Updates to Dusttracker-AI: Inclusion of Post-Fire Burn Scar Training Cases

Atmospheric dust, composed of tiny (~0.1-1.0 μm), dry particles suspended in the air, poses significant threats to human health, transportation, and agriculture. Recent research by Yu and Ginoux (2022) revealed that over half of large wildfires globally are followed by increased dust emissions. This finding aligns with previous studies demonstrating the Dust RGB’s ability to detect dust originating from post-fire burn scars. To better understand these events, investigations have examined critical factors influencing dust lofting, considering both land surface characteristics (soil type, drought severity, vegetation cover, and soil moisture) and atmospheric conditions (boundary layer instability, wind shear, and wind speed). Analysis of dust events caused by post-fire burn scars, identified through weather service advisories and validated using the Dust RGB, revealed extremely dry soil conditions (U.S Drought Monitor - moderate to severe drought) and minimal vegetation (Enhanced Vegetation Index below 0.2) in these areas, making them highly susceptible to dust lofting. Burn scars can remain dust sources for weeks to months due to slow vegetation recovery. Given the substantial risks associated with dust emitted from burn scars, our ongoing research aims to improve the Dusttracker-AI algorithm developed at the NASA Short-term Prediction Research and Transition (SPoRT) Center to identify these small-scale events. To address this, we have expanded the model's training data to include dust cases from southern Nebraska, aiming to improve its ability to detect burn scar-generated dust and enhance forecaster awareness of this hazard.