Impacts of recurring ice jams on channel geometry and geomorphology
in a small high-boreal watershed
Étienne Boucher a,
, Yves Bégin b
, Dominique Arseneault c
a
Centre d'études nordiques and Département de géographie, Université Laval, Québec, Canada G1V 0A6
b
Centre d'études nordiques and Centre Eau, Terre & Environnement, Institut national de la recherche scientifique, Québec, Canada G1K 9A9
c
Centre d'études nordiques and Département de biologie, chimie et géographie, Université du Québec  Rimouski, Rimouski, Canada G5L 3A1
a b s t r a c ta r t i c l e i n f o
Article history:
Received 23 October 2008
Received in revised form 2 February 2009
Accepted 3 February 2009
Available online 20 February 2009
Keywords:
Ice jams
Fluvial geomorphology
Hydraulic geometry
Dendrogeomorphology
Ice scars
High boreal
River ice jams are generally perceived as significant erosive events and are well known to impact both
channel morphology and geometry. However, the extent of these impacts and the frequency of events
required to maintain erosion-induced morphologies remain unexplored in most cold region watersheds. In
this study, we investigated downstream variations in channel width, cross-sectional area, depth, and
geomorphological characteristics in a small high-boreal basin. We coupled these observations to
dendrochronological data on ice jam frequency. Our results show that channels affected by ice erosion
appear enlarged and present an important retreat of the upper bank. Such enlarged channels present a
typical two-level, ice-scoured morphology when ice jams recur more often than once every 5 years. By
contrast, channels appear unaffected when ice jams are less frequent. These results suggest that ice jams
maintain ice-scoured and enlarged morphologies once a minimal frequency-of-occurrence threshold is
exceeded. We therefore conclude that ice jam frequencies should be taken into account in order to better
define the role of ice as a geomorphological agent in cold environments.
 2009 Elsevier B.V. All rights reserved.
1. Introduction
The size and shape of stream channels and their capacity to contain
flows of various size and frequency vary according to the hydrologic
regime and sediment supply (Wolman and Miller, 1960; Harvey,
1969). This useful concept forms the basis of the hydraulic geometry
theory (Leopold and Maddock, 1953) from which the following
principles can be drawn: (i) alluvial channels are maintained by
relatively frequent flows reaching the bankfull stage, and (ii) downstream
changes in channel characteristics at bankfull are related to
increasing discharge or contributory area. Hydraulic geometry relations
(i.e., functions relating discharge to channel characteristics)
were extensively used by engineers and geomorphologists to predict
channel size and to manage streams and floodplains of the temperate
climate (Singh, 2003).
Whether or not these hydraulic geometry relations hold for cold
region watersheds is much debated, mainly because of ice effects.
From an examination of 24 rivers in Alberta (Canada), Smith (1979,
1980) proposed that repeated episodes of ice abrasion and gouging of
the banks during breakup ice jams could be responsible for the
maintenance of enlarged and high capacity bankfull channels. More
specifically, these results suggested a region-specific discharge (Q)
threshold marking the occurrence of mobile ice and ensuing ice
scouring (Fig. 1A). Once this threshold is exceeded, ice abrasion on
banks and bed causes a step change in geometric properties of
channels such as width, cross-sectional area, or depth (Smith, 1980).
More recently, McNamara (2000) and Best et al. (2005) reinforced the
hypothesis of Smith (1979) by reporting a downstream hydraulic
geometry anomaly in the Kuparuk watershed, Alaska, that corresponds
to a transition from bedfast to floating ice (Fig. 1B).
However, in a discussion of the results of Smith (1979), Kellerhals
and Church (1980) identified some potential sources of error, such as
confounding fluvial terraces and genetic floodplains. These authors
insisted on the fact that additional processes such as channel
entrenchment, channel icings, and backwater effects behind ice
jams might have engendered geometric unconformities in Albertan
rivers. Moreover, the study of Smith (1979) did not include any data
describing how frequent ice jams must be to maintain such enlarged
channel forms. Smith (1980) suggested that, when a stream acquires
the capacity to fragment and mobilize its ice cover, enlarged
morphologies systematically develop. This may be an overly simplistic
assumption considering the fact that ice jam frequencies vary
importantly among stream reaches in a watershed (Beltaos, 1996;
Boucher, 2008). We must also consider that Smith (1979) did not
provide any geomorphological description, so the variety of processes
leading to channel enlargement remains unclear.
Geomorphology 108 (2009) 273­281
 Corresponding author. Present address: Centre Eau, Terre & Environnement, Institut
national de la recherche scientifique, Québec, Canada G1K 9A9. Tel.: +1 418 654 3823;
fax: +1 418 654 2600.
E-mail address: etienne.boucher@ete.inrs.ca (É. Boucher).
0169-555X/$ ­ see front matter  2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomorph.2009.02.014
Contents lists available at ScienceDirect
Geomorphology
journal homepage: www.elsevier.com/locate/geomorph
Despite the important implications of the findings of Smith (1979)
for fluvial geomorphology, river management, and engineering, very
few additional studies intended to investigate how ice dynamics
influence channel morphology and hydraulic geometry relations. An
important part of the literature on river ice impacts presented
geomorphological descriptions of fluvial landscapes that have apparently
been shaped by ice (Dionne, 1974, 1976, 1978; Mackay et al.,
1974; Hamelin, 1979; Prowse and Gridley, 1993; Dyke, 2000; Prowse,
2001; Ettema, 2002; Smith and Pearce, 2002; Walker and Hudson,
2003). However, few studies examined the spatial extent of these
landforms along channels (Ettema, 2002)--a necessary aspect to
assess the impact of ice dynamics on hydraulic geometry relations.
Moreover, the minimal frequency of ice-scouring events to maintain
such landforms remains largely unexplored, mainly because historical
data on ice jam occurrence is sparse, anecdotal, and highly sitespecific
in most cold region watersheds. Thus, additional data is
needed on bankfull geometry in rivers that undergo significant ice
effects during breakup in order to verify if hydraulic geometry
relations hold in these environments.
In this study, we investigated downstream variations of channel
geometry and geomorphology in a small, high-boreal watershed that
experiences frequent spring ice jams. We used tree-ring chronologies
of ice jams, cross-sectional analysis, aerial photographs, and detailed
geomorphological descriptions to determine if geometric and geomorphologic
properties of channels relate to variations in the
frequency of ice-scouring events.
2. Study area
The Necopastic River (53°73 N., 78°28 W.) is an ungauged, highboreal
watercourse located 40 km west of Radisson, James Bay,
northern Québec. Its small hydroclimatically homogeneous watershed
drains an area of 250 km2
, and the bankfull discharge in the lower
reach is about 25 m3
/s (estimated from a rating curve). Climate
normals (1971­2001) at the La Grande Rivire A station displays a
typical high-boreal continental climate with mean annual temperatures
around -3.1 °C (Environment Canada, 2008). Minimum
(-23.2 °C) and maximum (13.2 °C) monthly averages are reached
in January and July, respectively. The total annual precipitation
is 684 mm, with 37% (248 mm) falling as snow. The months of
December, January, and February each have an average of less than
Fig. 1. Schematic representation of an idealized Q threshold separating between icescoured
and normal channels. Adapted from (A) Smith (1980) and (B) Best et al.
(2005).
Fig. 2. The Necopastic watershed. Studied sites are represented by circles and squares. Black symbols denote sites where ice-scouring events have occurred in the past (see Fig. 3).
White circles represent wide floodplain channels with no ice-scouring evidence (cluster 1, see Results). Black circles correspond to entrenched channels that experienced frequent ice
jams (cluster 2). Black squares refer to entrenched sites that display an ice-scoured morphology (cluster 3). The shaded gray zone corresponds to the extent of lithologically
homogeneous fluvioglacial deposits. These Holocene deposits are mainly composed of sands and silts. The symbology described here applies to the rest of the paper.
274 É. Boucher et al. / Geomorphology 108 (2009) 273­281
1 day with a temperature over 0 °C. In contrast with rivers of the coldtemperate
climate where extreme ice jam events frequently occur
during winter warm spells (Beltaos and Prowse, 2001; Beltaos and
Burrell, 2003), ice jamming is most likely to occur during the spring
flood on the Necopastic River.
Regional vegetation is characterized by the codominance of black
spruce [Picea mariana (Mill. BSP)] and Jack pine [Pinus banksiana
(Lamb)] (Arseneault and Sirois, 2004). In riparian areas, black spruce
shares its dominance with eastern larch [Larix laricina (Du Roi) K.
Koch] and balsam fir [Abies balsamea (L.) Mill] (Arseneault and Sirois,
2004; Bouchon and Arseneault, 2004; Boucher et al., 2006; Arseneault
et al., 2007). Ice scouring tolerant shrubs such as Alnus viridis ssp.
crispa (Ait.) Turrill. Salix planifolia Pursh. and Betula glandulosa
Michx cover the riverbanks and floodplains of most streams in the
area.
3. Materials and methods
3.1. Studied sites
The Necopastic River watershed can be separated into three classes
following major increases in drainage area (upper basin, mid-basin,
lower basin) and into two groups according to the type of ice cover
breakup (i.e., thermal versus mechanical breakups) (Fig. 2). At each
site, ice scars on shrubs (Fig. 3) indicated the presence of ice jams. A
total of 39 sites were chosen along the Necopastic River: 16 in the
upper basin, 16 in the mid-basin, and 7 in the lower basin. All sites are
incised into geologically homogeneous Holocene fluvioglacial deposits
(Vincent, 1985) (Fig. 2) and correspond to relatively rectilinear
stream sections of constant slope, width, depth and cross sectional
area. Sites recently flooded by the Canadian beaver (Castor canadensis
Kuhl.) or located immediately downstream or upstream of a sill were
avoided.
3.2. Geometric and geomorphologic descriptions of sites
During fieldwork, we first identified the bankfull stage that we
defined as the height of the relatively flat depositional surface
adjacent to the river (Wolman and Leopold, 1957; Williams, 1978).
On the Necopastic River, these depositional surfaces (in some cases)
correspond to the height of the valley flat (Fig. 4A). However, with
entrenched channels, we carefully distinguished the floodplains from
the less frequently flooded terraces. In such sites, genetic floodplains
are narrow and form a lower bench that is traceable along the river
channel (Fig. 4B).
At each site, one representative cross section was mapped using a
Leica TC-705 theodolite (precision=2 mm2 ppm). These cross
sections were orientated perpendicularly to streamflow direction and
encompassed the full valley width. We then computed an algorithm in
MATLAB R14
to interpolate between measurements and calculate
bankfull channel widths (Wbkf), cross sectional areas (Abkf), depths
(Dbkf), and flat widths (Wflat) (Fig. 4). An entrenchment ratio
(ER=Wflood/Wbkf where Wflood =valley width at 2× Dbkf) was
calculated after Rosgen (1996). Theodolite measurements were finally
used to evaluate the slope of the water surface and the bankfull
stage. The slope values in each site were obtained by regressing a
minimum of five water elevation measurements on the distance
downstream. The slope gradient was very low (b0.001 m/m) in most
sites (Fig. 5).
In addition to the field data, supplementary valley-flat width
measurements were performed on each Strahler orders using
1:10,000 aerial photos (Société d'Énergie de la Baie James, 1984).
Although multiple lithologies are encountered throughout the basin
and might account for some variations, our intention was to get a
general portrait of how flood-prone widths (i.e., valley flats) vary
downstream. Flat widths correspond, on aerial photography, to the
easily recognizable zone colonized by Alnus viridis, Salix planifolia,
and Betula glandulosa (Bouchon and Arseneault, 2004; Boucher et al.,
2006). Three width measurements were averaged in each stream
section of the basin. Overall, 78 averaged widths were calculatedFig. 3. Shrubs scarred by drift ice.
Fig. 4. Geometric properties and identification of the bankfull stage. (A) Stable and
(B) entrenched channels. Wbkf, Wflat, Abkf, and Dbkf refer to bankfull width, valley flat
width, bankfull cross-sectional area, and bankfull depth, respectively.
275É. Boucher et al. / Geomorphology 108 (2009) 273­281
in first--(N=57), second--(N=15), third--(N=3), fourth--(N=2),
and fifth--(N=1) order streams.
All geometric measurements (Wbkf, Wflat, Abkf, Dbkf) were plotted
against the contributive area (C) instead of discharge (Q) because the
Necopastic River is ungauged. Contributive area measurements were
performed within PHYSITEL (Royer et al., 2006). The PHYSITEL
software determines the drainage structure of a basin from a Digital
Elevation Model (DEM). A DEM (cell=20×20 m) of the Necopastic
River watershed was constructed from SRTM elevation data (http://
srtm.usgs.gov/). Hydraulic geometry power functions and equations
were computed in SPSS
for Windows.
Geomorphological features and processes were described at each
site using the reconnaissance survey technique (Thorne et al., 1996;
Thorne,1998). Data collected include valley description (valley height,
shape, angle, coupling of channel to floodplain, valley floor type,
surficial geology, number of terraces, extent of flood deposits, channel
platform, lateral activity, etc.), channel description (dimensions, flow
type, bed sediments, bed armouring, width controls), and left and
right bank description (bank height, material, slope, shape, vegetation,
status and processes of erosion and accumulation, etc.).
Geometric data and geomorphologic descriptions were incorporated
into a two-step cluster analysis (TSCA) to reveal groupings among
sites, following a log-likelihood distance measure. Geometric data was
detrended using a nonlinear regression model. Only residuals were
considered in order to remove any downstream change attributable to
drainage area. Furthermore, only significantly discriminating geometric
or geomorphic variables (pb0.05 for t or 2
tests) will be
presented and discussed here.
3.3. Tree-ring reconstructed ice jam frequencies
Trees bordering the Necopastic River present multiple ice scars
(Boucher, 2008), especially at mid-basin. These scars form on trees
when drift ice raised by jams comes in contact with riparian trees.
These marks can be dendrochronologically dated and exhaustive
event chronologies can be constructed (Payette, 1980; Smith and
Reynolds, 1983; Hupp, 1988; Boucher, 2008). On the Necopastic River,
scars were sampled in 15 out of the 16 mid-basin sites. None were
sampled in the lower basin because this section is part of a protected
area. The number of trees sampled varied considerably between sites
(Table 1). To overcome this problem, Boucher (2008) established an
iterative sampling-with-replacement (ISR) procedure to determine if
the at-a-site sampling effort was sufficient. Three sites presented an
insufficient sampling and were not considered here (Table 1).
In the field, when ice marks were well defined, cross sections were
taken in the middle of each scar. Otherwise, when multiple scars were
present on the trunk, cross sections were taken at multiple heights in
order to date every possible scar on the tree. Trees with a diameter
b10 cm were never sampled. Individuals were sampled on both banks
and were separated by a minimal distance of one tree height to avoid
redundancy in the record. In the laboratory, cross sections were finely
sanded and tree rings were counted from the last year of growth
(2005 or 2006) to the center. Dead samples were cross-dated with
master chronologies existing in the area. Ice-scouring events were
dated with the precaution of not replicating events found at multiple
heights on a same individual. Scars that were formed after year A.D.
2003 were excluded because newly formed damages are difficult to
observe in the field.
For each site, ice jam frequency was calculated by dividing the
number of events recorded by the period available for recording. Sites
were considered available for recording when at least one tree had
grown over 3 cm of radius. We attributed a frequency of "zero" to one
mid-basin site that did not show any geomorphological nor tree-ring
evidence of ice scouring (Fig. 2).
4. Results
In the Necopastic River watershed, an important part (N50%) of the
variations in Wbkf and Abkf is explained by the downstream increase in
contributive area (Fig. 6A,B). However, only a weak relationship is
observed with channel depth (Dbkf) (Fig. 6C). Furthermore, from a
detailed examination of power functions modeling downstream
changes in width and cross-sectional area (Fig. 6A,B), considerable
variability is observed at mid-basin (for contributive areas of 100 to
150 km2
). For example, in the upper basin, average Wbkf and Abkf are
7.41.6 m and 7.93.6 m2
, respectively. At mid-basin, these values
respectively rise to 19.74.8 m and 24.28.7 m2
, therefore
approaching values of the lower basin (i.e., 19.33.5 m and 23.4
5.7 m2
). Observed downstream channel geometry variations are not
related to changes in channel gradient (rs b0.2, pN0.05 with Wbkf,
Abkf, and Dbkf).
Mid-basin variations in Wbkf and Abkf are significantly related to ice
jam frequencies (rs =0.61 and 0.54, respectively; Fig. 7A,B), while
Fig. 5. Water slope values as a function of contributive area. Each dot represents a study site.
Table 1
Characteristics of study sites where dendrochronological sampling was conducted;
missing sites were eliminated because of insufficient sampling (Boucher, 2008).
Site Date of site availability
(year A.D.)
Number of trees
sampled (N)
Frequency
(N y-1
)
1 1869 18 0.66
2 1945 15 0.41
3 1871 11 0.14
4 1936 20 0.24
5 1914 29 0.51
6 1798 9 0.23
7 1856 7a
0.20
10 1865 8 0.25
11 1917 5a
0.13
12 1877 15 0.23
13 1870 14 0.18
14 1879 8 0.23
16 ­ ­ 0
Mean 0.26
a
All trees bearing scars have been exhaustively sampled in these sites.
276 É. Boucher et al. / Geomorphology 108 (2009) 273­281
variations in channel depth are not. Narrow channels are associated
with low ice jam frequencies (~0­0.2 event year-1
on the abscises) in
comparison with larger channels (~0.2­0.6 event year-1
on the
abscises).
Wflat tend to decrease in the downstream direction (Fig. 8A). This
contrasts with the previously described tendency of bankfull channels
to become larger at mid-basin (Fig. 6A,B). Upstream sites present
relatively wide valley flats (114.7 m) in comparison with narrower
flood corridors of the mid (6.72.5 m) and lower (82.9 m) basins.
The latter situation is also visible on 1:10,000 aerial photographies.
Wflat are the narrowest in orders 4b and 5 where frequent ice jams
occur (Fig. 8B).
Following the TSCA, studied sites were classified into three groups
according to the most influential morphological and geometrical
variables. The first group (cluster 1) is found almost exclusively in the
upper basin (Fig. 2) and represents wide floodplain channels (Fig. 9A).
Field observations confirm the existence of extensive depositional
areas adjacent to the channel in this group (Fig.10A,B). High ER values
(N2.4) point out that these sites are vertically stable (Fig. 9B). Slightlybelow-zero
residual widths and cross-sectional areas (Fig. 9C)
indicate that the hydraulic geometry curve fits observed values quite
well. Thus, we can deduce that stream sections of clusters 1 and 2 are
probably laterally stable. Moreover, ice scouring activity is not present
in these channels, as evidenced by the absence of ice scars on the
vegetation (Fig. 9D) and no obvious ice-scoured morphology (Fig. 9E).
The two other groups (clusters 2 and 3) represent entrenched
channels of the mid and lower basins. In both groups, low Wflat
(b10 m) values and small ER ratios (b2) are indicative of incised
channels (Fig. 9A,B). The omnipresence of ice scars reveals that ice
jams occurred in all sites (Fig. 9D). However, a major difference
between clusters 2 and 3 is that, whilst the former shows no particular
evidence of lateral erosion (Fig. 10C,D), the latter depicts abnormally
large channels (Fig. 9C) and presents a distinct ice-scoured morphology
(Fig. 9E). Ice-scoured channels are characterised by a discontinuous
and partly eroded genetic floodplain forming a narrow bench
(Fig. 10E,F(i)). This bench is covered with heavily scarred shrubs installed
on freshly deposited alluvium. A steep eroded talus (Fig.10F(ii))
(about 5014 cm high) separates the flat depositional surface from a
higher terrace. Mechanical erosion on this talus has excavated tree and
shrub roots living on the higher terrace (Fig. 10F(iii)). Recent alluviums
are often found at the base of scarred trees, indicating that these
terraces might slowly aggrade.
Finally, the ice-scoured morphology (cluster 3) is most often
associated with sites that experienced frequent ice jams in the past.
Where ice jam events occurred at least once every 5 years (frequency
Fig. 6. Downstream changes in geometric properties of channels (Wbkf, Abkf, Dbkf).
Symbols correspond to those used in Fig. 2.
Fig. 7. Relationship between channel geometry at mid-basin and ice jam frequencies.
Only sites where the sampling was sufficient were considered (see Table 1). Symbols
correspond to those used in Fig. 2.
Fig. 8. Downstream changes in valley flat widths in the Necopastic watershed. (A)
Valley flat widths measured in the studies sites. (B) Mean (S.D.) valley flat widths
measured on 1:10,000 aerial photographs for each Strahler order. Black squares
represent the contributive area of each Strahler order. Order 4 was divided into 4a and
4b because ice scouring occurs in the latter but not in the former.
277É. Boucher et al. / Geomorphology 108 (2009) 273­281
~0.2 event year-1
), most sites developed an ice-scoured cross section
(Fig. 7A,B).
5. Discussion
5.1. Are ice jams really causing enlargement?
In the Necopastic River watershed, recurring ice jams seem to
influence the geometric and geomorphologic properties of channels,
resulting in anomalously large and heavily scoured sites. However, as
pointed out by Best et al. (2005), further discussion is needed to
determine whether channel enlargement is truly caused by ice jams
or, on the contrary, if ice movements resulting in jams are just
permitted in channels that are already enlarged by other environmental
processes. Among the factors other than ice that could possibly
influence downstream variations in channel width are longitudinal
changes in slope, lithology, and climate (Singh, 2003). In this study,
the sampling strategy was specifically designed to minimize the
influence of these variables on channel geometry. First, the Necopastic
River basin can be considered hydroclimatically homogeneous
because of its small size (b250 km2
). Second, sampling sites are all
located within a Holocene fluvioglacial terrace complex composed of
sandy silts (Fig. 2). As with other small streams flowing in these
geologically uniform environments, the Necopastic is just reworking
"previously-sorted" Holocene deposits; consequently, floodplain
(modern) and terrace (inherited) sediments are indistinct granulometrically
(Boucher et al., 2006). Third, downstream variations in
channel slope are not related to geometric properties of channel in our
system, and the gradient of all studied sites is very low (bthan
0.001 mm- 1
) (Fig. 4). Hence, downstream channel geometry
variations are more likely to be attributable to characteristics of the
hydrological regime (including ice effects) than to "permanent"
variables affecting geomorphological processes in the watershed.
5.2. Ice jam frequency (F) thresholds
This study only partly supports the hypothesis of Smith (1979)
regarding the role of ice jams as significant erosive processes in coldenvironment
rivers. We agree with Smith (1979, 1980) that ice jams
may alter channel geometry and geomorphology, especially when
channels are entrenched. However, in the Necopastic River, thresholds
in the frequency (F) of ice jams seem to have a greater influence on
channel geomorphology compared to discharge (Q) or contributive
area (C) thresholds documented in earlier works (Smith, 1979, 1980;
McNamara, 2000; Best et al., 2005). Such thresholds imply that, once a
critical value is exceeded, streams systematically display an enlarged
or scoured morphology (Fig. 1A,B) regardless of the intersite
variations in ice jam frequency. In the Necopastic watershed,
mechanically induced ice jams become common when the drainage
area exceeds 120 km2
(Fig. 2). However, downstream of this C
threshold, all sites do not appear systematically scoured or enlarged.
On the contrary, many mid-basin and almost all lower basin sites
appear uninfluenced by ice (i.e., cluster 2) (Figs. 6A,B and 10C,D)
despite the abundant tree-ring evidence for ice jamming in these
environments. As demonstrated earlier, the enlarged morphology
(cluster 3) becomes omnipresent in sites where ice jams recur more
often than once every 5 years (Fig. 7A,B).
An important implication of these findings is that hydraulic
geometry relations might only hold in sites where ice jams are less
frequent than the F threshold. In other words, downstream variations
in ice jam frequency hinder the precise estimation of geometric
properties from hydraulic geometry curves in ice jam-prone basins.
For example, in the Necopastic River watershed, the downstream
increase in contributive area does not explain the large geometric
variability occurring at mid-basin (Fig. 6A,B). Furthermore, downstream
changes in geometric properties would probably be more
accurately described from hydraulic geometry curves if ice-scoured
channels (cluster 3) were not considered (Fig. 6A,B). To verify this, R2
values for these models were recomputed after removing cluster 3
sites. As expected, the explained variance jumped to 0.838 and 0.704
for Wbkf and Abkf, respectively.
Finally, channel entrenchment and basin hydrography are two
factors that explain why the ice jam F threshold is exceeded only in
some mid-basin sites and not downstream. Ice jams are well known to
occur more frequently where channel morphology precludes ice
movement, a very common situation in medium-sized flat reaches
(Beltaos, 1996). At mid-basin, ice movements are constrained in a
Fig. 9. Geometric and geomorphologic variables discriminating between the three
clusters (1, 2, 3) produced by the TSCA. Only variables that contribute significantly to
discriminate between clusters are shown [pb0.05 for t-tests (A, B, C) and 2
tests (D, E)].
Cluster 1 represents wide floodplain channels. Cluster 2 corresponds to entrenched/
narrow-floodplain channels. Cluster 3 includes entrenched/ice-scoured channels.
278 É. Boucher et al. / Geomorphology 108 (2009) 273­281
Fig. 10. Graphical and pictorial representations of the three most common geomorphological environments encountered in the Necopastic River: (A, B) wide floodplain channels; (C, D) entrenched/narrow floodplain channels, and (E, F)
entrenched/ice-scoured channels. Genetic floodplains (benches), erosion talus, and ice-flood terraces with ice-scarred trees are identified as i, ii, and iii, respectively.
279É.Boucheretal./Geomorphology108(2009)273­281
highly entrenched channel. Consequently, ice rafts frequently accumulate
in nearby sections during mechanical breakups. In the lower
basin, ice rafts possibly move farther downstream because of
increased channel sizes and higher discharge. Thus, jam-prone
sections tend to distance from one another so that F thresholds are
probably rarely crossed in the lower basin.
5.3. Bank scouring processes
In sites where the F threshold is crossed, repeated ice-scouring
events caused the riverbanks to retreat (Fig. 12A,B). Scouring in these
entrenched channels is probably very severe because the system's
energy cannot dissipate laterally. Moreover, the riverbank's top is
extremely vulnerable to erosion as it is often free of snow and partly
unfrozen at the moment of breakup (Fig. 11A). For its part, the lower
bench appears constricted and very narrow but has not been
completely eroded. A first explanation is that water levels rise much
higher than the bankfull stage during ice jams. From a detailed
analysis of ice scar heights in this watershed, it was shown that water
levels rise, in average, 1 m above the bankfull stage during these
events (Boucher, 2008). Remnants of the initial ice cover may also be
attached to the riverbank toe, therefore conferring additional
protection to the lower bench (Smith, 1980; Ettema, 2002). Nevertheless,
the resulting two-level channel is similar to the one documented
by Smith (1980).
During flood recession, sediments are deposited on the lower
bench forming an inclined surface of loose material (Figs. 10F(iii)
and 12C). These recently deposited sediments are very sensitive to
erosion and are rapidly reworked into a lower bench by subsequent
floods during the ice-free period (Fig. 12C). The surface of the lower
bench roughly corresponds to the bankfull stage. Although vegetation
encroaches and stabilizes this aggradational surface, fluvial erosion
steepens and eventually undercuts the lower bench. Between each
ice-scouring event, benches can experience several depositional
episodes during open-channel floods.
Finally, ice dynamics may explain the presence of wide floodplains
in the upper basin. In high-boreal regions, severe winters generate
thick ice covers that often remain still during breakup in low-order
streams. An important consequence is that spring flood waters
frequently overtop the ice cover in the upper basin and inundate
the full flood-prone area (Fig. 11B). The ice cover then melts in situ
from thermal exchanges with flood waters and the atmosphere. In
Fig. 11. The Necopastic River just before spring breakup in 2006. (A) In downstream
sites, the top of riverbanks is often free of snow (white arrows), therefore augmenting
their vulnerability to erosion. (B) In upstream sites, flood waters are forced on top of the
ice cover and inundate the full flood-prone width without eroding the banks.
Fig. 12. Bank retreat mechanisms in portions of the Necopastic River that are frequently
affected by ice jams. (A) Initial stage, (B) ice jam erosion, (C) deposition of sediments on
the lower bench during flood recession, and (D) reworking of fresh alluviums by
hydrometeorological floods occurring during the ice-free period in these entrenched
channels.
280 É. Boucher et al. / Geomorphology 108 (2009) 273­281
combination with flood events occurring during the ice-free period,
this phenomenon probably contributes to widening the flood-prone
area. A similar process was documented for the Kuparuk basin (Best
et al., 2005) and is possibly representative of many small streams of
the cold areas (Kellerhals and Church, 1980).
6. Conclusion
Our results demonstrate that hydraulic geometry relations lead to
imprecise width (Wbkf) and cross-sectional area (Abkf) estimations in
the Necopastic River, a small stream that experiences frequent ice
jams. However, the hypothesis of Smith (1979) regarding the role of
ice jams as important and generalized erosive events in high-latitude
rivers is only partly supported by our data. In our study, ice jams have
to occur at least once every 5 years to maintain an enlarged and icescoured
morphology. Thus, our data suggests that some frequency-ofoccurrence
thresholds might have to be crossed for ice jams to become
geomorphologically significant in the Necopastic River basin. Indeed, a
frequency­magnitude approach would be more useful than a
presence/absence assessment to quantify the geomorphic impacts
of these extreme events on riparian environments.
Among the visible geomorphic impacts on the Necopastic River is a
"two-level" channel occurring in frequently ice-scoured sites. Our
study provided new insights on the mechanisms forming these
enlarged two-level channels. It is suggested that both ice jams and the
yearly open-channel flood might be involved in shaping these
landforms. First, ice-jam-related flooding laterally erodes the riverbank.
Second, alluviums are deposited on the lower part of the
riverbank during flood recession. Third, sediments are reworked into a
flat bench by hydrometeorological floods occurring during the ice-free
period.
Acknowledgements
We would like to thank Marilou Girard-Thomas, Mathieu Roy, and
Marc-André Robin for the field assistance. This research could not
have been successful without the technical assistance of Sylvain Jutras,
Patrick Gagnon and Alain Rousseau. We gratefully acknowledge the
following organisations for their financial support: NSERC (National
Science and Engineering Research Council of Canada), FQRNT (Fonds
Québécois de Recherche sur la Nature et les Technologies); HydroQuébec,
Ouranos, ArcticNet and the Northern Student Training
Program of the Department of Indian and Northern Affairs of Canada.
The town of Radisson (James Bay) also provided lodging facilities for
this study. The quality of the original manuscript was improved by the
valuable comments of two anonymous reviewers.
References
Arseneault, D., Sirois, L., 2004. The millenial dynamics of a boreal forest stand from
buried trees. Journal of Ecology 92, 490­504.
Arseneault, D., Boucher, É., Bouchon, É., 2007. Asynchronous forest-stream coupling in a
fire-prone boreal landscape: insights from woody debris. Journal of Ecology 95,
789­801.
Beltaos, S., 1996. River Ice Jams. Water Resources Publications, Highland Ranch,
Colorado, USA. 390 pp.
Beltaos, S., Prowse, T.D., 2001. Climate impacts on extreme ice jam events in Canadian
rivers. Hydrological Sciences Journal 46, 157­182.
Beltaos, S., Burrell, B.C., 2003. Climate change and river ice breakup. Canadian Journal of
Civil Engineering 30, 145­155.
Best, H., McNamara, J.P., Liberty, L., 2005. Association of ice and river channel
morphology determined using ground-penetrating radar in the Kuparuk River,
Alaska. Arctic, Antarctic and Alpine Research 37, 157­162.
Boucher, É., 2008. Analyse hydro-climatique et géomorphologique des déglacements
mécaniques de la rivire Necopastic au Québec nordique. Department of
Geography, Laval University, Québec, Canada. 153 pp.
Boucher, É., Arseneault, D., Hétu, B., 2006. Late Holocene development of a floodplain
along a small meandering stream, northern Quebec, Canada. Geomorphology 80,
267­281.
Bouchon, E., Arseneault, D., 2004. Fire disturbance during climate change: failure of
postfire forest recovery on a boreal floodplain. Canadian Journal of Forest Research
34, 2294­2305.
Dionne, J.-C., 1974. How drift-ice shapes the St-Lawrence. Canadian Geographic Journal
88, 4­9.
Dionne, J.-C., 1976. Le glaciel de la région de La Grande rivire, Québec subarctique.
Revue de Géographie de Montréal 30, 133­153 (in French).
Dionne, J.-C., 1978. Le glaciel en Jamésie et en Hudsonie, Québec subarctique. Revue de
Géographie physique et Quaternaire 32, 3­70 (in French).
Dyke, L.D., 2000. Shoreline permafrost along the Mackenzie River. In: Dyke, L.D., Brooks,
G.R. (Eds.), The Physical Environment of the Mackenzie Valley, North-West
Territories; A Base Line for the Assessment of Environmental Change. Geological
Survey of Canada, Ottawa, Canada, pp. 187­195.
Environment Canada, 2008. Canadian Climate Normals and Averages, 1971­2000.
http://www.climat.meteo.ec.gc.ca/climate_normals/index_e.html?.
Ettema, R., 2002. Review of alluvial-channel responses to river ice. Journal of Cold
Regions Engineering 16, 191­217.
Hamelin, L.E., 1979. The bechevnik: a river bank feature from Siberia. The Musk Ox 25,
70­72.
Harvey, A.M., 1969. Channel capacity and the adjustment of streams to hydrologic
regime. Journal of Hydrology 8, 82­98.
Hupp, C.R., 1988. Plant ecological aspects of flood geomorphology and paleoflood
history. In: Baker, V.R., Kochel, R.C., Patton, P.C. (Eds.), Flood Geomorphology. John
Wiley & Sons, New-York, pp. 335­356.
Kellerhals, R., Church, M., 1980. Comments on "Effect of channel enlargement by river
processes on bankfull discharge in Alberta, Canada" by D.G. Smith. Water Resources
Research 16, 1131­1134.
Leopold, L.B., Maddock, T., 1953. The hydraulic geometry of stream channels and some
physiographic implications. U.S. Geological Survey Professional Paper No. 252,
Washington, DC, pp. 1­56.
Mackay, D.K., Sherstone, D.A., Arnold, K.C., 1974. Channel Ice Effects and Surface Water
Velocities from Aerial Photography of Mackenzie River Break-up, Hydrologic
Aspects of Northern Pipeline Development. Glaciology Division, Water Resources
Branch. Department of the Environment, Ottawa, Canada, pp. 71­107.
McNamara, J.P., 2000. Bankfull Flow, Hydraulic Geometry, and River Ice in a
Northern River. AWRA Water Resources in Extreme Environments, Anchorage,
AK, pp. 191­196.
Payette, S., 1980. Les grandes crues glacielles de la rivire aux feuilles (NouveauQuébec):
une analyse dendrochronologique. Le Naturaliste Canadien 107, 215­225
(in French).
Prowse, T.D., 2001. River-ice ecology: hydrologic, geomorphic and water-quality
aspects. Journal of Cold Regions Engineering 15, 1­16.
Prowse, T.D., Gridley, N.C., 1993. Environmental Aspects of River Ice. NHRI Science
Report no. 5. National Hydrology Research Institute, Saskatoon, SK, Canada. 155 pp.
Rosgen, D., 1996. Applied River Morphology. Wildlife Hydrology, Pagosa Springs, CO.
390 pp.
Royer, A., Rousseau, A.N., Savary, S., Fortin, J.-P., Turcotte, R., 2006. PHYSITEL, un SIG pour
la mise en place de modles hydrologiques, 2e Symposium Scientifique Ouranos,
Montréal.
Singh, V.P., 2003. On the theories of hydraulic geometry. International Journal of
Sediment Research 18, 196­218.
Smith, D.G., 1979. Effects of channel enlargement by river ice processes on bankfull
discharge in Alberta, Canada. Water Resources Research 15, 469­475.
Smith, D.G., 1980. River ice processes: thresholds and geomorphologic effects in
northern and mountain rivers. In: Coates, D.R., Vitek, J.D. (Eds.), Thresholds in
Geomorphology. Allen & Unwin, London, pp. 323­343.
Smith, D.G., Reynolds, D.M., 1983. Tree scars to determine the frequency and stage of
high magnitude river ice drives and jams. Canadian Water Resources Journal 8,
77­94.
Smith, D.G., Pearce, C.M., 2002. Ice jams-caused fluvial gullies and scour holes on
northern river flood plains. Geomorphology 42, 85­95.
Société d'Énergie de la Baie James, 1984, Aerial photographs, 1:10 000, Sept. 17, Lines
152 to 266.
Thorne, C.R., 1998. Stream Reconnaissance Handbook: Geomorphological Investigation
and Analysis of River Channels. John Wiley, New York. 133 pp.
Thorne, C.R., Allen, R.G., Simon, A., 1996. Geomorphological river channel reconnaissance
for river analysis, engineering and management. Transactions of the Institute
of British Geographers 21, 469­483.
Vincent, J.S., 1985. Surficial Geology, Radisson, Québec, 1:100 000. Geological Survey of
Canada. "A" Series Map 1591A.
Walker, J.H., Hudson, P.F., 2003. Hydrologic and geomorphic processes in the Colville
River delta, Alaska. Geomorphology 56, 291­303.
Williams, G.P., 1978. Bank-full discharge of rivers. Water Resources Research 14,
1141­1154.
Wolman, M.G., Leopold, L.B., 1957. River floodplains: some observations on their
formation. U.S. Geological Survey Professional Paper no. 282-C, Washington, DC,
pp. 87­109.
Wolman, M.G., Miller, J.P., 1960. Magnitude and frequency of forces in geomorphic
processes. Journal of Geology 68, 54­74.
281É. Boucher et al. / Geomorphology 108 (2009) 273­281