A hybrid kriging/land-use regression model with Asian culture-specific sources to assess NO2 spatial-temporal variations

Tsun Hsuan Chen, Yen Ching Hsu, Yu Ting Zeng, Shih Chun Candice Lung, Huey Jen Su, Hsing Jasmine Chao, Chih Da Wu

Research output: Contribution to journalArticle

Abstract

Kriging interpolation and land use regression (LUR) have characterized the spatial variability of long-term nitrogen dioxide (NO2), but there has been little research on combining these two methods to capture small-scale spatial variation. Furthermore, studies predicting NO2 exposure are almost exclusively based on traffic-related variables, which may not be transferable to Taiwan, a typical Asian country with diverse local emission sources, where densely distributed temples and restaurants may be important for NO2 levels. To advance the exposure estimates in Taiwan, a hybrid kriging/LUR model incorporates culture-specific sources as potential predictors. Based on 14-year NO2 observations from 73 monitoring stations across Taiwan, a set of interpolated NO2 values were generated through a leave-one-out ordinary kriging algorithm, and this was included as an explanatory variable in the stepwise LUR procedures. Kriging interpolated NO2 and culture-specific predictors were entered in the final models, which captured 90% and 87% of NO2 variation in annual and monthly resolution, respectively. Results from 10-fold cross-validation and external data verification demonstrate robust performance of the developed models. This study demonstrates the value of incorporating the kriging-interpolated estimates and culture-specific emission sources into the traditional LUR model structure for predicting NO2, which can be particularly useful for Asian countries.

Original languageEnglish
Article number113875
JournalEnvironmental Pollution
Volume259
DOIs
Publication statusPublished - 2020 Apr

Fingerprint

Spatial Analysis
Land use
Taiwan
Nitrogen Dioxide
Model structures
Restaurants
Interpolation
Nitrogen
Monitoring
Research

All Science Journal Classification (ASJC) codes

  • Toxicology
  • Pollution
  • Health, Toxicology and Mutagenesis

Cite this

Chen, Tsun Hsuan ; Hsu, Yen Ching ; Zeng, Yu Ting ; Candice Lung, Shih Chun ; Su, Huey Jen ; Chao, Hsing Jasmine ; Wu, Chih Da. / A hybrid kriging/land-use regression model with Asian culture-specific sources to assess NO2 spatial-temporal variations. In: Environmental Pollution. 2020 ; Vol. 259.
@article{de9079fe271d4aec9fc59d2bb4d5b95e,
title = "A hybrid kriging/land-use regression model with Asian culture-specific sources to assess NO2 spatial-temporal variations",
abstract = "Kriging interpolation and land use regression (LUR) have characterized the spatial variability of long-term nitrogen dioxide (NO2), but there has been little research on combining these two methods to capture small-scale spatial variation. Furthermore, studies predicting NO2 exposure are almost exclusively based on traffic-related variables, which may not be transferable to Taiwan, a typical Asian country with diverse local emission sources, where densely distributed temples and restaurants may be important for NO2 levels. To advance the exposure estimates in Taiwan, a hybrid kriging/LUR model incorporates culture-specific sources as potential predictors. Based on 14-year NO2 observations from 73 monitoring stations across Taiwan, a set of interpolated NO2 values were generated through a leave-one-out ordinary kriging algorithm, and this was included as an explanatory variable in the stepwise LUR procedures. Kriging interpolated NO2 and culture-specific predictors were entered in the final models, which captured 90{\%} and 87{\%} of NO2 variation in annual and monthly resolution, respectively. Results from 10-fold cross-validation and external data verification demonstrate robust performance of the developed models. This study demonstrates the value of incorporating the kriging-interpolated estimates and culture-specific emission sources into the traditional LUR model structure for predicting NO2, which can be particularly useful for Asian countries.",
author = "Chen, {Tsun Hsuan} and Hsu, {Yen Ching} and Zeng, {Yu Ting} and {Candice Lung}, {Shih Chun} and Su, {Huey Jen} and Chao, {Hsing Jasmine} and Wu, {Chih Da}",
year = "2020",
month = "4",
doi = "10.1016/j.envpol.2019.113875",
language = "English",
volume = "259",
journal = "Environmental Pollution",
issn = "0269-7491",
publisher = "Elsevier Limited",

}

A hybrid kriging/land-use regression model with Asian culture-specific sources to assess NO2 spatial-temporal variations. / Chen, Tsun Hsuan; Hsu, Yen Ching; Zeng, Yu Ting; Candice Lung, Shih Chun; Su, Huey Jen; Chao, Hsing Jasmine; Wu, Chih Da.

In: Environmental Pollution, Vol. 259, 113875, 04.2020.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A hybrid kriging/land-use regression model with Asian culture-specific sources to assess NO2 spatial-temporal variations

AU - Chen, Tsun Hsuan

AU - Hsu, Yen Ching

AU - Zeng, Yu Ting

AU - Candice Lung, Shih Chun

AU - Su, Huey Jen

AU - Chao, Hsing Jasmine

AU - Wu, Chih Da

PY - 2020/4

Y1 - 2020/4

N2 - Kriging interpolation and land use regression (LUR) have characterized the spatial variability of long-term nitrogen dioxide (NO2), but there has been little research on combining these two methods to capture small-scale spatial variation. Furthermore, studies predicting NO2 exposure are almost exclusively based on traffic-related variables, which may not be transferable to Taiwan, a typical Asian country with diverse local emission sources, where densely distributed temples and restaurants may be important for NO2 levels. To advance the exposure estimates in Taiwan, a hybrid kriging/LUR model incorporates culture-specific sources as potential predictors. Based on 14-year NO2 observations from 73 monitoring stations across Taiwan, a set of interpolated NO2 values were generated through a leave-one-out ordinary kriging algorithm, and this was included as an explanatory variable in the stepwise LUR procedures. Kriging interpolated NO2 and culture-specific predictors were entered in the final models, which captured 90% and 87% of NO2 variation in annual and monthly resolution, respectively. Results from 10-fold cross-validation and external data verification demonstrate robust performance of the developed models. This study demonstrates the value of incorporating the kriging-interpolated estimates and culture-specific emission sources into the traditional LUR model structure for predicting NO2, which can be particularly useful for Asian countries.

AB - Kriging interpolation and land use regression (LUR) have characterized the spatial variability of long-term nitrogen dioxide (NO2), but there has been little research on combining these two methods to capture small-scale spatial variation. Furthermore, studies predicting NO2 exposure are almost exclusively based on traffic-related variables, which may not be transferable to Taiwan, a typical Asian country with diverse local emission sources, where densely distributed temples and restaurants may be important for NO2 levels. To advance the exposure estimates in Taiwan, a hybrid kriging/LUR model incorporates culture-specific sources as potential predictors. Based on 14-year NO2 observations from 73 monitoring stations across Taiwan, a set of interpolated NO2 values were generated through a leave-one-out ordinary kriging algorithm, and this was included as an explanatory variable in the stepwise LUR procedures. Kriging interpolated NO2 and culture-specific predictors were entered in the final models, which captured 90% and 87% of NO2 variation in annual and monthly resolution, respectively. Results from 10-fold cross-validation and external data verification demonstrate robust performance of the developed models. This study demonstrates the value of incorporating the kriging-interpolated estimates and culture-specific emission sources into the traditional LUR model structure for predicting NO2, which can be particularly useful for Asian countries.

UR - http://www.scopus.com/inward/record.url?scp=85077325785&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85077325785&partnerID=8YFLogxK

U2 - 10.1016/j.envpol.2019.113875

DO - 10.1016/j.envpol.2019.113875

M3 - Article

C2 - 31918142

AN - SCOPUS:85077325785

VL - 259

JO - Environmental Pollution

JF - Environmental Pollution

SN - 0269-7491

M1 - 113875

ER -