Enhanced radiative cooling by large aerosol particles from wildfire-driven thunderstorms

Abstract

Large wildfires can generate pyrocumulonimbus (pyroCb) clouds, injecting massive quantities of smoke aerosols into the upper troposphere and lower stratosphere (UT/LS), where they persist for months and affect climate. The radiative effects of pyroCb aerosols, however, remain poorly understood because of limited direct measurements. Here, we present in situ aircraft measurements of 5-day-old pyroCb smoke, addressing a critical observational gap in aerosol evolution from freshly emitted to weeks-to-months-aged states. The sampled smoke primarily contained unusually large aerosol particles (500 to 600 nanometers in diameter), formed through cloud processing and efficient coagulation in the UT/LS. Compared to smaller particles in typical non-pyroCb smoke, these large particles increase outgoing radiation by 30 to 36%, substantially enhancing atmospheric radiative cooling. Climate models may greatly underestimate this cooling effect by assuming smaller aerosol sizes for pyroCb smoke. As pyroCb events become more frequent, accurately representing their aerosol properties is essential for improving climate projections.

Fig. 1.. Satellite observations of pyroCb activity and smoke plume transport.

(A) Active fire in New Mexico on 16 June 2022 and the resultant pyroCb event. Purple dots indicate active fire locations. (B to D) Day 1 smoke plumes shown in GOES Advanced Baseline Imager (ABI) True Color, OMPS Aerosol Index, and GOES Cirrus Channel imagery. The white contour lines in (B) outline the smoke regions where the aerosol index is ≥1, as shown in (C). (E and F) Smoke plumes 3 and 5 days after pyroCb injection in GOES Cirrus Channel imagery. The smoke plumes are highlighted within red ovals. The black square in (B) represents the area of the image where the pyroCb occurred on day 0 in (A). The aircraft’s launch location in Salina, Kansas, is denoted by a yellow star, and the flight track on day 5 is shown in blue in (F).

Fig. 2.. Size distribution of wildfire smoke aerosols from the 2022 New Mexico pyroCb event.

(A) Heatmap illustrating the time series of aerosol size distribution. Color contours denote aerosol concentration in units of dN/dlogDp (cm⁻³) as a function of diameter (left axis). The flight track altitude is overlaid as a brown curve (right axis). Smoke plumes are numbered from 1 to 3 on the basis of descending aerosol concentration. (B) Composition-resolved volume size distribution within the three smoke plumes depicted in (A). The panel shows four particle types: internally mixed sulfate and organics, biomass burning, mineral dust, and others. This analysis incorporates chemical composition data from 697 particles measured by the PALMS-NG instrument, combined with size distribution data from the DPOPS. (C) Aerosol number size distributions for the three smoke plumes depicted in (A), as well as for background conditions at the same altitude, between 18:07 and 18:25 UTC. The gray-shaded area indicates the typical mode diameter range for non-pyroCb smoke aerosols (200 to 300 nm). The lower cutoff size for the DPOPS instrument is indicated by the vertical gray dashed line. All size distributions are under ambient conditions.

Fig. 3.. Modeled coagulation growth of aerosols in the UT/LS.

(A) Growth of a unimodal size distribution with an initial size mode of 130 nm. (B) Growth of a bimodal size distribution with initial size modes at 130 and 260 nm. The second, larger size mode represents the effects of cloud processing on particle growth during convection. Both simulations were carried out for 5 days, and the figure depicts the evolution of the size distribution over this period. Both simulations also account for aerosol coagulation and dilution under upper tropospheric conditions using a realistic dilution rate for the UT/LS. Table S1 summarizes the parameters used in the simulations. All size distributions are under ambient conditions.

Fig. 4.. TOA instantaneous radiative forcing of sampled pyroCb aerosols compared to typical non-pyroCb smoke and modeled wildfire aerosols.

(A) TOA radiative forcing (shortwave and longwave) for the three measured pyroCb smoke plumes. Error bars on shortwave forcing reflect uncertainties across five different RI scenarios. The mass concentration is calculated assuming a particle density of 1400 kg m⁻³ (27). (B) Total (shortwave + longwave) TOA radiative forcing for the three pyroCb smoke plumes compared to typical non-pyroCb smoke aerosols with a 250-nm mode diameter and a modal width of 1.4. Non-pyroCb cases are adjusted to the same mass as the corresponding pyroCb cases (i.e., mass-equivalent). Error bars reflect uncertainties across five different RI scenarios (fig. S8). Dotted lines connect paired cases to aid comparison within the same RI scenario. The red “RF ratio” values show the ratio of the radiative forcing of pyroCb smoke to that of the corresponding non-pyroCb smoke, with the average displayed above and the range in brackets. (C) Total TOA radiative forcing for aerosol distributions with number-mode diameters of 100, 150, 200, 250, 300, 350, 400, and 600 nm (with corresponding modal widths of 1.55, 1.5, 1.45, 1.4, 1.35, 1.3, 1.25, and 1.2, respectively), as well as measured pyroCb smoke 1 with a 523-nm number-mode diameter. The progressively narrower widths with increasing diameter are consistent with coagulation being the dominant growth mechanism. The gray-shaded area shows the typical wildfire aerosol sizes in regional or global climate models (–39), while the pink-shaded area represents the size range of pyroCb aerosols measured in this work and the hypothetical 600-nm case representing more extreme pyroCb conditions. Error bars reflect uncertainties from five RI scenarios. All radiative forcing results shown in this figure are derived from the offline radiative transfer calculations.

Fig. 5.. Changes in AOD at a 600-nm wavelength and instantaneous TOA shortwave radiative forcing from large pyroCb aerosols in regional meteorology-chemistry model simulations.

Panels show the simulated differences in (A) AOD at a 600-nm wavelength and (B) instantaneous TOA clear-sky shortwave radiative forcing between pyroCb smoke aerosols (500-nm mode diameter, 1.3 modal width; pyroCb experiment) and default wildfire aerosols in the model (140-nm mode diameter, 1.6 modal width; control experiment), averaged over 16 to 21 June 2022. Simulations were performed using the WRF-GC model, an online two-way coupling of the WRF regional meteorological model and the GC chemical transport model (47). Both pyroCb and control experiments used the same biomass burning emissions from the Global Fire Assimilation System (GFAS) inventory. Emissions from the New Mexico fires contributed to 73% of the total fire emissions in the study domain from 16 to 21 June, with emissions on 16 June alone accounting for 67% of the total during that period (fig. S12). In the pyroCb experiment, all biomass burning aerosols are treated as pyroCb aerosols upon emission, providing an upper limit to the simulated differences.