Content uploaded by Daniel Fortier
Author content
All content in this area was uploaded by Daniel Fortier on Nov 20, 2014
Content may be subject to copyright.
Geomorphology of thermo-erosion gullies
– case study from Bylot Island, Nunavut,
Canada
Godin Etienne & Fortier Daniel
Département de Géographie – Université de Montréal, Montréal, Québec, Canada
Center for Northern Studies, (CEN) – Laval University, Quebec, Quebec, Canada
ABSTRACT
In the valley of glacier C-79 on Bylot Island, snowmelt water runoff is creating thermo-erosion of permafrost wetlands.
This process contributes to the rapid formation of gullies in ice-wedge polygons. One gully had been observed since
1999 and had a growth of 748 m since then. The geomorphology of this gully is characterized by an active thermo-
erosion zone near the gully head, a poorly-active zone near the outlet and a moderately active zone in-between.
Feedback mechanisms contribute to the erosion processes governing the development of the gully, accelerating erosion
at its head and stabilizing it at its outlet. Erosion features such as sinkholes, collapses and baydjarakhs were
consequently observed in the gully. Thermo-erosion processes have remained active and have had an impact on the
ecosystem for more than a decade.
RÉSUMÉ
La thermo-érosion induite par la fonte du couvert nival cause la dégradation du pergélisol dans la vallée glaciaire C-79
sur l’Ile Bylot. Ce processus contribue à la formation rapide de réseaux de ravinement dans les polygones à coin de
glace. L'observation d'un ravin depuis sa formation en 1999 et de son évolution jusqu’à 748 m en 2009, révèle trois
types de zones d’érosion caractérisant sa géomorphologie : une zone de thermo-érosion très active en amont, une zone
intermédiaire, et une zone faiblement active à proximité de l’exutoire. Les mécanismes de rétroaction amplifient l’érosion
en tête de ravin et la stabilisent à l’exutoire. Des composantes géomorphologiques tels que des puits, effondrements et
baydjarakhs sont conséquemment observées dans le ravin. Le déclenchement du processus de thermo-érosion
souterrain exerce un impact sur l'écosystème depuis plus d'une décennie.
1 INTRODUCTION
In 1999, thermo-erosion processes triggered the
development of a sub-kilometric sized gully network in ice-
wedges polygons located in the valley of glacier C-79
(Bylot Island, NU, Canada). Development of this gully has
been closely monitored since its inception (Fortier et al.,
2007). Several forms of erosion and permafrost
degradation were observed in the gully system. It was
observed that these forms evolved with time, some
remained active for a number of years, other were
deactivated very rapidly after a few years. For over ten
years, the gully head section has migrated upstream while
the gully outlet has remained stable. The geomorphology
of the central section between the gully inlet and outlet
has also evolved over the years. The objectives of this
paper are to characterize the geomorphology of three
sections of a thermo-erosion gully with different ages: 1)
the head of the gully characterized by very active thermo-
erosion processes 2) the central part of the gully
characterized by low to moderate thermo-erosion activity
and gullying 3) the outlet of the gully characterized by very
low thermo-erosion activity, gullying and permafrost
degradation.
2 STUDY SITE
The study site is located on the south-western plain of
Bylot Island (73° 09’N –79° 57’ W) at about 85 km north-
west of the village of Mittimatalik (Pond Inlet) (Figure 1).
Figure 1 : Bylot Island is located in the Eastern Canadian
Arctic Archipalego (Nunavut). Pond Inlet is located 85 km
south-east from the study site.
Climate normal from Pond Inlet (72° 40 N, 77° 58 W) for
the period 1971-2000 indicate a mean annual air
temperature of -15.1°C, with 190 mm of annual
precipitation, 145 mm of which falls as snow (Environment
Canada, 2002). The active layer depth in peaty-silt is
about 40-50 cm and is a few decimeters deeper in coarse
1540
grained materials (Fortier et al., 2006). The permafrost
thickness in the area is estimated to be over 400 m (Smith
et al. 2000; Young et al. 1986). The study site is located in
the valley of Glacier C-79. The valley is about 15 km long
and 5 km wide and oriented ENE – WSW.
Plateaus up to 500 m high form the valley walls. Two
glaciers give rise to a braided river flowing down the valley
(Figure 2). This river is the highest stream order in this
hydrographical system and forms a delta as it ends in the
Navy Board Inlet. (Horton, 1945).
Ice-wedge polygons terraces aggraded on each side of
the outwash plain during the Late-Holocene (Fortier et al,
2004). The slopes are deeply incised and alluvial fans
Figure 2 : A digital elevation model centered on the C-79
Glacier valley (CDED, 2003). The gully case-study is
located at less than 1 km south of the braided river. The
proglacial river is flowing towards the Navy Board Inlet.
are formed at the bottom of the valley walls on the
terraces. Previous cryostratigraphic studies showed that
the central portion of the terraces comprises 2 to 4 meters
of ice-rich, fine to coarse aeolian sediments mixed with
poorly decomposed peat (Fortier et al. 2006). Periglacial
features are abundant on the terrace with well-developed
ice-wedges polygons, thermokarstic ponds and lakes and
about half a dozen pingos (Figure 3).
The studied gully is oriented sub-perpendicular to the
proglacial river conformably to the gentle slope of the
terrace. Water flow in the gully resumes with snowmelt
water run-off and continues throughout the summer due to
drainage of the surrounding wetlands. The gully outlet
connects to a lake-discharge stream which is flowing
toward the proglacial river 1 km downstream.
3 METHODS
During the 2009 fieldwork, a detailed survey of the gully
geometry was done using a differential GPS. The unit
used to characterize the gully was a Trimble DGPS
(model Pathfinder Pro XRS with a TSC1 data collector).
Differential correction was applied to the DGPS data using
GPS Pathfinder Office v3.10 and the Thule (Greenland)
base station (located 496 km from Glacier C-79 valley) as
a reference. Differential correction report indicates that
99.9% of positions (x,y,z) have an effective accuracy
between 0.5m and 1m. Gully contour, erosion landforms
and active geomorphological processes were
georeferenced in the DGPS database during 2009 survey.
The gully was visited almost yearly between 1999 and
2009. Markers were installed along the boundaries of
active thermo-erosion and gullying zones. The positions of
the markers were georeferenced and integrated into a
GIS (ESRI’s ArcGIS v9.3.1). Gully metrics (e.g. area,
length) were calculated directly in ArcGIS using the
‘Calculate Geometry’ tool. The geometry of a dozen
cross-sectional transects were measured along the length
Figure 3 : Oblique Aerial view of the study site (2009). (A)
Thermo-erosion gully. Flow direction in the gully is
indicated by arrows. (B) The proglacial river is flowing
from the glacier to the sea (arrow); (C) ice wedges
polygons and ponds are widespread on the terrace. (D)
Gully head (inlet); (E) gully outlet.
of the gully. A Trimble DGPS was used to make precise
positional lectures of gully transect from one side to the
other side of the gully channel. The distribution of
transects covers the length of the gully.
4 RESULTS
50 m
B
C
A
N
D
E
1541
4.1 Distribution of erosion processes and landforms in
the gully.
In 2009, in the vicinity of the gully head we observed the
presence of sinkholes feeding water to tunnels formed in
the permafrost. Later in the summer, some tunnels
collapsed and created very steep new gully walls. Several
streams were flowing into the gully in this area. Active
layer slumping and exposures of ground ice were
widespread in these newly formed gully branches.
Thermo-erosion, the process of rapid heat transfer that
occurs between flowing water and frozen ground or ice,
was the main process of permafrost degradation. A few
thermokarstic ponds were localized on the polygonal
terrace close to the margin of the gully. The gully area
formed during the previous summers and located
downstream of the gully-head zone was exempt of
sinkholes and tunnels except at the location of a few
intermittent streams flowing on the ice-wedge polygon
terrace and captured after retrogressive erosion of the
gully walls. Thermo-erosion in this section of the gully has
a much more limited impact than in the gully head section.
In this section of the gully, exposures of ground ice were
rare. The gully walls were not as steep as in the gully
head section and the slopes were evolving towards
stabilization. Plant and mosses had colonized some of
these slopes. Ground ice exposed along the gully walls
the previous year commonly evolved into retrogressive
thaw slumps. Thawing of ice-rich permafrost soils and
melting of ground ice promoted thaw settlement, ground
subsidence and eventually collapses. Sediment transport
in the gully channel was significant due to high sediment
input in the gully head zone.The downstream area near
the outlet is the oldest part of the gully system and was
formed eight to ten years ago. It is characterized by stable
and low angle vegetated slopes (gully walls), drained
polygon centers along the gully margin, stabilized to very
weakly active retrogressive thaw slump, and very to totally
(flat) degraded baydjarakhs. In enlarged sections of the
gully channel, alluvial levees were formed over the years.
The levees often contained small pools.
4.2 Gully geometry
The main axis of the gully is 748 m long; the cumulative
length of the gully network, considering all branches and
relict channels, is 2572 m. The area that was directly
affected by thermo-erosion over the ten year period is
25000 m2. Figure 4 shows a schematic of the gully
contours and the localization of the pools within in the
gully channel. The general direction of water flowing in the
gully is from the head toward the outlet (NW) following a
gentle slope of approximately 3 meters over 748 m
(Figure 5). Water flows out of the gully in a small stream
draining the terrace and ending in the proglacial river
about 1 km downstream. The angular layout of the gully
system is essentially due to the degradation of the ice
wedges forming the polygons.The localization of three
typical cross-sections of the gully is shown on Figure 4.
TR1, TR2, and TR3 represent cross-sections of the gully
head, central and outlet sections respectively (Figure 5).
Error bars represented on each point illustrate the 1 meter
ellipsoid maximum spatial error from the DGPS recording
unit. Transects width is increasing from 5.2 m near the
head (TR1) to 9.9 m at gully outlet (TR3). Gully depth is
decreasing from 4.4 m at (TR1) to 1 m at (TR3). The
intermediate transect (TR2) is having in-between values
both for width and depth.
4.3 Gully evolution from 1999 to 2009
The location of the gully head for four periods (1999,
2000-2001, 2002-2005 and 2006-2009) is shown on figure
4. The development of the gully was extremely rapid
during the first year (390 m) and about 50 m year-1 (102 m
total) the second and third year (2000, 2001). The
progression continued to slow down considerably during
the 2002-2005 (≈ 32 m year-1) and the 2006-2009 (≈ 38 m
year-1) periods.
4.4 Gully geomorphological forms and processes
Several forms of erosion can be associated with the
development and evolution of the gully over the period of
observations. Some of these forms were the direct result
of the thermo-erosion process (sinkholes, tunnels, gully
head and surface lowering), some were triggered by
permafrost degradation processes that followed thermo-
erosion (retrogressive thaw-slump, tunnel collapse and
active layer slumping) and others were related to fluvial
processes in the gully channel (levees and pools) (Table
1).
4.4.1 Sinkholes
Sinkholes were found exclusively at gully head where the
thermo-erosion processes were active. Sinkholes
promoted the infiltration of streams running on the surface
of the polygons into the permafrost (Figure 6). Sinkholes
were connected to the gully by a tunnel network subdued
to the geometry of the ice wedges.
4.4.2 Gully head
Gully head were points of active thermo-erosion where
water penetrated in the gully network by way of waterfalls
(Figure 8). Gully heads were essentially present in the
high-erosion activity zone and to a lesser extent in the
intermediate-erosion activity zone. They were absent from
the low-activity zone downstream. Gully heads were
deactivated during the summer when runoff became
insignificant.
4.4.3 Retrogressive thaw slump (RTS)
Retrogressive thaw slumps (RTS) were observed from the
gully head to the gully outlet but were more common in
the intermediate zone. RTS walls were steep and arcuate
(Figure 10). The ground affected by active RTS zone was
chaotic and poorly drained. A general gradient of activity
was observed with the more active RTS located close to
the gully-head area and RTS evolving towards
stabilization downstream. Stabilized RTS did not have
ground ice exposure, the slopes of the valley wall had
reach or were close to equilibrium, were colonized by
plants and the ground was better drained than in
upstream RTS zones.
1542
Figure 4 : DGPS-based map of the gully. The gully area is 25000 m2. The main axis of the gully is 748 m long and its
cumulative length including all the branches is 2572 m. TR1, TR2 and TR3 are typical cross-sections of the gully head,
central section and outlet zones respectively. Active erosion zones for four time frames are represented for 1999, 2000-
2001, 2002-2005 and 2006-2009 periods.
Figure 5 : Typical cross-sections of the gully head (TR1),
central section (TR2), and outlet (TR3) zones (see figure
4 for localization of the cross-sections).
4.4.4 Pools
Pools were observed in the gully channel downstream of
active RTS zones (Figure 9). These pools were formed
and contained by the development of alluvial levees
following high snowmelt water discharge in the gully.
Table 1: Geomorphological forms and processes found in
the gully.
Name n
Area
(m
2
)
Figure
Sinkhole 9 50
6
Tunnel 3 N/A
7
Gully Head 15 N/A
8
Pool 25 2350
9
Retrogressive Thaw
Slump 135 4153
10
Collapse 11 1030
11
Surface Lowering 26 917
N/A
Baydjarakhs 5 44
12
Ponds 6 991
13
6
8
10
12
14
16
0 5 10
Gully Altitude (m)
Gully Width (m)
TR1
TR2
TR3
1543
Figure 6 : Sinkhole in ice wedge polygon. The ladder to
the left indicates the scale. The arrow in the stream
indicates water direction toward the sinkhole.
Figure 7 : Exposed tunnel in permafrost. The broken line
indicates the ceiling of the tunnel.
Figure 8 : Gully head in ice wedge polygons forming falls,
(person for scale).
Figure 9 : The arrow in the lower left corner indicates the
water flow direction in the gully. (A) Pool in enlarged
portion of the gully channel. (B) An alluvial levee
containing the pool.
Figure 10 : Active retrogressive thaw slump due to the
degradation of an exposed ice wedge.
Figure 11 : A fresh collapse following thermo-erosion of
ice wedge polygons.
A
B
4 m
4 m
5 m
4 m
1544
Figure 12 : Degraded baydjarakh (A). Xeric plant species
such as Cassiope tetragona colonized the baydjarakh (B)
Zone evolving towards stabilization with Senecio
congestus plants.
Figure 13 : The broken line at (A) and (B) indicate
thermokarstic ponds near the gully network. The pond at
(B) was formed at an ice wedge intersection. The arrow in
the upper-right corner indicates the geographic north.
4.4.5 Surface lowering
Surface lowering consists of local terrain subsidence
adjacent to the gully margins. This phenomenon was
mainly due to the joint action of both conductive and
convective heat transfer following water flow over the
peaty surface of the polygons. Zones subject to surface
lowering were lower than the surrounding ground not
submitted to surface run-off and with gentle slopes
conformable to the direction of the water flow. Surface
lowering was not as common near the gully head in the
high-erosion activity zone and the intermediate-erosion
activity zone, but more frequent in the low-erosion activity
zone. Drainage of low-center polygons following gully
formation was very often associated with this
phenomenon.
4.4.6 Collapse
Tunnel collapse and associated active layer slumps were
observed essentially in the intermediate-erosion activity
zone and to a lesser extent in the low-erosion activity
zone (Figure 11). Collapses create baydjarakhs which
became better developed with time in the low thermo-
erosion activity zone.
4.4.7 Baydjarakh
The baydjarakh, also known as thermokarst mound (van
Everdingen, 1998) is the result, at the study site, of
thermo-erosion of the ice wedges forming the boundaries
of ice wedge polygons. The polygon center, in this case
composed of ice-rich material, is then exposed on all
sides. The formation process of this form in the context of
the current study begins when active thermo-erosion
degrades one or more sides of an ice-wedge polygon
under the effects of convective heat transfer. Once the
polygon boundary (ice wedges) are thermo-eroded,
thermo-erosion action on the baydjarakh is negligible.
Baydjarakhs located in the intermediate-erosion activity
zone degrade slowly by conductive heat transfer from the
slopes and the top of the polygon center. Degradation of
ground ice of the polygon center promotes surface
subsidence. The final stage of degradation is achieved by
fluvial erosion in the gully channel. There are very few
baydjarakhs in the gully; those observed were in the low-
activity erosion zone where water flow was negligible
(Figure 12).
Figure 14 : Proportion of each landform in relation to
zones of erosion. The erosion activity level zone in the
gully is indicated by H = High, I = Intermediate, L = Low
erosion activity.
4.4.8 Thermokarstic pond
Low-center polygons with ponds were observed on the
terrace along some sections of the gully channel (Figure
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
SinkHole
Gully Head
Collapse
Pond
RTS
S. Lowering
Baydjarakh
Pool
Proportion %
Gully geomorphological feature
H
I
L
B
A
3 m
10 m
A
B
N
1545
13).Thermo-erosion and retrogressive sub-aerial erosion
can lead to drainage of these ponds in the gully. This
process locally and momentarily enhances gullying of the
surrounding permafrost until the pond is completely
drained.
Figure 15 : Zone of high thermo-erosion where entry
points to the gully for stream running on the polygon
terrace are numerous: this area is characterized by
sinkholes and gully heads.
5 DISCUSSION
5.1 The function of time on zonal categorization
The zones of erosion and permafrost degradation in the
gully were defined by the active processes at work for a
given year. As time passes, the speed and amplitude at
which the processes have acted on gully features have
changed. The spatial delineation of each zone is therefore
dynamic and is dependent on the speed and importance
of the thermo-erosion and other processes of permafrost
degradation on the gully geomorphology.
Figure 16 : The zone of intermediate thermo-erosion is
dominated by tunnel collapses and retrogressive thaw
slumps. Permafrost degradation in this area will eventually
initiate the development of baydjarakh.
Figure 17 : The zone of low/null thermo-erosion process is
the largest zone of this gully and is characterized by the
presence of baydjarakhs, pools, alluvial levees, surface
lowering and channel enlargement.
The delineation is also function of when the survey is
accomplished, which is a snapshot of the gully system for
a given moment. To illustrate this concept, we may
consider that during the first year after gully initiation, the
whole gully was heavily under the effect of thermo-
erosion, and sinkholes and gully head were very active.
On the contrary, not enough time has passed for ice-
wedge polygon boundary and the polygon center to
develop baydjarakhs. In the current case, ten year after
gully formation, the localization of baydjarakhs is in the
oldest part of the gully: the high thermo-erosion zone is
progressively evolving into a low to null thermo-erosion
zone, while passing by the intermediate stage.
5.2 Effects of positive and negative feedback processes
on gully development
5.2.1 Observed positive feedback mechanisms
Feedback mechanisms are an important factor in gully
formation. Positive feedback mechanisms are contributing
to accelerate gullying under the action of thermo-erosion,
particularly in the high thermo-erosion zone and near the
upstream boundary of the intermediate zone. Water
infiltration in sinkholes initiates the melting of an ice-
wedge polygon boundary toward the gully main axis. This
new path for water in a gently sloped polygon terrace will
capture water and maintain or enhance the thermo-
erosion process in its immediate area. A stabilized section
of the gully can experience the reactivation of the thermo-
erosion processes under the effects of thermal and
mechanical action of pond drainage and cause new
retrogressive thaw slumps and ice-wedges exposition. A
winter season where a thick snow blanket accumulates or
if a rapid snowmelt happens during late spring create
positive feedbacks effects due to the warmer ground
temperature related to the insulating effect of snow and
especially due to the above average and rapid input of
water in the gully system.
1546
5.2.2 Observed negative feedback mechanisms
On the other hand, negative feedback mechanisms are
affecting the area of low thermo-erosion action near the
gully outlet and the lower boundary of the intermediate
zone. Retrogressive thaw slump or collapse material
transported by water can be deposited near the gully
outlet and can contribute to the formation of alluvial levees
and meanders. Alluvial levees retain water in pools and
prevent its free flowing to the active part of the gully
channel. Water circulating in the meanders is at a
distance to ice-rich polygons which prevent further from
thermo-erosion action. Enlargement of gully walls
channels by retrogressive thaw slumps decrease the
possibilities of rapid thermo-erosion of walls downstream.
Sediments that are not removed by water in large
channels near the gully outlet contribute to deactivation of
old thaw slumps and to slope stabilization. Drainage of
stabilized thaw-slump and slope colonization by plants
then contribute to the development of an insulation layer
and eventually to permafrost recovery and final
stabilization of old sections of the gully.
6 CONCLUSIONS
Our long term observations of the gully case-study on
Bylot Island revealed that geomorphic features in the gully
are function of the level of thermo-erosion currently active
near these features. The quantification and localization of
each geomorphic feature makes possible to characterize
the gully in function of its thermo-erosion level of activity.
The gully walls are steeper in active thermo-erosion zone
and very gently sloped in older stabilized part. Gully
evolution in the past 10 years shows that the speed of
gully development is not linear and that the last few years
of progression were much slower than the first year. This
indicates that the threshold reached to trigger
underground thermo-erosion processes during the first
year can have impacts on the ecosystem for decades.
Negative feedback mechanisms contribute to stabilization
of old gully section whereas positive feedback
mechanisms ensure reactivation of gullying processes
over the years.
ACKNOWLEDGEMENTS
We would like to warmly thank the following individuals
and organisation for the help, contributions and support to
our project: PCSP, Parks Canada – Sirmilik National
Parks, NSTP, NSERC Fortier, NSERC-USRA Godin,
NSERC (M. Sc.) Godin, Gilles Gauthier (Laval
University), Northern Studies Center, ESRI, Alexandre
Guertin-Pasquier, Rachel Thériault, Naïm Perrault, Esther
Lévesque, Josée Turcotte.
REFERENCES
CDED [038C04, 038C05, 048D01, 048D08]. (2003).
Sherbrooke, Quebec: Government of Canada, Natural
Resources Canada, Earth Sciences Sector, Centre for
Topographic Information. Available:
http://www.geobase.ca/geobase/en/data/cded/index.ht
ml (Accessed July 07, 2010).
Environment Canada. 2002. Canadian Climate
Normals,1971–2000. Pond Inlet. Environment
Canada, Atmospheric Environment Service,
Downsview, Ontario [Web, Online, 17 February 2010]
Fortier D, Allard M, Shur Y. 2007. Observation of rapid
drainage system development by thermal erosion of
ice wedges on Bylot island, Canadian Arctic
Archipelago. Permafrost and Periglacial Processes,
18(3): 229-243.
Fortier D, Allard M, Pivot F. 2006. A late-Holocene record
of loess deposition in ice-wedge polygons reflecting
wind activity and ground moisture conditions, Bylot
Island, eastern Canadian Arctic. Holocene, 16(5): 635-
646.
Fortier D, Allard M. 2004. Late Holocene syngenetic ice-
wedge polygons development, Bylot Island, Canadian
Arctic Archipelago. Canadian Journal of Earth
Sciences, 41(8): 997-1012.
Horton RE. 1945. Erosional Development of Streams and
their Drainage Basins; Hydrophysical Approach to
Quantitative Morphology. Geological Society of
America Bulletin, 56(3): 275-370.
Smith, S. and Burgess, M.M., 2000. Ground temperature
database for northern Canada; Geological Survey of
Canada Open File Report #3954, 57 p.
van Everdingen R, ed. 1998 revised May 2005. Multi-
language glossary of permafrost and related ground-
ice terms. Boulder, CO: National Snow and Ice Data
Center/World Data Center for Glaciology.
Young, S and Judge, A.S., 1986. Canadian permafrost
distribution and thickness data collection: a
discussion; Proceedings of National Student
Conference on Northern Studies, W.P Adams and
P.G. Johnson eds., Ottawa, ON, CA, 1:223-228.
1547
