A framework integrating affinity propagation algorithm and spatial bivariate analysis for enhanced identification and...

Area of Science:

• Environmental Science
• Geochemistry
• Spatial Analysis

Background:

• Accurate identification of soil heavy metals (SHMs) sources and their spatial distribution is critical for effective pollution mitigation.
• Traditional receptor models face challenges in precisely categorizing sources and determining dispersion trends.

Purpose of the Study:

• To develop and validate a comprehensive framework for optimizing the traditional approach to tracing SHMs sources in industrial regions.
• To accurately locate SHMs source areas and identify their dispersion tendencies.

Main Methods:

• A novel framework integrating a receptor model with random forest (RF) for source apportionment.
• Utilizing bivariate local indicator of spatial association (BLISA) combined with the affinity propagation (AP) algorithm for source localization and dispersion analysis.

Main Results:

• SHMs were attributed to mixed sources: equipment manufacturing/agriculture (59.0%), geological background (30.5%), and heavy industry emissions (10.5%).
• Soil Cd and Pb sources were near specific industries, exhibiting multi-site diffusion influenced by proximity.
• Cr, Cu, and Zn sources concentrated in urban industrial zones, transitioning from point to nonpoint sources, influenced by industrial agglomeration.

Conclusions:

• The enhanced framework accurately identifies SHMs source locations and dispersion patterns.
• This approach significantly improves regional soil pollution management strategies.