An increasing trend of larger and more intense wildfires emerges in recent observations. The extreme events known to generate pyrocumulonimbus (pyroCb) firestorms can inject a large amount of smoke aerosols into the stratosphere. The smoke plume stays longer (e.g., weeks to months) in the stratosphere than in the troposphere, and hence can lead to climate impacts by imposing negative radiative forcing to the Earth system at hemispheric scale and by changing stratospheric ozone and circulation. Although their importance is widely recognized, representing pyroCb firestorms in global Earth system models (ESMs) remains a challenge due to limitations such as the coarse grid spacing, missing important physical processes, and uncertainties in fire heat and aerosol emission estimations.

In this study, we built a new global multiscale wildfire simulation framework by leveraging on the new US Department of Energy (DOE) Energy Exascale Earth System Model (E3SM) capabilities (i.e., non-hydrostatic dynamic core, regionally refined mesh (RRM; 3 km at California and 100 km outside), and interactive chemistry/aerosol) and the satellite-based high-resolution (hourly, 500 m) wildfire observational dataset. We simulated the 2020 Creek Fire with this E3SM California RRM (CARRM) framework. The initial model sensitivity test suggests that E3SM-CARRM can produce the pyroCb with a cloud top up to ~15 km as observed (see Figure). Further analyses reveal that fire-induced vertical moisture transport and the high-resolution wildfire emission dataset are key to capture the more realistic pyroCb. Furthermore, we will elaborate on the fire-induced strong surface convergence, lightning, and vortex in this case study. Our simulations here will help better understand the pyroCb mechanisms, sensitivities to different driving factors, as well as the resulting climate impacts.

Figure: Altitude-time plots of radar reflectivity from (left) NEXRAD observations (Lareau et al. (2022) Fig. 7a) and (right) the E3SM-CARRM simulation. Note that the observational radars are sensitive to particulate ash and debris at millimeter to centimeter scale, whereas the model diagnoses the radar reflectivity of clouds. This comparison is only good for qualitative purpose.