Multivariate spatial patterns of ambient PM_2.5 elemental concentrations in Eastern Massachusetts

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Spatial Analysis

Background:

• Spatial variations in fine particulate matter (PM2.5) composition are critical for effective emissions control and public health strategies.
• Previous research has primarily focused on spatial modeling for exposure assessment, with limited use of multivariate clustering for particle composition patterns.

Purpose of the Study:

• To identify and characterize spatial patterns of ambient PM2.5 elemental concentrations in Eastern Massachusetts using a multivariate clustering approach.
• To investigate the influence of air pollution sources and geodemographic variables on these spatial patterns.

Main Methods:

• Applied multivariate clustering to analyze spatial patterns of 11 elemental components (S, K, Ca, Fe, Zn, Cu, Ti, Al, Pb, V, Ni) in ambient PM2.5.
• Utilized land use, population density, and daily traffic data to characterize identified clusters.
• Employed R-squared values to quantify the effectiveness of variables in discriminating between site clusters.

Main Results:

• Identified varying numbers of spatial clusters for different PM2.5 elements, ranging from 2 clusters (Fe, Zn, V, Ni) to 12 clusters (K).
• Found significant variations in cluster patterns across different PM2.5 components.
• Population density, land use, and traffic emerged as significant variables for characterizing clusters, with high R-squared values indicating strong discriminatory power.

Conclusions:

• The multivariate clustering approach effectively reveals distinct spatial patterns in PM2.5 elemental composition.
• Geodemographic factors and land use significantly influence and help characterize these spatial variations.
• Findings enhance the ability to model source emissions and pollution regimes, improving both between- and within-area variability assessments.