E47 FireWx Wildfire Aerosol Concentrations and Their Impact on Severe Weather in North America: Historical Analysis and AI Predictions

Tuesday, 14 January 2025
Hall C (New Orleans Ernest N. Morial Convention Center)
Mayra Ivelisse Oyola-Merced, University of Wisconsin-Madison, Madison, WI; and K. Wilson, S. Kluge, M. McKellar, P. A. Bartels, and D. S. Sproles

The frequency and intensity of wildfires in North America have escalated due to climate change and human activities, leading to significant increases in biomass burning aerosol concentrations. This study analyzes the trends and patterns of wildfire activity and associated aerosol emissions, specifically focusing on black carbon (BC) and organic carbon (OC) concentrations over the United States and Canada. Using 43 years of MERRA-2 reanalysis data (1980-2023) and over two decades of MODIS Fire Radiative Power (FRP) observations, we revisit the climatology of BC and OC and explore their temporal and spatial distribution.

Our analysis reveals a substantial increase in aerosol concentrations over the past decades, particularly during the summer and fall, driven by intensified wildfire activity. The study highlights significant regional hotspots, with the western United States and Canada showing the most pronounced increases. We also examine the repercussions of the 2023 wildfire season, which exhibited record-breaking aerosol emissions and FRP anomalies.

In addition, we investigate the potential influence of transported smoke on severe weather events over the last 20 years. The study specifically explores how smoke impacts critical ingredients for severe weather formation, such as Convective Available Potential Energy (CAPE), Convective Inhibition (CIN), Lifted Condensation Level (LCL), and wind shear. Our findings suggest that transported smoke can significantly modulate these parameters, potentially altering the intensity and occurrence of severe weather.

Finally, we employ a hybrid machine learning approach, integrating Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks, to forecast future biomass burning activities and aerosol concentrations for the next decade. The model demonstrates strong predictive capabilities, suggesting a continued upward trend in aerosol emissions, which poses ongoing challenges for air quality and public health in the region.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner