Source-Specific Exposure Assessment by Using Bayesian Spatial Multivariate Receptor Modeling

• Activity Number: 266 - Recent Advances in Spatial-Temporal Modeling and Its Applications
• Type: Topic Contributed
• Date/Time: Tuesday, August 4, 2020 : 1:00 PM to 2:50 PM
• Sponsor: Korean International Statistical Society
• Author(s): Eun Sug Park*
• Companies: Texas A&M Transportation Institute
• Keywords: Nonnegative factor analysis models ; Latent variable models; Model non-identifiability; Source apportionment; Multiple air pollutants; Multiple monitoring sites

Abstract

A major difficulty with assessing source-specific health effects is that source-specific exposures cannot be measured directly; rather, they need to be estimated by a source apportionment method such as Positive Matrix Factorizations (PMF). The uncertainty in estimated source-specific exposures (source contributions) has been largely ignored in previous studies. Also, most previous studies examining health effects of source-specific air pollution have used monitor-specific source contribution estimates as an indicator of individual exposures, which are subject to non-ignorable spatial misalignment error. We present a Bayesian spatial multivariate receptor modeling (BSMRM) approach that incorporates spatial correlations in multisite multipollutant data into the estimation of source composition profiles and contributions. The BSMRM can predict unobserved source-specific exposures at any location and time along with their uncertainty, which can greatly reduce spatial misalignment errors. The proposed method is illustrated with real multipollutant data obtained from multiple monitoring sites. Maps of estimated source-specific exposures are also presented.