PNNL

Abstract

In this study, we demonstrate the utility of conditional probability functions (CPFs), potential source contribution functions (PSCFs), and hierarchical clustering analysis to identify the source region and transport pathways of hydrocarbons measured at five photochemical assessment monitoring stations (PAMS) near the Houston ship channel from June to October 2003. Over 50 volatile organic compound (VOC) concentrations were measured on the hourly collected samples. Routine surface observations of wind directions measured at each of the receptor sites were used extensively. We show that VOCs with similar CPF patterns likely have common transport pathways. This was established with the multivariate technique, which uses the hierarchical clustering analysis to allow clusters of groups of VOCs to form with similar CPF patterns. This method revealed that alkenes, and in particular those with geometric isomers such as cis-/trans-2-butene and cis-/trans-2-pentene, have similar CPF patterns. The alkane isomers often show CPF patterns among themselves, and similarly, aromatic compounds often show similar patterns among themselves too. We also show how trajectory information can be used in conjunction with the PSCF analysis to produce a graphic analysis suggesting specific source areas for a given VOC. The use of these techniques in the chemically and meteorologically complex environment of Houston, Texas, suggests its further utility in other areas with relatively simpler conditions.