Wildfires Now Simulated to Spawn Their Own Storms

Scientists have achieved a major breakthrough: they have successfully simulated wildfire-driven thunderstorms within global Earth system models—capturing key fire-weather feedbacks that until now eluded climate and fire scientists. The advance marks the first time models have convincingly generated pyrocumulonimbus clouds triggered by extreme wildfires, reliably reproducing real cases such as California’s 2020 Creek Fire and parts of the 2021 Dixie Fire.

The research team, led by Ziming Ke of the Desert Research Institute, embedded a multiscale wildfire–atmosphere interaction framework into a climate model, enabling the simulation of smoke, moisture, heat fluxes, and convection dynamics in a coupled fashion. Their model reproduced the observed timing, height, and intensity of the Creek Fire’s thunderhead, and also simulated multiple pyroCb cells from the Dixie Fire under varied meteorological conditions. Key to success was accurately representing water vapour transport from the surface to higher altitudes—a process that earlier models failed to capture.

Traditional climate models lacked the resolution or physics to capture the formation of pyroCbs, leaving a blind spot in understanding how extreme wildfires influence atmospheric processes and climate. The new approach bridges that gap by refining model meshes in convection-permitting domains and dynamically coupling fire emissions, atmospheric heating, and cloud microphysics. As one member of the team put it, the work represents a “first-of-its-kind breakthrough” in Earth system modelling.

Parallel efforts have pushed the frontier further. A study by Qing Wang and co-authors employed fully coupled, high-resolution simulations to probe the life cycle of a pyroCb, identifying the competing influence of fuel moisture and a mechanism dubbed Self-Amplifying Fire-Induced Recirculation. SAFIR refers to precipitation-driven downdrafts that spin back into the fire, intensifying it under weak wind conditions. The study offers a new mechanistic lens through which to predict when a pyroCb might intensify sharply.

The need for such modelling gains urgency in a climate era where wildfires are becoming more frequent, larger, and more intense. Researchers estimate that between tens to hundreds of pyroCb events occur globally each year—with expectations that this frequency will rise as climate change strengthens fire regimes. PyroCb clouds can loft smoke and aerosols into the stratosphere, alter cloud albedo, disrupt ozone chemistry, and feed back on weather and radiative forcing across regions.

The modelling breakthrough also enables fresh insight into extreme event coupling. The team’s simulations suggest that in fire-prone regions, detecting early thermodynamic and moisture thresholds might hint at looming pyroCb onset—information that could improve forecasts and emergency response. In effect, the models now allow scientists to explore “what if” scenarios: how a given fire might evolve under variable humidity, wind shear, or terrain influences.