® .Geomorphology 40 2001 57­90
www.elsevier.comrlocatergeomorph
Peak discharge estimates of glacial-lake outburst floods and
AnormalB climatic floods in the Mount Everest region, Nepal
Daniel A. Cenderelli)
, Ellen E. Wohl
Department of Earth Resources, Colorado State Uniersity, Fort Collins, CO 80225, USA
Received 9 March 2000; received in revised form 8 January 2001; accepted 13 January 2001
Abstract
® .Glacial-lake outburst floods GLOFs in the Mount Everest region of Nepal on 3 September 1977 and 4 August 1985
dramatically modified channels and valleys in the region by eroding, transporting, and depositing large quantities of
sediment for tens of kilometers along their flood routes. Prior to this research, the GLOF discharges had not been determined
® .and the hydrology of AnormalB climatic floods SHFFs: seasonal high flow floods was not known. A one-dimensional
step-backwater flow model was utilized, in conjunction with paleostage indicators, to estimate the peak discharges of the
GLOFs and SHFFs and to reconstruct the hydrology and hydraulic conditions of the GLOFs at 10 reaches and SHFFs at 18
reaches. The most reliable GLOF and SHFF peak discharge estimates were upstream from constrictions where there was
critical-depth control.
The peak discharge of the 1977 GLOF at 8.6 km from the breached moraine was approximately 1900 m3
rs. At 7.1 km
downstream from the breached moraine, the 1985 GLOF discharge was estimated at 2350 m3
rs. At 27 km downstream from
the breached moraine, the 1985 GLOF attenuated to an estimated discharge of 1375 m3
rs. The peak discharges of SHFFs
ranged from 7 to 205 m3
rs and were positively correlated with increasing drainage area. The GLOF discharges were 7 to 60
times greater than the SHFF discharges with the greatest ratios occurring near the breached moraines. The downstream
decline in the ratio between the GLOF discharge and SHFF discharge is the result of the downstream attenuation of the
GLOF and the increased discharge of the SHFF because of increased contributing drainage area and the increased effects of
monsoonal precipitation at lower elevations. q 2001 Elsevier Science B.V. All rights reserved.
Keywords: Glacial-lake outburst floods; Seasonal high flow floods; Paleoflood hydrology; Step-backwater modeling; Peak discharge
estimates; Mount Everest
1. Introduction
The Mount Everest region of Nepal experienced
two extraordinary floods on 3 September 1977 and 4
)
Corresponding author. Department of Geological Sciences,
University of Alabama, Box 870338, Tuscaloosa, AL 35487-0338,
USA.
August 1985 when glacial lakes, situated immedi-
ately downstream from alpine glaciers, breached their
moraine dams. Both glacial-lake outburst floods
® .GLOFs dramatically modified channel and valley
morphology for tens of kilometers downstream from
® .the source area Fig. 1 . Although the discharges of
both GLOFs have been described as several orders
of magnitude larger than typical annual floods in the
region, the peak discharges of the 1977 and 1985
0169-555Xr01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
® .PII: S0169-555X 01 00037-X
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9058
GLOFs along the upper segments of their flood route
had not been systematically quantified prior to the
research described in this paper. In the Mount Ever-
est region, annual peak discharges are caused by the
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 59
combination of seasonal snowmelt, glacier meltwa-
®ter, and monsoonal precipitation in this paper, these
floods are referred to as seasonal high flow floods,
.SHFFs . However, the magnitude, duration, and tim-
ing of SHFFs are poorly understood because streams
in the region have never been gaged.
In the absence of gaging stations or direct stream
measurements, paleoflood hydrology techniques can
be used to estimate the peak discharge of floods.
Paleoflood hydrology is the study of floods that
occurred in the absence of direct measurements or
prior to when hydrological records were collected
® .from a given stream Baker, 1987; Jarrett, 1990 .
Paleoflood hydrology studies in ungaged watersheds
can provide invaluable information for water-re-
source managers, hydrologists, and fluvial geomor-
phologists concerned with past and potential flood-
ing in an area. Additionally, geomorphologists
assessing fluvial processes and the geomorphic ef-
fects of extreme floods require accurate discharge
estimates in order to calculate local flow hydraulics.
Paleoflood hydrology identifies and uses erosional
and depositional features produced by past floods to
delineate the probable water-surface elevation of the
floods and to estimate the magnitude of those floods.
Evidence left by a flood is referred to as paleostage
® .indicators PSIs and includes scour lines, scars on
vegetation, silt lines, debris accumulations, slackwa-
ter sediments, and boulder bars. Nonflooded surfaces
along the flood route can also be considered paleo-
stage indicators. Where there are PSIs along a given
channel reach, the flood discharge can be recon-
structed by calculating water-surface profiles using
the step-backwater method in conjunction with the
PSIs. Numerous studies in a variety of environments
have effectively demonstrated the use of the step-
backwater method to estimate the peak discharges
®and flow hydraulics of ungaged floods e.g., Ely and
Baker, 1985; Jarrett and Malde, 1987; Webb et al.,
1988; Wohl, 1992a, 1992b, 1995; O'Connor, 1993;
O'Connor and Baker, 1992; O'Connor et al., 1986;
Rathburn, 1993; Grimm et al., 1995; House and
Pearthree, 1995; Waythomas et al., 1996; Benito,
.1997 . Of these studies, several used the step-back-
water method to reconstruct floods from dam fail-
ures, such as the Pleistocene Lake Missoula glacier-
® . ®dam failure flood s O'Connor and Baker, 1992;
.Benito, 1997 , the Pleistocene Lake Bonneville flood
® .Jarrett and Malde, 1987; O'Connor, 1993 , and
® .Aniakchak Caldera flood Waythomas et al., 1996 .
Prior to the work described here, no studies have
quantified the annual peak discharges of SHFFs on
streams in the Mount Everest region or systemati-
cally quantified the peak discharges of the 1977 and
1985 GLOFs along their flood routes. The purposes
® .of this paper are to i demonstrate the application of
the step-backwater method, used in conjunction with
geomorphic evidence, in estimating the magnitude of
previously unquantified GLOFs and SHFFs in the
® .Mount Everest region of Nepal and ii improve the
understanding of flood hydrology in this remote,
mountainous region. For the 1977 and 1985 GLOFs,
multiple reaches along the GLOF were selected to
assess the downstream changes in the discharge
magnitude of the flood wave. The SHFFs character-
ize the hydrology of AnormalB climatic floods in the
region and provide a baseline to which the GLOF
discharges can be compared.
2. Study area
2.1. Geology and geomorphology
The Mount Everest region is located in eastern
® .Nepal Fig. 2 . The area is underlain primarily by
® .Precambrian gneiss and granite Vuichard, 1986 ,
lies within the High Himalaya Physiographic
Province, and is characterized by extremely high
relief. The four major valleys in the area, Bhoti Kosi,
® . ® . ® .Fig. 1. a Upstream view of the Imja Khola valley foreground and the Nare Khola valley arrow showing the erosional and depositional
effects of the 1977 GLOF. The river valley bottom in the foreground is approximately 50 to 70 m wide. The stone wall pastures on the right
® . ® .valley margin A are the lower outskirts of the village Pangboche and the mountain in the background is Ama Dablam. b Upstream view
of the Bhoti Kosi valley showing the erosional and depositional effects of the 1985 GLOF. This segment of the river valley is approximately
® .16 km downstream from the breached moraine. The village of Thamo A is located on a glaciofluvial terrace along the left valley margin.
® .A nearly completed hydroelectric dam B that was located on the fan was destroyed by the 1985 GLOF.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9060
Fig. 2. Map of the study area showing the locations of study reaches for the 1977 GLOF, 1985 GLOF, older GLOFs, and tributaries. N
designates 1977 study reaches, L designates 1985 study reaches. For the older GLOF study reaches: UBKsupper Bhoti Kosi and
UDKsupper Dudh Kosi. For the tributary study reaches: IKsImja Khola, KJKsKyajo Khola, KKsKyashar Khola, and TKsThame
Khola.
Dudh Kosi, Imja Khola, and Khumbu Khola, are
deeply incised with valley floors 4000 m lower than
® .the surrounding mountains Fig. 1 . The Nepalese
term AKosiB indicates a major stream or large river,
whereas the term AKholaB indicates a smaller stream.
Valleys at elevations higher than 3400 to 3600 m
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 61
®were glaciated during the Pleistocene Fushimi, 1977,
.1978 , are distinctly U-shaped, and have boundaries
consisting of bedrock, coarse-grained colluvium, or
coarse-grained glaciofluvial sediment. Below 3400 to
3600 m, valleys are V-shaped and have boundaries
consisting primarily of bedrock and secondarily ter-
races comprised of coarse-grained sediment. Alpine
glaciers are typically present at elevations above
4500 m and have been, for the most part, in retreat
from their Little Ice Age maximum positions, creat-
ing numerous moraine-and glacier-dammed lakes
® .Mayewski and Jeschke, 1979; Fushimi et al., 1985 .
2.2. Climate and hydrology
Hydrology in the study area is strongly influenced
by monsoonal precipitation and late springrearly
summer snowmelt. Eighty percent of the total annual
precipitation occurs between June and September
with the most intense precipitation typically occur-
®ring from mid-July to mid-August Ageta, 1976;
.Brower, 1991 . The high mountain topography, how-
ever, creates a rain-shadow effect that reduces the
intensity and amount of monsoonal precipitation with
®increasing elevation Ageta, 1976; Zimmermann et
.al., 1986 . For example, from June to September
1974 in the Mount Everest region, the total precipita-
tion was 1100 mm at 2700 m, 685 mm at 3900 m,
® .and 428 mm at 4400 m Ageta, 1976 . Although not
quantified, discharge in the study area is character-
ized by low flow from late fall to early spring and
high flow from late spring to early fall because of
the combined runoff produced by snowmelt, glacier
meltwater, and monsoonal precipitation.
2.3. The 1977 and 1985 glacial-lake outburst floods
On 3 September 1977, a series of ice-cored
moraine dams failed below the Nare Glacier, sending
a flood surge down the Nare Khola, Imja Khola, and
® . ®Dudh Kosi valleys Fig. 2 Buchroithner et al.,
1982; Fushimi et al., 1985; Zimmermann et al.,
.1986 . This flood caused extensive erosion and depo-
sition for 35 km downstream from the source area,
destabilized valley side slopes, and destroyed bridges
® . ®and trails Fig. 1a Ives, 1986; Zimmermann et al.,
.1986 . Based on a field survey, the volume of water
released by the lake was estimated to be 500,000 m3
® .Fushimi et al., 1985 , although Buchroithner et al.
® . 3
1982 reported a lake volume of 5,000,000 m
based on remote sensing imagery. At a gaging sta-
tion located 90 km from the breached moraine, the
1977 GLOF had an estimated peak discharge of 800
m3
rs that lasted about 1 h and a total flood duration
® .of approximately 6 h Fig. 3 .
On 4 August 1985, a moraine-dammed lake lo-
cated below the Langmoche Glacier failed when an
ice avalanche from the glacier plunged into the lake,
triggering a surge wave that breached the moraine
®Vuichard and Zimmermann, 1986, 1987; Ives,
Fig. 3. Discharge hydrograph of the Dudh Kosi at the Rabuwa Bazar gaging station from 26 August 1977 to 4 September 1977. The increase
in discharge of the Dudh Kosi on 3 September 1977 is due to the 1977 GLOF that originated 90 km upstream from the gaging station
® .modified from Zimmermann et al., 1986 .
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9062
.1986 . The volume of water released was estimated
3 ®to be 5,000,000 m Ives, 1986; Vuichard and Zim-
.mermann, 1986, 1987 . The gaging station that
recorded the discharge of the 1977 GLOF was not in
operation at the time of the 1985 GLOF. Vuichard
® .and Zimmermann 1987 estimated that the 1985
outburst flood had a peak discharge of 1600 m3
rs at
a distance of 2 km from the breached moraine. Based
® .Fig. 4. a Downstream view of reaches N2 and N1, located about 8.3 km downstream from the breached moraine, showing the geomorphic
® .effects of the 1977 GLOF. Reach N1 is located on the Imja Khola background , approximately 500 m downstream from reach N2. Note the
extensive erosion of the coarse-grained glaciofluvial terraces along the valley side margins and deposition of cobbles and boulders across the
® .valley bottom. Valley in the background is approximately 125 m wide and the valley in the foreground is approximately 60 m wide. b
Upstream view of reach N3 showing the geomorphic effects of the 1977 GLOF. Reach N3 is located on the Imja Khola at approximately
11.5 km downstream from the breached moraine. The valley is relatively narrow, ranging in width between 30 and 45 m, with valley
boundaries consisting primarily of bedrock and coarse-grained glaciofluvial terraces or colluvium.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 63
on discussions with residents at various villages along
® .the GLOF route, Vuichard and Zimmermann 1987
estimated that the GLOF peak discharge had a dura-
tion of approximately 1 h and that the total GLOF
duration was between 6 and 8 h. The resulting flood
caused considerable erosion and deposition for 40
km along the Langmoche Khola, Bhoti Kosi, and
® .Dudh Kosi Figs. 1b and 2 resulting in the destruc-
tion of several bridges, tens of houses, and a nearly
®completed hydroelectric power plant Ives, 1986;
.Vuichard and Zimmermann, 1987 . Using pho-
tographs taken before and after the flood, Vuichard
® .and Zimmermann 1987 estimated that approxi-
mately 3,000,000 m3
of sediment were eroded and
deposited along the flood route; approximately 70%
of this total occurred in the first 16 km of the flood
route.
The 1977 and 1985 GLOFs produced an assort-
ment of erosional and depositional features along
® .their flood routes Figs. 4 and 5 . Erosion of valley
slopes and valley bottoms was the predominant geo-
morphic process in narrow, steep valleys with
boundaries comprised of coarse, unconsolidated sedi-
® .ment Figs. 6 and 7 . Deposition was dominated by
®coarse sediment primarily boulder-, cobble-, and
.pebble-size particles and typically occurred in wider,
less steep reaches across the entire valley bottom
where flow energy was reduced and flow was di-
verging. Deposition typically occurred at locations
where the channel andror valley widened, upstream
and downstream of obstructions, and along the mar-
® .gins of channel bends Figs. 6 and 7 . The deposits
produced by the 1977 and 1985 GLOFs were deposi-
®tional macroforms Baker, 1978, 1984; Church and
.Jones, 1982 and consisted primarily of expansion
bars and longitudinal bars and secondarily of point
®bars, pendant bars, and imbricate clusters Figs. 6
.and 7 . Fine-grained sedimentation was minimal
along the 1977 and 1985 GLOF routes, but did occur
in areas upstream of constrictions where ponding
occurred or along channel and valley margins where
®flow energy was reduced andror recirculating Fig.
.7A, C and F . The sedimentologic characteristics of
the 1977 and 1985 GLOF deposits have been de-
® .scribed elsewhere by Cenderelli 1998 , Cenderelli
® . ® .and Cluer 1998 , and Cenderelli and Wohl 1998 .
In general, along the upper 10 km of the 1977 GLOF
route and the upper 16 km of the 1985 GLOF route,
erosion and deposition were more pronounced than
below these distances from the breached moraines
®compare Fig. 4a to b, Fig. 5a and b to 5c and d, Fig.
.6a to b and Fig. 7A­C to D­F .
2.4. Eidence of older, preiously undocumented
glacial-lake outburst floods
During field reconnaissance of the upper Bhoti
Kosi and upper Dudh Kosi drainages, cobble-boulder
deposits located adjacent to and 3 to 5 m above the
present-day channel on the surface of lower
® .glaciofluvial terraces were identified Figs. 2 and 8 .
These deposits were interpreted to have been de-
posited by GLOFs of an unknown age; however, the
heavily lichen-encrusted cobbles and boulders indi-
cate that these deposits are considerably older than
the 1977 and 1985 GLOFs. The upper Bhoti Kosi
GLOF deposits appear to be younger than the upper
® .Dudh Kosi GLOF deposits compare Fig. 8a to b ,
but considerably older than the 1977 and 1985
GLOFs. The older GLOF deposits are clast-sup-
ported and composed primarily of cobbles and boul-
ders that are moderately imbricated. No attempt was
made to trace the GLOF deposits to the source area.
3. Methods
3.1. Selection of study reaches
Ten reaches along various streams in the study
area were selected to estimate the peak discharges of
both GLOFs and SHFFs using the step-backwater
® .method Fig. 2 . Two reaches were studied along the
® .1977 GLOF route reaches N1 and N3 , six reaches
®were studied along the 1985 GLOF route reaches
.L2, L1, L4, L5, L7, and L8 , and two reaches were
studied where evidence of older GLOFs were identi-
® .fied reaches UDK and UBK . Eight other reaches
® .TK, IK, KYK, KJ, L3, L6, N2, and N4 were
selected to estimate the peak discharge of SHFFs for
® .comparative purposes Fig. 2 . The physical charac-
teristics of the study reaches are summarized in
Table 1. The criteria used to select these reaches
® . ® .were i accessibility to the stream, ii relatively
straight and uniform or gradually narrowing reaches
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9064
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 65
so that the floods could be accurately modeled using
® .the step-backwater method, and iii the presence of
multiple paleostage indicators produced by the GLOF
and SHFF to delineate the water-surface elevation of
the floods. Although reaches N2, N4, L3, and L6 are
located along the 1977 or 1985 GLOF routes, the
GLOF peak discharges were not calculated at these
reaches because they were not appropriate for model-
ing using the step-backwater method. At reach N2,
GLOF deposits were interpreted to be fluvially re-
worked, noncohesive, sediment gravity-flow de-
posits. At reach N4, there was no geomorphic evi-
dence of the 1977 GLOF. Reaches L3 and L6 are
located in hydraulic settings that caused flow during
the GLOF to be rapidly varied. Although reaches
N2, N4, L2, and L6 were not well suited for model-
ing the GLOF using the step-backwater model, they
were appropriate for modeling the SHFFs.
3.2. Characterizing channel and alley geometry
To accurately estimate discharge using the step-
backwater method, the channelrvalley geometry
must be adequately characterized and representative
of the channelrvalley conditions at the time of the
®flood Bailey and Ray, 1966; O'Connor and Webb,
.1988 . In this study, reaches were surveyed using
standard transit-stadia rod techniques to characterize
the channel and valley geometry. Depositional fea-
tures, erosional features, andror nonflooded surfaces
were identified, surveyed, and mapped at each
cross-section to constrain the flood stage of the
GLOFs and SHFFs. Cross-sections were surveyed
perpendicular to the assumed flow direction, placed
at locations that best characterized the channel ge-
ometry of the modeled reaches, and placed at loca-
tions so that channel geometry changes between
successive cross-sections were relatively gradual
® .Figs. 6 and 7 . Along the 1977 and 1985 outburst-
flood routes, the surveyed channel and valley geome-
try were assumed to represent the geometry of the
channel and valley during peak flow. Although seg-
ments of the modeled reaches experienced consider-
®able erosion and deposition during the GLOFs in
.particular, reaches N1, L2, L3, L1, and L4 , these
geomorphic processes are assumed to have occurred
in close proximity to peak flow and were minimal
during the receding limb of the GLOFs. Large quan-
tities of coarse-grained sediment introduced to the
channel by the GLOFs, and SHFFs are incapable of
reworking that sediment and modifying the geometry
®of the main channel Cenderelli and Wohl, in re-
.view .
3.3. Paleostage indicators used in this study
At all of the study sites, multiple paleostage indi-
cators of the GLOFs and SHFFs were identified and
surveyed at numerous cross-sections along a given
modeled reach. Upper surfaces of boulder and cobble
bars, scour lines, and the lowest elevation of non-
flooded surfaces were used to define the peak stages
of the 1977 and 1985 GLOFs, as well as the older
GLOFs identified in the study area. Upper surfaces
of boulder and cobble bars were assumed to repre-
sent a minimum peak stage of the outburst floods.
® .Fig. 5. a Cross valley view of the middle portion of reach L2 showing the geomorphic effects of the 1985 GLOF. The valley in the center
of the photo is approximately 110 m wide and the flow direction is from left to right. Reach L2 is located on the Langmoche Khola,
® .approximately 7.1 km downstream from the breached moraine. b Downstream view of middle segment of reach L1, which is located on
the Bhoti Kosi approximately 11 km downstream from the moraine breached in 1985, showing the geomorphic effects of the GLOF. In the
foreground the valley width ranges from 150 to 225 m. Deposition was the primary geomorphic process as expansion bars comprised
® .primarily of cobbles and boulders were deposited across the valley bottom. c Downstream view of reach L5, which is located
approximately 22.1 km downstream from the moraine breached in 1985 along the Dudh Kosi near the village of Jorsale. The valley slopes
consist of near vertical bedrock walls, coarse-grained colluvium, or coarse-grained glaciofluvial terraces. In the foreground on the right bank
® .and background on the left bank arrow , note the bank erosion and deposition of cobbles and boulders on the surface of the lower
® .Glaciofluvial terraces. The foot bridge extending across the valley is approximately 70 m long. d Upstream view of reach L8 along the
1985 GLOF route, which is located on the Dudh Kosi near the village of Phakding approximately 26.7 km downstream from the breached
moraine. The foot bridge extending across the valley is approximately 80 m long. The valley slopes consist of near vertical bedrock walls,
coarse-grained colluvium, or coarse-grained glaciofluvial terraces. Erosion of valley side slopes was minimal, but the surfaces of the
glaciofluvial terrace were eroded. Deposition was minimal and consisted primarily of longitudinal bars comprised of cobbles and boulders
deposited on the surfaces of lower glaciofluvial terraces.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9066
® . ® .Fig. 6. Geomorphic maps of reach N1 A and reach N3 B showing the distribution of erosional and depositional features produced by the
1977 GLOF.
® .However, Stewart and LaMarche 1967 , Costa
® . ® .1983 , and Carling 1987 documented the surface
of longitudinal bars at the flood-water surface or just
®below the water surface. Additionally, Jarrett in
.review conducted a systematic study of floods in the
western USA, showing that the top of boulder bars
may be just below the water surface, at the water
surface, or protrude above the water in some cases.
Thus, the top of cobble-boulder bars in this study
may also represent the maximum peak stage of the
GLOFs or, in some instances, slightly overestimate
the peak stage of the GLOFs.
Scour lines and nonflooded surfaces identified
along the study reaches were considered to represent
the maximum peak stage of the outburst flood. Scour
lines may be considerably higher than the actual
water surface if undercutting of the side slope caused
the surface above the water surface to fail, so caution
should be exercised when using this type of PSI.
Nonflooded surfaces, such as alluvial fans and
glaciofluvial terraces, were identified as such if sedi-
ment was not deposited on these surfaces and the
surfaces were not eroded during the GLOF. How-
ever, water can overtop a surface without depositing
sediment or eroding that surface. Thus, nonflooded
surfaces do not absolutely delineate the maximum
elevation of a given flood, but were probably a
reasonable assumption in most situations.
Streams in the study area have relatively high
concentrations of sediment in suspension because the
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 67
headwaters of most streams are occupied by glaciers
® .and underlain by glacial sediment Fig. 2 . During
the field season, the streams had a distinct, milky-
white-gray color. Along the cobble-boulder margins
of the main channel, a well-defined powdery-white-
gray colored silt-clay line is present and was used to
delineate the water-surface elevation of the SHFFs.
The silt-clay PSIs identified at reaches N1, N2, N3,
and N4 represent the largest SHFF that has occurred
since the GLOF in 1977, whereas the silt-clay PSIs
identified at reaches L2, L3, L1, L4, L6, L5, L7, and
L8 represent the largest SHFF that has occurred
since the GLOF in 1985. For the older GLOF and
tributary study reaches, the ages of the SHFFs that
produced the silt-clay PSIs are unknown.
3.4. Discharge calculations
3.4.1. Step-backwater modeling
®The computer program HEC-RAS Hydrologic
.Engineering Center-River Analysis System , devel-
oped by the Hydrologic Engineering Center of the
® .U.S. Army Corps of Engineers 1995 , was used to
perform the step-backwater calculations and estimate
the peak discharges of the SHFFs and GLOFs. The
hydraulic theory of step-backwater analysis and its
application to natural channels have been discussed
® . ® .in detail by Chow 1959 , Bailey and Ray 1966 ,
® . ® . ® .Davidian 1984 , Hoggan 1989 , O'Connor 1993
® .and O'Connor and Webb 1988 and will only be
briefly reviewed here.
The step-backwater method calculates a one-di-
mensional, energy-balanced, water-surface profile
that is a function of discharge, channel roughness,
®and channel geometry Chow, 1959; Bailey and Ray,
1966; Davidian, 1984; Hoggan, 1989; O'Connor,
1993; O'Connor and Webb, 1988; Hydrologic Engi-
.neering Center, 1995 . Flow can be modeled as
subcritical, supercritical, or both subcritical and su-
® .percritical mixed flow regime . For subcritical flow,
the step-backwater calculations begin at the furthest
downstream cross-section and proceed upstream;
whereas for supercritical flow, step-backwater calcu-
lations begin at the furthest upstream cross-section
and proceed downstream. To perform the step-back-
water calculations for subcritical flow, a starting
water-surface elevation needs to be estimated at the
furthest downstream cross-section by assuming ei-
® .ther i a known water-surface elevation defined by a
® . ®PSI, ii critical depth which is calculated by itera-
tively determining the water-surface elevation at the
cross-section so that the specific energy is at a
. ® . ®minimum , or iii normal depth in which the energy
slope should be entered; however, bed slope can be
substituted for energy slope because these slopes are
.equal for normal depth or uniform flow conditions .
A known water-surface elevation is selected if a PSI
has been identified at the furthest downstream
cross-section. Critical depth is selected at channel
transitions, such as abrupt steepening in slope or
narrowing of the channelrvalley that causes flow to
change from subcritical to supercritical flow. Imme-
diately upstream from the channel transition, flow is
critical and the total energy head is at a minimum.
The critical depth or critical water-surface elevation
can be determined by the equation:
HsWSqa 2
r2 g, 1® .2 2
where H is the total energy head, WS is the water
surface elevation, and a 2
r2 g is the velocity head.2 2
® . ®Solving Eq. 1 is an iterative procedure calculated
.in the HEC-RAS program in which WS values are
assumed and changed until a minimum H value is
obtained. Normal depth is selected if the known
water-surface elevation or critical depth criteria are
not met. Normal depth is calculated using Manning's
equation. If normal depth is selected, the user must
enter the energy slope, which can be approximated
by entering the channel slope at the furthest down-
stream cross-section. Regardless of the initial starting
water-surface elevation, the water-surface profile will
typically converge to a single profile for a specified
discharge if the step-backwater calculations are car-
ried upstream for an adequate distance, typically
®three or four cross-sections Bailey and Ray, 1966;
.O'Connor and Webb, 1988 .
For a specified discharge and assumed friction
and form energy losses, the step-backwater method
iteratively calculates an energy-balanced, water-
surface elevation between the surveyed cross-sec-
tions. The elevation of the PSIs at cross-sections
along a surveyed reach are then compared to the
computed water-surface elevation at the respective
cross-sections. The water-surface profile is adjusted
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9068
® . ® . ® . ® . ® . ® .Fig. 7. Geomorphic maps of reach L2 A , reach L1 B , reach L4 C , reach L5 D , reach L7 E , and reach L8 F showing the
distribution of erosional and depositional features produced by the 1985 GLOF.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 69
® .Fig. 7 continued .
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9070
® . ® .Fig. 8. a Downstream view of the older GLOF deposits identified along the upper Bhoti Kosi reach UBK . The valley width ranges
® .between 50 and 70 m. Although not easily seen in the photograph, the cobbles and boulders are encrusted by lichens. b Side view of an
® .older GLOF deposit identified along the upper Dudh Kosi reach UDK . Field of view is 60 m wide.
by varying discharge until the computed water-
surface profile best matches the PSIs at the modeled
reach.
When applying the step-backwater method to nat-
® .ural channels, the basic assumptions are that i flow
is relatively steady or constant along the surveyed
® .reach, ii flow is gradually varied between succes-
® . ® .sive cross-sections, iii flow is one dimensional, iv
® .slopes are less than 10%, and v the energy slope
between successive cross-sections is constant across
®the cross-section Bailey and Ray, 1966; Davidian,
1984; O'Connor and Webb, 1988; Hoggan, 1989;
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 71
Table 1
Summary of survey parameters and Manning's n coefficients for the study reaches of the GLOFs and SHFFs
Study Distance from Reach Drainage Reach Number Average Initial Adjusted Initial Channel n
a c
reach breached elevation area length of cross gradient channel n channel n overbank n for SHFF
2 b b® . ® . ® .moraine m km m sections for GLOF for GLOF for GLOF
® .km
1977 GLOF
N1 8.6 3840 339 385 8 0.070 0.100 0.160 0.300 0.110
N3 11.5 3790 365 340 10 0.087 0.100 0.140 0.200 0.120
1985 GLOF
L2 7.1 4080 41 1000 10 0.053 0.100 0.125 0.300 0.105
L1 10.9 3710 295 1565 19 0.060 0.100 0.140 0.300 0.105
L4 15.6 3440 356 775 9 0.071 0.100 0.140 0.300 0.110
L5 22.1 2770 1093 490 10 0.033 0.100 0.110 0.200 0.090
L7 24.7 2700 1143 270 9 0.030 0.100 0.100 0.200 0.085
L8 26.7 2580 1151 750 16 0.027 0.100 0.100 0.200 0.075
Older GLOFs
UBK 4040 211 290 8 0.039 0.100 0.100 0.200 0.090
UDK 3340 271 130 7 0.056 0.100 0.110 0.200 0.100
Other reaches
KJK 3360 20 35 5 0.061 0.085
d
N2 8.2 3940 33 245 10 0.115 na na na 0.150
KK 2840 44 45 6 0.052 0.105
TK 3590 45 70 7 0.080 0.115
d
L3 9.4 3790 221 335 6 0.056 na na na 0.100
IK 3340 306 115 8 0.056 0.110
d
L6 20.7 2840 402 65 6 0.055 na na na 0.105
d
N4 2840 685 55 5 0.020 0.080
a
Nsreaches along 1977 GLOF route; Lsreaches along the 1985 GLOF route; UBKsUpper Bhoti Kosi; UDKsUpper Dudh Kosi;
MDKsMiddle Dudh Kosi; IKsImja Khola; KJKsKyajo Khola; KKsKyashar Khola; TKsThame Khola. Refer to Fig. 2 for reach
locations.
b
® .Determined from method described by Arcement and Schneider 1989 .
c
® . 0.38 y0.16
Determined using Jarrett's 1984 equation, ns0.32S R .
d
Along 1977 or 1985 GLOF route, but reach not appropriate to model the GLOF using the step-backwater method.
.Hydrologic Engineering Center, 1995 . The peak
discharges of the GLOFs were estimated to have
lasted at least 1 hr; and because the length of the
reaches modeled were between 270 and 1600 m, the
relatively steady flow assumption was probably sat-
® .isfied Table 1 . Cross-sections were selected along a
given reach so that the gradually varied flow as-
® .sumption was satisfied Figs. 6 and 7 . Flow at a
given cross-section was probably not one dimen-
sional for the GLOFs or SHFFs; however, one-di-
mensional flow was assumed and was probably the
dominant flow direction, satisfying this assumption
in the step-backwater model. Seventeen of the eigh-
teen reaches modeled had reach-averaged slopes less
than 10%. Although the energy slope is probably not
uniform across a cross-section or between successive
cross-sections in these high-gradient channels, the
deviations are assumed to be small.
3.4.2. Selecting energy-loss coefficients
Selecting appropriate channel and flood plain
® .roughness coefficients Manning's n and channel
expansionrcontraction coefficients are important
components to accurately modeling flow conditions
along a surveyed reach using the step-backwater
method. For extreme floods, such as the 1977 and
1985 GLOFs, considerable energy losses because of
turbulence and sediment transport must be taken into
account using Manning's n to estimate total energy
® .losses Trieste and Jarrett, 1987 . If the above-men-
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9072
tioned factors are not considered when selecting
Manning's n, total energy losses are typically under-
estimated, which in turn causes flood discharges to
be overestimated because flow is incorrectly mod-
® .eled as supercritical Trieste and Jarrett, 1987 . As
® . ® .pointed out by Jarrett 1984, 1987 , Trieste 1992
® .and Trieste and Jarrett 1987 , flow is primarily
subcritical in natural channels for extreme floods;
and supercritical flow only occurs locally with lim-
® .ited spatial extent. Trieste and Jarrett 1987 recom-
mended that if flow is computed as being supercriti-
cal along a modeled reach, the initially selected
Manning's n should be increased until flow is sub-
critical and approaching critical along the modeled
reach. This adjustment of Manning's n accounts for
other energy losses associated with an extreme flood.
For the GLOFs modeled in this study, flow was
assumed to be subcritical and approaching critical
flow.
For the GLOFs, initial Manning's n values for
® . ® .the channel n and overbank areas n werec ob
determined using the visual method described by
® . ® .Arcement and Schneider 1989 Table 1 . Using
these initial Manning's n values to perform the
step-backwater calculations for the GLOFs caused
the assumption of subcritical flow to be violated at
different segments of seven of the ten GLOF reaches
modeled. For these seven reaches, the initially se-
lected Manning's n values for the main channel
were increased until flow conditions were subcritical
®or near critical for the entire modeled reach Table
.1 . Manning's n was not adjusted from the initial
selected values for overbank areas.
For the SHFFs, Manning's n was estimated using
® .Jarrett's 1984 equation,
ns0.32 S0.38
Ry0 .16
2® .f
® .where R is hydraulic radius m and S is the energyf
slope. In this study, bed slope was substituted for
energy slope. Jarrett's equation was developed for
discharges ranging from 0.34 to 128 m3
rs, hydraulic
radii ranging from 0.15 to 2.10 m, and slopes rang-
® .ing from 0.002 to 0.04 Jarrett, 1984 . Although the
slopes and discharges of channels in this study, in
most cases, did not fall within the limits for which
the equation was designed, the predicted Manning's
n from Jarrett's equation are probably reasonable.
® .For example, Marcus et al. 1992 , in a study of
high-gradient channels in Alaska in which several
channels were not within the data limits of Jarrett's
equation, showed that Jarrett's equation only slightly
overpredicted Manning's n and provided the best
estimate of Manning's n when compared to visual
estimates and other equations.
Energy losses resulting from channel expansion
and contraction are taken into account in the step-
backwater method by selecting expansion and con-
traction coefficients. Reaches were selected and
cross-sections placed so that the assumption of grad-
ually varied flow between cross-sections was satis-
® .fied for the most part Figs. 6 and 7 . Contraction
and expansion coefficient values were assigned 0.1
and 0.3, respectively, as recommended by the Hydro-
® .logic Engineering Center 1995 for gradual flow
transitions.
Sensitivity analyses were performed on the se-
lected flow-resistance and energy-loss coefficients to
assess how variations in the selected values affect
the water-surface profiles and estimated peak dis-
charges. To assess the influence of the selected
energy-loss coefficients on the calculated discharges,
the following sensitivity analyses were performed:
® .i the Manning's n values selected using the meth-
ods described above were varied by 10% and 25%,
® .ii Manning's n values for the overbank areas were
changed to the same value as the main channel
® .Manning's n, and iii contraction and expansion
coefficients were changed from 0.1 and 0.3 to 0 and
0.5, 0 and 0, and 0.3 and 0.7, respectively.
4. Step-backwater modeling results
4.1. 1977 Glacial-lake outburst flood
4.1.1. Reach N1
Reach N1 is located 8.6 km downstream from the
® .breached moraine Fig. 2 . The physical character-
istics of the reach are summarized in Table 1. The
GLOF caused considerable erosion and deposition
® .along this segment of the valley Figs. 4a and 6A .
The starting water-surface elevation was determined
using the normal-depth method. The initially selected
main channel Manning's n was increased 60% to
®obtain subcritical flow for the outburst flood Table
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 73
.1 . The reconstructed water-surface profile through
reach N1 for the discharge of 1900 m3
rs was rea-
sonably well constrained at each of the eight sur-
veyed cross-sections by the bar deposit and non-
® .flooded surface PSIs Fig. 9A . The computed criti-
cal-depth profile closely matches the boulder PSIs
® .along reach N1 Fig. 9A , suggesting that modeling
flow as critical may have been a better assessment of
hydraulic conditions at reach N1. Table 2 summa-
rizes the reach-averaged hydraulics of the modeled
GLOF.
4.1.2. Reach N3
Reach N3 is located 11.5 km downstream from
® .the breached moraine Fig. 2 . Erosion and deposi-
tion by the GLOF was minimal along this reach
because the bedrock and coarse-grained colluvium
valley boundaries resisted erosion for the most part
and because the steep, narrow valley produced hy-
draulic conditions that transported most of the GLOF
® .sediment through the reach Figs. 4b and 6B . Imme-
diately downstream from XS2, the valley narrows
® .and steepens Fig. 6B , providing a critical-depth
control for the starting water-surface elevation of the
GLOF. Undisturbed, pre-1977-age vegetation and
poorly developed soil along the left margin of the
bedrock ledge at XS2 were used to delineate the
® .maximum stage of the GLOF Fig. 9B . Using the
critical-depth method and the maximum stage PSI at
XS2, the GLOF discharge was estimated at 1500
m3
rs. Upstream from the critical-depth control, the
initially selected main channel Manning's n value
was increased 40% so that flow was subcritical for
® .the modeled reach Table 1 . The reconstructed wa-
ter-surface profile upstream from the critical-depth
control was constrained by deposit and nonflooded
®surface PSIs at five of the nine cross-sections Fig.
.9B . Table 2 summarizes the reach-averaged hy-
draulics of the modeled GLOF.
4.2. Glacial-lake outburst flood
4.2.1. Reach L2
Reach L2 is located 7 km downstream from the
® .breached moraine Fig. 2 . Reach boundary condi-
tions and geomorphic features produced by the GLOF
are illustrated in Figs. 5a and 7A. At the furthest
downstream cross-section, the valley steepens
® .abruptly providing a critical-depth control Fig. 7A .
A small deposit of pebbles and cobbles just below a
nonflooded surface at XS10 were used to delineate
the maximum stage of the GLOF. Using the critical-
depth method and the PSIs at XS10, the GLOF
3 ® .discharge was estimated at 2350 m rs Fig. 10A .
Upstream from the critical-depth control, the initially
selected main channel Manning's n was increased
®25% so that the modeled flow was subcritical Table
.1 . Using this approach, the reconstructed water-
surface profile upstream from the critical-depth con-
trol section was constrained by the nonflooded PSIs,
but was at least 1 m higher than the deposit PSIs
® . ® .Fig. 9. Computed water-surface profile and comparison with PSIs for the 1977 GLOF at reach N1 A and at reach N3 B .
()D.A.Cenderelli,E.E.WohlrGeomorphology40200157­9074
Table 2
® .Summary of reach-averaged and ranges of in parentheses below the reach average hydraulic variables of the modeled GLOF reaches
Study Total Channel discharge Energy slope Total velocity Channel Total flow Channel flow Total hydraulic Channel hydraulic Total Froude Channel Froude
a 3® . ® .reach discharge m rs mrs velocity width width radius radius number number
3® . ® . ® . ® . ® . ® .m rs m m m m m
N1 1900 1624 0.071 3.98 5.30 121 49 3.70 5.79 0.64 0.68
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .1401­1856 0.045­0.089 3.28­4.48 4.72­5.73 109­133 40­60 3.41­4.26 5.13­6.71 0.49­0.74 0.56­0.75
N3 1500 1456 0.071 5.85 6.26 44 32 4.93 6.37 0.74 0.77
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .1381­1491 0.032­0.118 4.53­7.57 4.77­8.25 34­61 19­39 4.02­5.89 5.61­7.83 0.52­0.94 0.54­1.00
L2 2350 1863 0.049 4.24 6.07 122 45 4.75 6.50 0.60 0.73
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .1371­2337 0.033­0.093 2.94­8.29 5.17­8.65 46­180 28­62 3.42­5.56 5.68­7.46 0.40­1.00 0.62­0.97
L1 2250 1914 0.060 4.42 5.92 113 49 4.58 6.36 0.63 0.72
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .1615­2249 0.029­0.125 3.27­7.91 4.33­7.96 45­222 29­103 2.99­5.75 4.28­7.84 0.53­1.00 0.51­1.00
L4 2275 1608 0.072 4.07 5.62 146 40 4.43 6.61 0.58 0.66
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .986­2252 0.054­0.118 2.49­8.20 3.90­8.55 41­264 25­56 3.33­5.40 4.97­7.93 0.41­1.00 0.52­0.98
L5 1725 1593 0.0330 4.68 5.41 68 43 4.88 6.47 0.63 0.65
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .1299­1725 0.008­0.055 3.12­5.94 3.71­6.85 54­80 25­57 3.64­7.18 4.98­9.36 0.38­0.83 0.35­0.94
L7 1575 1499 0.031 4.87 5.53 67 44 4.37 5.73 0.70 0.70
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .1370­1535 0.018­0.047 3.58­6.11 4.51­6.62 50­82 37­52 3.65­5.07 5.15­6.22 0.49­0.88 0.56­0.88
L8 1375 1302 0.029 4.57 4.98 66 47 4.16 5.19 0.66 0.66
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .1063­1375 0.014­0.057 3.41­5.76 3.69­6.56 41­97 32­61 3.18­5.07 4.28­6.14 0.48­0.83 0.48­0.89
UBK 400 342 0.036 2.93 3.72 64 30 2.04 2.95 0.64 0.67
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .261­385 0.011­0.054 2.07­3.81 2.60­4.27 46­77 18­36 1.63­2.57 2.39­3.83 0.42­0.87 0.42­0.89
UDK 700 670 0.056 4.22 4.48 59 42 2.52 3.13 0.76 0.80
® . ® . ® . ® . ® . ® . ® . ® . ® . ® .638­684 0.035­0.084 3.31­5.21 3.78­5.27 46­77 26­50 2.05­2.83 2.71­4.24 0.59­0.93 0.60­0.99
Total refers to the entire cross sections modeled and includes the main channel and overbank areas.
Channel refers to the main channel of flow.
a
N refers to reaches modeled along the 1977 GLOF route, L refers to reaches modeled along the 1985 GLOF route, UBK and UDK are older GLOF reaches modeled on the
upper Bhoti Kosi and upper Dudh Kosi.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 75
® . ® . ® .Fig. 10. Computed water-surface profile and comparison with PSIs for the 1985 GLOF at reach L2 A , reach L1 B , reach L4 C , reach
® . ® . ® .L5 D , reach L7 E , and reach L8 F .
® .Fig. 10A . This indicates some uncertainty in the
water-surface elevation upstream from the critical-
depth control. Table 2 summarizes the reach-aver-
aged hydraulics of the modeled GLOF.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9076
4.2.2. Reach L1
Reach L1 is located approximately 11 km down-
® .stream from the breached moraine Fig. 2 . Reach
boundary conditions and geomorphic features pro-
duced by the GLOF are illustrated in Figs. 5b and
®7B. At XS1, the valley narrows and steepens Fig.
.7B , providing critical-depth control for the starting
water-surface elevation. Upstream from XS1, the
crest of a longitudinal bar at XS2 was used to
delineate the stage of the GLOF. Using the critical-
depth method combined and the PSI at XS2, the
estimated GLOF discharge was 2250 m3
rs. Up-
stream from the critical-depth control section, the
initially selected main channel Manning's n value
® .was increased 40% Table 1 so that the modeled
flow was subcritical. For the estimated discharge of
2250 m3
rs, the reconstructed water-surface profile
corresponded fairly well with the deposit and non-
® .flooded surface PSIs Figs. 7B and 10B . Table 2
summarizes the reach-averaged hydraulics of the
modeled GLOF.
4.2.3. Reach L4
Reach L4 is located approximately 15.5 km
® .downstream from the breached moraine Fig. 2 .
Reach boundary conditions and geomorphic features
produced by the GLOF are illustrated in Figs. 1b and
7C. The narrow valley between XS2 and XS1 pro-
vided a critical-depth control for the starting water-
surface elevation. A nonflooded surface at XS1,
combined with a thin veneer of cobbles and boulders
overlying a lower glaciofluvial terrace at XS2, were
used to delineate the maximum stage of the GLOF
® .Fig. 7C . Using the critical-depth method in con-
junction with the PSIs at XS1 and XS2, the esti-
3 ® .mated GLOF discharge was 2275 m rs Fig. 10C .
Upstream from the critical-depth control, the initially
selected main channel Manning's n value was in-
creased 40% so that the modeled flow was subcriti-
cal. Upstream from the critical-depth control to XS4,
the reconstructed water-surface profile corresponded
® .reasonably well with the PSIs Fig. 10C . Upstream
from XS4, the computed water-surface profile was
0.68 to 0.77 m lower than the highest deposit PSIs at
® .XS9 and XS5, respectfully Fig. 10C . The poor
match between the water-surface profile and PSIs
upstream from XS4 is partly attributed to modeling
flow through this expanding segment of the reach
® .Fig. 7C . Table 2 summarizes the reach-averaged
hydraulics of the modeled GLOF.
4.2.4. Reach L5
Reach L5 is located approximately 22 km down-
® .stream from the breached moraine Fig. 2 . Reach
boundary conditions and geomorphic features pro-
duced by the GLOF are illustrated in Figs. 5c and
7D. The starting water-surface elevation was deter-
mined using the normal-depth method. The initially
selected main channel Manning's n was increased
®10% so that the modeled flow was subcritical Table
.1 . Through reach L5, the reconstructed water-surface
profile, at a discharge of 1725 m3
rs, was fairly well
constrained by the deposit and nonflooded surface
PSIs with the best matches occurring at XS6, XS9,
® .and XS10 Fig. 10D . Table 2 summarizes the
reach-averaged hydraulics of the modeled GLOF.
4.2.5. Reach L7
Reach L7 is located on the Dudh Kosi, approxi-
mately 24.7 km downstream from the breached
® .moraine Fig. 2 . Reach boundary conditions and
geomorphic features produced by the GLOF are
illustrated in Fig. 7E. Nonflooded surfaces were the
primary PSIs identified along reach L7. Deposit PSIs
at high elevations were lacking along reach L7 with
the exception of XS9. A distinct scour line on a large
® .boulder greater than 15 m in diameter with etch-
ings was identified between XS2 and XS3. The
starting water-surface elevation at XS1 was deter-
mined using the normal-depth method. The main
channel Manning's n did not have to be increased
from its initially selected value of 0.10 to attain
subcritical flow for the GLOF using the step-back-
water method. The reconstructed water-surface pro-
file through reach L7 was fairly well constrained by
® .the PSIs Fig. 10E and was associated with a dis-
charge of 1575 m3
rs. Table 2 summarizes the
reach-averaged hydraulics of the modeled GLOF.
4.2.6. Reach L8
Reach L8 is located on the Dudh Kosi, approxi-
mately 26.7 km downstream from the breached
® .moraine Fig. 2 . Reach boundary conditions and
geomorphic features produced by the GLOF are
illustrated in Fig. 7F. The starting water-surface
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 77
elevation was determined using the normal-depth
method. The main channel Manning's n did not have
to be increased from its initially selected value of
0.10 to attain subcritical flow for the GLOF using
the step-backwater method. The reconstructed
water-surface profile through reach L8 adequately
matched the longitudinal bar deposit and nonflooded
surface PSIs and was associated with a discharge of
3 ® .1375 m rs Fig. 10F . The computed critical-depth
profile more closely matched the deposit PSIs along
® .reach L8 Fig. 10F , suggesting that modeling flow
as critical may have been a better assessment of
hydraulic conditions at reach L8. Table 2 summa-
rizes the selected reach-averaged hydraulic parame-
ters of the modeled GLOF through reach L8.
4.3. Older outburst floods
4.3.1. Upper Bhoti Kosi, UBK
The upper Bhoti Kosi reach is located in the
® .upper Bhoti Kosi drainage Fig. 2 at an elevation of
4040 m. Reach boundary conditions and geomorphic
features produced by the GLOF are illustrated in Fig.
8a. The starting water-surface elevation at the fur-
thest downstream cross-section was determined us-
ing the normal-depth method. The main channel
Manning's n was not increased from its initially
selected value of 0.10 to attain subcritical flow for
the GLOF using the step-backwater method. The
reconstructed water-surface profile through the upper
Bhoti Kosi reach matched the cobble-boulder deposit
® .PSIs fairly well Fig. 11A and was associated with a
discharge of 400 m3
rs. Table 2 summarizes the
selected reach-averaged hydraulic parameters of the
modeled older GLOF through reach UBK.
4.3.2. Upper Dudh Kosi, UDK
The upper Dudh Kosi reach is located in the
® .upper Dudh Kosi drainage basin Fig. 2 at an eleva-
tion of 3340 m. Reach boundary conditions and
geomorphic features produced by the GLOF are
illustrated in Fig. 8b. The starting water-surface ele-
vation at the furthest downstream cross-section was
determined using the normal-depth method. The main
channel Manning's n was increased 10% to attain
subcritical flow for the GLOF. The main channel
Froude numbers ranged from 0.59 to 0.93 with a
reach-averaged main channel Froude number of 0.76.
Through the upper Dudh Kosi reach, the recon-
structed water-surface profile at a discharge of 700
m3
rs was fairly well constrained by the deposit PSIs
® .Fig. 11B . Table 2 summarizes the selected reach-
averaged hydraulic parameters of the modeled older
GLOF through reach UDK.
4.4. Seasonal high flow floods
The peak discharges of SHFFs were estimated at
the ten GLOF reaches modeled and at eight addi-
tional reaches. These reaches are located in several
different streams at varying elevations with various
® .drainage areas Table 1; Fig. 2 . The reaches range
in length from 35 to 1565 m, have channel widths
ranging from 10 to 30 m, and have average gradients
® . ® .Fig. 11. Computed water-surface profile and comparison with PSIs for the older GLOFs at reach UBK A and at reach UDK B .
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9078
® .ranging from 0.020 to 0.115 Table 1 . The modeled
reaches can generally be classified as riffles with
channel beds of cobbles-boulders andror bedrock
with isolated cobbles and boulders. Distinct white-
® .Fig. 12. Computed water-surface profile and comparison with PSIs for the SHFFs at the 1977 GLOF study reaches A­D , 1985 GLOF
® . ® . ® .study reaches E­L , older GLOF reaches M and N , and tributary reaches O­R .
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 79
® .Fig. 12 continued .
gray silt lines along the channel margins were used
to delineate the peak stage of the SHFF. The study
reaches IK, KK, TK, N3, L2, L1, and L4 are located
immediately upstream from distinct stable steps, al-
lowing the critical-depth method to be used to deter-
mine the starting water-surface elevation and SHFF
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9080
® .Fig. 12 continued .
discharge in conjunction with geologic evidence. At
the study reaches KYK, N2, N1, N4, L3, L6, L5, L7,
and L8, the initial starting water-surface elevation
was determined using the normal-depth method. All
reaches were modeled as subcritical flow, and aver-
aged reach Froude numbers ranged from 0.66 to
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 81
0.82. As illustrated in Fig. 12, the computed water-
surface profiles are in close agreement with the PSIs
that delineate SHFFs along the 18 reaches. For the
18 reaches modeled, the SHFFs estimated peak dis-
charges ranging between 7 and 205 m3
rs, reach-
averaged velocities ranging from 1.68 to 4.25 mrs,
reach-averaged hydraulic radiuses ranging from 0.41
to 2.16 m, and reach-averaged flow widths ranging
® .from 10.02 to 31.97 m Table 3 .
4.5. Quality of step-backwater modeling
4.5.1. Glacial-lake outburst floods
The uncertainty of the GLOF discharge estimates
based on Abest matchB water-surface profiles ranged
® .from y42% to 36% Table 4 . These uncertainty
discharge estimates were determined by reconstruct-
ing the water-surface profile along each modeled
® .reach so that i the deposit PSIs were above the
®water-surface profile conservative, lower discharge
. ® .estimate and ii the nonflooded surface PSIs were
®below the water-surface profile liberal, upper
.discharge estimate . Although the Abest matchB
discharge estimate provided the most reasonable
discharge estimate because it utilized both deposit
and nonflooded surface PSIs, the lower and upper
discharge estimates provided a potential range of
discharge estimates based on the geomorphic evi-
dence.
The most reliable GLOF discharge estimates,
based on reach selection and the best matches be-
tween the computed water-surface profiles and geo-
morphic evidence, were obtained at reaches N1, N3,
®L2, L1, and L4 Figs. 9A, B, 10A, B and C, respec-
. ®tively . Of these five reaches, four reaches N3, L2,
.L1, and L4 were just upstream from constrictions
where there was critical-depth control and geomor-
phic evidence to define the maximum stage of the
® .GLOFs Figs. 6B, 7A, B and C, respectively . As
® .pointed out by Jarrett 1987 , the critical-depth
method for estimating discharge in high-gradient
streams is preferred because the estimated discharge
is independent of Manning's n. When the discharge
estimate calculated from the critical-depth method at
the critical-depth control was applied to the remain-
der of those reaches, the water-surface profiles at
reaches N3, L2, L1, and L4 were in reasonably close
agreement with the PSIs at their respective reaches
® .Figs. 9B, 10A, B and C, respectively . Note that
upstream from the critical-depth control at reaches
N3, L2, L1, and L4, flow was modeled as subcriti-
cal, but approaching critical-flow conditions.
Although the step-backwater calculations may be
accurate for the 1985 GLOF at reaches L5, L7, and
L8, the discharge estimates are sensitive to the se-
lected energy-loss coefficients, principally Manning's
n. Recall that Manning's n was used not only to
account for energy losses associated with grain
roughness, but also used to account for additional
energy losses from turbulence, erosion, and sediment
transport that occurred during the GLOFs. Because
there is uncertainty in assigning energy-loss coeffi-
cients, underestimating or overestimating the energy-
loss coefficients in the model can result in inaccurate
discharge estimates. Additionally, at these reaches,
either fewer PSIs clearly define the maximum stage
of the GLOF or PSIs are not in complete agreement
® .with each other Fig. 10D, E and F . Although the
match between the water-surface profile and PSIs is
acceptable, subtle inconsistencies between the
water-surface profile and PSIs suggest the actual
flow conditions were not being completely and accu-
rately modeled.
For the older GLOF reaches there is a reasonable
match between the PSIs and computed water-surface
® .profiles Fig. 11A and B ; however, there is uncer-
tainty as to whether the channel geometry had been
modified since the older GLOFs occurred at the
upper Bhoti Kosi and upper Dudh Kosi reaches.
Thus, the older GLOF discharge estimates are based
on the present channelrvalley geometry and not
necessarily the channelrvalley geometry during peak
flow conditions of the older GLOFs. Another source
of uncertainty is that both the upper Bhoti Kosi reach
and upper Dudh Kosi reach gradually expand in the
downstream direction. This gradual expansion intro-
duces error in the computed water-surface profile
because flow is diverging.
4.5.2. Seasonal high flow floods
The step-backwater modeling of the SHFFs at the
various reaches seems to accurately represent those
flows. At the older GLOF and tributary reaches, the
silt-clay PSIs delineate the largest SHFF that has
occurred at those reaches. The abundant seasonal
high flow flood PSIs at the reaches modeled were
()D.A.Cenderelli,E.E.WohlrGeomorphology40200157­9082
Table 3
® .Summary of reach-averaged and ranges of in parentheses below the reach average hydraulic variables of the modeled SHFF reaches
Study Drainage Reach Discharge Energyslope Velocity Flow width Hydraulic radius Froude number
a 3® . ® . ® . ® .reach area elevation m rs mrs m m
2® . ® .km m
KJK 20 3360 7 0.071 1.68 10.02 0.41 0.82
® . ® . ® . ® . ® .0.040­0.105 1.44­2.08 7.60­13.65 0.33­0.53 0.65­1.00
N2 33 3940 25 0.119 1.84 17.49 0.74 0.66
® . ® . ® . ® . ® .0.085­0.201 1.64­2.05 13.31­27.81 0.48­0.90 0.56­0.83
L2 41 4080 40 0.063 2.25 18.77 0.94 0.71
® . ® . ® . ® . ® .0.037­0.120 1.73­3.37 10.13­30.89 0.69­1.26 0.57­1.00
KK 44 2840 25 0.073 1.98 15.16 0.74 0.69
® . ® . ® . ® . ® .0.041­0.157 1.55­2.31 9.24­19.70 0.48­0.89 0.54­1.00
TK 45 3590 35 0.082 2.32 14.32 0.97 0.71
® . ® . ® . ® . ® .0.030­0.155 176­2.91 10.81­20.08 0.78­1.26 0.44­1.00
UBK 211 4040 60 0.033 2.17 21.52 1.19 0.60
® . ® . ® . ® . ® .0.008­0.054 1.29­2.51 15.16­29.85 0.88­1.51 0.33­0.78
L3 221 3790 70 0.057 2.69 19.52 1.24 0.74
® . ® . ® . ® . ® .0.035­0.97 2.23­3.36 16.53­26.33 1.08­1.50 0.60­0.96
UDK 271 3340 90 0.054 2.68 24.07 1.37 0.70
® . ® . ® . ® . ® .0.018­0.090 1.97­3.35 14.31­41.65 0.86­1.93 0.44­0.89
L1 295 3710 85 0.059 2.60 26.02 1.24 0.72
® . ® . ® . ® . ® .0.026­0.114 1.98­3.39 16.59­51.97 0.78­1.58 0.50­1.00
IK 306 3340 100 0.068 2.74 23.94 1.33 0.72
® . ® . ® . ® . ® .0.035­0.125 1.93­3.64 15.86­31.03 1.08­1.69 0.53­1.00
N1 339 3840 135 0.073 3.11 28.91 1.45 0.80
® . ® . ® . ® . ® .0.046­0.107 2.46­3.61 20.89­42.09 1.21­1.83 0.60­0.90
L4 356 3440 85 0.080 3.08 19.39 1.33 0.82
® . ® . ® . ® . ® .0.060­0.119 2.57­3.62 6.15­25.99 1.19­1.52 0.72­1.00
N3 365 3790 135 0.091 3.51 20.76 1.70 0.80
® . ® . ® . ® . ® .0.042­0.142 2.68­4.69 12.76­31.10 1.49­1.99 0.58­1.00
L6 402 2840 90 0.058 2.95 18.27 1.55 0.73
® . ® . ® . ® . ® .0.018­0.096 1.96­3.72 15.11­21.58 1.37­1.90 0.43­0.94
N4 685 2840 165 0.040 4.25 12.39 2.16 0.76
® . ® . ® . ® . ® .0.025­0.062 3.65­5.11 10.44­13.36 1.99­2.28 0.62­0.93
L5 1093 2770 195 0.034 3.10 31.97 1.94 0.68
® . ® . ® . ® . ® .0.014­0.068 1.98­4.61 19.17­48.91 1.45­2.61 0.44­0.99
L7 1143 2700 200 0.035 3.31 28.46 2.02 0.73
® . ® . ® . ® . ® .0.013­0.059 2.34­4.19 22.27­33.34 1.52­2.37 0.46­0.94
L8 1151 2580 205 0.029 3.34 31.82 1.87 0.75
® . ® . ® . ® . ® .0.011­0.047 2.45­4.20 17.78­49.00 1.38­2.50 0.49­0.95
a
N refers to reaches modeled along the 1977 GLOF route, L refers to reaches modeled along the 1985 GLOF route, UBKsupper Bhoti Kosi, UDKsupper Dudh Kosi,
IKsImja Khola, TKsThame Khola, KJKsKyajo Khola, KKsKyashar Khola.
()D.A.Cenderelli,E.E.WohlrGeomorphology40200157­9083
Table 4
Summary of best match discharges and bracketing discharges for the 1977, 1985, and older glacial-lake outburst floods
Study Best match peak Lower bracketing Percent change Upper bracketing Percent change Lower bracketing Percent change Upper bracketing Percent change
a
reach discharge peak discharge in peak peak discharge in peak peak discharge in peak peak discharge in peak
estimate estimate for discharge estimate for discharge estimate for discharge estimate for discharge
3® .m rs deposit PSI estimate deposit PSI estimate nonflooded PSI estimate nonflooded PSI estimate
3 3 3 3® . ® . ® . ® .m rs m rs m rs m rs
N1 1900 1200 y37 1900 0 1900 0 2400 26
N3 1500 1300 y13 1500 0 1500 0 2000 33
L2 2350 2200 y6 2350 0 2350 0 2700 15
L1 2250 1300 y42 2250 0 2250 0 3100 38
L4 2275 1500 y34 2900 28 2275 0 2800 23
L5 1725 1100 y36 1725 0 1725 0 2350 36
L7 1575 1575 0 1575 0 1575 0 1850 17
L8 1375 900 y35 1375 0 1375 0 1700 24
UBK 400 250 y38 400 0 na na na na
UDK 700 600 y14 700 0 875 25 875 25
a
N refers to reaches modeled along the 1977 GLOF route, L refers to reaches modeled along the 1985 GLOF route, UBKsupper Bhoti Kosi, and UDKsupper Dudh Kosi.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9084
usually in close agreement with the computed
water-surface profile. The uncertainty of the SHFF
discharge estimates based on Abest matchB water-
® .surface profile ranged from y43% to 43% Table 5 .
These uncertainty discharge estimates were deter-
mined by reconstructing the water-surface profile
along each modeled reach so that silt-clay PSIs were
®above the water-surface profile conservative, lower
.discharge estimate and below the water-surface pro-
® .file liberal, upper discharge estimate .
® . ®Although the use of Jarrett's 1984 equation Eq.
® ..2 to estimate Manning's n for the channels stud-
ied is debatable because many of the slopes and
discharges fell outside of the data range used to
develop the equation, the SHFFs modeled were usu-
ally close to critical flow using the Manning's n
value calculated from Jarrett's equation. It has been
suggested that in high-gradient, mountainous chan-
®nels, flow is usually close to critical Jarrett, 1984;
Trieste, 1992; Trieste and Jarrett, 1987; Tinkler,
.1996; Grant, 1997 . For most of the SHFFs modeled
in this study, flows were close to critical, which
suggests that the Manning's n calculated using Jar-
rett's equation reasonably estimated the channel
®roughness of streams in the study area Fig. 12,
.Table 3 .
Perhaps the greatest uncertainty in modeling the
SHFFs is the channel geometry. Although the flows
during the field season were low, they were high
enough to prevent channel crossing and surveying
the topography within the channel. To compensate
for the lack of surveying information within the
channel, for each modeled reach the channel was
® .assumed to be i rectangular between the left and
right bank water's edge at the time of measurement
® .and ii at least 1 m lower than the water-surface
elevation at the time of measurement based on visual
estimates and probing. For the latter assumption, the
channel-bottom elevation was additionally lowered
by the elevation difference between the SHFF PSI
and the water surface at the time of measurement.
This elevation adjustment to the channel bottom was
necessary because the elevation difference between
the SHFF PSI and water surface at the time of
Table 5
Summary of best match discharges and bracketing discharges for the seasonal high flow floods
Study Best match peak Lower bracketing Percent change in Upper bracketing Percent change in
a
reach discharge estimate peak discharge peak discharge peak discharge peak discharge
3® .m rs estimate for estimate estimate for estimate
deposit PSI deposit PSI
3 3® . ® .m rs m rs
N1 135 110 y19 170 26
N3 135 105 y22 165 22
L2 40 35 y13 50 25
L1 85 65 y24 110 29
L4 85 60 y29 110 29
L5 195 150 y23 205 5
L7 200 180 y10 235 18
L8 205 180 y12 260 27
UBK 60 50 y17 80 33
UDK 90 65 y28 120 33
KJK 7 4 y43 10 43
N2 25 20 y20 30 20
KK 25 20 y20 30 20
TK 35 30 y14 40 14
L3 70 65 y7 75 7
IK 100 80 y20 115 15
L6 90 65 y28 105 17
N4 165 150 y9 190 15
a
N refers to reaches modeled along the 1977 GLOF route, L refers to reaches modeled along the 1985 GLOF route, UBKsupper Bhoti
Kosi, UDKsupper Dudh Kosi, IKsImja Khola, TKsThame Khola, KJKsKyajo Khola, KKsKyashar Khola.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 85
measurement increased with increasing drainage area.
This surveying adjustment was consistently applied
to each modeled reach in an effort to minimize error
or at least be consistent on the amount of error
introduced. Because of the lack of channel bottom
surveying information, the discharges of the SHFFs
may be underestimated or overestimated. Precise
characterization of the channel bottom probably had
minimal effects on the estimated GLOF discharges
because of the greater flow depths associated with
the GLOFs.
4.5.3. Sensitiity analysis on the selection of friction-
and energy-loss coefficients
To assess the influence or sensitivity of the se-
lected Manning's n values and the contractionrex-
pansion coefficient values on the step-backwater
method, these values were varied. Varying contrac-
tion and expansion coefficient values of 0 and 0.5, 0
and 0, and 0.3 and 0.7, respectively, while keeping
Manning's n values constant had little effect on the
computed step-backwater water-surface profile. For
example, using conservative contraction and expan-
sion coefficient values of 0.3 and 0.7, respectively,
for abrupt flow transitions, discharge did not change
or decreased by only 4.35%. These results suggest
that the energy losses associated with contraction
and expansion have a minimal effect on the esti-
mated discharges, regardless of the values selected.
Varying Manning's n for the main channel had
the greatest influence on the estimated discharges.
Increasing the main channel Manning's n by 10%
caused the estimated discharge to decrease by 5.5%
to 9.1% for the GLOFs and decrease by 5.9% to
10.0% for the SHFFs. Increasing the main channel
Manning's n by 25% caused the estimated discharge
to decrease by 13.7% to 20.0% for the GLOFs and
14.3% to 20.7% for the SHFFs. The changes in the
SHFF discharges that occurred in response to the
changes in Manning's n are similar in magnitude to
the Manning's n changes documented by Wohl
® .1998 in a study of high-gradient channels. The
results in this study show that the estimated dis-
charge values are influenced by the selection of the
main channel Manning's n. With the exception of
reach L2, reducing the n values predicted by Jarrett's
® . ® ® ..1984 equation Eq. 2 by 10% caused sections of
the modeled SHFF not to be subcritical.
Adjusting the overbank Manning's n by y10%,
y25%, 10%, and 25%, while keeping the main
channel constant had a minimal effect on the esti-
mated GLOF discharges. For example, increasing
overbank Manning's n by 25% either did not change
the discharge or decreased the discharge estimate by
only 4.2%. Similarly, decreasing overbank Manning's
n by 25%, either did not change the discharge or
increased the total discharge estimate by only 6.3%.
These results indicate that the selected overbank
Manning's n only has a minimal effect on the
outburst flood estimated discharge.
A uniform Manning's n for the main channel and
overbank areas caused the estimated GLOF dis-
charges to change; however, the magnitude and
change in discharge were variable between reaches.
®In general, reaches with larger overbank areas re-
.aches L1, L2, L3, L4, and L6 had greater changes in
® .discharge ranging from y4.3% to 15.6% , whereas
®reaches with smaller overbank areas reaches N1,
.N3, L5, L7, and L8 only had a minimal change in
®the estimated discharges ranging from 1.7% to
.10.1% .
In summary, the sensitivity analyses performed
illustrate that varying contraction coefficients, expan-
sion coefficients, and overbank Manning's n only
had a minimal effect on the estimated discharges in
this study. In contrast, varying the main channel
Manning's n had the greatest effect on the estimated
GLOF and SHFF discharges. Increasing the main
channel Manning's n by 10% and 25% caused a
similar magnitude decrease in the estimated dis-
charge. These results indicate that of the energy
losses quantified in the step-backwater model, the
selection of the main channel Manning's n had the
greatest influence or control on the estimated dis-
charges of the GLOFs and SHFFs determined by
using the step-backwater method.
5. Flood hydrology in the Mount Everest region
5.1. Seasonal high flow floods
The SHFF discharge estimates at the 18 reaches
quantify for the first time the flood hydrology from
seasonal climatic conditions in the Mount Everest
® .region Tables 3 and 5 . The SHFF discharge esti-
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9086
Fig. 13. Plot of the GLOF and SHFF peak discharges vs. drainage
area. The solid line delineates the envelope of the maximum
discharges of the SHFFs for various drainage areas.
mates only quantify the largest SHFF that has oc-
curred at a given study site because the silt-clay PSIs
only delineate the maximum SHFF. The SHFF dis-
charge estimates provide a baseline to which the
magnitude of the GLOF discharges can be compared.
A plot of the SHFF discharges vs. drainage area
shows that as contributing drainage area increases
® .the discharge of the SHFFs increases Fig. 13 .
Hydrology in the study area is strongly influenced
by monsoonal precipitation and late springrearly
® .summer snowmelt and glacier runoff. Ageta 1976
® .and Zimmermann et al. 1986 suggested that the
high mountain topography in the study area creates a
rain-shadow effect that reduces the intensity and
amount of monsoonal precipitation with increasing
elevation. The estimated SHFF discharges in the
study area show a progressive, gradual increase with
® .increasing drainage area Fig. 13 . This trend of
gradual, increasing discharges with increasing
drainage area suggests that extreme flooding from
intense monsoonal precipitation is unlikely in the
study area because the high mountain topography
impedes and minimizes the influx of intense mon-
soonal precipitation into the study area drainage
basin.
5.2. Glacial-lake outburst floods
The estimated peak discharges of the 1977, 1985,
and older GLOFs were considerably larger than the
® .SHFFs Fig. 13; Table 6 . As drainage area increases
or distance from the GLOF source increases, the
difference in magnitude between the GLOFs and
® .SHFFs decreases Fig. 12; Table 6 . At locations
near the GLOF source, the discharge ratio between
the GLOF and SHFF is greatest and progressively
decreases downstream as the GLOF attenuates and
the SHFF increases. At the reaches where the 1977
and 1985 GLOF discharges were estimated, the
GLOF discharges were 7 to 60 times greater than the
® .SHFF discharges Fig. 13; Table 6 .
Comparing the peak discharge estimates of the
1985 GLOF and SHFF along the 1985 GLOF route
perhaps best illustrates changes in the peak dis-
charges of the GLOF and SHFF with respect to
distance downstream from the breached moraine and
Table 6
Summary of discharge estimates at reaches where both the GLOF and SHFF were modeled
Study Distance from breached GLOF discharge, SHFF discharge, Ratio of QGLOF
a
reach moraine Q Q to QGLOF SHFF SHFF
3 3® . ® . ® .km m rs m rs
N1 8.6 1900 135 14
N3 11.5 1500 135 11
L2 7.1 2350 40 59
L1 10.9 2250 85 27
L4 15.6 2275 85 27
L5 22.1 1725 195 9
L7 24.7 1575 200 8
L8 26.7 1375 205 7
UBK not known 400 60 7
UDK not known 700 90 8
a
N refers to reaches modeled along the 1977 GLOF route, L refers to reaches modeled along the 1985 GLOF route, UBKsupper Bhoti
Kosi, and UDKsupper Dudh Kosi.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 87
increasing drainage area. For the 1985 GLOF, the
peak discharge of the GLOF ranged from 2250 to
2400 m3
rs along the upper 16 km of the flood route
®and was 30 to 60 times greater than SHFF Fig. 14;
.Table 6 . The lack of a decreasing trend in the GLOF
discharge along the upper 16 km of the 1985 GLOF
route reflects the uncertainty in the GLOF discharge
® .estimates for these reaches Fig. 14; Table 6 . Addi-
tionally, the direction of error associated with the
GLOF discharge estimates at reaches L2, L1, and L4
influences the discharge relationship between these
® .reaches Table 6 . At 27 km downstream from the
breached moraine, the peak discharge of the 1985
GLOF attenuated to 1375 m3
rs and was seven times
® .greater than SHFF Fig. 14; Table 6 . The down-
stream decline in the ratio of the 1985 GLOF peak
discharge to the SHFF peak discharge is the result of
the downstream attenuation of the GLOF peak dis-
charge and the increased peak discharge of SHFF
because of increased contributing drainage area and
the increased effects of monsoonal precipitation at
lower elevations. The most distinct changes in the
discharge ratio between the GLOF and SHFF occur
at major confluences: the Langmoche Khola and
Bhoti Kosi confluence and the Bhoti Kosi and Dudh
® .Kosi confluence Fig. 14 . Immediately downstream
from these confluences, the peak discharge of the
SHFF approximately doubled because of the consid-
erable increase in drainage area, resulting in a sharp
decrease in the discharge ratio between the GLOF
discharge and SHFF discharge.
The 1977 GLOF had an estimated peak discharge
of 1900 m3
rs at reach N1, which is approximately
8.6 km downstream from the breached moraine. At
reach N1, the peak discharge of the GLOF was about
14 times greater than the peak discharge of the SHFF
® .at that location Table 6 . At reach N3, approxi-
mately 3 km downstream from reach N1 and 11.5
km downstream from the breached moraine, the
1977 GLOF attenuated to 1500 m3
rs and was 11
times greater than the peak discharge of the SHFF at
® .that location Fig. 13; Table 6 . It was mentioned
earlier that the peak discharge of the 1985 GLOF
was larger than the 1977 GLOF because of the lack
of paleoflood evidence for the 1977 GLOF down-
stream from the Dudh Kosi and Bhoti Kosi conflu-
® .ence Fig. 2 . The 1977 and 1985 GLOF discharge
estimates support this assumption. At reach N3, the
® .Fig. 14. A Peak discharges of the six GLOF study reaches and
® .eight SHFF study reaches along the 1985 GLOF route. B Ratio
of the GLOF peak discharge to the SHFF peak discharge at the six
study reaches along the 1985 GLOF route.
1977 GLOF had a peak discharge of 1500 m3
rs,
which is 9.4 km upstream from the Dudh Kosi and
Bhoti Kosi confluence. The 1977 GLOF discharge at
reach N3 is less than the 1985 GLOF discharge at
® .reaches L5 and L7 Tables 2 and 4 , which are
located on the Dudh Kosi downstream from the
® .Dudh Kosi and Bhoti Kosi confluence Fig. 2 . The
difference between the 1977 GLOF discharge at
reach N3 and the 1985 GLOF discharge at reach L5
and reach L7 becomes even greater if one takes into
account that the 1977 GLOF discharge probably
attenuated downstream from reach N3. Thus, the
1985 GLOF, which was larger in magnitude than the
1977 GLOF downstream from the Dudh Kosi and
Bhoti Kosi confluence, either eroded, reworked, or
buried the evidence of the 1977 GLOF below the
Dudh Kosi and Bhoti Kosi confluence.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9088
For the older GLOFs, the estimated GLOF dis-
charges are seven to eight times greater than the
SHFFs. Specifically, at the upper Bhoti Kosi reach,
the estimated discharge of the GLOF was 400 m3
rs,
approximately seven times greater than the SHFF at
® .that location Fig. 13; Table 6 . For the upper Dudh
Kosi reach, the estimated discharge of the GLOF
was 700 m3
rs, approximately eight times greater
® .than the SHFF at that location Fig. 13; Table 6 .
The presence of the older GLOF deposits indicates
that glacial-lake outburst floods are a recurring geo-
morphic process in the Mount Everest region.
6. Conclusions
The paleoflood hydrology techniques utilized in
this study provided an effective means for assessing
the hydrology of floods in the Mount Everest region
of Nepal. Prior to this study, the flood hydrology of
the 1977 and 1985 GLOFs and SHFFs in the Mount
Everest region had not been established. A step-
backwater model was used to construct the water-
surface profiles of the GLOFs and SHFFs based on
the geomorphic evidence of the GLOF and SHFF
flood stages to estimate peak discharges. The upper
surfaces of cobble and boulder bars, scour lines, and
the lowest elevation of nonflooded surfaces were
used to delineate the flood stage of the GLOFs. A
white-gray silt-clay line along the channel margins
®produced from glacier meltwater and snowmelt
.runoff was used to define the water-surface eleva-
tion of the SHFFs. The most reliable GLOF and
SHFF peak discharge estimates were upstream from
constrictions where there was critical-depth control.
Our assumption that flow was approaching critical
flow during flow modeling provided a useful ap-
proach to reducing the uncertainty in selecting Man-
ning's n values and estimating the peak discharges
of the GLOFs and SHFFs along the high-gradient
streams in this region.
The 1977 GLOF had an estimated peak discharge
of 1900 and 1500 m3
rs at 8.6 and 11.5 km down-
stream from the breached moraine, respectively. The
peak discharge of the 1985 GLOF ranged from 1375
to 2350 m3
rs, with the greatest peak discharge
estimate occurring at the reach closest to the breached
moraine at 7.1 km downstream from the breached
® .moraine reach L2 and the lowest peak discharge
estimate occurring at the reach furthest from the
breached moraine at 27 km downstream from the
® .breached moraine reach L8 . The maximum dis-
charges of the 1977 and 1985 GLOFs were probably
larger at the breached moraines than the peak dis-
charges estimate at reaches N1 and L2. The presence
of older GLOF features along streams in other
drainage basins indicate that extreme flooding from
GLOFs is a recurring event in the Mount Everest
region. The SHFF peak discharge estimates in the
Mount Everest region ranged from 7 to 205 m3
rs
and were positively correlated with increasing
drainage area. The estimated SHFF discharges were
7 to 60 times less than the GLOF discharge estimates
with lower ratios occurring furthest from the breached
moraines because of the attenuation of the GLOFs as
they progressed downstream and the increase in the
SHFF discharges as contributing drainage area in-
creased. The uncertainty of the GLOF and SHFF
peak discharge estimates varied among the study
® .reaches and ranged from y43% to 43% Table 4
based on whether a conservative or liberal approach
was used to evaluate the PSIs. Sensitivity analyses of
the selected Manning's n and contractionrexpansion
coefficients in the step-backwater model indicate that
the Manning's n coefficient had the greatest influ-
ence on the estimated discharges, whereas the con-
tractionrexpansion coefficients had only a minimal
effect on the estimated peak discharges.
Acknowledgements
This project was funded for the most part by the
® .National Science Foundation Grant CMS-9320876 .
Additional funding was provided by the Geological
Society of America, the Department of Earth
®Resources at Colorado State University Chevron
.Research Grant , and the Lary Burns Memorial
Scholarship. Many thanks are extended to Robert D.
Jarrett for numerous comments, discussions, and in-
sights about flow hydraulics and flow modeling of
extreme floods in high-gradient streams. We ac-
knowledge the thoughtful comments provided by
David R. Butler and Richard A. Marston on an
earlier version of the manuscript.
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­90 89
References
Ageta, Y., 1976. Characteristics of precipitation during monsoon
season in Khumbu Himal. J. Jpn. Soc. Snow Ice 38, 84­88.
Arcement, G.J., Schneider, V.R., 1989. Guide for selecting man-
ning's roughness coefficients for natural channels and flood-
plains. U. S. Geol. Surv. Water-Supply Pap. 2339. United
States Government Printing Office, Washington D.C., pp.
1­38.
Bailey, J.F., Ray, H.A., 1966. Definition of stage-discharge rela-
tion in natural channels by step-backwater analysis. U. S.
Geol. Surv. Water-Supply Pap. 1869A. United States Govern-
ment Printing Office, Washington D.C., pp. 1­24.
Baker, V.R., 1978. Large-scale erosional and depositional features
of the Channeled Scabland. In: Baker, V.R., Nummedal, D.
® .Eds. , The Channeled Scabland. National Aeronautics and
Space Administration, Washington, DC, pp. 81­115.
Baker, V.R., 1984. Flood sedimentation in bedrock fluvial sys-
® .tems. In: Koster, E.H., Steel, R.J. Eds. , Sedimentology of
Gravels and Conglomerates. Can. Soc. Pet. Geol., Mem., vol.
10, Canadian Society of Petroleum Geologists, Calgary, Al-
berta, Canada, pp. 87­98.
Baker, V.R., 1987. Paleoflood hydrology and extraordinary flood
events. J. Hydrol. 96, 79­99.
Benito, G., 1997. Energy expenditure and geomorphic work of the
cataclysmic Missoula flooding in the Columbia River Gorge,
USA. Earth Surf. Processes Landforms 22, 457­472.
Brower, B., 1991. Sherpa of Khumbu: People, Livestock, and
Landscape. Oxford Univ. Press, Delhi, 202 pp.
Buchroithner, M.F., Jentsch, G., Wanivenhaus, B., 1982. Monitor-
®ing of recent geological events in the Khumbu area Himalaya,
.Nepal by digital processing of Landsat MSS data. Rock
Mech. 15, 181­197.
Carling, P.A., 1987. Hydrodynamic interpretation of a boulder
berm and associated debris-torrent deposits. Geomorphology
1, 53­67.
Cenderelli, D.A., 1998. Glacial-lake outburst floods in the Mount
Everest region of Nepal: Flow processes, flow hydraulics, and
geomorphic effects. Ph.D. Dissertation Colorado State Univer-
sity, Ft. Collins, 247 pp.
Cenderelli, D.A., Cluer, B., 1998. Depositional processes and
sediment supply in resistant-boundary channels: examples from
® .two case studies. In: Tinkler, K.J., Wohl, E.E. Eds. , Rivers
over Rock: Fluvial Processes in Bedrock Channels. American
Geophysical Union Geophysical Monograph 107, Washington,
DC, pp. 105­131.
Cenderelli, D.A., Wohl, E.E., 1998. Sedimentology and clast
orientation of deposits produced by glacial-lake outburst floods
in the Mount Everest region, Nepal. In: Kalvoda, J., Rosen-
® .field, C.L. Eds. , Geomorphological Hazards in High Moun-
tain Areas. Kluwar Academic Publishers, The Netherlands, pp.
1­26.
Cenderelli, D.A., Wohl, E.E., in review. Flow hydraulics and
geomorphic effects of glacial-lake outburst floods in the Mount
®Everest region of Nepal submitted to Earth Surf. Processes
.Landforms .
Chow, V.T., 1959. Open-Channel Hydraulics. McGraw-Hill, New
York, 677 pp.
Church, M., Jones, D., 1982. Channel bars in gravel-bed rivers.
® .In: Hey, R.D., Bathurst, J.C., Thorne, C.R. Eds. , Gravel-Bed
Rivers. Wiley, New York, pp. 291­324.
Costa, J.E., 1983. Paleohydraulic reconstruction of flash-flood
peaks from boulder deposits in the Colorado Front Range.
Geol. Soc. Am. Bull. 94, 986­1004.
Davidian, J., 1984. Computation of water-surface profiles in open
channels. U. S. Geol. Surv. Tech., Water-Resour. Invest.,
Book 3. United States Government Printing Office, Washing-
ton D.C., Chap. A15, 48 pp.
Ely, L.L., Baker, V.R., 1985. Reconstructing paleoflood hydrol-
ogy with slackwater deposits, Verde River, Arizona. Phys.
Geogr. 5, 103­126.
® .Fushimi, H., 1977. Glaciations in the Khumbu Himal 1 . J. Jpn.
Soc. Snow Ice 39, 60­67.
® .Fushimi, H., 1978. Glaciations in the Khumbu Himal 2 . J. Jpn.
Soc. Snow Ice 40, 71­77.
Fushimi, H., Ikegami, K., Higuchi, K., 1985. Nepal case study:
® .catastrophic floods. In: Young, G.J. Ed. , Techniques for
Prediction of Runoff from Glacierized Areas. Int. Assoc.
Hydrol. Sci. Publ., vol. 149, IAHS Press, Wallingford, Osford-
shire, pp. 125­130.
Grant, G.E., 1997. Critical flow constrains flow hydraulics in
mobile-bed streams: a new hypothesis. Water Resour. Res. 33,
349­358.
Grimm, M.M., Wohl, E.E., Jarrett, R.D., 1995. Coarse-sediment
distribution as evidence of an elevation limit for flash flood-
ing, Bear Creek, Colorado. Geomorphology 14, 199­210.
Hoggan, D.H., 1989. Computer-Assisted Floodplain Hydrology
and Hydraulics. McGraw-Hill, New York, 518 pp.
House, P.K., Pearthree, P.A., 1995. A geomorphic and hydrologic
evaluation of an extraordinary flood discharge estimate: Bronco
Creek, Arizona. Water Resour. Res. 31, 3059­3073.
Hydrologic Engineering Center, 1995. HEC-RAS, River Analysis
System, version 1.1, U. S. Army Corps of Engineers, Davis,
CA.
Ives, J.D., 1986. Glacial Lake Outburst Floods and Risk Engineer-
ing in the Himalaya: A Review of the Langmoche Disaster,
Khumbu Himal, 4 August 1985. International Centre for Inte-
grated Mountain Development Occasional Paper No. 5, 42 pp.
Jarrett, R.D., 1984. Hydraulics of high-gradient streams. J. Hy-
draul. Eng. 110, 1519­1539.
Jarrett, R.D., 1987. Errors in slope-area computations of peak
discharges in mountain streams. J. Hydrol. 96, 53­57.
Jarrett, R.D., 1990. Paleohydrologic techniques used to define the
spatial occurrence of floods. Geomorphology 3, 181­195.
Jarrett, R.D., in review. Reliability of paleostage indicators and
® .flood height used in paleoflood studies submitted to Science .
Jarrett, R.D., Malde, H.E., 1987. Paleodischarge of the late Pleis-
tocene Bonneville flood, Snake River, Idaho, computed from
new evidence. Geol. Soc. Am. Bull. 99, 127­134.
Marcus, M.G., Roberts, K., Harvey, L., Tackman, G., 1992. An
evaluation of methods for estimating Manning's n in small
mountain streams. Mountain Res. Dev. 12, 227­239.
Mayewski, P.A., Jeschke, P.A., 1979. Himalayan and trans-
( )D.A. Cenderelli, E.E. WohlrGeomorphology 40 2001 57­9090
Himalayan glacier fluctuations since AD 1812. Arctic Alpine
Res. 11, 267­287.
O'Connor, J.E., 1993. Hydrology, hydraulics, and geomorphology
of the Bonneville flood. Geol. Soc. Am., Spec. Pap. 274, 83
pp.
O'Connor, J.E., Baker, V.R., 1992. Magnitudes and implications
of peak discharges from glacial Lake Missoula. Geol. Soc.
Am. Bull. 104, 267­279.
O'Connor, J.E., Webb, R.H., 1988. Hydraulic modeling for paleo-
flood analysis. In: Baker, V.R., Kochel, R.C., Patton, P.C.
® .Eds. , Flood Geomorphology. Wiley, New York, pp. 393­402.
O'Connor, J.E., Webb, R.H., Baker, V.R., 1986. Paleohydrology
of pool-and-riffle pattern development: Boulder Creek, Utah.
Geol. Soc. Am. Bull. 97, 410­420.
Rathburn, S.L., 1993. Pleistocene cataclysmic flooding along the
Big Lost River, east central Idaho. Geomorphology 8, 305­
319.
Stewart, J.E., LaMarche, V.C., 1967. Erosion and deposition
produced by the flood of December 1964 on Coffee Creek
Trinity County, California. U. S. Geol. Surv. Prof. Pap. 422K.
United States Government Printing Office, Washington D.C.,
pp. 1­22.
Tinkler, K., 1996. Critical flow in rockbed streams with estimated
values of Manning's n. Geomorphology 20, 147­164.
Trieste, D.J., 1992. Evaluation of supercriticalrsubcritical flows
in high-gradient channel. J. Hydraul. Eng. 118, 1107­1118.
Trieste, D.J., Jarrett, R.D., 1987. Roughness coefficients of large
® .floods. In: James, L.G., English, M.J. Eds. , Irrigation and
Drainage Division Specialty Conference, Irrigation Systems
for the 21st Century, Portland, Oregon, Proceedings. Am. Soc.
Civ. Eng., New York, pp. 32­40.
Vuichard, D., 1986. Geological and petrographical investigations
for the mountain hazards project, Khumbu Himal, Nepal.
Mountain Res. Dev. 6, 41­52.
Vuichard, D., Zimmermann, M., 1986. The Langmoche flash-
flood, Khumbu Himal, Nepal. Mountain Res. Dev. 6, 90­94.
Vuichard, D., Zimmermann, M., 1987. The 1985 catastrophic
drainage of a moraine-dammed lake, Khumbu Himal, Nepal:
Cause and consequences. Mountain Res. Dev. 7, 91­110.
Waythomas, C.F., Walder, J.S., McGimsey, R.G., Neal, C.A.,
1996. A catastrophic flood caused by drainage of a caldera
lake at Aniakchak Volcano, Alaska, and implications for
volcanic hazards assessment. Geol. Soc. Am. Bull. 108, 861­
871.
Webb, R.H., O'Connor, J.E., Baker, V.R., 1988. Paleohydrologic
reconstruction of flood frequency on the Escalante River,
south-central Utah. In: Baker, V.R., Kochel, R.C., Patton, P.C.
® .Eds. , Flood Geomorphology. Wiley, New York, pp. 403­418.
Wohl, E.E., 1992a. Bedrock benches and boulder bars: Floods in
the Burdekin Gorge of Australia. Geol. Soc. Am. Bull. 104,
770­778.
Wohl, E.E., 1992b. Gradient irregularity in the Herbert Gorge of
northeastern Australia. Earth Surf. Processes Landforms 17,
69­84.
Wohl, E.E., 1995. Estimating flood magnitude in ungauged moun-
tain channels, Nepal. Mountain Res. Dev. 15, 69­76.
Wohl, E.E., 1998. Uncertainty in flood estimates associated with
roughness coefficient. J. Hydraul. Eng. 124, 219­223.
Zimmermann, M., Bichsel, M., Kienholz, H., 1986. Mountain
hazards mapping in the Khumbu Himal, Nepal, with Prototype
Map, scale 1:50,000. Mountain Res. Dev. 6, 29­40.