Search

Deep convection induced by large forest fires is an efficient mechanism for transport of aerosol particles and trace gases into the upper troposphere and lower stratosphere (UT/LS). For many pyro-cumulonimbus clouds (pyroCbs) as well as other cases of severe convection without fire forcing, radiometric observations of cloud tops in the thermal infrared (IR) reveal characteristic structures, featuring a region of relatively high brightness temperatures (warm center) surrounded by a U-shaped region of low brightness temperatures. We performed a numerical simulation of a specific case study of pyroCb using a non-hydrostatic cloud resolving model with a two-moment cloud microphysics parameterization and a prognostic turbulence scheme. The model is able to reproduce the thermal IR structure as observed from satellite radiometry. Our findings establish a close link between the observed temperature pattern and small-scale mixing processes atop and downwind of the overshooting dome of the pyroCb. Such small-scale mixing processes are strongly enhanced by the formation and breaking of a stationary gravity wave induced by the overshoot. They are found to increase the stratospheric penetration of the smoke by up to almost 30 K and thus are of major significance for irreversible transport of forest fire smoke into the lower stratosphere.

A fire storm that occurred on 28 May 2001 and devastated the town of Chisholm, ~150 km north of Edmonton, Alberta, induced a violent fire-invigorated cumulonimbus cloud. This pyro-cumulonimbus (pyro-Cb) had overshooting tops of 2.5–3 km above the tropopause, and injected massive amounts of smoke into the lower stratosphere. Fortunately, this event occurred under good coverage of radar, rain gauge, lightning and satellite measurements, which allowed in-depth documentation of the event, and gave us an opportunity to study the cloud top morphology and microstructure, precipitation and cloud electrification of the pyro-Cb. The combination of heat and smoke created a cloud with extremely small drops, which ascended rapidly in violent updrafts. There appeared to be little freezing up to the homogeneous freezing isotherm level of −38°C. A cloud with such small and short-lived highly supercooled drops is incapable of producing precipitation except for few large graupel and hail, which produced the observed radar echoes and charged the cloud with positive lightning. The small cloud drops froze homogeneously to equally small ice particles, for which there is no mechanism to aggregate into precipitation particles, and which hence remain in the anvil. The lack of significant precipitation implies that only a small fraction of the smoke is scavenged, so that most of it is exhausted through the anvil to the upper troposphere and lower stratosphere. Comparisons with other cases suggest that a pyro-Cb does not have to be as violent as the Chisholm case for precipitation to be strongly suppressed. However, this level of convective vigor is necessary to create the overshooting updraft that injects the smoke into the lower stratosphere.

Wildland fires in boreal regions have the potential to initiate deep convection, so-called pyro-convection, due to their release of sensible heat. Under favorable atmospheric conditions, large fires can result in pyro-convection that transports the emissions into the upper troposphere and the lower stratosphere. Here, we present three-dimensional model simulations of the injection of fire emissions into the lower stratosphere by pyro-convection. These model simulations are constrained and evaluated with observations obtained from the Chisholm fire in Alberta, Canada, in 2001. The active tracer high resolution atmospheric model (ATHAM) is initialized with observations obtained by radiosonde. Information on the fire forcing is obtained from ground-based observations of the mass and moisture of the burned fuel. Based on radar observations, the pyro-convection reached an altitude of about 13 km, well above the tropopause, which was located at about 11.2 km. The model simulation yields a similarly strong convection with an overshoot of the convection above the tropopause. The main outflow from the pyro-convection occurs at about 10.6 km, but a significant fraction (about 8%) of the emitted mass of the smoke aerosol is transported above the tropopause. In contrast to regular convection, the region with maximum updraft velocity in the pyro-convection is located close to the surface above the fire. This results in high updraft velocities >10 m s−1 at cloud base. The temperature anomaly in the plume decreases rapidly with height from values above 50 K at the fire to about 5 K at about 3000 m above the fire. While the sensible heat released from the fire is responsible for the initiation of convection in the model, the release of latent heat from condensation and freezing dominates the overall energy budget. Emissions of water vapor from the fire do not significantly contribute to the energy budget of the convection.

The Chisholm forest fire that burned in Alberta, Canada, in May 2001 resulted in injection of substantial amounts of smoke into the lower stratosphere. We used the cloud-resolving plume model ATHAM (Active Tracer High resolution Atmospheric Model) to investigate the importance of different contributing factors to the severe intensification of the convection induced by the Chisholm fire and the subsequent injection of biomass smoke into the lower stratosphere. The simulations show strong sensitivity of the pyro-convection to background meteorology. This explains the observed coincidence of the convective blow-up of the fire plume and the passage of a synoptic cold front. Furthermore, we performed model sensitivity studies to the rate of release of sensible heat and water vapor from the fire. The release of sensible heat by the fire plays a dominant role for the dynamic development of the pyro-cumulonimbus cloud (pyroCb) and the height to which smoke is transported. The convection is very sensitive to the heat flux from the fire. The emissions of water vapor play a less significant role for the injection height but enhance the amount of smoke transported beyond the tropopause level. The aerosol burden in the plume has a strong impact on the microphysical structure of the resulting convective cloud. The dynamic evolution of the pyroCb, however, is only weakly sensitive to the abundance of cloud condensation nuclei (CCN) from the fire. In contrast to previous findings by other studies of convective clouds, we found that fire CCN have a negative effect on the convection dynamics because they give rise to a delay in the freezing of cloud droplets. Even in a simulation without fire CCN, there is no precipitation formation within the updraft region of the pyroCb. Enhancement of convection by aerosols as reported from studies of other cases of convection is therefore not found in our study.