Within-City Variation in Ambient Carbon Monoxide Concentrations: Leveraging Low-Cost Monitors in a Spatiotemporal...

Area of Science:

• Environmental Health Sciences
• Epidemiology
• Air Quality Monitoring

Background:

• Ambient carbon monoxide (CO) exposure is linked to cardiovascular disease, but epidemiological data is limited.
• Existing regulatory monitors lack the resolution to capture fine-scale CO variations.

Purpose of the Study:

• To create a daily, high-resolution urban-scale ambient CO exposure prediction model.
• To assess the impact of novel data sources on CO modeling accuracy.

Main Methods:

• Developed a spatiotemporal statistical algorithm using regulatory and low-cost monitor data in Baltimore, MD.
• Incorporated high-resolution meteorological data, satellite imagery, and copollutant concentrations.
• Validated model performance using spatial and temporal cross-validation.

Main Results:

• Achieved strong spatial (R² $$) and temporal (R² $$) prediction accuracy.
• Identified CO hotspots linked to traffic, population density, and industrial sources.
• Observed rapid CO concentration decreases away from major roadways.

Conclusions:

• The spatiotemporal modeling framework is robust and reliable, even without novel parameters.
• This approach can be extended to resource-limited settings with increasing availability of low-cost air monitors.
• Facilitates comprehensive epidemiological research on CO exposure and health outcomes.