. The EXPANSE project is funded by the European Union's Horizon 2020 research and innovation programme under grant agreement No 874627. The content of this article is not officially endorsed by the European Union. The EXPOSOME-NL project is funded through the Gravitation programme of the Dutch Ministry of Education, Culture, and Science and the Netherlands Organization for Scientific Research (NWO grant number 024.004.017). The authors declare no competing financial interest. Publisher Copyright: {\textcopyright} 2022 The Author(s)",

year = "2022",

month = oct,

doi = "10.1016/j.envint.2022.107485",

language = "English",

volume = "168",

journal = "Environment International",

issn = "0160-4120",

publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Europe-wide air pollution modeling from 2000 to 2019 using geographically weighted regression

AU - Shen, Youchen

AU - de Hoogh, Kees

AU - Schmitz, Oliver

AU - Clinton, Nicholas

AU - Tuxen-Bettman, Karin

AU - Brandt, Jørgen

AU - Christensen, Jesper H.

AU - Frohn, Lise M.

AU - Geels, Camilla

AU - Karssenberg, Derek

AU - Vermeulen, Roel

AU - Hoek, Gerard

N1 - Funding Information: This work was supported by EXPANSE and EXPOSOME-NL projects. The EXPANSE project is funded by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 874627. The content of this article is not officially endorsed by the European Union. The EXPOSOME-NL project is funded through the Gravitation programme of the Dutch Ministry of Education, Culture, and Science and the Netherlands Organization for Scientific Research (NWO grant number 024.004.017). The authors declare no competing financial interest. Funding Information: This work was supported by EXPANSE and EXPOSOME-NL projects. The EXPANSE project is funded by the European Union's Horizon 2020 research and innovation programme under grant agreement No 874627. The content of this article is not officially endorsed by the European Union. The EXPOSOME-NL project is funded through the Gravitation programme of the Dutch Ministry of Education, Culture, and Science and the Netherlands Organization for Scientific Research (NWO grant number 024.004.017). The authors declare no competing financial interest. Publisher Copyright: © 2022 The Author(s)

PY - 2022/10

Y1 - 2022/10

N2 - Previous European land-use regression (LUR) models assumed fixed linear relationships between air pollution concentrations and predictors such as traffic and land use. We evaluated whether including spatially-varying relationships could improve European LUR models by using geographically weighted regression (GWR) and random forest (RF). We built separate LUR models for each year from 2000 to 2019 for NO2, O3, PM2.5 and PM10 using annual average monitoring observations across Europe. Potential predictors included satellite retrievals, chemical transport model estimates and land-use variables. Supervised linear regression (SLR) was used to select predictors, and then GWR estimated the potentially spatially-varying coefficients. We developed multi-year models using geographically and temporally weighted regression (GTWR). Five-fold cross-validation per year showed that GWR and GTWR explained similar spatial variations in annual average concentrations (average R2 = NO2: 0.66; O3: 0.58; PM10: 0.62; PM2.5: 0.77), which are better than SLR (average R2 = NO2: 0.61; O3: 0.46; PM10: 0.51; PM2.5: 0.75) and RF (average R2 = NO2: 0.64; O3: 0.53; PM10: 0.56; PM2.5: 0.67). The GTWR predictions and a previously-used method of back-extrapolating 2010 model predictions using CTM were overall highly correlated (R2 > 0.8) for all pollutants. Including spatially-varying relationships using GWR modestly improved European air pollution annual LUR models, allowing time-varying exposure-health risk models.

AB - Previous European land-use regression (LUR) models assumed fixed linear relationships between air pollution concentrations and predictors such as traffic and land use. We evaluated whether including spatially-varying relationships could improve European LUR models by using geographically weighted regression (GWR) and random forest (RF). We built separate LUR models for each year from 2000 to 2019 for NO2, O3, PM2.5 and PM10 using annual average monitoring observations across Europe. Potential predictors included satellite retrievals, chemical transport model estimates and land-use variables. Supervised linear regression (SLR) was used to select predictors, and then GWR estimated the potentially spatially-varying coefficients. We developed multi-year models using geographically and temporally weighted regression (GTWR). Five-fold cross-validation per year showed that GWR and GTWR explained similar spatial variations in annual average concentrations (average R2 = NO2: 0.66; O3: 0.58; PM10: 0.62; PM2.5: 0.77), which are better than SLR (average R2 = NO2: 0.61; O3: 0.46; PM10: 0.51; PM2.5: 0.75) and RF (average R2 = NO2: 0.64; O3: 0.53; PM10: 0.56; PM2.5: 0.67). The GTWR predictions and a previously-used method of back-extrapolating 2010 model predictions using CTM were overall highly correlated (R2 > 0.8) for all pollutants. Including spatially-varying relationships using GWR modestly improved European air pollution annual LUR models, allowing time-varying exposure-health risk models.

KW - Geographically and temporally weighted regression

KW - Land-use regression

KW - Random forest

KW - Spatially varying coefficient

KW - Spatiotemporal variation

UR - http://www.scopus.com/inward/record.url?scp=85136571829&partnerID=8YFLogxK

U2 - 10.1016/j.envint.2022.107485

DO - 10.1016/j.envint.2022.107485

M3 - Article

C2 - 36030744

AN - SCOPUS:85136571829

SN - 0160-4120

VL - 168

JO - Environment International

JF - Environment International

M1 - 107485

ER -

Europe-wide air pollution modeling from 2000 to 2019 using geographically weighted regression

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Corrigendum to “Europe-wide air pollution modeling from 2000 to 2019 using geographically weighted regression” [Environ. Int. 168 (2022) 107485](S0160412022004123)(10.1016/j.envint.2022.107485)

Cite this