Diagnostics for Determining Influential Species in the Chemical Mass Balance Receptor Model

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Geochemistry

Background:

• Airborne particulate matter (PM) source apportionment is crucial for environmental and health assessments.
• The Chemical Mass Balance (CMB) model is widely used to estimate PM contributions from various sources.
• Identifying influential chemical species is vital for accurate CMB model application and error reduction.

Purpose of the Study:

• To evaluate existing and develop new diagnostics for the CMB model.
• To compare the performance of different diagnostics using simulated data with controlled errors.
• To identify the most effective diagnostic for improving CMB model accuracy and identifying influential species.

Main Methods:

• Investigated standard regression diagnostics, including single-row deletion diagnostics.
• Developed novel nondeletion diagnostics based on the pseudo-inverse of the source composition matrix.
• Generated simulated datasets to assess diagnostic responses to random error.
• Introduced and validated the modified pseudoinverse matrix (MPIN) diagnostic.

Main Results:

• The modified pseudoinverse matrix (MPIN) diagnostic was identified as the superior choice for CMB model application.
• MPIN integrates information from both deletion and nondeletion diagnostics.
• MPIN can identify influential species using only source profiles, enabling proactive analysis.
• Established thresholds for MPIN values to classify elements as influential (1-0.5), ambiguous (0.3-0.5), or noninfluential (≤0.3).

Conclusions:

• The MPIN diagnostic offers a comprehensive approach to identifying influential chemical species in CMB modeling.
• MPIN facilitates pre-sampling analysis and targeted remedial actions for influential species.
• Recommendations are provided for the interpretation and application of MPIN within CMB software.
• The MPIN diagnostic enhances the reliability and accuracy of particulate matter source apportionment studies.