Pyrocb Smoke Rises and Persists in the Global Stratosphere: Constraints on Injected Smoke Mass, Black Carbon Abundance, and Ozone Reaction Rates with Organic Coatings

Intense wildfire-driven thunderstorms, known as pyrocumulonimbus (pyroCb) injected large amounts of smoke into the stratosphere over British Columbia, Canada on August 12, 2017. While many smoke properties, including composition, morphology, lifetime, optical properties, and ozone uptake rate, were poorly known in the past, for this pyroCb event, optical measurements from satellites, balloon-borne particle counters, and lidars provided excellent constrains on various smoke properties. Satellite data from Stratospheric Aerosol and Gas Experiment III on International Space Station (SAGE III-ISS) and the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) for the first time clearly showed pyroCb smoke rising from 12 km to 22 km within a few weeks after the initial injection. In this study we used a sectional aerosol model coupled with a climate model (CESM-CARMA) to simulate the temporal and spatial distribution of the smoke. Constrained by observations, our simulations suggest that the emitted smoke mass (BC with organic coating) was about 0.5 Tg, the BC mass fraction of the smoke was around 2%, the smoke particles were likely fractals, and the reaction probability (unitless) between ozone and the organic coating was about 10^-6. The simulations show the smoke-induced heating leads to vertical transport of relatively ozone-poor air creating a local ozone minimum, which was 50% lower than background. The simulations show that the 2017 August pyroCb smoke warmed the stratospheric air locally by 7 K. Our study provides strong support for model simulations of nuclear winter, in which smoke from burning targets is injected into the stratosphere and then self-lofted far above the tropopause, producing multi-year lifetimes for the smoke. It also shows that to model the climate response to smoke, black carbon with a thick organic coating would give accurate results.