GEOPHYSICAL RESEARCH LETTERS, VOL. 30, NO. 10, 1542, doi:10.1029/2002GL016820, 2003
Transport of forest fire smoke above the tropopause by supercell convection
Michael D. Fromm1 and René Servranckx2
Received 20 December 2002; revised 5 March 2003; accepted 17 March 2003; published 31 May 2003.
[1] A recent letter [Fromm et al., 2000] postulated a link
between boreal forest fire smoke and observed stratospheric
aerosol enhancements in 1998. Therein a case was made that
severe convection played a role in the cross-tropopause
transport. A similar occurrence of stratospheric aerosol
enhancements in the boreal summer of 2001 was the
stimulus to investigate the causal mechanism more deeply.
Herein, we show a detailed case illustrating the unambiguous
creation of a widespread, dense smoke cloud in the upper
troposphere and lower stratosphere (UT/LS) by a Canadian
forest fire and explosive convection in May 2001. This event
is the apparent common point for several downstream
stratospheric mystery cloud observations. Implications of
this finding and those pertaining to the boreal summer of
1998 are that convection and boreal biomass burning have an
under-resolved and under-appreciated impact on the upper
troposphere, lower stratosphere, radiative transfer, and
INDEX TERMS: 0305 Atmospheric
atmospheric chemistry.
Composition and Structure: Aerosols and particles (0345, 4801);
0368 Troposphere—constituent transport and chemistry; 3314
Meteorology and Atmospheric Dynamics: Convective processes.
Citation: Fromm, M. D., and R. Servranckx, Transport of forest
fire smoke above the tropopause by supercell convection, Geophys.
Res. Lett., 30(10), 1542, doi:10.1029/2002GL016820, 2003.
1. Introduction
[2] The tropopause is generally regarded as a strong
inhibitor of upward transport across material surfaces
because of the large change in lapse rate between the
well-mixed troposphere and the very stable stratosphere. It
is known that convective turrets can bulge through the
tropopause, but the only natural, explosive force that is
regarded to have sufficient energy to penetrate the tropopause and transport material deep into the stratosphere is
that of volcanic eruptions. It has been postulated, though,
that firestorms spawned by use of nuclear weaponry would
be sufficiently energetic and widespread to breach the
tropopause and cause a ‘‘nuclear winter’’ [Turco et al.,
1983]. More recently, Fromm et al. [2000] qualitatively
linked numerous satellite observations of anomalous stratospheric aerosol layers in the boreal summer of 1998 to forest
fires, and proposed that extreme convection was the mechanism for the troposphere-to-stratosphere transport (TST).
The evidence for the unusual stratospheric aerosols came
primarily from vertical profiles measured by the Polar
Ozone and Aerosol Measurement III (POAM III) [Lucke
1
Computational Physics, Inc., Springfield, VA, USA.
Canadian Meteorological Centre, Meteorological Service of Canada,
Montreal, Canada.
2
Copyright 2003 by the American Geophysical Union.
0094-8276/03/2002GL016820$05.00
49
et al., 1999] and Stratospheric Aerosol and Gas Experiment
II (SAGE II) [Mauldin et al., 1985]. One intense forest fire
blowup in Canada on 3 August 1998 was shown to be
responsible for long-range transport of aerosols [Hsu et al.,
1999], CO and O3 [Forster et al., 2001], and NOX [Spichtinger et al., 2001] to western Europe.
[3] It is expected that a surface warming trend at high
northern latitudes will lead to an increase in boreal fires and
their effects; such a trend toward increased burning has
already been detected [Stocks et al., 1998]. Warming in the
boreal biome might also increase the occurrence of firerelated TST of smoke and other emissions. Thus this newly
discovered mode of TST might play an important role in the
redistribution of material between the troposphere and
stratosphere, with a consequently substantial radiative,
chemical, and cloud impact.
2. Case Study: Chisholm Fire, 28/29 May 2001
[4] In the northern summer of 2001 over 200 observations
of stratospheric aerosol layers were made by POAM III.
Roughly half of these measurements were made at potential
temperatures between 380 and 460 K (up to 9 km above the
tropopause), clearly above what Holton et al. [1995] term the
‘‘lowermost stratosphere.’’ Using a back trajectory technique
similar to Fromm et al. [2000], we found that air parcels from
a number of the stratospheric layers traced back to northwestern Canada on or about 29 May.
[5] On 23 May 2001 a forest fire named the Chisholm
fire (55N, 114W, roughly 160 km north of Edmonton,
Alberta) started. This fire was investigated in depth because
of its man-made origin and its impact on developed property
[ASRD, 2001]. The severity of the fire-risk conditions and
the fire itself increased until 28 May [ASRD, 2001], when an
approaching cold front, a strong low-level jet, and consequent rapid spread of flame resulted in the eruption of
convection at Chisholm. Figure 1 shows views from three
nadir-viewing satellite sensors of the conditions on 28 May
before the convective development. NASA’s Earth Probe
Total Ozone Mapping Spectrometer (EP-TOMS) aerosol
index (AI) [Hsu et al., 1999], shows the Chisholm plume
(Figure 1a) snaking generally northwest from the source
fire. The average and maximum value of enhanced AI
pixels at this time was 1.7 and 5.6, respectively. For
corroboration we use the Sea-viewing Wide Field of View
Sensor (SeaWiFS) true-color imagery in a manner similar to
Falke et al. [2001]. The SeaWiFS image at 20 UTC (Figure
1b) also shows the Chisholm smoke plume, a gray-tan cloud
amidst white water-ice clouds, oriented southeast to northwest, fanning out from Chisholm. There is also a smaller
smoke plume southeast of the Chisholm fire. Figure 1c
shows a false-color red-green-blue (RGB) image from the
Advanced Very-High Resolution Radiometer (AVHRR) at
14 UTC (6 hours earlier than the SeaWiFS image). AVHRR
- 1
49 - 2
FROMM AND SERVRANCKX: CONVECTION AND SMOKE ABOVE TROPOPAUSE
Figure 2. Early stage of convection, Chisholm fire, and
smoke, 00 UTC 29 May. (a), AVHRR false-color RGB. (b),
AVHRR 10.7 mm Tb.
Figure 1. Satellite images of Chisholm fire and smoke on
28 May, before onset of convection. (a) EP TOMS aerosol
index, approximately 18 UTC. Chisholm is at the red
asterisk. (b) SeaWiFS true-color, 20 UTC. Chisholm is at
the red square. (c) AVHRR false-color RGB of channels 3,
2, and 1, 14 UTC. (d) AVHRR channel 4 (10.7 mm) Tb, 14
UTC.
data have been used extensively to detect fire hotspots and
smoke [e.g., Li et al., 2000]. In this letter, the AVHRR RGB
composites use red, green, and blue for channels 3, 2 and 1
(3.7, 0.9, and 0.6 mm, respectively). The distinct fire hot
spots (red) and the fanning plume of the Chisholm smoke
against the clear-sky background are evident. The Chisholm
smoke plume is distinctly less reflective than the whiter
water-ice clouds in the image. Note that there is no
apparent plume southeast of Chisholm at 14 UTC, as seen
in the later SeaWiFS image. Presumably this fire had
started or grown substantially only in the 6 hours between
images. The accompanying AVHRR IR (10.7 mm,
channel 4) image (Figure 1d) of brightness temperature
(Tb), indicative of the altitude of opaque emitting surfaces
such as cloud tops, reveals no substantial vertical cloud
development at Chisholm.
[6] During the convective phase, between roughly 23
UTC 28 May and 05 UTC 29 May (local time is UTC
minus 6 hours), there were reports of a funnel cloud and
peak wind gust of 92 km/hr [ASRD, 2001]. Carvel radar
(54N, 114W, near Edmonton) echoes of the thunderstorm
at 0130 UTC indicated the top of the backscattering column
at 13 km [ASRD, 2001]. The radiosonde data from Stony
Plain (50 km west of Edmonton) on 29 May 00 UTC placed
the tropopause at 11 km [Archives of the Meteorological
Service of Canada, Environment Canada]. Hence this convection had extreme vertical development, penetrating the
tropopause. The AVHRR view of the Chisholm area soon
after onset of convection is shown in Figure 2. The RGB
image at 00 UTC 29 May (Figure 2a) reveals more and
larger hotspots than Figure 1c. The Chisholm fire hotspot, in
the center of the image, has increased in size by roughly a
factor of four. A hotspot north of Chisholm, at roughly
56N, is evident at 00 UTC and not in Figure 1c, indicating
a new or growing fire. Another indication of fire spread and
intensification is red-shaded pixels in Figure 2b, the IR
view. Pixels above 28C were not observed earlier in the
day (Figure 1). It is apparent from the RGB and IR image
that convective vertical development is occurring. There is a
substantial line of deep cloud southwest of Chisholm, and
active cumulonimbus growth on the northern flank of the
Chisholm hotspot and the new fire north of Chisholm.
Figure 3, which shows the AVHRR view at 02 UTC,
captures a pair of mature thunderstorms. The RGB image
(Figure 3a) shows the flattening of the anvil cloud near the
tropopause over the two convective columns near the
Chisholm fire and the hotspot north of Chisholm. Strong
illumination and shadowing highlights two features on the
southernmost cell’s anvil. One feature is either an overshooting top or a plume [Levizzani and Setvak, 1996],
indicative of supercell energetics. The second feature is
concentric waveforms, indicating the presence of gravity
waves excited by vigorous vertical currents at the base of
the stratosphere. The accompanying IR Tb image (Figure
3b) shows that the lowest temperature on the anvils was
between 60 and 67C, which is lower than air temperature in the surrounding environment as obtained from the
Figure 3. Mature supercell convection, Chisholm fire, and
smoke, 02 UTC 29 May. (a), AVHRR false-color RGB. (b),
AVHRR 10.7 mm Tb.
FROMM AND SERVRANCKX: CONVECTION AND SMOKE ABOVE TROPOPAUSE
49 - 3
Figure 5. Radiosonde profiles for 12 UTC, 29 May at
three sites closest to the smoke and water-ice clouds
illustrated in Figure 4.
Figure 4. Clouds and smoke on 29 May, after collapse of
convection. (a) TOMS AI, approximately 1800 UTC. (b)
SeaWiFS, 21 UTC. (c) AVHRR RGB, 14 UTC. (d) AVHRR
10.7 mm Tb. Locations of Fort Smith, Norman Wells, and
Cambridge Bay radiosonde sites are the black, red and blue
circles, respectively.
radiosonde data: tropopause of 58C and minimum temperature (Tmin) of 61C. There is a ‘‘warm’’ core in the
center of the southernmost anvil, another recognized signature of supercell convection [McCann, 1981]. The Tb
difference between the warm core and the flanking cold arc
near the southern rim is as much as 15C. Yet the RGB
image clearly shows illumination along the western edge of
the warm core and shadowing toward the eastern flank.
Thus the plume-feature particles in the warm core are higher
than the surrounding cloud top. These clues indicate one or
a combination of two scenarios: 1) the cloud penetrated
above the Tmin, is in thermal equilibrium with the ambient
air, and resides at an altitude in the lower stratosphere at
which the lapse rate has reversed, and is thus warmer than
the cloud below, or 2) the warm core indicates a dip in the
water-ice cloud top into the upper troposphere and the
visible material in the warm core is composed of particles
too small to be opaque at 10.7 mm, which may indicate a
predominance of smoke particles.
[7] These clues from ground- and space-based observations form a compelling view of an extreme convective
event in concert with very energetic boreal forest fires. The
interaction between the fire and convection is dynamic;
intense and expanding surface heating destabilizes the air
which in turn generates wind and lightning that both
aggravate existing fires and start new fires. The result is a
pair of supercell storms that at maturity protrudes considerably through the local tropopause and excites intense
gravity waves at this interface.
[8] By the morning of 29 May the cold front had passed
through Chisholm [ASRD, 2001] and the convection of the
prior evening had collapsed. The AI map (Figure 4a) shows
an expansive aerosol cloud. The pixel values over a large
area at the center of the cloud are very high (12.8 units),
much higher than any value on 28 May. The average pixel
value of enhanced AI is 6.5. Such high AI values signify
extreme atmospheric or satellite-viewing conditions, a high
altitude aerosol cloud, or aerosol over a bright surface
[McPeters et al., 1998]. The SeaWiFS image at 21 UTC
(Figure 4b) shows a broad smoke cloud. The SeaWiFS and
AI smoke shroud covers an area with underlying or commingled water-ice clouds, but also overlies otherwise clear
skies. The RGB AVHRR image at 14 UTC (Figure 4c)
reveals the smoke to be brownish gray, markedly darker than
the neighboring water-ice clouds, as it was on 28 May. Note
also the unusually flat texture of the smoke, which is
distinctive in comparison to water-ice cloud textures. This
broad, smooth cloud signature may well be evidence that the
cloud resides in the very stable lower stratosphere. The RGB
image also shows shadowing at the western edge of the
smoke cloud, indicating that the smoke here is higher than
the underlying deck of water-ice clouds. The IR Tb image
(Figure 4d) shows that in a broad area, the emitting temperature is below 55C, the 12 UTC tropopause temperature.
The two AVHRR panels of Figure 4 are annotated with
symbols showing the location of the three closest radiosonde
Figure 6. Link between UTLS smoke cloud and supercell convection. Contours of GOES 8 10.7 mm 56C Tb
in 3-hour intervals on 29 May.
49 - 4
FROMM AND SERVRANCKX: CONVECTION AND SMOKE ABOVE TROPOPAUSE
profiles at 12 UTC, illustrated in Figure 5. Figure 5 also
shows an orange-shaded area corresponding to the same
shaded Tb range ( 55 to 60C) in Figure 4d. Two of the
three radiosonde profiles recorded a Tmin (the top of the
tropopause region) ‘‘warmer’’ than 55C. The Fort Smith
Tmin, 64C, is at 12 km. Thus the low Tb (the minimum
pixel value is 59.8C) cloud resides between roughly 10
and 13 km. This entire area is comprised of the distinctive
SeaWiFS smoke coloration and extreme values of AI.
[9] The SeaWiFS, TOMS, and AVHRR visible imagery
taken together show a broad, dense, distinctive smoke cloud
over an area, approximately 22,500 km2, that consists of
otherwise clear sky and water-ice cloud. This remarkable
cloud is substantially bigger than the Chisholm smoke
plume of 28 May. The central portion is unambiguously
opaque and tops out above the tropopause. The shadowing
in the AVHRR RGB image and the extreme values of the AI
both indicate that the smoke cloud top is even higher than
the opaque water-ice cloud (TOMS cannot ‘‘see’’ through
dense water-ice clouds [Newchurch et al., 2001]).
[10] The high, opaque cloud feature in Figure 4 can be
traced back to the extreme convection of the prior evening.
Geostationary Operational Environmental Satellite (GOES)
8 IR (10.7 mm) imagery provides a link between the midday
29 May cloud and the convective phase. Figure 6 shows the
contour of the 56C Tb in 3-hour intervals, confirming
that the cold (i.e. deep) cloud tops in the post-convective
phase (Figure 4) had their origin in the Chisholm supercell
convection of the prior evening (Figure 3). Thus the connection between boreal forest fire, extreme convection, and
extensive UTLS-level smoke observed approximately 15
hours later is complete.
3. Summary and Conclusions
[11] We conclude that under certain conditions extreme
convection is sufficiently energetic to efficiently transport a
large amount of material from the planetary boundary layer
to the UTLS. The TST is likely a product of direct upward
advective transport and gravity wave breaking. We also
conclude that the postulation of Fromm et al. [2000], that
such a combination of intense forest fire and convection was
the cause for a substantial stratospheric aerosol increase in
1998, is reasonable and proved.
[12] The implications of this phenomenon are considerable. Noting the events of 2001 and 1998, it is apparent that
this previously undetected form of TST operates with some
regularity in the middle and high northern latitudes. Thus
the radiative, chemical, and cloud microphysical impact
attributable to this TST mechanism may well be significant.
Another implication is that by using the traceability of inert
smoke aerosols, new constraints on the power of convection
to rapidly transport and redistribute planetary boundary
layer constituents can now be gained. Exploring these issues
can be facilitated and studying other occurrences of the
unique cold (IR) and gray (visible) cloud signature in nadirviewing satellite imagery such as GOES, AVHRR, and
similar imagers.
[13] In addition to new findings, this work has also led to
new questions. It is not known precisely how high the
smoke cloud was on 29 May. Also not known is the
precise composition of the smoke and water-ice cloud,
especially the highest, opaque cloud on 29 May. Our
investigations to date indicate strongly that this is a
recurring phenomenon that occurs at many degrees of
intensity, with consequently variable atmospheric impact.
These recent discoveries of stratospheric intrusions caused
by boreal forest fires and convection should create the
impetus to examine the historical record, especially the
satellite era, more closely.
[14] Acknowledgments. We are grateful to David Larko of NASA
for providing TOMS AI data. We thank Brian Stocks of the Canadian
Forest Service for providing facts about the Chisholm fire. We also thank
Richard Bevilacqua (NRL) for suggesting valuable improvements to the
manuscript, and Julie Dion for formatting the figures. SeaWiFS data are
provided by Orbimage Corporation via NASA GSFC. AVHRR data are
provided by the NOAA Satellite Active Archive.
References
ASRD, Final Documentation Report - Chisholm Fire (LWF-063), Alberta
Sustainable Resource Development, Forest Protection Division, ISBN
0-7785-1841-8, 2001.
Falke, S., R. Husar, and B. Schichtel, Fusion of SeaWiFS and TOMS
satellite data with surface observations and topographic data during extreme aerosol events, J. Air & Waste Management Assoc., 51, 1579 –
1585, 2001.
Forster, C., et al., Transport of boreal forest fire emissions from Canada to
Europe, J. Geophys. Res., 106, 22,887 – 22,906, 2001.
Fromm, M. D., J. Alfred, K. Hoppel, J. Hornstein, R. Bevilacqua, E. Shettle, R. Servranckx, Z. Li, and B. Stocks, Observations of boreal forest fire
smoke in the stratosphere by POAM III, SAGE II, and lidar in 1998,
Geophys. Res. Lett., 27, 1407 – 1410, 2000.
Holton, J., P. Haynes, M. McIntyre, A. Douglass, R. Rood, and
L. Pfister, Stratosphere-troposphere exchange, Rev. Geophys, 33, 403 –
439, 1995.
Hsu, N. C., J. Herman, J. Gleason, O. Torres, and C. Seftor, Satellite
detection of smoke aerosols over a snow/ice surface by TOMS, Geophys.
Res. Lett., 26, 1165 – 1168, 1999.
Levizzani, V., and M. Setvak, Multispectral, high resolution satellite observations of plumes on top of convective storms, J., Atmos, Sci., 53,
361 – 369, 1996.
Li, Z., S. Nadon, and J. Cihlar, Satellite detection of Canadian boreal forest
fires: development and application of an algorithm, Int. J. Rem. Sens., 21,
3057 – 3069, 2000.
Lucke, R., et al., The Polar Ozone and Aerosol Measurement (POAM) III
instrument and early validation results, J. Geophys. Res., 104, 18,785 –
18,799, 1999.
Mauldin, L., N. Zaun, M. McCormick, J. Guy, and W. Vaughn, Stratospheric Aerosol and Gas Experiment II Instrument: A Functional Description, Opt Eng., 24, 307 – 312, 1985.
McCann, D., The enhanced-V, a satellite observable severe storm signature.
NOAA Tech. Memo. NWS NSSFC-4, National Severe Storms Forecast
Center, Kansas City, Mo. (NTIS- #PB81-230328), 31 pp., 1981.
McPeters, R., et al., Earth Probe Total Ozone Mapping Spectrometer
(TOMS) Data Products User’s Guide, NASA Technical Publication
1998-206895, 70 pp., 1998.
Newchurch, M., X. Liu, J. Kim, and P. Bhartia, On the accuracy of Total
Ozone Mapping Spectrometer retrievals over tropical cloudy regions,
J. Geophys. Res., 106, 32,315 – 32,326, 2001.
Spichtinger, N., M. Wenig, P. James, T. Wagner, U. Platt, and A. Stohl,
Satellite detection of a continental-scale plume of nitrogen oxides from
boreal forest fires, Geophys. Res. Lett., 28, 4579 – 4582, 2001.
Stocks, B., et al., Climate change and forest fire potential in Russian and
Canadian boreal forests, Climatic Change, 38, 1 – 13, 1998.
Turco, R., A. Toon, T. Ackerman, J. Pollack, and C. Sagan, Nuclear Winter:
Global Consequences of Multiple Nuclear Explosions, Science, 222,
1283 – 1297, 1983.
M. D. Fromm, Computational Physics, Inc., Springfield, VA, USA.
R. Servranckx, Canadian Meteorological Centre, Meteorological Service
of Canada, Montreal, Canada.