Using a land use regression model with machine learning to estimate ground level PM2.5

Wong, Pei-Yi; Lee, Hsiao-Yun; Chen, Yu‐Cheng; Yu-Ting, Zeng; Chern, Yinq-Rong; Chen, Nai-Tzu; Lung, Shih-Chun Candice; Su, Huey‐Jen; Wu, Chih-Da

doi:10.1016/j.envpol.2021.116846

Cited by 89 publications

(51 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In recent years, studies on PM 2.5 mainly focus on its source and composition (Martuzevicius et al, 2004;Xu et al, 2014), spatio-temporal distribution (Lu et al, 2017;Mi et al, 2020) and concentration prediction (Mao et al, 2012;Yang et al, 2018a;Zhang et al, 2018;Stafoggia et al, 2019;Wong et al, 2021) to seek measures to reduce the negative effects of PM 2.5 by clarifying its chemical composition and sources. At the same time, many factors have been proved to have a certain influence on PM 2.5 concentration.…”

Section: Literature Reviewmentioning

confidence: 99%

Spatio-Temporal Evolution and Correlation Analysis of Urban Land Use Patterns and Air Quality in Pearl River Delta, China

Hong

Qiu

Yang

et al. 2021

Front. Environ. Sci.

View full text Add to dashboard Cite

Urban air quality, which is related to the health of local residents of the Pearl River Delta Region, China, (PRD) has been a hot topic among the masses and academic circles. In addition to economic growth, China’s rapidly increasing urbanization rate has also brought great pressure on urban air quality, in the Region, where due to its huge economic size and population, urban air quality has become the focus of local residents. This study first analyzed the spatio-temporal trends and correlation of land use and PM2.5 average annual concentration, which represents air quality in the PRD from 2000 to 2018, and according to Moran index, PM2.5 concentration in this area has spatial correlation in the study time. Secondly, the spatial error model of the PRD was constructed by using spatial effect, spatial modeling theory and spatial model estimation. The spatial-temporal evolution and influencing factors of PM2.5 concentration were discussed in fixed overall effect. The results showed that the area of cultivated land in the PRD decreased and the area of artificial surface increased year by year, while the PM2.5 concentration increased first and then decreased. In addition, the area of cultivated land and forestland were negatively correlated, while the area of grassland and water body were positively correlated with PM2.5 concentration. The conclusion of this study has a certain theoretical reference for urban land use planning and air quality assurance.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Spatio-Temporal Evolution and Correlation Analysis of Urban Land Use Patterns and Air Quality in Pearl River Delta, China

Hong

Qiu

Yang

et al. 2021

Front. Environ. Sci.

View full text Add to dashboard Cite

show abstract

“…In addition, the Environment Resource Datasets 43 are publicly available from open government data. This dataset was obtained by the Environmental Protection Administration of Taiwan, which determined ambient pollutants and temperatures at 76 monitoring stations across Taiwan, from 1993 to 2013.…”

Section: Methodsmentioning

confidence: 99%

“…These were selected based on weak correlations (Pearson's correlation coefficients <0.3) of target pollutants with ten other monitored air pollutants: sulfur dioxide (SO2); ozone (O3); carbon monoxide (CO); carbon dioxide (CO2); nitrogen oxides (NOX); nitrogen monoxide (NO); nitrogen dioxide (NO2); particulate matter <10 μm in size (PM10); particulate matter <2.5 μm in size (PM2.5); and methane (CH4) (Supplementary Table S1). Daily air quality data were collected at 76 monitoring stations from July 1, 1993, to December 31, 2013, and maintained by the EPA 43 .…”

Section: Exposure Modelingmentioning

confidence: 99%

Long-term ambient hydrocarbon exposure and incidence of urinary bladder cancer

Zhang

Tsai

Kok

et al. 2022

Preprint

View full text Add to dashboard Cite

BackgroundParticulate matter and volatile organic compounds, including total hydrocarbons (THCs), are major ambient air pollutants. The primary nonmethane hydrocarbons (NMHCs) originate from vehicle emissions. Studies on the association between air pollution and urinary bladder cancer (UBC) have revealed contradictory results.ObjectivesThe present study investigated whether long-term exposure to ambient hydrocarbons increases UBC risk among people aged ≥20 years in Taiwan.MethodsLinkage dataset research with longitudinal design was conducted on 600,666 cancer-free individuals from 2000–2013; 12 airborne pollutants were determined. Several Cox models considering potential confounders were employed. The study outcomes were invasive or in situ UBC incidence over time. The targeted pollutant concentration was divided into three tertiles: T1/T2/T3.ResultsThe mean age of the individuals at risk was 42.8 (SD, 15.7), and 50.2% were men. The mean daily average over 10-years of airborne THC concentration was 2.24 ppm (SD, 0.14), and NMHC was 0.29 ppm (SD, 0.09). There was a dose-dependent increase in UBC at follow-up. The incidence of UBC cases per 100,000 people by T1/T2/T3 exposure to THC was 60.3, 203.7, and 450.8, respectively; it was 180.2/202.4/453.8 per 100,000, corresponding to T1/T2/T3 exposure to NMHC, respectively. Without controlling for confounding air pollutants, the adjusted hazard ratio (adj.HR) was 1.77 (95% CI, 1.69–1.87) per 0.14 ppm increase of THC; after controlling for PM2.5, adj.HR was even higher at 2.10 (95% CI, 1.99– 2.22). The adj.HR was 1.38 (95% CI, 1.31–1.44) per 0.09 ppm increase in ambient NMHC concentration. After controlling for SO2 and CH4, the adj.HR was 1.16 (95% CI, 1.11–1.21). Sensitivity analyses showed that the UBC development risk was not sex-specific or influenced by diabetes status.DiscussionLong-term exposure to THC and NMHC may be a risk factor for UBC development. Acknowledging the pollutant sources can inform risk management strategies.

show abstract

“…The spatiotemporal LUR model was a hybrid two-stage model integrating a static LUR model and a multiple linear-regression-based meteorological factor regression (MFR) model for more accurate spatiotemporal predictions [41]. Finally, an LUR model was integrated with machine learning algorithms to improve the prediction accuracy, where the linear relationships between air pollutants and explanatory variables are replaced by nonlinear relationships explored by machine learning [15,16]. For instance, non-parametric LUR models were developed with the support of a random forest model and a generalized additive model for predicting spatial distributions of ambient total particulate concentrations [42], and additive regression smoother-based LUR models were developed for investigating agglomeration and infrastructure effects on air pollutants [43].…”

Section: Literature Reviewmentioning

confidence: 99%

“…The applications and advantages of LUR models have been reviewed in the next section. In recent years, a series of new models have been developed based on LUR to improve prediction capacity, such as dimensionality reduction for explanatory variables [13], spatiotemporal LUR modelling [14] and the integration of LUR and machine learning algorithms [15,16], as reviewed in the next section.…”

Section: Introductionmentioning

confidence: 99%

Land Use Quantile Regression Modeling of Fine Particulate Matter in Australia

Song

2022

Remote Sensing

View full text Add to dashboard Cite

Small data samples are still a critical challenge for spatial predictions. Land use regression (LUR) is a widely used model for spatial predictions with observations at a limited number of locations. Studies have demonstrated that LUR models can overcome the limitation exhibited by other spatial prediction models which usually require greater spatial densities of observations. However, the prediction accuracy and robustness of LUR models still need to be improved due to the linear regression within the LUR model. To improve LUR models, this study develops a land use quantile regression (LUQR) model for more accurate spatial predictions for small data samples. The LUQR is an integration of the LUR and quantile regression, which both have advantages in predictions with a small data set of samples. In this study, the LUQR model is applied in predicting spatial distributions of annual mean PM2.5concentrations across the Greater Sydney Region, New South Wales, Australia, with observations at 19 valid monitoring stations in 2020. Cross validation shows that the goodness-of-fit can be improved by 25.6–32.1% by LUQR models when compared with LUR, and prediction root mean squared error (RMSE) and mean absolute error (MAE) can be reduced by 10.6–13.4% and 19.4–24.7% by LUQR models, respectively. This study also indicates that LUQR is a more robust model for the spatial prediction with small data samples than LUR. Thus, LUQR has great potentials to be widely applied in spatial issues with a limited number of observations.

show abstract

Using a land use regression model with machine learning to estimate ground level PM2.5

Cited by 89 publications

References 45 publications

Spatio-Temporal Evolution and Correlation Analysis of Urban Land Use Patterns and Air Quality in Pearl River Delta, China

Spatio-Temporal Evolution and Correlation Analysis of Urban Land Use Patterns and Air Quality in Pearl River Delta, China

Long-term ambient hydrocarbon exposure and incidence of urinary bladder cancer

Land Use Quantile Regression Modeling of Fine Particulate Matter in Australia

Contact Info

Product

Resources

About