High-spatiotemporal-resolution mapping of PM2.5 traffic source impacts integrating machine learning and
Lingling Lv1, Peng Wei2, Jingnan Hu2
1State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Sciences, Beijing 100012, PR China; School of the Environment and Safety Engineering, Jiangsu University, Zhenjiang 212013, PR China.
Environment International
|January 9, 2024
Summary
This study introduces a new method combining machine learning and emission data to accurately map fine particulate matter (PM2.5) pollution from traffic sources. It identifies pollution hotspots for better air quality management.
Area of Science:
- Environmental Science
- Atmospheric Chemistry
- Data Science
Background:
- Traffic emissions significantly contribute to fine particulate matter (PM2.5) pollution.
- Existing receptor models struggle with high-resolution PM2.5 source apportionment due to limited observational data.
Purpose of the Study:
- To develop a novel method for accurately estimating PM2.5 traffic source impacts with high spatiotemporal resolution.
- To address limitations in current source apportionment techniques, particularly in data-scarce regions.
Main Methods:
- Integrated machine learning (Extreme Gradient Boosting) with chemical transport models to optimize pollutant concentration fields.
- Employed an emission-based Integrated Mobile Source Indicator (IMSI) method for PM2.5 traffic source apportionment.
- Utilized multisource data for model development and validation.
Main Results:
- The Extreme Gradient Boosting model achieved high accuracy (R values 0.87-0.92) in predicting traffic-related pollutants (NO2, CO, EC), reducing errors by 50-67%.
- The IMSI method successfully mapped daily PM2.5 traffic source impacts, revealing significant spatial heterogeneity and identifying pollution hotspots.
- A strong correlation (R=0.79) was found between IMSI-derived impacts and receptor model results at a Beijing site.
Conclusions:
- The integrated approach provides a robust tool for estimating PM2.5 traffic source impacts with high spatiotemporal resolution.
- This method offers valuable insights for air pollution management, especially in regions lacking detailed PM2.5 composition data.
- Findings support the development of precise and timely air pollution control strategies.
