A source area model incorporating simplified atmospheric dispersion and advection at fine scale for population air pollutant exposure assessment

Ainslie, Bruce; Steyn, D. G.; Su, Jason; Buzzelli, Michael; Bräuer, Michael; Larson, Tim; Rucker, Magdalena

doi:10.1016/j.atmosenv.2007.12.021

Cited by 29 publications

(13 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, we combined GAM with cokriging of spatial residuals to estimate summer and winter NO 2 and NO x pollution surfaces in Southern California, assuming that spatial variations in concentrations are substantially influenced by both local variation due to proximity to emission sources and by global variation due to atmospheric transport (Ainslie et al, 2008;Beelen et al, 2009; Liu et al, 2009). Local means were predicted by GAM, while spatial residuals from GAM were assumed to be second-order stationary (Gartan and Guyon, 2010) and modeled through cokriging with global residuals (representing global variations) at sampling locations nearby.…”

Section: Introductionmentioning

confidence: 99%

Use of generalized additive models and cokriging of spatial residuals to improve land-use regression estimates of nitrogen oxides in Southern California

Wilhelm

et al. 2012

Atmospheric Environment

View full text Add to dashboard Cite

Land-use regression (LUR) models have been developed to estimate spatial distributions of traffic-related pollutants. Several studies have examined spatial autocorrelation among residuals in LUR models, but few utilized spatial residual information in model prediction, or examined the impact of modeling methods, monitoring site selection, or traffic data quality on LUR performance. This study aims to improve spatial models for traffic-related pollutants using generalized additive models (GAM) combined with cokriging of spatial residuals. Specifically, we developed spatial models for nitrogen dioxide (NO2) and nitrogen oxides (NOx) concentrations in Southern California separately for two seasons (summer and winter) based on over 240 sampling locations. Pollutant concentrations were disaggregated into three components: local means, spatial residuals, and normal random residuals. Local means were modeled by GAM. Spatial residuals were cokriged with global residuals at nearby sampling locations that were spatially auto-correlated. We compared this two-stage approach with four commonly-used spatial models: universal kriging, multiple linear LUR and GAM with and without a spatial smoothing term. Leave-one-out cross validation was conducted for model validation and comparison purposes. The results show that our GAM plus cokriging models predicted summer and winter NO2 and NOx concentration surfaces well, with cross validation R2 values ranging from 0.88 to 0.92. While local covariates accounted for partial variance of the measured NO2 and NOx concentrations, spatial autocorrelation accounted for about 20% of the variance. Our spatial GAM model improved R2 considerably compared to the other four approaches. Conclusively, our two-stage model captured summer and winter differences in NO2 and NOx spatial distributions in Southern California well. When sampling location selection cannot be optimized for the intended model and fewer covariates are available as predictors for the model, the two-stage model is more robust compared to multiple linear regression models.

show abstract

Section: Introductionmentioning

confidence: 99%

Use of generalized additive models and cokriging of spatial residuals to improve land-use regression estimates of nitrogen oxides in Southern California

Wilhelm

et al. 2012

Atmospheric Environment

View full text Add to dashboard Cite

show abstract

“…As an alternative to measurements, investigations of road noise and health generally use models to estimate noise exposures (Babisch et al, 2005; Beelen et al, 2008; Bluhm et al, 2007; Calixto et al, 2003; de Kluizenaar et al, 2007; van Kempen et al, 2002). Similarly, some studies of traffic-generated air pollution use dispersion and/or “land use regression” models to estimate concentrations (Ainslie et al, 2008; Jerrett et al, 2005a; Su et al, 2008). However, these air pollution models often require spatially dense monitoring or extensive data on emissions and meteorology prior to model development.…”

Section: Introductionmentioning

confidence: 99%

The spatial relationship between traffic-generated air pollution and noise in 2 US cities

Allen

Davies

Cohen

et al. 2009

Environmental Research

148

View full text Add to dashboard Cite

Traffic-generated air pollution and noise have both been linked to cardiovascular morbidity. Since traffic is a shared source, there is potential for correlated exposures that may lead to confounding in epidemiologic studies. As part of the Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air), 2-week NO and NO2 concentrations were measured at up to 105 locations, selected primarily to characterize gradients near major roads, in each of 9 US communities. We measured 5-min A-weighted equivalent continuous sound pressure levels (Leq) and ultrafine particle (UFP) counts at a subset of these NO/NO2 monitoring locations in Chicago, IL (N = 69 in December 2006; N = 36 in April 2007) and Riverside County, CA (N = 46 in April 2007). Leq and UFP were measured during non-“rush hour” periods (10:00–16:00) to maximize comparability between measurements. We evaluated roadway proximity exposure surrogates in relation to the measured levels, estimated noise–air pollution correlation coefficients, and evaluated the impact of regional-scale pollution gradients, wind direction, and roadway proximity on the correlations. Five-minute Leq measurements in December 2006 and April 2007 were highly correlated (r = 0.84), and measurements made at different times of day were similar (coefficients of variation: 0.5–13%), indicating that 5-min measurements are representative of long-term Leq. Binary and continuous roadway proximity metrics characterized Leq as well or better than NO or NO2. We found strong regional-scale gradients in NO and NO2, particularly in Chicago, but only weak regional-scale gradients in Leq and UFP. Leq was most consistently correlated with NO, but the correlations were moderate (0.20–0.60). After removing the influence of regional-scale gradients the correlations generally increased (Leq–NO: r = 0.49–0.62), and correlations downwind of major roads (Leq–NO: r = 0.53–0.74) were consistently higher than those upwind (0.35–0.65). There was not a consistent effect of roadway proximity on the correlations. In conclusion, roadway proximity variables are not unique exposure surrogates in studies of endpoints hypothesized to be related to both air pollution and noise. Moderate correlations between traffic-generated air pollution and noise suggest the possibility of confounding, which might be minimized by considering regional pollution gradients and/or prevailing wind direction(s) in epidemiologic studies.

show abstract

“…Some are beginning to address the manner in which sampling sites are selected, the treatment of time in what is typically a cross‐sectional study design, the cost associated with the sampling campaigns needed to parameterise models for reliable estimates, the transferability of models among urban regions and finally the validation of model estimates (itself a function of cost). In almost all of these points the underlying criticism of LUR is its lack of a physical‐theoretical basis; that the models are data(base) driven and despite such consistent and strong results there is little a priori consideration of atmospheric processes behind the sampling, modelling and output (Ainslie et al 2008; Su et al 2008). A better handle on wind speed and direction, for example, could minimise the sampling requirements and related costs, allow for incorporation of temporal variations and raise the transferability of models from one region to another.…”

Section: Discussionmentioning

confidence: 99%

A political ecology of scale in urban air pollution monitoring

Buzzelli

2008

Trans Inst British Geog

View full text Add to dashboard Cite

Recent developments in air pollution epidemiology are generating new data on human air pollution exposure and health effects in the city. Based on regression mapping or land use regression (LUR), these data also provide a new window on the urban political ecology of scale. Air pollution itself is understudied in the emergent urban political ecology literature and in this case it is aided by -indeed must rely upon -GIS/spatial analysis. A case study of Vancouver illustrates the scalar contradictions in regulatory versus LUR measurements and estimates leading us finally to consider the wider implications for urban political ecology and environmental governance.key words air pollution GIS scale urban political ecology Vancouver

show abstract

A source area model incorporating simplified atmospheric dispersion and advection at fine scale for population air pollutant exposure assessment

Cited by 29 publications

References 37 publications

Use of generalized additive models and cokriging of spatial residuals to improve land-use regression estimates of nitrogen oxides in Southern California

Use of generalized additive models and cokriging of spatial residuals to improve land-use regression estimates of nitrogen oxides in Southern California

The spatial relationship between traffic-generated air pollution and noise in 2 US cities

A political ecology of scale in urban air pollution monitoring

Contact Info

Product

Resources

About