® .Geomorphology 28 1999 141­168
The drainage of the Lake Ha!Ha! reservoir and downstream
geomorphic impacts along Ha!Ha! River, Saguenay area, Quebec,
Canada
G.R. Brooks )
, D.E. Lawrence 1
Geological Surey of Canada, 601 Booth Street, Ottawa, ON, Canada K1A 0E8
Received 8 November 1997; received in revised form 12 October 1998; accepted 31 October 1998
Abstract
On July 18­21, 1996, a severe rainstorm caused widespread flooding along the north shore of the St. Lawrence River,
southern Quebec, Canada, particularly along rivers that drain the area just south of the Saguenay­Lake St. Jean region. At
the Lake Ha!Ha! reservoir, inadequate available capacity to spill during the storm at the outlet dam resulted in the
overtopping and erosion of a nearby earthfill saddle dyke. A new outlet formed at the site of the dyke and drained the
reservoir over a period of many hours decreasing its area from 8.1 to 4.7 km2
. Estimates of discharge range from 910 to
3 y1 3 y1 ®1380 m s at the site of the eroded dyke to 1080 to 1260 m s at a location 27 km downstream about 8 km above the
.mouth of the river . The uncontrolled drainage of the Lake Ha!Ha! reservoir increased flooding along the lower 35 km of
Ha!Ha! River where flooding was already in progress because of the rainstorm runoff. The flooding caused extensive
® . ®geomorphic impacts along the river. Long sections of the river totalling 25 km experienced significant widening locally up
. ® . ® .to 280 m and channel incision locally up to 20 m while two reaches 6 and 4.5 km long experienced up to several metres
of aggradation. In general, the slope of the valley was the most important variable affecting whether or not the energy of the
flow was above or below the erosive threshold of the valley bottom. Locally, a permanent channel diversion now exists
where the drainage divide between the main river course and a small ravine was overtopped and extensively eroded.
Communities, infrastructure, and industry located along the river were extensively damaged by the flood waters. The effects
of flooding along Ha!Ha! River demonstrate that rivers on the Canadian Shield can undergo severe geomorphic changes
caused by very high-magnitude flooding. q 1999 Elsevier Science B.V. All rights reserved.
Keywords: flooding; dam breach; Canadian Shield; Quebec; Canada
1. Introduction
The uncontrolled drainage of a reservoir by the
failure of a dam 2
can produce a flash flood of
)
Corresponding author. Tel.: q1-613-996-4548; Fax: q1-613-
992-0190; E-mail: gbrooks@gsc.nrcan.gc.ca
1
Tel.: q1-613-995-7644; Fax: q1-613-992-0190; E-mail:
tlawrence@gsc.nrcan.gc.ca.
magnitude well beyond the `natural' flow regime of
a river or stream. Such floods can have catastrophic
effects on the geomorphology downstream in the
valley bottom and on communities and infrastructure
situated within the flood path. Numerous floods
2
The term `dam' is used here in a broad sense, and includes all
types of dams, embankments and dykes.
0169-555Xr99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
® .PII: S0169-555X 98 00109-3
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168142
caused by dam failures have been documented
worldwide with the flood magnitude in part being
®proportional to the impoundment size see Costa,
.1988 . Breaches in dams can occur for a variety of
reasons, including foundation failure, the effects of
®piping, inadequate capacity for spilling due to a
deficiency in either the spillway design or the opera-
.tion of the spillway gates , differential settlement,
®and poor construction and maintenance Jansen,
.1983 .
Flooding from a dam failure has not caused a
major disaster in Canada. Nonetheless, some signifi-
cant floods have occurred from dam failures, al-
though these are poorly documented. As examples,
on April 7, 1912, a dam owned by the Erindale
Power on the Credit River, near Toronto, Ontario,
Fig. 1. Map of the Saguenay area showing Ha!Ha! River and the Lake Ha!Ha! reservoir, consisting of Lake Ha!Ha! and Little Lake Ha!Ha!.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 143
breached during spring flooding as overtopping wa-
ter eroded the leeside of the dam causing failure of a
section of the concrete core, 40 m wide and 4.6 m
® .high Longley, 1912 . This breach was caused by the
inadequate capacity of the spillway and undoubtedly
compounded the flooding already underway, al-
though this is not mentioned in the description of the
failure. Piping lead to the breaching of the Scott
Experimental Farm Dam No. 2 in Saskatchewan, on
April 21, 1948 and resulted in flooding downstream
®without `significant' damage Lukey and Noonan,
. ® .1986 . Lapointe 1986 mentions that a dam `burst'
in 1966 killed three people at Saint-Joseph River,
near Quebec City, Quebec. Seepage problems led to
the September 18, 1985 failure of the Beloeil dam,
located in a sparsely populated area 120 km north of
Quebec City, Quebec. The failure drained Lake Be-
loeil and led to severe flooding and channel erosion
along a 7-km long reach of Little Pikauba River
® .Robitaille and Dubois, 1989; Robert and Pare, 1992 .´
Notably, all of these failures involved small dams
less than 15 m high.
In July 1996, eight dams along Chicoutimi, Sables
and Ha!Ha! rivers in the Saguenay area, Quebec,
were overtopped and breached by the floodwaters
®from a severe rainstorm described below; see Brooks
et al., 1997; CSTGB, 1997; INRS-Eau, 1997; Brooks
.and Lawrence, 1998; Fig. 1 . At all eight sites, the
impoundments were at least partially drained, how-
ever, seven of the reservoirs were relatively small,
essentially shallow, back-flooded reaches of river
channel. The breaches were small or occurred slowly,
over many hours. The draining of these reservoirs
did not contribute significantly to the downstream
flooding already in progress. The exception, how-
ever, occurred with the draining of the Lake Ha!Ha!
® .reservoir Fig. 1 which greatly accentuated the rain-
fall runoff along Ha!Ha! River.
The drainage of the Lake Ha!Ha! reservoir is the
most significant Canadian dam breach documented
in terms of downstream geomorphic and human im-
pacts. The flood also represents an important geo-
morphic event because it triggered an extreme flood
event along a Canadian Shield river system where
the effects of such flooding are poorly documented.
The purpose of this paper is to review events leading
to the draining of Lake Ha!Ha!, provide estimates of
the resulting flood discharge, and summarize the
downstream geomorphic impacts of the flood along
Ha!Ha! River.
2. The storm
Between July 18­21, 1996, a severe rainstorm
stalled over the Gulf of St. Lawrence and record
®amounts of rain fell on southern Quebec see Milton
.and Bourque, 1997 . The greatest accumulation of
® .rain greater than 200 mm fell in an area south of
the Lake St. Jean­Saguenay River area, Quebec
® .Fig. 2 . Orographic processes and the counter-clock-
wise rotation of the storm system were responsible
for the concentration of rain in this area. Individual
locations within this zone reported receiving accu-
®mulations of, for example, 210.9 mm Portages-des-
. ®Roches station 061001 , 271.9 mm Pikauba station
. ®061022 , and 279.4 mm Riviere-aux-Ecorces station`
. ® .061020 Milton and Bourque, 1997 . Although these
totals are for a 4 day period, most of the rain in this
area fell within a 36 h period beginning at about
0800 h on July 19 and continuing until about 2000 h
July 20.
This rainfall, in combination with the near-
®saturated ground from antecedent July rainfall En-
.vironment Canada, 1996 and the generally thin and
discontinuous overburden blanketing the bedrock of
the Laurentian Highlands, produced widespread
flooding throughout the north shore area of the St.
Lawrence River in southern Quebec. The most se-
vere flooding occurred along rivers flowing north-
wards into the Saguenay Valley whose headwater
areas are located within the zone of greater than 200
® .mm accumulation Fig. 2; see CSTGB, 1997 . The
flooding damaged or destroyed homes, cottages,
farms, businesses and infrastructure, and impaired
® .local industry CSTGB, 1997 .
3. Lake Ha!Ha!
The Lake Ha!Ha! reservoir, 12 km long and with
a surface area of 8.1 km2
, is located about 34 km
® .above the mouth of Ha!Ha! River Fig. 3 . It consists
of Lake Ha!Ha! and Little Lake Ha!Ha! which oc-
® .cupy separate, but connected lake basins Fig. 3 .
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168144
®Fig. 2. Map showing rainfall accumulation in southern Quebec between 0800 h July 18 and 0800 h July 21, 1996 after Milton and Bourque,
.1997, reproduced with permission .
The contributing area of the reservoir represents
® 2
37.5% of the Ha!Ha! River drainage basin 608 km ;
.CSTGB, 1997 . Before the water level was raised by
® .the construction of a dam and two dykes see below ,
the Lake Ha!Ha! and Little Lake Ha!Ha! basins
contained separate lakes.
Prior to the July 1996 rainstorm, the level of Lake
Ha!Ha! reservoir was controlled by a concrete grav-
ity dam located at the north end of Lake Ha!Ha!
® . ®Figs. 3­5 . Two earthfill saddle dykes known as
.the `Cut-away' and `Rive-gauche' dykes , located
across the head of separate bays just to the south of
® .the dam, also retain lake water Figs. 3 and 4 .
Attributes of the dam and two dykes are summarized
in Table 1. Most significantly, the elevation of the
® .crest of the dam control structure is higher than the
two earthfill dykes; the dam, Cut-away and Rive-
gauche dykes have crests at 381.06, 380.65, 381.0 m
® .a.s.l., respectively Table 1 .
The purpose of the reservoir was to maintain the
level of discharge of Ha!Ha! River for use by two
small `run of the river' power dams along the lower
part of the river and a paper mill located in the town
of La Baie at the river mouth. The dam was built in
1950 at the site of an older dam originally con-
® .structed in the early 1920s Le Quotidien, 1996 .
®Owned by Consolidated-Stone now Abitibi Consoli-
.dated , the spillway of the dam consists of four
stoplog sluices, each 4.56 m high and 3.66 m wide,
the levels of which are controlled by inserting and
removing 0.3 m square logs. The elevation of the sill
®of the sluice gates i.e., the base with all of the
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 145
Fig. 3. Map of the pre-flood course of Ha!Ha! River from the river mouth to Lake Ha!Ha!, showing the kilometre distances referred to in the
text. Also shown are the location of the dam and two dykes at Lake Ha!Ha!.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168146
Fig. 4. Map of the northern portion of Lake Ha!Ha! reservoir prior to the flood event showing the location of the dam which controlled the
lake level, and the Cut-away dyke.
. ®stoplogs removed is 376.5 m a.s.l. 4.56 m below
.the 381.06 m a.s.l. crest of the dam . The reservoir
level was normally maintained between about 379
® .and 380 m a.s.l. Le Quotidien, 1996 .
4. The drainage of Lake Ha!Ha!
At the Lake Ha!Ha! meteorological station
® .706347 , located near the head of Lake Ha!Ha!,
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 147
Fig. 5. Post-flood view of the concrete gravity dam that controlled the level of Lake Ha!Ha!. Lake basin is in the foreground. For scale, the
four stoplog sluices within the dam are 4.56 m high.
251.4 mm of rainfall were recorded between 0300
and 0400 h July 19, 1996 and 1000 h July 21, 1996
® .Milton and Bourque, 1997 . At 1100 h July 19, the
lake level was 380.08 m a.s.l., having fallen slightly
® . ®over the previous 2 h CSTGB, 1997 . All reported
water levels and stoplog sluice heights were made
originally in feet relative to a scale extending above
the spillway sill. These have been converted to eleva-
tions by adding the converted height in S.I. units to
® .the elevation of the spillway sill see Table 1 . All
.should be regarded as approximate elevations. At
some time shortly after this, the influx of runoff into
the reservoir caused the lake level to rise gradually
® .CSTGB, 1997 . The water level eventually rose to
an elevation of 380.91 m a.s.l. or 0.15 m below the
crest of the dam, as indicated by a water line formed
by organic matter on the side of the dam and debris
lines on the ground at the outer margins of the
® .structure. CSTGB 1997 reports a maximum water
level of 380.77 m; our level may be slightly higher
because of wave actions.
When the lake reached its maximum level, the
Cut-away dyke was being overtopped by up to 0.26
m of water, an estimate based on the maximum
difference in elevation between the water level on
® .the dam and the crest of the saddle dyke Table 1 .
Water spilling from the lake incised into the dyke
and subsequently into the Quaternary deposits of the
® .saddle underlying the dyke Fig. 6 . A new outlet
channel, about 130 m wide and 14 m deep, was
formed at the site of the dyke. This new channel
extends 2 km downvalley and joins the pre-flood
® .Ha!Ha! River channel Figs. 7­9 . The erosion of
the new outlet reduced the level of Lake Ha!Ha! by
about 13 m, although the level of Little Lake Ha!Ha!
was reduced by only about 2 m because of a sill
®located at the narrows between the two basins Fig.
.3 . Lake Ha!Ha! decreased from a pre-flood surface
()G.R.Brooks,D.E.LawrencerGeomorphology281999141­168148
Table 1
® .Summary of dam and dyke attributes at the Lake Ha!Ha! reservoir after CSTGB, 1997
Structure Year in Type Maximum Length Number, type and Maximum Elevation of Elevation of Critical
® . ® . ® . ® .service height m m dimensions m outlet crest m a.s.l. the spillway maximum
® . ® .of sluices capacity sill m a.s.l. level m a.s.l.
3 y1® .m s
a c
Dam 1950 concrete 8.2 106.3 4 stoplog 250 381.06 376.5 380.45
gravity 4.56=3.66
b
Cut-away dyke 1926 earthfill 2­3 162 na na 380.65 na 380.45
Rive-gauche dyke 1926 earthfill 2.8 38 na na 381 na 380.45
a
This apparently is a new structure built at the site of an older one which was constructed in conjunction with the two dykes.
b
® .There is a disagreement about the age of the Cut-away dyke, INRS-Eau 1997 reports that it was built in 1917.
c
® X
.Estimated by subtracting the 4.56 m 15 depth of the sluice gates from the top of the dam.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 149
Fig. 6. Photostereogram of the pre-flood morphology of the dyke and saddle where the new channel was eroded causing the drainage of the
®Lake Ha!Ha! reservoir aerial photographs 22D3-92; -93, June 15, 1994; reproduced with permission Photocartotheque Quebecoise,` ´ ´
.Ministere des ressources naturalles, Quebec .` ´
2 2 ®area of 5.9 km to a residual lake of 2.6 km the
pre-flood area is based on the 1:50,000 scale NTS
map Ferland whereas the post-flood area is from a
®July 29, 1996 multispectral SPOT satellite image 20
..m pixel resolution . Little Lake Ha!Ha! was reduced
in area only slightly from 2.2 to 2.0 km2
because of
the lower drop in water level. Overall, 59=106
m3
of water is estimated roughly to have drained from
® 6 6 3
the reservoir 55=10 and 4=10 m from the
Lake Ha!Ha! and Little Lake Ha!Ha! basins, respec-
tively, based upon the mean of the pre- and post-lake
.surface areas and the change in water levels . The
dam, which formerly controlled the level of the Lake
Ha!Ha! reservoir, remained completely intact after
the drainage, but no longer impounding any water.
The drainage of the Lake Ha!Ha! reservoir began
during the afternoon of Saturday, July 20, 1996.
® .CSTGB 1997 estimates that the reservoir level
began to overtop the Cut-away dyke at about 0200 h
Saturday. At 0600 h, Ha!Ha! River was in flood
because of runoff from the severe rainstorm; the
contribution from the overflow of the Cut-away dyke
® .was likely minimal see below . By this time, a
bridge located about 10 km above the river mouth
® .had been washed out Le Quotidien, 1996 . By mid-
morning, the residents of the village of Boilleau,
located along the valley bottom 1 to 6 km down-
® .stream of the dam Fig. 4 , were evacuated from
their homes because of flooding downstream. At
1400 h, the Cut-away dyke is alleged to have `burst'
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168150
because of the overtopping, and by 1430 h, the
reservoir level is reported to have fallen to 380.06 m
® .a.s.l. Le Quotidien, 1996 , from the maximum level
of 380.92 m a.s.l. `Massive' erosion of the dyke
occurred between 1600 and 1700 h; by 1800 h the
water level had dropped to 379.55 m a.s.l., and to
® .379.09 m a.s.l. by 1845 h Le Quotidien, 1996 .
Based upon the drawdown rate between 1800 and
1845 h, the water level in Lake Ha!Ha! probably
dropped below the level of the sluice gates late
Saturday night, eventually falling another 8 m to
about 368 m a.s.l. sometime Sunday morning when
the rapid erosion of the new outlet terminated.
® .CSTGB 1997 estimates that the lake emptied in 18
h, a time interval that is generally consistent with our
interpretation. The drainage of Lake Ha!Ha!, thus,
occurred over a period of many hours, but not catas-
trophically as in a period of many tens of minutes or
within few hours. The duration of drainage probably
reflects that roughly three quarters of the 14 m
incision at the site of the Cut-away dyke eroded into
®surficial material of the saddle compact, massive,
® . .matrix-supported silty-sand , glacigenic diamicton
rather than entirely of earthfill dyke material. Also
significant is the high ratio of pre-flood height to
width of the topography across the saddle into which
® .the escaping waters incised see Fig. 7a .
With the stoplog sluices completely open, the
maximum spilling capacity of the dam impounding
the Lake Ha!Ha! reservoir is 250 m3
sy1
, well above
the maximum inflow into the lake during the July
3 y1 ®rainstorm, estimated to be 160 m s CSTGB,
.1997 . As observed on July 28, 1996, after the
drainage of the reservoir, three of the stoplog sluices
were set at levels of 380.16 m a.s.l., the fourth
378.94 m a.s.l.; which represents the outlet state
during the rainstorm. Initially during the storm, out-
flow occurred through only one of the four stoplog
® .Fig. 7. The new outlet at the Lake Ha!Ha! reservoir that was eroded through the Cut-away dyke, a looking downstream from above the
® .lake basin and b looking upstream into the drained lake basin. The outlet is about 130 m wide at the site of the dyke. In both pictures,
visible locations of the remnant dyke are marked with arrows.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 151
® .Fig. 7 continued .
®sluices the lake level was 380.08 m a.s.l. at 1100 h
.July 19; see above until the reservoir level later rose
to 380.16 m a.s.l. and began to overtop the other
®three sluices. The crest of the Cut-away dyke was
only 0.49 m above the level of the three highest
sluices, based upon elevations of the stoplogs and
® . .dyke crest Table 1 . Because the sluices had a
maximum capacity above the influx of runoff into
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168152
Fig. 8. Vertical multi-spectral video image of the northern portion of the Lake Ha!Ha! reservoir showing the new outlet and a portion of the
® .drained lake bed, taken August 6, 1996 courtesy of Canada Centre of Remote Sensing .
the reservoir, but were only partially opened, the
overtopping of the Cut-away dyke is the result of
inadequate available spilling capacity rather than in-
sufficient maximum spilling capacity at the control
® .dam CSTGB, 1997 .
5. Estimates of discharge
Streamflow records for Ha!Ha! River during and
immediately after the July 18­21 rainstorm are not
available because the gauging station located about 8
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 153
®Fig. 9. Photostereogram of the post-flood outlet of the Lake Ha!Ha! reservoir eroded through the Cut-away dyke aerial photographs
.Q96304-90, -91; reproduced with permission Photocartotheque Quebecoise, Ministere des ressources naturalles, Quebec . The outlet channel` ´ ´ ` ´
extending downvalley was created by the escaping water. Note the dry knickpoints branching from the channel incised into the subaerial
lake bed.
km above the river mouth was destroyed by the
flood. Quantitative estimates of the discharge of
the flood, calculated by three independent methods
® .with a fourth from CSTGB 1997 , are listed in
Table 2.
The estimates of discharge in Table 2 range from
910 to 7650 m3
sy1
. Of these, the estimate based
upon the empirical relationship between lake volume
®and peak flow produced the largest discharge 7650
3 y1 .m s ; Table 2 . This estimate is 5.5 to 8.5 times
larger than the other three estimates, and thus is
interpreted to be unreasonably high.
The estimates of discharge based upon lake draw-
® 3 y1.down 1380 m s , runoff modellingrdrawdown
® 3 y1. ®910 m s , and slope-area method 1080 to 1260
3 y1. ® .m s are reasonably close Table 2 . These esti-
mates, however, are only generally comparable be-
cause: the estimate of drawdown is based upon an
early stage of lake drainage and does not necessarily
® .represent the peak discharge see Table 2 ; the esti-
()G.R.Brooks,D.E.LawrencerGeomorphology281999141­168154
Table 2
Estimates of the flood discharge along Ha!Ha! River
Method Discharge estimate Ratio to historic Comments
3 y1® .m s maximum instantaneous
3 y1 a® .flow 114 m s
6 3
Empirical relationship between 7650 67 Calculated using a volume of 22.1=10 m which is
b 2
lake volume and peak flow based on the 5.9 km pre-flood surface area of Lake Ha!Ha! and
a 3-m drop in lake level reflecting the maximum
height of the dyke, in combination with a 2-m drop in the
2 ®2.2 km pre-flood surface area of Little Lake Ha!Ha! a sill
.between the basins limited a further decrease in level .
Since a large portion of the incision occurred into surficial
material underlying the dyke, to estimate the peak
discharge based on the total drained volume is deemed
beyond the applicability of the empirical equation.
c
Drawdown of reservoir 1380 12 This estimate may be smaller than peak discharge since this
represents an early stage of the incision.
e
Slope-area method 1080­1260 9.5­11.1 Discharge represents a combination of rainfall runoff and
® .d
continuity and Manning equations drainage from Lake Ha!Ha!.
f
Runoff modellingrdrawdown 910 8 Based upon the combination of hydrologic modelling of
runoff into Lake Ha!Ha! and the reconstruction of the
reservoir drawdown.
a
® .From the 1976­1995 discharge record R. Couture, Milieu Hydrique, Environnement et Faune Quebec, pers. comm., July 1997 .
b 0.48 ® . ® .Based on the equation Q s1730 V modified from Costa 1988 following Desloges et al. 1989 .max
c
® .Based on a reported drop in the water surface of 0.46 m which is reported to have occurred between 1800 and 1845 h, July 20, 1996 Le Quotidien, 1996 .
d
® .Based on a surveyed cross-section of the flood channel located 27 km downstream of Lake Ha!Ha! 8.4 km above the river mouth and which was not significantly modified by
the flood.
e
The data used for this calculation is as follows: hydraulic radius and mean depth, 4.6 m; width, 131 m; Manning's n, 0.036 and 0.043; slope of the high water surface, 0.0007
® .this slope is lower than that on Fig. 10 and is believed to reflect a backwater effect from bridge located 1.5 km downstream that was later washed out .
f
® .This estimate is from CSTGB 1997 .
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 155
mate from runoff modellingrdrawdown represents
the peak discharge at the new outlet of Lake Ha!Ha!;
and the slope-area estimate relates to peak discharge
® .27 km downstream 8 km above the river mouth
and thus incorporates rainfall runoff from other por-
tions of the drainage basin. That the discharge near
the river mouth could have been similar to or lower
than that at the lake outlet might reflect downstream
®attenuation of the flood wave Richards, 1982; Costa,
.1988 , if the rainfall runoff forms a small component
of the peak flow at the slope-area cross-section.
Overall, the available data indicate that the discharge
was in the 910 to 1380 m3
sy1
range at the new
outlet of Lake Ha!Ha! and 1080 to 1260 m3
sy1
along the lower part of the river.
Significantly, all four estimates of discharge are
nearly an order-of-magnitude or more larger than the
previously recorded maximum instantaneous dis-
® 3 y1
charge for Ha!Ha! River 114 m s occurring at
.1500 h, May 9, 1983; see Table 2 . This large
difference clearly indicates that the drainage of the
Lake Ha!Ha! reservoir, in combination with the
runoff from the rainstorm, created an exceptionally
high-magnitude flood along the river.
5.1. Discussion
Based upon a simulation model of runoff, the
peak discharge at the mouth of Ha!Ha! River, result-
ing exclusively from the rainstorm, is estimated to be
3 y1 ® .384 m s CSTGB, 1997; INRS-Eau, 1997 . Al-
though over 3 times larger than the previous maxi-
3 y1 ®mum instantaneous discharge of 114 m s Table
.2 , this discharge is roughly one-half to one-third
less than the 910 to 1380 m3
sy1
range of the
estimates of discharge arising from the lake drainage
contained in Table 2. The drainage of the Lake
Ha!Ha! reservoir from the erosion of the dyke and
saddle, obviously severely compounded the flooding
problem downstream to the river mouth. The flood-
ing along Ha!Ha! River below the reservoir, there-
fore, represents a combination of a dam-break flood
superimposed on a severe `natural' hydrological
event caused by the rainstorm in progress at the
same time the reservoir was draining. The accentua-
tion of a severe rainstorm-generated flood already in
progress by a dam failure is not unique to Lake
Ha!Ha!, having also occurred, for example, at the
® .South Fork dam, PA 1889 , the Lower Otay dam,
® . ® .CA 1916 , and the Canyon Lake dam, SD 1972
® .see Jansen, 1983 .
6. Downstream impacts of the flood
The flooding downstream of the Cut-away dyke
had severe, but variable, geomorphic impacts along
Ha!Ha! River. These effects, outlined for seven suc-
® .cessive reaches reference to the river mouth; Fig. 3 ,
are discussed in terms of change in the width and
incisionraggradation of the channel. The pre- and
post-flood widths of the channel were measured at
250 m intervals, respectively, from Quebec Ministere`
de L'Energie et des Ressources 1:20,000 scale maps
®`La Baie', 22D02-200-0101; `Ferland', 22D02-200-
.0201; and `Boilleau', 22D07-200-0101 and post-
flood aerial photographs taken on July 30, 1996
® .Table 3 . Cartographic error present in the pre-flood
channel boundaries on these maps was determined to
be low compared to pre-flood aerial photographs and
was deemed acceptable for our purposes. Depths of
incision or aggradation reported are based upon field
observations, the interpretation of oblique 35 mm
aerial photographs, andror figures within INRS-Eau
® .1997 .
6.1. km 35.5 to 33.5-- erosion of a new channel
Immediately below the overtopped dyke, a com-
pletely new channel, 2 km long, was eroded into the
saddle and valley bottom and joined the pre-flood
® .channel at km 33.5 Fig. 8 . The new channel ranges
Table 3
List of post-flood aerial photographs along Ha!Ha! River used in this study
Source Flight line Aerial photograph numbers Date of photography Range of scale
Photocartotheque Quebecoise Q96304 53­91 July 30, 1996 1:12,000­14,100`
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168156
Fig. 10. Graphs depicting the pre-flood and post-flood downstream changes in channel width along Ha!Ha! River and the pre-flood
® .longitudinal gradient of the valley. The erosion and deposition designations the latter zones are also shaded across the top of the channel
width graph refer to the dominant geomorphic process along the valley bottom within a given reach.
from 60 to 144 m wide, with a mean width of 98 m
® .Fig. 10 . Reflecting the high-magnitude of the out-
flow from Lake Ha!Ha!, the post-flood mean width
is approximately 4.5 times greater than the pre-flood
channel of the river downstream between km 33.5
and 27.5. Incision along the new channel decreases
downstream from a maximum of about 14 m in the
area of the eroded dyke to a negligible change at the
®junction of new and pre-flood channels INRS-Eau,
.1997 .
6.2. km 33.5 to 27.5-- aggradation in the alley
bottom
From km 33.5 to 27.5, the flood waters swept
® .along the relatively gently-sloped 0.0016 valley
bottom and caused minor to negligible widening of
the channel; the pre-flood meandering planform of
the river essentially was preserved. The pre-flood
®and post-flood widths ranged from 16 to 36 m mean
. ® .22 m and 12 to 42 m mean 26 m; Fig. 10 ,
respectively, with some of this difference probably
partially reflecting minor error in measurement, on
the map, or minor aggradation of the channel.
The most significant erosion along the entire reach
occurred at km 33.5 to 33 where the floodplain was
partially dissected, and at km 28 where an elongated
scour hole, 35 m wide and 150 m long, was eroded
® .into the floodplain Fig. 11a . Large tracts of mature
coniferous trees were knocked over between km 30.5
® .to 29.5. At km 28 Fig. 11a , where the forest
encroaches upon the channel, flood waters cut across
® .several well-defined meander loops Fig. 3 .
In contrast, extensive deposition occurred along
the valley bottom adjacent to the river, particularly
® .Fig. 11. Geomorphic effects along Ha!Ha! River between km 33.5 and 27.5: a trees on the floodplain knocked over by flood water and an
® . ® .elongated scour hole, view looking upvalley from about km 29.5 GSC photograph 1997-42HH , and b overbank deposition on the valley
® . ® .bottom light-shaded areas , looking upvalley of km 32.5 GSC photograph 1997-42II . Photographs taken on July 28, 1996.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 157
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168158
between km 33.5 to 32.5 and 31 to 29.5 where up to
® .2 m of sediment was deposited Fig. 11b . The
sediments, derived from erosion of the new channel
immediately upstream, form broad fine sand sheets
that aggraded wide, low-lying areas of the valley
bottom. Deposition also occurred along the narrower
parts of the valley bottom and within the forested
areas of the floodplain where the vegetation trapped
the sediment.
6.3. km 27.5 to 16-- moderate channel widening
Moderate widening of the channel occurred along
this reach where the pre-flood morphology consisted
of predominately non-alluvial sections of river con-
®fined within a narrow valley bottom short bedrock
® .and alluvial meandering planform sections are also
.present . The post-flood channel ranges from 22 to
® .111 m wide mean 54 m compared to the 10 to 54
® .m wide mean 24 m pre-flood channel. Channel
widening, in general, was limited by the confined
morphology of the valley bottom. Negligible to up to
several metres of incision occurred along this reach
® .of the river INRS-Eau, 1997 .
6.4. km 16 to 12.5-- erosion of a new channel
Extensive alteration of the channel occurred be-
tween km 16 and 12.5. Prior to the flood, the river
profile from km 13 to 16 was controlled by a bedrock
outcrop which formed the Chute-a-Perron rapids`
® .Figs. 3 and 12 . During the flood, water overtopped
a low divide along the right side of the river, situated
at about km 13, and flowed into a ravine that pro-
® .vided a shorter route downvalley Fig. 12 . Subse-
quently, along this new course, deep incision and
extensive lateral erosion into late Pleistocene sand
and clay­silt deposits ensued and propagated up-
stream to about km 16. The section of channel
leading to the rapids was left hanging on the valley-
Fig. 12. Post-flood aerial photograph of the reach from km 16 to 12.5 reach which was severely altered by the flood, as mentioned in the
®text aerial photograph Q96304-74, taken July 30, 1996; reproduced with permission Photocartotheque Quebecoise, Ministere des` ´ ´ `
.Ressources naturalles, Quebec . Flow of the river is from right to left.´
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 159
® .side Fig. 13 . The entire flow of the river is now
carried down the new channel and re-joins the pre-
® .flood channel at km 12.5 Figs. 3 and 12 .
Along this reach, the depth of incision decreases
upstream gradually from a maximum of about 20 m
®at km 13 to several metres at km 16 INRS-Eau,
.1997 . The post-flood channel is considerably wider
® .72 to 190 m, mean 136 m than the pre-flood
® .channel 12 to 40 m, mean 24 m along the entire
® .reach Fig. 10 . A section of the pre-flood channel at
Fig. 13. View looking downstream from about km 13 along the new channel course that follows the route of a pre-flood ravine. The original
® .channel marked by a white arrow which leads to the now-abandoned Chute-a-Perron rapids, `hangs' about 20 m above the valley bottom.`
Up to 20 m of incision occurred locally along this portion of the river. Photograph taken on July 28, 1996.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168160
the Chute-a-Perron rapids is preserved, albeit some-`
what widened and with a new branch extending
® .downstream from it Fig. 12 . The combination of
major incision and widening of the channel along
this reach resulted in the reworking of a considerable
volume of unconsolidated sand, silt and clay, esti-
® . 6
mated by INRS-Eau 1997, Table 2 to be 6.3=10
m3
.
6.5. km 12.5 to 8-- aggradation in the alley bottom
The geomorphic effects of the flood between km
12.5 and 8 were primarily depositional. The large
volume of sediment derived from the km 16 to 12.5
reach, in combination with the relatively low slope
® . ® .of the valley 0.002 between km 12.5 and 8 Fig. 3 ,
resulted in extensive aggradation of the valley bot-
tom and buried the pre-flood channel. The sand
deposited along the valley bottom thins gradually
®downstream from a maximum of about 6 m INRS-
.Eau, 1997 at km 12.5 to its limit at about km 8.5
where the pre-flood channel margins again become
recognizable.
Between km 12.5 to 8.5, the width of the post-
® .flood 33 to 133 m, mean 77 m is considerably
®wider than the pre-flood channel 18 to 36 m, mean
.26 m, Fig. 10 . This reflects the development of a
new channel on the aggraded valley bottom rather
than the erosion of the pre-existing margins of the
channel. At km 10.5 to 10, a large sheet of sand, up
to several metres thick, was deposited in a broad,
low-lying area of the valley bottom and buried the
pre-flood channel and a large area along the right
®side of the valley bottom including the foundations
.of several homes; Fig. 14 . From km 8.5 to 8, the
subaerial margins of the channel essentially are pre-
served, but the post-flood width is greater than the
®pre-flood 33 to 40 m vs. 24 to 28 m, respectively;
.Fig. 10 . This apparent widening probably reflects
minor aggradation of the river bed. Beginning at km
8 and extending downstream, the river bed is eroded
into the pre-flood bottom of the channel.
Fig. 14. Extensive sand aggradation along the right side of the valley at km 10.5 to 10. The post-flood channel flows on fresh alluvium that
® .covers the valley bottom. Photograph taken on July 28, 1996 GSC photograph 1997-42KK .
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 161
6.6. km 8 to 4.5-- moderate channel widening
From about km 9.5 and extending downstream to
the top of a canyon at km 5, Ha!Ha! River is
confined within a narrow, relatively straight valley
bottom. Beginning at km 8 and extending to km 5,
the flood caused extensive erosion into late Pleis-
® .tocene sediments alluvium underlain by diamicton
and widened and deepened the channel, completely
destroying the pre-flood channel. The mean width of
® .the post-flood channel 34 to 88 m wide, mean 68 m
®is about 3 times greater than the pre-flood channel 8
.to 34 wide m, mean 24 m; Fig. 10 ; the widening
was restricted by the narrow morphology of the
valley bottom which confines the river, the material
strength of the deposits now exposed along the river,
and, locally, by exposed bedrock. The depth of
incision increases downstream gradually from km 8
® .to a maximum of about 15 m at km 6 Fig. 15 , and
then declines, eventually grading to the top of the
® .canyon. Downstream along the canyon km 5 to 4.5 ,
the bedrock was extensively scoured of vegetation
® .grasses, mosses, small shrubs and small, thin pock-
ets of overburden by the flood.
Notably, a small concrete dam located at km 5
was breached and the reservoir drained and partially
® .aggraded with sediment Fig. 3 . The powerhouse,
® .situated at the base of the canyon km 4.5 , was
destroyed.
6.7. km 4.5 to 0-- major channel widening
The flood caused major geomorphic change along
the lower 4.5 km of the river. Previously, this por-
tion of the river followed an irregular meandering
® .course sinuosity 1.2 that alternated between single
®channelled and divided reaches the latter flowing
around small islands representing stabilized mid-
.channel bars; Fig. 16 . Between km 4.5 and 3.25,
Fig. 15. The incised and widened channel, about 15 m deep, looking upstream from km 6. Exposed alluvium, diamicton and bedrock, and
bouldery lag form the perimeter of the channel. Note the homes damaged by undermining along the right bank. Photograph taken on July
® .28, 1996 GSC photograph 1997-42NN .
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168162
® . ® . ® .Fig. 16. Maps showing a the pre-flood and b post-flood channel width along the lower 4 km of Ha!Ha! River see Fig. 3 . The pre-flood
® .map is based on Quebec Ministere de L'Energie et des Ressources map `La Baie' 22D02-200-0101 , 1:20,000 scale while the post-flood`
®map was obtained from a rectified mosaic of multi-spectral video images acquired on August 6, 1996 courtesy of Canada Centre for
.Remote Sensing .
widening and deepening of the river destroyed the
pre-flood channel, similar to conditions that occurred
upstream between km 8 and 5. In places, the allu-
vium was completely stripped and the post-flood
river flowed on exposed marine sediments. The
post-flood channel ranges from 62 to 145 m wide
® . ®mean 105 m compared to 24 to 32 m wide mean
. ® .28 m for the pre-flood channel Fig. 10 ; up to
®about 6 m of incision occurred locally INRS-Eau,
.1997 . The largest amount of widening occurred at a
sharp bend within a wider section of valley bottom
® .Fig. 16 .
Below km 3.25, the valley bottom broadens
markedly. Here, extensive widening of the channel
occurred through lateral erosion of the floodplain
® .and low terraces Figs. 16 and 17a . The post-flood
® .channel ranges from 92 to 281 m wide mean 160 m
® .compared to 15 to 110 m mean 47 m for the
® .pre-flood channel Figs. 3 and 16 , but of note, the
wider sections of the pre-flood channel reflect either
the subdivision of the channel by vegetated islands
® .measured as total width or a small reservoir im-
pounded by a dam at km 1.8. The post-flood mor-
phology of the channel, in places, is multi-chan-
nelled and resembles a braided planform. Major
®widening occurred adjacent to the small dam km
.1.8 where overflow of the left abutment resulted in
major lateral erosion and incision of the floodplain
formed a new channel adjacent to, and lower than,
® .the dam Fig. 17a . This erosion drained the small
reservoir, damaged the dam, severed the penstocks,
and destroyed the powerhouse.
At the river mouth, 9.3=106
m3
of sediment,
®derived from erosion upstream INRS-Eau, 1997;
.Table 2 , was deposited within Baie des Ha!Ha! as a
broad sheet on the tidal flat and considerably in-
® .creased its subaerial extent at low tide Fig. 17b .
Debris, from buildings undermined and washed away
by flood waters, was also carried out and deposited
® .on the tidal flat Fig. 17b .
6.8. Discussion
® .Significant widening of the channel Fig. 10
®along much of Ha!Ha! River specifically, from km
.35.5 to 33.5, 27.5 to 12.5 and 8 to 0 is evidence that
the flow energy of the flood exceeded the erosive
threshold of the pre-flood perimeter of the channel
® .and associated valley bottom floodplain, terraces .
®In the recent flood literature following Baker and
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 163
.Costa, 1987 , flow energy of a flood is expressed
® .quantitatively in terms of unit stream power v :
vsg QSrw 1® .
® y2 .where, v is unit stream power W m , g is
®specific weight of water assumed to be 9800 N
y3 . ® 3
m , the value for clear water , Q is discharge m
y1 . ®s , S is energy slope assumed to be reasonably
similar to the pre-flood valley slope; see Magilligan,
. ® .1988 , and w is width of the flood flow m .
Along Ha!Ha! River the downstream variation of
® .the slope, discharge and width in Eq. 1 differ
significantly. Valley slope changes downstream by
® .two orders-of-magnitude 0.2 to 0.0016 and reflects
the irregular profile of the river as the channel
traverses bedrock, non-alluvial, and bedrock-con-
® .trolled alluvial and non-alluvial reaches Fig. 3 . The
discharge estimates at the outlet of Lake Ha!Ha! are
reasonably similar to that made 27 km downstream
® .km 8.4; see Table 2 and suggest that no pro-
nounced downstream change occurred in the peak
discharge. The downstream variation in the width of
® .flow is more complicated to assess because: i along
many sections of the valley bottom the maximum
width of flow is difficult to measure accurately in the
post-flood aerial photographs because vegetation ob-
® .scures the flood margins; ii low-energy backwater
areas can exist adjacent to the general flow in reaches
of the valley bottom that significantly widen and
® .narrow abruptly; and iii along reaches that experi-
enced major erosion the hydraulic geometry of the
flood flow was altered by lateral bank erosion andror
incision. However, a comparison of the pre-flood
® .width Fig. 10 to either the width of the post-flood
channel or to the approximate maximum width of
flooding, whichever was more appropriate, indicates
that the width of flow changes downstream along the
river by no more than an order-of-magnitude. The
difference ranges from several 10 s to several 100 s
of metres. Overall, along Ha!Ha! River, the change
in valley gradient is the most important variable that
controls the downstream variation in unit stream
power. The importance of gradient affecting geomor-
phic response to flooding is well recognized in the
® .literature e.g., Kochel, 1988 .
The importance of the change in valley gradient
along Ha!Ha! River can be demonstrated along a
reach between km 10 and 5 where a marked transi-
tion at about km 8 occurs from major aggradation
and non-erosion, to major erosion. This occurs along
a narrow, very uniform section of valley. If it can be
® ® .assumed that the discharge no major tributary s
.join , flow width and, composition and roughness of
the valley bottom are reasonably constant, then the
downstream increase in valley slope is isolated as the
critical variable that raises the unit stream power
beyond the erosive threshold of the valley bottom.
Downstream along this 5 km long section of valley
bottom, the gradient of the pre-flood valley increases
®gradually by about 11 times 0.002 to 0.023; Fig.
.10 . The unit stream power of the successive reaches
®immediately above and below km 8 approximately
.where the gradient of the valley begins to steepen is
y2 ®163 to 190 and 651 to 760 N m using slopes of
® .0.002 and 0.008 Fig. 9 , the discharge data of 1080
3 y1 ® .to 1260 m s Table 2 , and a flow width of 130 m
.from the cross-section surveyed at km 8.5 . These
ranges of unit stream power fall on either side of the
300 N my2
threshold for major erosion of the
® .channel suggested by Miller 1990 and Magilligan
® .1992 , and correspond to the observed zones of
aggradationrnon-erosion and erosion along the river.
Widening of the channel during the flood repre-
sents an adjustment in the morphology of the river
that most readily dissipates the local unit stream
® ® ..power see Eq. 1 . The most extensive widening of
the channel occurred below km 3.25, where the
valley bottom broadens markedly and a floodplain
was present along the river. This degree of widening
could not occur to the same extent along the eroded
reaches elsewhere along the river because the chan-
nel generally is confined within a narrow valley
bottom. The pre-existing width of a valley bottom is
thus a limiting control on the magnitude of widening
that can occur along reaches where the erosive
threshold is exceeded. However, along reaches where
the valley bottom margins are composed of Quater-
nary deposits, additional factors also control the
degree of widening. These relate to the material
®strength of the sediments e.g., material size distribu-
tion, cohesiveness, compaction, internal stratigraphic
.relations and vegetative binding and the height of
the eroding bank. The difference in material strength
of the banks is undoubtedly a significant factor in the
major widening that occurred between km 16 and
12.5 relative to, for example, between km 8 and 5. In
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168164
the latter case, the river was eroding into a compact
®diamicton overlain by a gravel alluvium deposit Fig.
.15 vs. initially into sand and then a clay­silt unit at
® .the former reach Fig. 13 .
The two major sections of the river that experi-
enced negligible erosion andror major aggradation
® .reaches km 33 to 27 and 12.5 to 8 , not surprisingly,
® .had the lowest pre-flood gradient 0.0016 and 0.002
® . ® .Fig. 17. The broad post-flood channel of Ha!Ha! River picture `a' , viewed downstream from km 3 GSC photograph 1997-42RR . The
® .pre-flood valley bottom floodplain and terraces has been extensively reworked; an arrow marks the remains of a dam severely damaged
® .during the flood. Tidal flat at the river mouth looking upstream into the Ha!Ha! River valley picture `b'; GSC photograph 1997-42TT .
Note the debris from destroyed buildings on the tidal flats and the considerable damage to residential and commercial areas within the City
of La Baie at the river mouth. Photographs taken on July 28, 1996.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 165
® .Fig. 17 continued .
of any reach along the river. Unit stream power at
® .these reaches Table 4 ranged locally from 65 to
y2 ® . y2 ®289 N m km 33 to 27 and 64 to 260 N m km
. y2
12.5 to 8 which are both below the 300 N m
erosive threshold, mentioned previously. Major ero-
sion did not occur along either reach which indicates
® y2
that the `real' erosive threshold be it 300 N m or
® . .marginally ? higher was not crossed, being a func-
tion of the local valley bottom roughness, vegetative
cover, and surficial material.
Major aggradation along the reaches of river be-
tween km 33 to 27 and 16 to 12.5 occur immediately
downstream of reaches that experienced major ero-
sion; specifically, the formation of the new channel
® .into the saddle at Lake Ha!Ha! reach km 35 to 33
and the diverted and deeply incised channel in the
Table 4
Estimate of unit stream power along the km 33­27 and 16­12.5 reaches
3 y1 b c d® . ® .Reach Valley slope Specific weight of Discharge m s Width range m Unit stream power
y3 a y2® . ® .water N m W m
e
km 33­27 0.0016 9800 910­1380 75­220 65­289
e
km 16.5­12 0.002 9800 1080­1260 95­330 64­260
a
Value for clear water.
b
See Table 2.
c
Estimated from post-flood aerial photographs.
d
® .See Eq. 1 in text.
e
Maximum width of flow is conservative. Irregularly shaped, low-lying areas of the valley bottom that were obviously back-flooded were
not included in measurement.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168166
® .area of the Chute-a-Perron rapids km 16 to 12.5 .`
The relatively low unit stream power is a significant
factor influencing the aggradation, but equally im-
portant is the presence of abundant sand-sized sedi-
ment being carried by the river and is thus available
for deposition where the energy of flow decreases.
Not all of the erosion and deposition along Ha!Ha!
River can be explained in terms of variation in unit
stream power. At km 16 to 13, the gradient of the
pre-flood valley averages 0.003, just slightly greater
than that of reaches km 33 to 27 and km 12.5 to 8,
yet this reach experienced major lateral erosion and
incision during the flood. In this instance, the effects
of the flood relate specifically to events along the
reach immediately downstream. As described above,
a major diversion of the river occurred at km 13 and
the river cut a new route across the drainage divide
between the main channel course and a ravine; previ-
ously, the river had been flowing over a bedrock
surface that controlled the baselevel of the reach
immediately upstream. Incision and lateral erosion,
thus, ensued along the new course and extended
several kilometres upstream severely altering a pre-
viously gently-sloped reach. This geomorphic change
is the product of a decrease in local baselevel, the
effects of which propagated upstream.
7. Implications for other rivers on the Canadian
Shield
The events along Ha!Ha! River demonstrate that
rivers on the Canadian Shield draining less than
1000 km2
can experience significant geomorphic
change from exceptionally high-magnitude floods.
These changes include major widening and aggrada-
tionrincision of the channel along alluvial and non-
alluvial reaches. Widening in places will be re-
stricted by narrow valley bottoms formed of bedrock
or Quaternary deposits. In the latter cases, variations
in widening will also reflect changes in the material
strength and height of the eroding valleyside de-
posits. The erosive threshold of the valley bottom
will be exceeded locally rather than uniformly along
the river course because of local variations in slope
as the river passes over bedrock, alluvial and non-al-
luvial surfaces. Small lake basins, however, are very
common along many rivers on the Canadian Shield
and these will dampen the downstream progression
of a flood wave. For example, the flood arising from
® .the 1985 failure of Beloeil dam see Section 1 was
significantly dampened by Lake Talbot located only
7 km downstream.
The occurrence of the channel diversion at km 13
may not be a phenomenon unique to Ha!Ha! River.
Drainage on the Canadian Shield follows the re-
gional slope of the land surface, but the stream
courses generally are controlled by bedrock struc-
ture, faulting and joint patterns, and are diverted in
®places by the presence of glacial deposits Bostock,
.1970 . The lack of a well-developed valley along
some reaches suggests that channel diversions could
occur during extreme floods along other Shield rivers
from the overtopping and incision of low within-basin
drainage divides formed of glacial sediments. Such
diversions could have severe and irreversible local
geomorphic impacts.
A number of homes were damaged or destroyed
along Ha!Ha! River by undermining of the river
banks despite, in many instances, the homes being
located above the level of inundation. At some river
settings, extensive bank erosion during very high-
magnitude floods can be a severe hazard to commu-
nities and the infrastructure located in valley bottoms
in addition to the obvious inundation problems.
8. Conclusions
The overtopping and erosion of the Cut-away
dyke caused the draining of the Lake Ha!Ha! reser-
voir, and significantly accentuated flooding along
Ha!Ha! River already in progress from a severe
rainstorm. The drainage of the reservoir occurred
over a period of many hours. The overtopping of the
dyke was caused by insufficient available capacity of
the spillway at the time of the rainstorm and because
the elevation of the dyke crest was below that of the
concrete dam controlling the reservoir level.
The flood discharge along Ha!Ha! River is esti-
mated to be 910 to 1380 m3
sy1
at the new lake
outlet and 1080 to 1260 m3
sy1
at a point 27 km
downstream. This is about an order-of-magnitude or
more larger than the previously recorded maximum
instantaneous flow.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168 167
The flooding produced morphologic changes
downstream of the reservoir. Long sections of the
Ha!Ha! River experienced significant widening and
incision of the channel because the energy in the
flood surpassed the erosive threshold of the valley
bottom. Major aggradation occurred along two
reaches where the flood energy was below threshold
and where extensive erosion happened immediately
upstream. Valley gradient was the most important
variable controlling whether energy of flow ex-
ceeded the erosive threshold of the valley bottom.
One reach underwent major incision and widening as
a result of the flood overtopping and eroding across
the drainage divide between the main course of the
channel and a small ravine. This resulted in a short
diversion of the channel where deep incision ensued
and propagated upstream. The morphologic changes
along Ha!Ha! River demonstrate that rivers on the
Canadian Shield can experience significant impacts
from high magnitude floods.
9. Postscript
Preliminary work to reconstruct a replacement
earthfill dyke across the new outlet channel of Lake
Ha!Ha! began within days of the waning of the
flood. Work on the dyke was underway in mid-
November, 1996 and scheduled to be completed by
®the end of December, 1996 Emile Soucie, pers.
.comm., November 13, 1996 . The new dyke is lo-
cated several hundred metres west of the remnants of
the Cut-away dyke, on ground that was originally
part of the reservoir bed. The dam that again controls
the reservoir level has been rebuilt and now incorpo-
rates a wide concrete overflow spillway.
Acknowledgements
®We thank Ko Fung and Terry Pultz Canada
.Centre for Remote Sensing for access to various
®remote sensing images, and Raimo Kallio Environ-
.ment Canada for hydrological discussions. Com-
ments on drafts of the manuscript by Christian Begin,´
Steve Evans, Ted Hickin, Raimo Kallio, John Vitek
and an anonymous journal reviewer are appreciated.
This work was partially funded by Emergency Pre-
paredness Canada. Geological Survey of Canada
contribution 1997196.
References
Baker, V.R., Costa, J.E., 1987. Flood power. In: Mayer, L., Nash,
® .D. Eds. , Catastrophic Flooding. Allen and Unwin, London,
pp. 1­21.
Bostock, H.S., 1970. Physiographic subdivisions of Canada. In:
® .Douglas, R.J.W. Ed. , Geology and Economic Minerals of
Canada. Geological Survey of Canada Economic Report No.
1, pp. 10­30.
Brooks, G.R., Lawrence, D.E., 1998. Geomorphic effects and
impacts of severe flooding: photographic examples from the
Saguenay area, Quebec. Geological Survey of Canada Miscel-
laneous Report 62.
Brooks, G.R., Lawrence, D.E., Fung, K., Begin, C., Perret, D.,´
1997. Flooding from the July 18­21, 1996 rainstorm in the
Saguenay area, Quebec: fluvial geomorphic effects and slope
stability along selected major river reaches. Geological Survey
of Canada Open File Report 3498, 81 pp.
Commission scientifique et technique sur la gestion des barrages
® .CSTGB , 1997. Rapport: Commission scientifique et tech-
nique sur la gestion des barrages. Quebec, Janvier 1997, 241
pp.qannexes.
Costa, J.E., 1988. Floods from dam failures. In: Baker, V.R.,
® .Kochel, R.C., Patton, P.C. Eds. , Flood Geomorphology.
Wiley, New York, pp. 439­463.
Desloges, J.R., Jones, D.P., Ricker, K.E., 1989. Estimates of peak
discharge from the drainage of ice-dammed Ape lake, British
Columbia, Canada. Journal of Glaciology 35, 349­354.
Environment Canada, 1996. Canadian climate summary. Environ-
ment Canada, Vol. 1, July 1996, 10 pp.
INRS-Eau, 1997. Simulation hydrodynamique et bilan sedi-´
mentaire des rivieres Chicoutimi et des Ha!Ha! lors des crues`
exceptionnelles de juillet 1996. Rapport INRS-Eau No. R487,
Travaux realises pour le compte de la Commission scientifiqué ´
et technique sur la gestion des barrages, 207 pp.
Jansen, R.B., 1983. Dams and public safety. U.S. Department of
the Interior, Bureau of Reclamation, Denver, CO, 332 pp.
Kochel, R.C., 1988. Geomorphic impact of large floods: reviews
and new perspectives on magnitude and frequency. In: Baker,
® .V.R., Kochel, R.C., Patton, P.C. Eds. , Flood Geomorphol-
ogy. Wiley, New York, pp. 169­187.
Lapointe, J., 1986. Safety at Quebec's dams. Proceedings of Dam
Safety Seminar, Edmonton, Alberta, September, 1986. BiTech
Publishers, pp. 31­35.
Le Quotidien, 1996. Stone decline toute responsabilite. Saturday,´
July 27, 1996.
Longley, F.F., 1912. The failure of the dam of the Erindale Power.
® .Engineering Record 65 17 , 457.
Lukey, A.F., Noonan, M.L., 1986. PFRA experience with Prairie
dams. Proceedings of Dam Safety Seminar, Edmonton, Al-
berta, September, 1986. BiTech Publishers, pp. 489­557.
( )G.R. Brooks, D.E. LawrencerGeomorphology 28 1999 141­168168
Magilligan, F.J., 1988. Variations in slope components during
large magnitude floods, Wisconsin. Annals of the Association
of American Geographers 78, 520­533.
Magilligan, F.J., 1992. Threshold and spatial variability of flood
power during extreme floods. Geomorphology 5, 373­390.
Miller, A.J., 1990. Flood hydrology and geomorphic effectiveness
in the central Appalachians. Earth Surface Processes and
Landforms 15, 119­134.
Milton, J., Bourque, A., 1997. Torrential rains of July 18 to 21,
1996, in the province of Quebec: analysis and interpretation of
meteorological and climatological data. Environment Canada,
September 9, 1997, 103 pp.
Richards, K., 1982. Rivers, form and process in alluvial channels.
Methuen, New York, 358 pp.
Robert, B., Pare, J.-J., 1992. Rupture du barrage du Lake Beloeil´
causes et consequences. Proceedings of the Canadian Dam
Safety Conference, Quebec City, Quebec, September 8­11,
1992, pp. 1­14.
Robitaille, A., Dubois, J.-M., 1989. A valley transformed. Geos
® .18 2 , 23­29.