Incorporating Time-Dependent Source Profiles Using the Dirichlet Distribution in Multivariate Receptor Models

Abstract

Multivariate receptor modeling is used to estimate profiles and contributions of pollution sources from concentrations of pollutants such as particulate matter in the air. The majority of previous approaches to multivariate receptor modeling assume pollution source profiles are constant through time. In an effort to relax this assumption, this article uses the Dirichlet distribution in a dynamic linear receptor model for pollution source profiles. The receptor model developed herein is evaluated using simulated datasets and then applied to a physical dataset of chemical species concentrations measured at the U.S. Environmental Protection Agency's St. Louis-Midwest supersite. Supplemental materials to this articles are available online.

Pollution source direction identification: embedding dispersion models to solve an inverse problem

“…() and Heaton et al . (). For examples of recent statistical developments in the modeling of wind‐related phenomena generally, see Herring and Genton () and Kestens and Teugels ().…”

“…(), and Heaton et al . (). The species we consider were measured between May 2001 and May 2003 and are a collection of metals typically associated with point sources.…”

“…Alternatively, the identification of a municipal incinerator is supported when large estimated contributions from the hypothesized incinerator occur on days where winds transport air from the known direction of an incinerator. Additionally, source direction identification can be an important tool for constructing prior distributions in a Bayesian PSA approach such as those considered by Lingwall et al (2008) and Heaton et al (2010). For examples of recent statistical developments in the modeling of wind-related phenomena generally, see Herring and Genton (2010) and Kestens and Teugels (2002).The complicated nature of atmospheric dispersion physics has been an obstacle to the development of sound statistical procedures for identifying source directions.…”

Multivariate Receptor Models for Spatially Correlated Multipollutant Data

“…Spiegelman and Park (2007) performed a jackknife evaluation of the uncertainty of the estimates of the source contribution and source composition matrices as a way of incorporating dependence in air pollution data into estimation. Lingwall, Christensen, and Reese (2008) developed Dirichlet-based Bayesian multivariate receptor modeling, and Heaton, Reese, and Christensen (2010) proposed a Dirichlet process model to incorporate time-varying source profiles in multivariate receptor models. Nikolov et al (2011) extended the multiplicative factor analysis model proposed by Wolbers and Stahel (2005) by imposing mixed models on the latent source contributions to include the covariate effects and to adjust for temporal correlation in the source contribution.…”

“…A comprehensive review of the field of receptor modeling can be found in the articles by Hopke (1991Hopke ( , 2003. Traditionally, multivariate receptor models have been used to resolve the observed air pollutant mixtures into contributions from individual sources (or source types) based on time series of multiple (or multivariate) air pollutants, such as volatile organic compounds (VOCs) or specific metal constituents of fine particulate matter (PM 2.5 ), at a receptor site (see e.g., Hopke 1985;Henry 1997a;Park, Guttorp, and Henry 2001;Wolbers and Stahel 2005;Hopke et al 2006;Heaton, Reese, and Christensen 2010).…”