This manuscript was prepared for Atmos. Chem. Phys. (Discuss.)
with version 7.7 of the egu LATEX class egu.cls.
Date: June 28, 2006
ACP-LOGO
Defintion of ’banner clouds’ based on time lapse movies
J. H. Schween1 , D. Reinert2 , V. Wirth2 , and J. Reuder3
1
Met. Inst. Univ. Munich
Uni Mainz
3
Uni Bergen
2
Abstract.
This article gives a definition of banner clouds to investigate them systematically. Banner clouds form on the leeward side of mountains resembling a flag or banner attached
to the mountain and do not appear on the windward side of
the mountain. They are real clouds, i.e. they consist of water droplets and not of snow blown off the mountain. They
are quasi stationary, the air is flowing through them. They
are not convective but driven by dynamical processes. This
definition is clarified by examples given as still images and
time lapse movies taken at Zugspitze mountain in the Bavarian Alps.
1 Introduction
The Glossary of Meteorology (Huschke, 1959; Glickman,
2000) defines a Banner cloud as “A cloud plume often
observed to extend downwind from isolated, sharp, often
pyramid-shaped mountain peaks, even on otherwise cloudfree days.” Well known examples of mountains where banner
clouds form regularly are the Matterhorn in the Swiss Alps
or the Mount Everest in the Himalayas. Although banner
clouds are a well known phenomenon there is surprisingly
limited knowledge about the physical mechanisms of their
formation. The Glossary of Meteorology (Huschke, 1959;
Glickman, 2000) states that “The physics of the formation
of such clouds is not completely understood”. Moreover exists no clear definition of the phenomenon although there are
many qualitative arguments about the origin of these clouds
(e.g. Douglas, 1928; Küttner, 1949; Beer, 1974; Hindman
and Wick, 1990) with probably the oldest Hann (1896) more
than 100 years ago. To our knowledge the only numerical
Correspondence to: J. H. Schween ([email protected])
approach to understand banner clouds was made by Geerts
(1992b).
Aside this theoretical considerations about the formation
of banner clouds and some drawings and photographs only
few scientific observations of banner clouds exist. Peppler
(1927) connected his observations of a single banner cloud
event to upper air wind observations and found significantly
higher wind speed at the mountain top level. In 1947 Küttner
(2000) measured temperatures and humidities inside and outside banner clouds at Zugspitze and observed that the air
inside a banner cloud is by 3–4 K warmer and by 40–70 %
more humid than outside. This means that the top of a banner
cloud is unstable but in contrast hereto the top of the cloud
usually appears rather smooth and laminar.
In December 2002 we installed an electronic camera close
to the summit of the Zugspitze which produces every day
a time lapse movie. We use them to identify banner cloud
situations and to investigate their temporal and spatial evolution. To classify this movies we developed a clear definition
for banner clouds. In this article we give this definition and
present a selection of stills and movies to show the different
forms of these clouds. To our knowledge these are the first
movies of banner clouds presented in scientific literature. A
detailed analysis of the mechanisms leading to the formation
of the cloud is not aim of this article. Instead we will give a
summary of the various existing theories. An improved understanding of the relevant physical processes leading to the
formation of banner clouds is only possible through the combination of measurements and fine scale modelling and will
be subject of forthcoming publications.
2
Existing Theories
There exist three basic theories to explain the formation of
the banner cloud at the leeward side of a mountain. According to them the condensation is due to
2
J.H.Schween: Banner clouds
cold dry air
windward side
low
pressure
strong wind
C. turbulent momentum transport between the flow over
and around the mountain and the air at the leeward side
of the mountain.
turbulence
warm humid air
cloud
horizontal
convergence
leeward side
Fig. 1. Schematic vertical section through a banner cloud showing
the main features.
1. the pressure reduction in the lee of the mountain crest
(Humphreys, 1920; Grant, 1944; Huschke, 1959; Beer,
1974)
2. contact cooling above the cold surface in the boundary
layer of the mountain (Humphreys, 1920)
3. adiabatic cooling in an updraft on the leeward side of the
mountain (Hann, 1896; Douglas, 1928; Hindman and
Wick, 1990; Geerts, 1992a,b; Glickman, 2000)
According to theory 1 the flow deformation and acceleration
at the crest leads to a pressure reduction and accordingly to
adiabatic cooling along stream lines. Geerts (1992a) pointed
out that even strong storms do not lead to a temperature reduction strong enough to explain the observed banner clouds.
The basic idea of theory 2 is that the cloud is formed as fog of
mixing between cold air from the boundary layer close to the
mountain and warm and humid ambient air. Geerts (1992a)
pointed out that in most cases there is no reason that the air
at the leeward side of the mountain is colder than around.
The most accepted theory is 3 which implies an updraft at
the leeward side of the mountain as shown schematically in
figure 1. Scorer (1972) identifies this as a special case of flow
separation with the ridge as the point of separation.
Assuming identical humidity distribution on both sides of
the mountain the updraft must reach deeper than the lifting
on the windward side of the mountain otherwise the cloud
would occur on both sides of the mountain. Again there are
different theories about the driving force of this updraft. According to them the updraft is driven by:
A. the pressure minimum in the lee of the mountain top
(Hann, 1896; Douglas, 1928; Banta, 1990; Hindman
and Wick, 1990; Glickman, 2000)
B. confluence of vortices at the leeward side (Geerts,
1992a,b)
Theory A assumes in a vertical section streamlines as shown
in figure 1. The dashed streamline can be interpreted as a part
of a vortex with horizontal axis. But considering the three dimensional structure it can be as well a secondary circulation
in a complex wake behind the mountain. It is the most accepted theory and is also given by the actual edition of the
glossary of Meteorology (Glickman, 2000).
Since in many cases banner clouds are observed at peaked
mountains, Geerts (1992b) presented a theory together with
simulations in which the confluence in the wake behind such
mountains leads to an up slope wind. Regarding a horizontal
section through the peak, the wake is formed by two horizontal vortices with opposite direction of rotation. This leads
to horizontal convergence at the leeward side which is compensated by an updraft. In a vertical section this secondary
circulation gives the same picture as in figure 1. Regarding
the three dimensional structure the two horizontal vortices
are part of a vortex tube following the ridges at the leeward
side of the mountain. Supposedly because this theory works
only for mountain peaks Geerts (1992a) restricted in his definition banner clouds to peaked mountains.
As far as we know theory C has never been considered.
At the ridge of the mountain a strong shear between the flow
above (solid stream line in figure 1) and the air at the leeward side (dashed stream line) is apparent. In this shear
layer Kelvin-Helmholtz-Instability leads to fast growth of
even small disturbances. This results in turbulent transport
of momentum across the shear layer. In figure 1 the air in
the dashed stream line at the top of the cloud experiences a
driving force to the right leading to the circulation indicated
by the dashed streamline.
We think that both, theory A and theory C, work together
as a driving force for the vortex in the lee of an obstacle. In
the case of a horizontal ridge the vortex is a vortex tube with
horizontal axis. A peak on this ridge leads to an up folding
of the vortex tube and convergence in the lee of that peak
enhances the updraft as described in theory B. Accordingly
the updraft is stronger in the lee of peaked mountains like the
Matterhorn thus making banner clouds there more probable.
3
Orography of Zugspitze
The summit of the Zugspitze (2962 m above sea level) is the
westernmost and highest peak of a larger mountain range
(Wetterstein Gebirge) with summits mainly around 2800 m.
To the west and North the average height of the mountains is
lower at around 2200 m.
Close to the summit is the observatory of the german
weather service (DWD) where aside many other measurements regular visual weather and cloud observations are
made. The electronic camera is installed at the west side of
J.H.Schween: Banner clouds
3
SW
W
Schneefernerkopf 2874 m
Schneefernerscharte 2700 m
Zugspitzeck
2816 m
Ehrwald
1000 m
UFS
2656 m
Fig. 2. Field of view of the camera with directions and the main
landmarks.
250
0
50
22
ridge forks into two with one running in NE-direction, and
the other running to the ESE for about 2k̇m slightly descending to 2700ṁ from where it continues further east.
The area enclosed by these ridges is called the Zugspitzplatt, a rough terrain mainly covered by ice, snow, boulders
and gravel. The Zugspitzplatt slopes in southeasterly direction from about 2600 m at the foot of the ridges to about
1700 m at the eastern edge of figure 3 where it drops 300 m
down to a narrow valley running to the east. In contrast to this
rather gentle slope to the southeast, the north and west sides
of the mountain are dominated by very steep rocks dropping
from around 2800 m down to 1700 m from where forested
slopes reach down to the valley floor at around 1000 m.
Banner clouds form depending on the meteorological situation and wind direction at all the ridges. We focus here
on the ridge running westward to Zugspitzeck, visible in the
field of view of the camera (fig. 2). At this ridge Küttner
(2000) made his measurements more than 50 years ago. Images of the camera are stored every 5 s, and are joined every night automatically. This gives at the standard rate of 25
frames per second a 125 times fold time lapse. These movies
are inspected visually for banner cloud episodes and appropriate sequences are cut out.
2
UFS
Since no clear definition of banner clouds exist in the scientific literature, we developed a definition based on our time
lapse movies of banner cloud events. The definition consists
of 5 points:
ST
50
22
SK
2500
47˚ 24'
Definition of “banner clouds”
00
ZE2750
47˚ 25'
4
25
22
50
25
00
ZZ750
2000
Zugspitzplatt
Zugspitzplatt
I. A banner cloud is a cloud closely connected to the orography, it forms in the wake at the leeward side of the
mountain and touches the ground at least at the top of
the ridge.
25
00
10˚ 59'
11˚ 00'
11˚ 01'
Fig. 3. Map of Zugspitze with the main landmarks Zugspitze summit (ZZ), Zugspitzeck (ZE), Schneefernerscharte (ST), Schneefernerkopf (SK). The viewing direction of the camera is indicated by
the solid black line and the 55◦ field of view by the dashed lines.
this observatory looking to the southwest (see figures 2 and
3) along the western ridge of the mountain. This ridge runs
for about 500 m to the southwest descending from 2960 m
to 2800 m then turning to the west and continuing for 500 m
at roughly the same level of 2800 m to the Zugspitzeck (ZE
in fig. 3). From here the ridge runs for about 2 km to the
south with the Schneefernerkopf at 2874 m (ZK in fig. 3) as
the highest peak and the Schneefernerscharte at 2700 m (ST
in fig. 3) as the deepest gap. East of Zugspitze summit the
II. It appears only on the leeward side of the mountain and
there are no clouds on the windward side, or the banner
cloud is at least not connected to them.
III. The cloud is made up of small water droplets and not of
snow crystals blown of the snow covered mountain.
IV. It is quasi stationary, i.e. it lives significantly longer
than the typical turnover time of the leeward vortex. The
air is flowing through the cloud.
V. The cloud is not convective. It is primarily dynamically
driven by the wind and not by buoyancy.
Point I resembles the common definition. In the cases
shown here the wake is mainly formed by a vortex with a
horizontal axis. Condensation occurs due to adiabatic cooling in the upward directed branch of this vortex. In the case
of peaked mountains like the Matterhorn the wake is formed
by a vortex tube along the leeward ridges with confluence at
4
J.H.Schween: Banner clouds
Fig. 4. Still image of movie 050827 from 27. August 2005 19:07–
20:20. Wind blowing from the north i.e. from right to left.
Fig. 5. Still image of movie 050805 from 5. August 2005 8:34–
9:14. Wind blowing from the north i.e. from right to left.
the leeward side of the mountain leading to a secondary circulation with up slope wind. The constraint to appear only
on the Leeward side (point II) distinguishes the cloud from
cap clouds which form on both sides of a mountain due to the
lifting and adiabatic cooling of the air when flowing over the
mountain. Blowing snow forming banners can be observed
many times in winter or at sufficiently high mountains. Since
these banners are not direct result of thermodynamical processes they are no clouds and excluded by point III. Point
IV ensures that not short events where cloud or fog frazzles
come close to the ridge are identified as a banner cloud. The
final point V is the most difficult one since there is often
a continuous transit between banner clouds and convective
clouds.
(frame 814) before it starts to dissolve. Obviously the cloud
is attached to the ridge and touches the ground also further
down the Zugspitzplatt (point I of the definition). There are
no clouds on the windward side of the ridge (point II of the
definition). The cloud frazzles which are passing through the
field of view and obscure the view in some case are originating from the ridge just in front. Beside that there is nearly no
snow lying on the mountain which could be blown off by the
wind, the opaqueness of the cloud indicates that it consists
of water droplets thus fulfilling point III. The cloud remains
visible for more than 30 minutes whereas the turnover time of
the vortex just behind the ridge is estimated to be 2 minutes
thus it is clearly quasi stationary (point IV of the definition).
The cloud shows no sign of convective movement (point V
of the definition) and is driven solely by the wind crossing
the ridge.
5
Examples
In the following movie sequences are shown and discussed
with regard to the above formulated definition. This document contains only still images of these movies. In the
caption below every image a link leads to the corresponding
movie file. Inside the movie the actual time (Central European Time = local average solar time −12 minutes) is shown
in the top right corner. This time and the frame number is
used to reference certain events in every sequence.
5.1
A typical banner cloud
Figure 4 shows a typical banner cloud at Zugspitze. The
movie starts at 19:07 with only some few cloud frazzles visible at Schneefernerkopf in the background. Then condensation starts at the ridge to Zugspitzeck showing clearly the updraft along the ridge. Around 19:38 (frame 375 of the movie)
a cloud has developed. It remains for 30 minutes till 20:11
5.2
Transition from banner cloud to convective cumulus
cloud
The example referenced in figure 5 shows how a dynamically
driven banner cloud transforms into a orographically induced
cumulus cloud. The movie starts with a cloud which is obviously a cap cloud visible on both sides above the ridge and
originating from the lifting of air above the ridge. This cloud
is obviously not a Banner cloud (violation of point II) but it
has disappeared completely by 8:45 (frame 132). Before at
around 8:37 (frame 58) a banner cloud develops at the ridge
below this cap cloud which does not reach the thickness of
the cloud of the previous example but it remains visible in the
field of view for 37 minutes. In the beginning of its lifetime
the cloud is driven dynamically by the wind coming over the
ridge. Towards the end of the sequence there is a transition
to more convective behavior with strong vertical movements
J.H.Schween: Banner clouds
Fig. 6. Still image of movie 051006 from 6. October 2005 14:07
– 16:22. Wind blowing from the south i.e. from left to right.
which is a violation of point V of our definition. Regarding
the mountains in the background one can observe how during the sequence more and more cumulus clouds develop and
grow. Thus this example shows how a dynamically driven
banner cloud transforms into a orographically induced convective cumulus cloud.
5.3
Different condensation levels
In the example presented in figure 6 different condensation
levels are apparent indicating the different mechanisms leading to the formation of the clouds. The movie starts at 14:07
when the banner cloud at the north side of the ridge existed,
except for short periods without a cloud, already for several
hours. At this stage the cloud at the ridge is obviously a banner cloud since it is attached to the ridge (point I), it is on the
leeward side (point II), it is opaque and must thus consist of
water droplets (point III), it lives longer than the timescale
of the leeward vortex (point IV), and it is obviously dynamically driven and not convective (point V).
What makes this example so interesting are the cumulus clouds above. They show a condensation level which
lies clearly higher than the condensation level of the banner cloud. There are two possible reasons for this. First the
air masses on the windward and leeward side could have different temperatures, water contents and stratification. Since
the cloud lives for several hours there must be a strong water vapour source on the leeward side which continuously
replaces the loss of vapour through the flow above the mountain. A second possible explanation is that the base of the
banner cloud indicates the lifting condensation level (LCL
see e.g. Banta, 1990, pp. 231) and the base of the cumulus
clouds indicate the cumulus condensation level (CCL).
5
The definition of the LCL assumes forced lifting and that
the air is lifted without mixing with the surrounding air. Condensation occurs at the level where a lifted air parcel becomes saturated. Since a banner cloud is due to forced lifting its base lies at the LCL. In contrast the definition of the
CCL assumes two things. First the air must be heated until
it becomes positively buoyant. And secondly the convective
processes due to the heating lead to a well mixed layer before
a cloud forms. That means the water vapour mixing ratio in
the layer below the cloud is constant. Since in general the
water vapour mixing ratio decreases with increasing height
mixing leads to lower vapour mixing ratios in the lower part
of the mixed layer. Both the heating and the mixing lead to
higher values for the CCL compared to the LCL. We performed three radio soundings for that special day from the
valley at Garmisch Partenkirchen about 9k̇m to the NE from
the summit of the Zugspitze. From these soundings we derived a LCL at 2500 m ASL i.e. 300 m below the ridge to the
Zugspitzeck whereas the CCL was significantly higher.
Beside this the movie shows from about 15:00 (frame
615) the development of more convective clouds behind the
Schneefernerscharte and Schneefernerkopf. The condensation level of these clouds is still around 2700 m ASL and thus
lower than the base of the cumulus clouds. From about 15:20
(frame 849) these clouds join with the upper cumulus clouds
while the banner cloud in the foreground at the ridge towards
the Zugspitzeck remains obviously dynamically driven. The
development of the clouds behind Schneefernerscharte and
Schneefernerkopf seems to be triggered by the insolation on
the western slopes of Schneefernerkopf. The insolation leads
to strong heating in the boundary layer there and accordingly to updrafts strong enough to form a convective cloud.
Later in the evening (not shown in the movie) these convective clouds disappear but the banner cloud in the foreground
remains longer demonstrating again its different origin.
5.4
A non banner cloud
An example for a non banner cloud is given in figure 7. It
starts like a banner cloud: fog forms on the Zugspitzplatt and
is drawn towards the ridge (9:46 frame 0 – 9:56 frame 110)
where it is turned over in the lee vortex. But then condensation starts on the windward side and a lifting cloud forms
which covers the whole ridge violating point II of the definition. It must be concluded that the cloud forming processes
on both side of the mountain are the same. Air is lifted on
both sides of the mountain from the same level and condensation occurs on both sides at the same level. This cloud is
thus not a banner cloud.
5.5
Transition from convective cumulus to banner cloud
Another interesting example is given in figure 8. It shows
a transition from a buoyancy driven cumulus cloud to a dynamically driven banner cloud. The sequence starts with a
6
Fig. 7. Still image of movie 030329 from 29. March 2003 9:46 –
10:19. Wind blowing from the north i.e. from right to left.
J.H.Schween: Banner clouds
Fig. 9. Still image of movie 050107 from 7. January 2005 15:50
– 16:11. Wind blowing from the north i.e. from right to left.
cloud.
5.6
Snow blown off the mountain
Figure 9 shows an example of snow blowing off the ridge
and is thus according to point III not a banner cloud. The
Glossary of Meteorology (Huschke, 1959; Glickman, 2000)
states ’. . . [a Banner ] cloud strongly resembles snow blowing off the peak (snow banner), and it is often difficult to
tell the difference’. It is obviously hard to decide wether the
movie shows a cloud consisting of droplets or snow blowing off from the ridge, but there are some hints: in the foreground of the camera appear larger ice crystals as short light
flashes. The snow banner appears more transparent than a
banner cloud consisting of water droplets. The banner becomes better visible if the illumination from the sun comes
more from the front.
Fig. 8. Still image of movie 050909 from 9. September 2005
17:01 – 18:07. Wind blowing from the south east i.e. from left to
right.
rather convective cloud on the west side of the mountain.
For a short moment around 17:07 (frame 62) some convective clouds are visible in the valley to the west. These clouds
show strong vertical movement and grow fast. But the cloud
at the ridge is capped and blown away by the strong wind
shear at the ridge. Later in the sequence around 17:38 (frame
427) the clouds in the valley disappear and the banner cloud
vanishes as well. On the one hand this cloud seems to be
driven mainly by buoyancy on the other hand its top is capped
by the wind shear and the cloud exists longer than the other
clouds in the vicinity. Thus it must be dynamically driven
and we therefore identify it, following point V, as a banner
6
Conclusions
We presented a definition for banner clouds based on time
lapse movies taken at the Zugspitze. The definition consists
of the following points: Banner clouds are attached to mountains (I) and form at the leeward side (II). They consist of
water droplets (III) and not snow crystals blown off from the
mountain. Air is flowing through the cloud but it remains
quasi stationary at the same place (IV). The processes leading to the formation of the cloud are dynamically driven (V).
The examples presented show that there is a continuous transition between dynamically driven banner clouds and convectively driven cumulus clouds.
The examples show also that banner clouds are not restricted to peaked mountains as postulated by Geerts (1992a).
Although confluence in the wake of a mountain peak may
J.H.Schween: Banner clouds
enhance the updraft on the leeward side, banner clouds can
form also on mountain ridges.
Acknowledgements. This work is funded by the German research
foundation (DFG) under RE1710/1-1. We thank M. Kristen and his
coworkers at the meteorological Observatory Zugspitze from the
DWD for allowing us to install the camera and for their efforts to
keep it running. The manufacturer of the camera Mobotix supplied
us with a free substitute. Mario Mech wrote the Perl scripts which
produce the daily movies. Alexandra Mittermeier looked through
most of the movies and cut the banner cloud sequences.
References
Banta, R.: The role of Mountain flows in making clouds, in: Atmospheric processes over complex terrain, edited by Blumen, W.,
pp. 229–283, American Meteorological Society, Boston/Mass,
1990.
Beer, T.: Atmospheric Waves, Adam Hilger Press, 1974.
Douglas, C.: Some alpine cloud forms, Quarterly Journal of the
Royal Meteorological Society, pp. 175–178, 1928.
Geerts, B.: The Origin of Banner Clouds: a Case of Scientific Amnesia ?, Bull. Austr. Meteor. and Ocean. Soc., 4, 6–9, 1992a.
Geerts, B.: The origin of banner clouds: A potential vorticity
perspective., in: Sixth Conference on Mountain Meteorology
(Preprint Volume), edited by Bean, J. R., pp. 97–98, American
Meteorological Society, http://www-das.uwyo.edu/∼geerts/cwx/
notes/chap08/banner.html, 1992b.
Glickman, T. S., ed.: Glossary of Meteorology, American Meteorlogy Society, Allen Press, http://amsglossary.allenpress.com/
glossary, 2000.
Grant, H. D.: Cloud and Weather Atlas, Cowand-McCann, 1944.
Hann, J.: Allgemeine Erdkunde, Gebrüder Stiepel, first edn., 1896.
Hindman, E. E. and Wick, E. J.: Air Motions in the Vicinity of
Mt. Everest as Deduced from Pilatus Porter Flights, Technical
Soaring, XIV, 52–56, 1990.
Humphreys, W. J.: Physics of the air, McGraw-Hill, first edn., 1920.
Huschke, R. E., ed.: Glossary of Meteorology, American Meteorlogy Society, Allen Press, 1959.
Küttner, J. P.: Meteorologische Überraschungen für wettererfahrene
Alpinisten, Die Alpen, XXV, 157–160, 1949.
Küttner, J. P.: Erinnerungen 1945–1948, in: Die Geschichte der
Wetterstation Zugspitze; Geschichte der Meteorologie, edited by
Wege, K., vol. 4, pp. 70–75, Selbstverlag des Deutschen Wetterdienstes, Offenbach/Main, 2000.
Peppler, W.: Hinderniswolken am Säntisgipfel, Das Wetter, 44,
212–213, 1927.
Scorer, R.: Clouds of the world, Stackpole Books, 1972.
7