Australia's Black Summer pyrocumulonimbus super outbreak reveals potential for increasingly extreme stratospheric smoke events

An intense, multi-day outbreak of pyrocumulonimbus (pyroCb) occurred during 29–31 December 2019 and 04 January 2020 in southeastern Australia. This Australian New Year Super Outbreak (ANYSO) injected roughly 1.0 Tg of smoke particle mass into the lower stratosphere, comparable to a moderate volcanic eruption. ANYSO followed the 2017 Pacific Northwest Event (PNE), which previously set the benchmark for extreme lower-stratospheric pyroCb injections, but was about three times smaller. ANYSO suggests potential for dramatic shifts in stratospheric composition, radiative balance, and regional circulation from extreme pyroCb events, motivating research into their atmospheric and climate impacts in a warming climate. PyroCbs differ from typical convection due to abundant smoke-derived nuclei, producing smaller hydrometeors, suppressed precipitation, and efficient anvil outflow that transports smoke to high altitudes. Extreme updrafts can rapidly inject smoke into the lower stratosphere, analogous to explosive volcanic eruptions. PyroCbs require hot, dry near-surface conditions to drive fire behavior and moisture aloft to support deep convection, typically peaking during late afternoon/evening in summer fire seasons of Australia, North America, and northern Asia. The introduction frames key questions: whether larger pyroCb outbreaks could rival major volcanic aerosol injections, whether their potential is increasing in a warming climate, and whether multiple large outbreaks could approach nuclear winter-like impacts. The study examines ANYSO’s unusual features (prolonged duration, strong nocturnal activity) and the evolution of the resulting stratospheric smoke plumes to identify indicators of extreme outbreaks and their climate relevance.

The study integrates multi-sensor satellite, weather radar, and meteorological analyses to quantify pyroCb characteristics, smoke injection, and plume evolution.

• Event segmentation and detection: The 38 pyroCb pulses were grouped into 18 sub-events anchored to 13 blow-up fire initiation points. PyroCb pulses were identified every 10 minutes using Himawari-8 AHI 11 µm brightness temperature (BT11) imagery with a −35 °C threshold for ice anvils. A daytime algorithm exploiting large 3.9–11 µm BT differences (>~50 K) was used to distinguish pyroCb microphysics from typical convection.
• Radar echo-top analysis: Australian BoM C-band weather radar (Bairnsdale, Captain’s Flat) echo-tops were computed as the highest altitude with ≥3 consecutive range bins >10 dBZ and >8 km altitude within search boxes aligned to AHI pyroCb anvils. Echo-top potential temperature was derived via interpolation from radiosonde profiles (primarily Melbourne, supplemented by Wagga Wagga). Echo-tops were compared to tropopause heights to determine stratospheric injections.
• Stratospheric smoke mass estimation (top-down, lidar + UVAI): CALIPSO/CALIOP 532 nm backscatter and depolarization were used 2–4 days post-cessation to define plume vertical bounds above the tropopause and compute average particle mass density Mp = βR/ε, with lidar ratio R = 40–70 sr and mass extinction coefficient ε = 3.0–6.0 m² g⁻¹. OMPS Nadir Mapper UV Aerosol Index (UVAI) with a threshold ≥15, validated against CALIOP profiles, delineated plume horizontal extent. Integrating Mp over plume depth and area yielded total stratospheric smoke mass for each phase. Sensitivities to R (up to 112 sr in ground-based lidar) and temporal sampling are discussed.
• Stratospheric smoke mass estimation (top-down, UVAI-only): Within ~24 h post-cessation, a UVAI–extinction AOD relationship (from Chisholm) was used to estimate per-pixel smoke mass (Mp = extAOD × Ap, using ε = 3.0–6.0 m² g⁻¹), integrated over UVAI ≥15 plume areas.
• Plume characterization and evolution: CALIOP vertical profiling captured initial plume depth; OMPS LP limb aerosol extinction at 997 nm (and 869 nm for long-term sAOD) characterized zonal/temporal evolution, diabatic lofting, persistence (e-folding decay times), and comparisons with volcanic plumes (e.g., Ulawun 2019, Raikoke 2019). UVAI mapped early plume areas and intensity.
• Fire activity and burned area: Blow-up fires were mapped using MODIS active fire detections (FRP >127 MW) and refined with Normalized-Difference Burn Index (NDBI) products (NSW and Victoria), Sentinel-2 imagery, and radar confirmation of stationary convective cores. Hourly burned area allocation for blow-ups used a 60/20/20% distribution around peak intensity. FRP time series from AHI via WFABBA were normalized regionally at hourly resolution.
• Bottom-up emissions and energy: Fuel consumption assumed 15,000–60,000 kg ha⁻¹ (5–20% of ~300,000 kg ha⁻¹ total biomass) over 530,000 ha of blow-up fires, with PM2.5 emission factors 16.9–38.8 g kg⁻¹ to estimate emitted smoke mass. Energy release was computed with effective heat of combustion 18,700 kJ kg⁻¹.
• Synoptic/mesoscale meteorology: Synoptic analyses used mean sea-level pressure, frontal positions, jet-level divergence, and precipitable water anomalies (vs. 1981–2010 climatology) to diagnose environments supporting pyroCb development, including prefrontal troughs and moisture transport sustaining nocturnal activity. Tropopause altitudes were mapped to assess injection levels.

• Scale and duration: ANYSO comprised 38 pyroCb pulses over 51 non-consecutive hours, with phase one lasting ~45 h and featuring nocturnal activity atypical of canonical pyroCb behavior. Ten blow-up fires (of 28 in phase one) generated 33 pulses; phase two added 5 pulses. Total blow-up fire area was ~530,000 ha.
• Stratospheric injections: Weather radar indicated 20 of 38 pulses (53%) reached the lower stratosphere; three penetrated into the stratospheric overworld (θ >380 K). Echo-top maxima reached ~16.7 km in phase two and exceeded the tropopause by up to ~2.3 km in phase one nocturnal events.
• Injected mass: Phase one injected ~0.2–0.8 Tg; phase two injected ~0.1–0.3 Tg; combined 0.3–1.1 Tg of smoke particle mass into the stratosphere, at least three times the 2017 PNE (0.1–0.3 Tg) and comparable to the initial plume from the 2008 Kasatochi eruption (0.2–0.5 Tg).
• Plume structure and area: CALIOP showed initial residual smoke layers averaging ~3.5 km depth above the tropopause for both phases. OMPS UVAI ≥15 yielded instantaneous stratospheric plume areas of ~1.6 million km² (phase one) and ~1.1 million km² (phase two). Estimated smoke mass densities were ~40–140 µg m⁻³ (phase one) and ~13–47 µg m⁻³ (phase two); UVAI maxima of 25–44 (lower than PNE’s 40–50+).
• Diabatic lofting and persistence: OMPS LP showed plume ascent from ~14–17 km to ~34 km within ~40 days, reaching unprecedented wildfire smoke altitudes and rivaling large volcanic sulfate plume heights. Stratospheric aerosol extinction (997 nm) between 20°S–90°S increased fivefold between 15–20 km shortly after ANYSO. E-folding decay times were ~120–150 days below 18 km and ~250–350 days above 19 km. Elevated extinction (>150% above background) persisted for at least 15 months.
• Stratospheric dynamics: Absorbing smoke led to diabatic heating, instability, and vortex-like, blob features rising to ~34 km (~950 K), inducing anticyclonic circulation anomalies and perturbing potential vorticity in the Southern Hemisphere stratosphere.
• Comparison with volcanoes: ANYSO’s maximum sAOD (869 nm) reached ~0.015, exceeded only by Raikoke 2019 (~0.025) since 2012; ANYSO’s lifetime (~15+ months) was comparable to Raikoke and Calbuco 2015 and longer than PNE (~10 months). ANYSO contributed substantially to 2019–2020’s 25-year high in global stratospheric aerosol loading. Mixing with residual Ulawun (2019) aerosols likely occurred, especially in the tropics.
• Fire energetics and context: The 13 blow-up fires released an estimated 1.3–5.1 × 10¹⁴ kJ (~32–127 Mt TNT; >2000× Hiroshima). Eucalyptus forests, with high flammability and spotting, facilitated extreme spread. Persistent prefrontal troughs and moisture anomalies supported continuous and nocturnal pyroCb activity, a key differentiator from prior events.
• Forecasting indicators: Slow-evolving synoptic patterns with sustained prefrontal environments (>24 h), deep dry mixed layers with mid-level moisture, strong jet dynamics, and concurrent extensive blow-up fire activity are critical indicators of potential super outbreaks.

The findings demonstrate that regional outbreaks of intense pyroCb activity can inject volcanic-scale aerosol mass into the lower stratosphere, profoundly affecting stratospheric composition, radiative balance, and dynamics. ANYSO’s prolonged duration, extensive nocturnal activity, and exceptional plume lofting validate and extend understanding of pyroCb processes and feedbacks: abundant absorbing smoke can drive diabatic heating, lofting plumes to middle-stratospheric altitudes (~34 km), altering potential vorticity and inducing anticyclonic anomalies. These results address the research questions by showing that pyroCb super outbreaks can approach or exceed moderate volcanic injections, substantially augment background stratospheric aerosol loading, and persist for over a year—thereby posing potential seasonal-to-hemispheric climate impacts. The comparisons with PNE and recent volcanoes contextualize the scale of ANYSO’s impact (sAOD, lifetime, altitude), while the identified synoptic-mesoscale combinations (persistent prefrontal troughs, jet dynamics, mid-level moisture) explain the anomalous duration and nocturnal intensity. The observed efficiency of surface-to-stratosphere smoke transport and discrepancies between top-down and bottom-up mass estimates suggest either efficient injection pathways or underestimation of emission factors during extreme events. Collectively, the results underscore the need to incorporate pyroCb outbreaks into assessments of the stratospheric aerosol system and climate feedbacks, including interactions with volcanic plumes.

This study characterizes the Australian New Year Super Outbreak (ANYSO) as a pyroCb super outbreak that produced two of the three largest stratospheric smoke plumes on record, with a combined 0.3–1.1 Tg smoke mass injection, plume ascent to ~34 km, hemispheric spread, and >15-month persistence. It reveals that slow-evolving synoptic patterns enabling sustained prefrontal environments, coupled with widespread intense fires, can drive continuous and nocturnal pyroCb activity leading to extreme stratospheric impacts on par with moderate volcanic eruptions. The work establishes quantitative indicators (UVAI thresholds, lidar/radar diagnostics, FRP/burned area metrics, synoptic patterns) for identifying and forecasting potential super outbreaks, and highlights implications for stratospheric dynamics, aerosol–radiation interactions, and validation of nuclear-winter-like smoke lofting mechanisms. Future research should: improve top-down mass estimates (better coincident CALIOP/OMPS sampling, refined lidar ratios and ε), reconcile top-down vs. bottom-up emissions through event-specific emission factors and fuel consumption, assess interactions between smoke and sulfate plumes, quantify radiative forcing and ozone-chemistry impacts, and develop predictive frameworks linking fire behavior, mesoscale/topographic effects, and synoptic forcing to pyroCb outbreak potential in a warming climate.

• Remote sensing uncertainties: CALIOP missed the core of the phase-one plume on 02 Jan, potentially underestimating backscatter and mass; phase-two sampling lacked tight CALIOP–OMPS coincidence (~13 h offset), and used the same UVAI ≥15 threshold despite temporal decay, potentially underestimating plume area/mass.
• Optical property assumptions: Lidar ratio (R = 40–70 sr; potentially higher 75–112 sr from ground-based lidars) and mass extinction coefficient (ε = 3–6 m² g⁻¹) introduce substantial uncertainty bounds on mass estimates.
• Radar echo-top biases: Weather radar can underestimate echo-tops by up to ~1 km and is insensitive to small smoke particles; large pyrometeors may contaminate signals.
• Bottom-up emissions: Fuel consumption fractions and PM2.5 emission factors from climatological means may be too low for extreme events, causing discrepancies vs. top-down mass (0.3–1.1 Tg vs. 0.1–1.2 Tg bottom-up); fuel heterogeneity and combustion completeness are imperfectly constrained.
• Burned area/FRP estimation: MODIS resolution/return intervals, cloud/smoke obscuration, parallax, radar coverage, and incident management activities can affect burned area timing and FRP normalization.
• Generalizability: ANYSO’s unique synoptic persistence and regional fuel characteristics (eucalyptus) may not generalize to all regions; small-scale meteorology, topography, and smoke–radiation feedbacks on convection need further quantification.