Montana Statewide Google Earth Engine-Based Wildfire Hazardous Particulate (PM2.5) Concentration Estimation

Area of Science:

• Environmental Science
• Public Health
• Remote Sensing

Background:

• Wildfires in Montana threaten public health via smoke and air pollution.
• Existing ground-based air quality sensors are sparse, limiting accurate air quality prediction.
• Elevated particulate matter (PM2.5) is linked to severe health issues, including mortality and chronic diseases.

Purpose of the Study:

• To develop a predictive model for estimating daily PM2.5 concentrations in Montana.
• To utilize satellite-derived data and machine learning within the Google Earth Engine environment.
• To create high-resolution (1 km) daily PM2.5 concentration maps from 2012-2023.

Main Methods:

• A random forest predictive model was developed using Google Earth Engine.
• Input variables included satellite-based aerosol optical depth (AOD), dewpoint temperature (DPT), relative humidity (RH), wind speed (WIND), wind direction (WDIR), pressure (PRES), and planetary-boundary-layer height (PBLH).
• Model validation was performed using 10-fold cross-validation (R²=0.572, RMSE=9.98 μg/m³) and held-out data (R²=0.487, RMSE=10.53 μg/m³).

Main Results:

• A validated model successfully estimated daily PM2.5 concentrations across Montana.
• Generated daily PM2.5 concentration maps at 1 km resolution for the period 2012-2023.
• An accessible application was developed using Google Earth Engine to display these maps.

Conclusions:

• The study provides valuable spatiotemporal insights into PM2.5 pollution trends in Montana.
• The developed PM2.5 concentration maps can inform community mitigation strategies.
• This tool aids in minimizing public exposure to hazardous wildfire smoke and air pollution.