Land Use Regression (LUR) models typically use fixed-site monitoring; here, we employ mobile monitoring as a cost-effective alternative for LUR development. We use bicycle-based, mobile measurements (∼85 h) during rush-hour in Minneapolis, MN to build LUR models for particulate concentrations (particle number [PN], black carbon [BC], fine particulate matter [PM2.5], particle size). We developed and examined 1224 separate LUR models by varying pollutant, time-of-day, and method of spatial and temporal smoothing of the time-series data. Our base-case LUR models had modest goodness-of-fit (adjusted R(2): ∼0.5 [PN], ∼0.4 [PM2.5], 0.35 [BC], ∼0.25 [particle size]), low bias (<4%) and absolute bias (2-18%), and included predictor variables that captured proximity to and density of emission sources. The spatial density of our measurements resulted in a large model-building data set (n = 1101 concentration estimates); ∼25% of buffer variables were selected at spatial scales of <100m, suggesting that on-road particle concentrations change on small spatial scales. LUR model-R(2) improved as sampling runs were completed, with diminishing benefits after ∼40 h of data collection. Spatial autocorrelation of model residuals indicated that models performed poorly where spatiotemporal resolution of emission sources (i.e., traffic congestion) was poor. Our findings suggest that LUR modeling from mobile measurements is possible, but that more work could usefully inform best practices.
National-scale empirical models for air pollution can include hundreds of geographic variables. The impact of model parsimony (i.e., how model performance differs for a large versus small number of covariates) has not been systematically explored. We aim to (1) build annual-average integrated empirical geographic (IEG) regression models for the contiguous U.S. for six criteria pollutants during 1979-2015; (2) explore systematically the impact on model performance of the number of variables selected for inclusion in a model; and (3) provide publicly available model predictions. We compute annual-average concentrations from regulatory monitoring data for PM 10 , PM 2.5 , NO 2 , SO 2 , CO, and ozone at all monitoring sites for 1979-2015. We also use~350 geographic characteristics at each location including measures of traffic, land use, land cover, and satellite-based estimates of air pollution. We then develop IEG models, employing universal kriging and summary factors estimated by partial least squares (PLS) of geographic variables. For all pollutants and years, we compare three approaches for choosing variables to include in the PLS model: (1) no variables, (2) a limited number of variables selected from the full set by forward selection, and (3) all variables. We evaluate model performance using 10-fold cross-validation (CV) using conventional and spatially-clustered test data. Models using 3 to 30 variables selected from the full set generally have the best performance across all pollutants and years (median R 2 conventional [clustered] CV: 0.66 [0.47]) compared to models with no (0.37 [0]) or all variables (0.64 [0.27]). Concentration estimates for all Census Blocks reveal generally decreasing concentrations over several decades with local heterogeneity. Our findings suggest that national prediction models can be built by empirically selecting only a small number of important variables to provide robust concentration estimates. Model estimates are freely available online.
Background: Physical inactivity and exposure to air pollution are important risk factors for death and disease globally. The built environment may influence exposures to these risk factors in different ways and thus differentially affect the health of urban populations.Objective: We investigated the built environment’s association with air pollution and physical inactivity, and estimated attributable health risks.Methods: We used a regional travel survey to estimate within-urban variability in physical inactivity and home-based air pollution exposure [particulate matter with aerodynamic diameter ≤ 2.5 μm (PM2.5), nitrogen oxides (NOx), and ozone (O3)] for 30,007 individuals in southern California. We then estimated the resulting risk for ischemic heart disease (IHD) using literature-derived dose–response values. Using a cross-sectional approach, we compared estimated IHD mortality risks among neighborhoods based on “walkability” scores.Results: The proportion of physically active individuals was higher in high- versus low-walkability neighborhoods (24.9% vs. 12.5%); however, only a small proportion of the population was physically active, and between-neighborhood variability in estimated IHD mortality attributable to physical inactivity was modest (7 fewer IHD deaths/100,000/year in high- vs. low-walkability neighborhoods). Between-neighborhood differences in estimated IHD mortality from air pollution were comparable in magnitude (9 more IHD deaths/100,000/year for PM2.5 and 3 fewer IHD deaths for O3 in high- vs. low-walkability neighborhoods), suggesting that population health benefits from increased physical activity in high-walkability neighborhoods may be offset by adverse effects of air pollution exposure.Policy implications: Currently, planning efforts mainly focus on increasing physical activity through neighborhood design. Our results suggest that differences in population health impacts among neighborhoods are similar in magnitude for air pollution and physical activity. Thus, physical activity and exposure to air pollution are critical aspects of planning for cleaner, health-promoting cities.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.