Exposure to source-specific VOCs around the Houston Ship Channel.

Atmospheric VOCs are emitted by both anthropogenic and biogenic sources leading to concentrations that can react to form secondary pollutants like ozone (O3) and secondary organic aerosol (SOA). Both VOCs and their reaction products have impacts on human and ecosystem health. Thus, understanding the sources that lead to population exposures across an urban area provides a basis for understanding the relationships between source emissions and adverse environmental impacts. This study explores the sources of VOCs in the area of the Houston Ship Channel in Texas using hourly measurements from 8 Assessment Monitoring Stations (PAMS) sites in 2012 and 2023. Spatial and temporal variations in ambient VOC levels were assessed using coefficients of divergence (CoD) and squared Pearson correlation coefficients. Seven source types were resolved, six of which were common to both years, reflecting a broadly consistent source structure across the decade despite changes in relative contributions. In 2012, a distinct biogenic source dominated by isoprene was identified. However, in 2023, strong covariance between isoprene and industrial olefins precluded resolution of a standalone biogenic factor. Instead, a polymer production-related source was resolved, consistent with the spatial concentration of petrochemical and polymer manufacturing activities and the limitations of dispersion normalization alone for highly reactive compounds. In 2012, petrochemical industries were the largest contributor to all 8 sites. However, in 2023, it was the smallest contributor except at one site with only 7 months of data. For most sites, no source type was dominant indicating a change in the industrial processes within the Ship Channel area. CoD values indicate substantial spatial heterogeneity in source-specific VOC concentrations given no CoD values below 0.20 and a few pairs with weak homogeneity. Several source types had r2 > 0.5 suggesting similar temporal patterns with fewer higher correlations in 2023 than in 2012. Thus, the patterns of CoD and r2 values support the assessment that exposures were primarily from local sources.