A regionalized national universal kriging model using Partial Least Squares regression for estimating annual PM2.5 concentrations in epidemiology

Abstract

Many cohort studies in environmental epidemiology require accurate modeling and prediction of fine scale spatial variation in ambient air quality across the U.S. This modeling requires the use of small spatial scale geographic or “land use” regression covariates and some degree of spatial smoothing. Furthermore, the details of the prediction of air quality by land use regression and the spatial variation in ambient air quality not explained by this regression should be allowed to vary across the continent due to the large scale heterogeneity in topography, climate, and sources of air pollution. This paper introduces a regionalized national universal kriging model for annual average fine particulate matter (PM2.5) monitoring data across the U.S. To take full advantage of an extensive database of land use covariates we chose to use the method of Partial Least Squares, rather than variable selection, for the regression component of the model (the “universal” in “universal kriging”) with regression coefficients and residual variogram models allowed to vary across three regions defined as West Coast, Mountain West, and East. We demonstrate a very high level of cross-validated accuracy of prediction with an overall R2 of 0.88 and well-calibrated predictive intervals. In accord with the spatially varying characteristics of PM2.5 on a national scale and differing kriging smoothness parameters, the accuracy of the prediction varies by region with predictive intervals being notably wider in the West Coast and Mountain West in contrast to the East.

The Seasonal and Diurnal Influence of Surrounding Land Use on Temperature: Findings from Seoul, South Korea

“…The PLS regression model is adopted in studies that examine the impact of land use on various environmental conditions [83][84][85]. The model is a robust multivariate regression method that predicts the dependent variable by extracting from the independent variables a set of orthogonal factors called latent variables which have the best explanatory power [86].…”

Ambient Air Pollution Exposure and Incident Adult Asthma in a Nationwide Cohort of U.S. Women

“…Participant home addresses were geocoded using ArcMap version 10 (ESRI, Redlands, CA), and the residential locations ( Figure 2) were used for prediction of outdoor pollutant concentrations. Ambient pollutant concentrations were estimated using geographic covariates in a universal kriging regression on annual averages derived from a national network of air pollution-monitoring stations (using U.S. Environmental Protection Agency [EPA] reference methods) in a model described previously (16 …”

Evaluation of the potential of retrofitting a coal power plant to oxy-firing using CFD and process co-simulation

“…Detailed introductions to the Kriging interpolation procedure can be found in many existing publications [35][36][37][38]. In the Kriging system, the response Y(x) is expressed as the following regression model: x ,x ,...,x n = x with x R p i ∈ , P is the number of the design variables.…”

“…However, this approach is quite complicated in its mathematical form. Kriging is a response surface method for spatial data interpolation and is widely used in the area of geology and aerology research [35][36][37][38]. Kriging uses spatial relationships of known points and their distribution to predict an unknown point, and it is a statistical, unbiased, and minimum variance predictor in which errors can be determined at specified points.…”