Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259LOC.3D x [10–12] 26.9.2009 6:13PM
Contributors
Irasema Alcántara-AyalaInstituto de Geografía, UNAMCircuito Exterior, Ciudad Universitaria04510, Coyoacán, México, D.F.México
Gilles Arnaud-FassettaUniversité Paris-DiderotCase Postale 700175205 Paris Cedex 13France
Perry BarteltAvalanches, Debris Flows and Rockfall Research UnitWSL Institute for Snow and Avalanche Research SLFFlüelastrasse 11CH-7260 Davos DorfSwitzerland
Gerardo BenitoLaboratorio de Geomorfología e HidrologíaCentro de Ciencias Medioambientales, CSICSerrano 115 dup.28006, MadridSpain
John BoardmanEnvironmental Change Institute,SoGE, South Parks RoadOxford, OX1 3QYUK
Lisa BorgattiDipartimento di Ingegneria delle Strutture, dei Trasporti,delle Acque, del Rilevamento, del TerritorioAlma Mater Studiorum Università di BolognaViale Risorgimento, 241036 BolognaItaly
William B. Bull6550 N. Camino KatrinaTucson, AZ, 85718-2022USA
Michael BründlWarning and Prevention Research UnitWSL Institute for Snow and Avalanche Research SLFFlüelastrasse 11CH-7260 Davos DorfSwitzerland
Etienne CossartUMR Prodig 8586 – CNRSUniversité Panthéon-Sorbonne (Paris 1)2 rue Valette,75005 ParisFrance
Michael CrozierVictoria University of WellingtonSchool of Geography, Environment and Earth SciencesPO Box 600WellingtonNew Zealand
Monique FortUniversité Paris Diderot (Paris7)Département de GéographieUFR GHSS, Case 700175205 Paris Cedex 13France
Thomas GladeDepartment of Geography and Regional ResearchUniversity of ViennaUniversitaetsstr. 7A-1010 ViennaAustria
Andrew S. GoudieSt Cross CollegeSt GilesOxford, OX1 3LZUK
x
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259LOC.3D xi [10–12] 26.9.2009 6:13PM
Avijit GuptaSchool of GeographyUniversity of LeedsLeeds, LS2 9JTUK
Francisco GutiérrezUniversidad de ZaragozaDepartamento de Ciencias de la TierraEdificio GeológicasC/. Pedro Cerbuna, 1250009 ZaragozaSpain
David HiggittNational University of SingaporeDepartment of GeographyAS2, #03-011 Arts Link, Kent Ridge117570 Singapore
Paul F. HudsonDepartment of Geography and the EnvironmentUniversity of Texas at AustinAustin, TX 78712USA
Gabi HufschmidtSchool of GeographyVictoria University of WellingtonP.O. Box 600WellingtonNew Zealand
Margreth KeilerDepartment of Geography and Regional ResearchUniversity of ViennaUniversitaetsstr. 7A-1010 ViennaAustria
Molly McGrawSoutheastern Louisiana UniversitySLU-10686Hammond, LA 70402USA
David PetleyDurham UniversityDepartment of Geography Science LaboratoriesSouth RoadDurham, DH1 3LEUK
Jürg SchweizerWarning and Prevention Research UnitWSL Institute for Snow and Avalanche Research SLFFlüelastrasse 11CH-7260 Davos DorfSwitzerland
Olav SlaymakerUniversity of British ColumbiaDepartment of Geography1984 West MallVancouver, V6T 1Z2Canada
Mauro SoldatiDipartimento di Scienze della TerraUniversità di Modena e Reggio EmiliaLargo S. Eufemia, 1941000 ModenaItaly
Jean-Claude ThouretLaboratoire Magmas et Volcans UMR 6524 CNRS etOPGCUniversité Blaise Pascal Clermont II5 rue Kessler63038 Clermont-Ferrand cedexFrance
Cees J. van WestenInternational Institute for Geo-Information Science andEarth Observation (ITC)Hengelosestraat 99PO Box 67500 AA EnschedeThe Netherlands
Heather VilesSchool of Geography and the EnvironmentSouth Parks RoadOxford, OX1 3QYUK
Harley J. WalkerDepartment of Geography and AnthropologyLouisiana State UniversityBaton Rouge, LA 70803USA
Yang XiaopingInstitute of Geology and GeophysicsChinese Academy of SciencesP.O. Box 9825Beijing 100029China
xiContributors
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 75 [75–86] 26.9.2009 5:35PM
7 Catastrophic landslides and sedimentarybudgets
Monique Fort, Etienne Cossart, and Gilles Arnaud-Fassetta
Landslides are a dominant geomorphic process affectingmountain slopes worldwide (see also Chapters 6 and 8).They represent a major sediment source that can supply alarge amount of unstable debris to river channels and mayaffect the fluvial sediment yield. A distinctive part of thegeomorphic evolution of active mountain belts, catastrophiclandslides generally develop very rapidly so that they areamong the most powerful natural hazards on Earth. By tem-porarily or persistently impounding river channels, they delayor block the delivery of sediments, affect the (dis)continuity ofthe cachment-scale sediment cascade, and exert a control overthe fluvial valley systems (Hewitt, 2002; Korup et al., 2004).As a consequence, landslides may generate indirect hazardsalong fluvial systems, so that they represent a major threat tosettlements, infrastructures and catchment management thatmay be felt over long distances from the unstable area (Plafkerand Eriksen, 1978; Li et al., 1986; Costa, 1991; Korup, 2005;Hewitt et al., 2008). Their magnitudes, together with theircausative factors, suggest the difficulty, if not theimpossibility, of preventing and/or to predicting them.In this brief overview, we shall firstly define catastrophic
landslides and their geomorphic impacts, with specialattention given to landslide-induced dams; we shall moveon to their influence on sediment budgets, at local andbasin-wide scales, before eventually considering a fewactions that may help to minimize the vulnerability andrisks for the potentially affected populations.
7.1 Catastrophic landslides: definition,modes of emplacement andgeomorphic significance
7.1.1 Definition
The term landslide covers a large array of features.Catastrophic landslides (also referred to as super-large or
giant landslides in the literature) are considered here as lowfrequency, massive rock slope failures characterized bytheir magnitude, the rapidity of their emplacement and bytheir spatial and temporal impacts on geomorphic systems.More specifically, the magnitude of catastrophic landslidesrefers either to the depth of the sliding place, to the scar areaor the total disturbed area, including the deposition zone, orto their volume (Table 7.1). The displaced material gener-ally covers at least five orders of magnitude between 105
and 1010 m3 (Evans et al., 2006; Hewitt et al., 2008).Determination of landslide volumes can be assessed
directly from field data. Alternatively, at a basin-widescale, digital elevation models (DEMs) are useful tools butthey require good positional accuracy (Korup, 2005). Aerialphotomeasurements (basedmostly on the total affected area)allow the production of an exhaustive inventory, and timeseries analysis, yet their use necessitates photogrammetricmodelling and ground truthing; also, the fast recovery by thevegetation in some (tropical) mountains may lead to anunderestimate of the final area/volume (Brardinoni andChurch, 2004; Koi et al., 2008) and introduce biases inmagnitude/frequency estimation. Despite this limitation, anumber of analyses of landslide magnitude/frequency rela-tions have shown they are scale invariant and they obey apower law of the general form N(A) ≈ A−b, where A islandslide area, N(A) the number of events of greater than agiven volume, and b is a constant (Hovius et al., 2007). Thisequation is used to quantify the distribution of landslides inspace and time, and hence their density and recurrence.
7.1.2 Occurrence and modes of emplacementof catastrophic landslides
Landslide occurrence depends on various controls: climate,slope steepness, relief amplitude, bedrock geology, failureplane orientation, vegetation and landuse cover, etc.
Geomorphological Hazards and Disaster Prevention, eds. Irasema Alcántara-Ayala and Andrew S. Goudie. Published byCambridge University Press. © Cambridge University Press 2010. 75
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 76 [75–86] 26.9.2009 5:35PM
TABL
E7.1Selected
catastrophiclandslidesofthetwentiethcenturyandtheirimpacts
Year
Location
Landslid
evo
lume(m
3)Trigg
erIm
pacts
Com
ments
1911
Usoirockslid
e(Tadjik
istan)
2.2×10
9M
7.4Pam
irearthq
uake
54killed;
LakeSarez,the
largestn
atural
land
slidedam
intheworld
(>60
kmlong
,>50
0m
depth,
volume17
×10
9m
3)across
theMurgabRiver
Con
cernsabou
tapo
ssiblefailu
reof
thedam,
thatwou
ldaffect>5millionpeop
lealon
gthe
Bartang
–Pyanj–AmuDariawatershed
1933
Diexi,S
ichu
anprov
ince
(China)
150×10
6M
7.5earthq
uake
>2,40
0killed;
dam
oftheMin
River
for45
days
(lakevo
lume:40
0×10
6m
3)
Drainingof
thelake
byov
erspilling;
thefloo
dprop
agated
>25
0km
downstream
(average
velocity
5.5–
7m/s)
1959
Malpasset,F
réjus
(France)
30×10
6Heavy
precipitatio
n42
3killed;
railw
aydamaged
alon
g2.5km
;50
farm
sdestroyed
Inadequateartificialdam,builtagainstw
eaken
bedrock(foliatedandtecton
ized
schists+
non-identified
faults)
1963
Vaion
trockslid
e(Italy)
270×10
6Precipitatio
n+high
water
levelfollowed
byrapidlake
draw
down
2000
killed;
rockslidefailedin
the15
0×10
6
m3Vaion
treservo
ir,resultin
gin
acatastroph
icfloo
d(city
ofLangarone
badly
damaged,tog
etherwith
4othervillages)
A10
0m
high
waveof
water
overtopp
edthe
doub
lycurved
arch
ofhy
droelectricpo
wer
dam;p
ossiblereactiv
ationof
arelict
land
slide;brittlefailu
reof
claysin
depth
1967
Tangg
udon
gdebrisslide,
Sichu
an(China)
68×10
6Dam
oftheYalon
gRiver;lakevo
lume60
0×
106m
3Catastrop
hicfailu
reof
thedam
1970
NevadoHuascaran
rock
debris/avalanche
(Peru)
30–50
×10
6M
7.7earthq
uake
>18
,000
killed;
city
ofYun
gaydestroyed
Average
velocity28
0km
/hr;samepeak
already
affected
bythesametype
offailu
rein
1962
(4,000
–5,00
0killed)
1980
Mou
ntStH
elensdebris
avalanche(U
SA)
2.8×10
9Volcanicerup
tion
5–10
killed,
mostp
eopleevacuated;
destructionof
infrastructure;SpiritL
ake
(258
×10
9m
3)dammed
Rotationalrockslide,follo
wed
by23
kmlong
debrisavalanche;averagevelocity
125km
/hr;d
amartificially
stabilizedby
anou
tlet
tunn
el19
83Thistledebrisslide
(USA)
21×10
6Sno
wmeltand
heavyrain
Destructio
nof
infrastructure;d
amof
the
Spanish
ForkRiver;lakevo
lume78
×10
6
m3
Heavy
econ
omiclosses
($US60
0million);
lake
perm
anently
drainedby
human
interventio
n19
83SaleMou
ntainland
slide,
Gansu
Province
(China)
30×10
6Noob
viou
strigger
237killed,
3villagesdestroyed;
toeof
the
displacedmassacross
the80
0mwidevalley
oftheBaxieriver
Mud
ston
eandloessmaterial;compo
site
land
slide,with
prog
ressivefailu
reand
increasing
velocity
upto
20m/s
1984
Mou
ntOntake(Japan)
36×10
6M
6.8earthq
uake
Traveld
istanceof
13km
downto
adjacent
valleys
Failure
ofthesoutheastern
flankof
MtO
ntake
volcano,
transformed
into
alahar(estim
ated
volume56
×10
6m
3);velocity
22–36
m/s
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 77 [75–86] 26.9.2009 5:35PM
1985
Bairaman
debris
avalanche(Papua
New
Guinea)
180×10
6M
7.1earthq
uake
210m
high
land
slidedam;lake(volum
e50
×10
6m
3)artificially
drained;
villagers
evacuated
Debrisflow
volume12
0×10
6m
3;average
velocity
20km
/hr;affected
39km
Bairaman
River
downto
Salom
onSea
1985
Catastrop
hiclahar,
NevadodelR
uiz
(Colum
bia)
90×10
6Sno
wmeltlahar
triggered
byvo
lcanicerup
tion
>23
,000
killed(buriedcity
ofArm
ero,
40km
distant);smalleruption(<5×10
6m
3magma
ejected),butlargeam
ountof
meltwater(38–
44×10
6m
3)released
in20–90
min
Lahar
velocity
intheLagun
illas
valley,50
to80
km/hr;peak
flow
velocities5–15
m/s;
laharreached10
4km
in4ho
urs;econ
omic
losses
($US1billion
);form
erdestructive
laharin
1845
,with
sameareasaffected
1987
ValPolaland
slide(Italy)
40×10
6Cum
ulativeand
unusually
severe
rainfall
27killedanddestructionof
manybu
ildings
($US40
0million);d
ammingof
theAdd
aRiver
valley
The
rock-valancheproduced
awaveof
muddy
water
2.7km
upstream
;velocity
248–31
0km
/h;spillw
aytunn
elconstructed
1991
Randa
rockfall
(Switzerland
)20
×10
6Meltin
gof
perm
afrost?
Dam
oftheVispa
River,fl
ooding
oftheRanda
village
Con
tinuo
usrockfalllastingseveralh
ours,w
ithexpo
nentialaccelerationbefore
final
mov
ement
1993
LaJoseph
inaland
slide
(Ecuador)
20–44
×10
6Rainfall
Dam
ofthePauteRiver;lakevo
lume17
7×10
6
m3;d
rained
outcatastrop
hically
;71killed,
mostp
eopleevacuatedbefore
lake
breako
ut
Flood
prop
agated
>60
kmdo
wnstream;p
eak
discharge:9,00
0–14
,000
m3/s;1
3×10
9m
3
solid
load;v
elocity
5–20
m/s;h
eavy
econ
omiclosses
($US14
7millions)
1999
Malpa
rockfallanddebris
flow
(Ind
ia)
1×10
6Highintensity
rainfall
221killed;
Malpa
River
dammed,followed
byan
outburstfloo
dFormerland
slides
atthesamesite;proximity
oftheMainHim
alayan
ThrustF
ault
1999
MtA
dams,Sou
thWestland
(New
Zealand
)
10–15
×10
6Noob
viou
strigger
Dam
form
edin
thelowerPoeruaRivergo
rges;
massive
fanh
eadaggradationatthe
mou
ntainrang
efron
t
Outbu
rstfl
oodwith
inlessthan
oneweek(peak
discharge1,00
0–3,00
0m
3/s);persistent
postfailu
reaggradation
2000
Yigon
gdebris
slideanddebrisflow
,TibetProvince(China)
300×10
6Excessive
rainfalland
snow
-meltw
aters
Landslid
edam
oftheYigon
gRiver;n
atural
breach
caused
catastroph
icfloo
d(peak
discharge12
0,00
0m
3/s);loss
ofprop
erty
andinfrastructure
downstream
toIndia
Failure
very
similarto
the1900
Yigong,
large
(5.1×10
8)land
slideandresulting
dam
and
lake
2007
Guinsaugo
nrockslide–
debrisavalanche,
LeyteIsland
,Philip
pines
15×10
9Nospecifictriggerbu
theavyrainfallin
the
precedingdays
Disintegrationof
sliddenmass;average
thickn
essof
debris10
m;run
outd
istance
3,80
0m;>
1,00
0casualtieswhen
Guinsaugo
nvillage,located
onflatvalley
floo
r,was
buried
under4–7m
deep
debris.
Failure
of45
0m
high
,forestedrock
slop
elocatedwith
inactiv
ePhilip
pine
faultzon
e(sheared
andbrecciated
rocks);run
out
distance
enhanced
byfrictio
nredu
ction
whendebrisranov
erfloo
dedpadd
yfields;
estim
ated
andsimulated
meanvelocity
27–
38m/sand35
m/srespectiv
ely.
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 78 [75–86] 26.9.2009 5:35PM
(Cruden and Varnes, 1996; Dikau et al., 1996). The world-wide distribution of catastrophic landslides shows a goodcorrelation with mean local relief greater than 1000m(which makes up over 5% of Earth’s land surface; Korupet al., 2007), with seismo-tectonically active mountains(e.g. Central and High Asia, New Zealand Alps, CoastalRanges) or volcanic belts (Pacific Rim), and with recentlydeglaciated mountain slopes (Evans and Clague, 1994).A large number of catastrophic rock failures consist of
rock avalanches (or sturzstroms; Hsü 1975) and/or rock-slides, even if other processes such as translational rock-slides, rotational slumps or spreads cannot be excluded.The mode of emplacement of rock avalanches can be verycomplex: it generally involves an initial failure from steepmountain walls of cohesive rock mass that descends severalhundreds or thousands of metres and is disintegrated andcrushed, a process that gives great momentum to the result-ing debris (Hewitt, 2002). The nature and extent of theresulting deposits depend on the geological setting andvalley morphology (Costa and Schuster, 1988). In topo-graphically confined settings, the deposits may pile up onand run up the opposing slope, or split into separate lobesboth upstream and downstream of the impacted slope, allsituations that favour the blockage of the valley and theformation of a lake. Alternatively, wherever there is nospatial confinement, the crushed rocks may travel longdistances and evolve, in the presence of water (or snow,or ice), into giant debris flows (Plafker and Eriksen, 1978;Fort, 1987; Shang et al., 2003). These failures may alsoincorporate a variety of earth materials entrained in theirpath, and the runout process may affect, or be affected by,the type of substrate (Hewitt, 2002; Schneider et al., 2004).Earthquakes and/or high intensity precipitation are the
most efficient triggering factors for rapid collapse. Thefailure site may be controlled by major tectonic lines(Figure 7.1; Fort, 2000), or be considered as a direct adjust-ment to high rates of rock uplift and correlated river inci-sion (Burbank et al., 1996). The failure may be preceded bya phase during which slow, deep-seated deformation leadsto the opening of tension cracks, weakening of the shearstrength of the rockmass, and eventually to high pore-waterpressures. Other unusual mechanisms such as acoustic flu-idization during sliding, or internal, self-accelerating rockfracture (Kilburn and Petley, 2003) may also favour large-scale rock slope failures (Hewitt et al., 2008). Additionally,anthropogenic actions may be the cause of large hillslopedestabilization (e.g. Vaiont reservoir; Table 7.1).These very large rock mass failures are generally con-
sidered as a major denudational process of active orogens,and as formative events influencing landscape development(Brunsden and Jones, 1984; Fort, 1988, 2000; Hewitt,
1988, 1998, 2002; Fort and Peulvast, 1995; Burbanket al., 1996; Korup et al., 2004, 2007). More specifically,they give rise to complex sedimentary assemblages and tospecific constructional and erosional landforms that create‘interrupted valley landsystems’ distinctive of theselarge-scale landslides (Hewitt, 2006).
7.2 Geomorphic impacts of catastrophiclandslides
Catastrophic landslides have direct geomorphic impacts atthe local, basin-wide and mountain-belt scales, as synthe-sized by Costa and Schuster (1988), Hewitt (2002, 2006)and Korup (2005). We want to stress here local and regionalimpacts, which are the most meaningful in terms of hazardsand risks for the population. Geomorphic impacts and theirconsequences on water and sediment fluxes and budgetswill be appreciated in considering different interactionsbetween landslides and fluvial systems.
7.2.1 Interactions with river systems
Three different situations are illustrated: partial blockage ofthe valley by the landslide, complete damming andupstream water ponding, and catastrophic collapse of thelandslide dam (Figure 7.2).In the case of partial blockage (Figure 7.2, case 1), the
landslide mass forces the river to divert its course to theopposite bank; this in turn leads to a change in transverseand longitudinal channel geometry, channel pattern andmorphology. Wherever the opposite bank consists of softmaterial (slope or alluvial deposits), this diversion triggersbank erosion and further destabilizes the entire hillslope(Figure 7.2B). Upstream of the landslide mass, which actsas the local base level, braiding of the river and aggradationpredominate. These may indirectly increase overflow andchannel avulsion-shifting frequencies, and flood hazardsfor the adjacent settlements (Figure 7.2A). Across the land-slide mass, the narrower cross section of the river favourserosion of the landslide debris and increases sedimentfluxes downstream, whereas the larger blocks exceedingthe competence of the stream power form a debris lag,which armours the channel bed and prevents further erosionduring regular high flows.When the volume of the landslide is sufficient to block
the valley entirely (Figure 7.2, case 2 and Figure 7.2D), alake will instantaneously start filling up, whereas the down-stream part is starved of water. Lake depth depends onvalley geometry and landslide height; the latter may reachhundreds of metres (e.g. Lake Sarez; Table 7.1). Landslidedams may be ephemeral (a few minutes to a few hours), or
Monique Fort et al.78
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 79 [75–86] 26.9.2009 5:35PM
FIGURE 7.1. The large, prehistoric Dhumpu–Kalopani rock avalanche and its morpho-sedimentary impacts (Nepal Himalayas). Thecollapsedmass failed along theNorth HimalayanDetachment Fault (NHDF) and blocked durably the Kali Gandaki River, which had carvedthe deepest gorges in the world across the Annapurna (8091m) and Dhaulagiri (8172m) ranges. The 23 km long, >200m deep Marphalake developed and filled in with sediments brought from upstream and from glaciated tributary valleys. The braided pattern of the rivercourse reflects the still persisting role of the rock-avalanche barrier in the denudational evolution of the mountain.
79Catastrophic landslides and sedimentary budgets
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 80 [75–86] 26.9.2009 5:35PM
short-term (a few weeks or months), or persistent (severalthousand years), a duration that directly influences theirpotentially catastrophic nature and their control on sedi-ment fluxes and budgets (Figure 7.1 and Figure 7.2E). Damlongevity is a function of its stability, which depends on thesize and shape of the dam, the characteristics of the geo-logical material composing the dam (material properties,grain size distribution), the volume and rate of water andsediment inflow to the newly formed lake, and the rate ofseepage through the dam (Costa and Schuster, 1988).Water ponding occurs up to the dam height, resulting inupstream backwater flooding that can be damaging forsettlements and infrastructures (Figure 7.2C). The rapidityof the water level rise depends on the river inflow versusthe size and shape of the inundated valley; it generallyleaves enough time for evacuation of the threatenedpopulations to take place.In contrast, catastrophic downstream flooding
(Figure 7.2, case 3) will occur following a rapid failure ofthe landslide dam, caused either by overtopping and imme-diate retrogressive incision across the dam (Figure 7.2F), orby increasing shear stress because of seepage at the base ofthe dam (that can be reinforced by the weight of theimpounded water mass upstream), or by displacementwaves triggered by landslide failures directly into thelake. The catastrophic nature of the (flash)flood, involvingthe release of a large amount of sediments and water, isrecorded by extreme peak discharges (usually back-calculated from field surveys and hydraulic equations)and high velocities that permit the transport of sediments
derived from the dam and from the upstream lacustrinereservoir. The geometry and extent of the coarse debriswedge formed immediately downstream of the dam(Figure 7.2G) suggest both the competence of the debrisladen waters (often in the form of debris flows) and theirrapid evolution in both time and space into hyper-saturatedflows and then into ‘normal’ high flows, in relation toprogressive deposition of the coarser material in the floodplain (Figure 7.2H).The catastrophic draining of a landslide-dammed lake
and associated flooding may create secondary impacts.Firstly, new landslides may develop, both upstream of thelake, following the rapid drawdown of the water, and acrossand downstream of the dam, where the propagation of theflood contributes over very large distances to a sudden riseof water, bank undercutting and hillslope instabilities(e.g. Yigong debris flow; Table 7.1), which may in turngive rise to new dams and flooded areas. Secondly, specificaggradational landforms may develop, such as ‘barrier-defended’ terraces upstream of the dam, with the sedimen-tary facies reflecting the architecture and sequences ofdebris inputs from the trunk river and adjacent tributariesand slopes (Hewitt, 2002). Thirdly, a distinctive morphol-ogy of ‘interrupted valleys’ appears as a legacy of pastcatastrophic events, consisting of partial obstructions bylandslide barriers (e.g. boulder accumulations resting inthe middle of river channels) or in persistent impoundmentsthat ‘create a chronically fragmented drainage system’
(Hewitt, 2006). This is the case for the Dhampu–Chooya–Kalapani rock avalanche that dammed the upper Kali
FIGURE 7.2. Types of geomorphic impacts of landslide dams on hydrosystems.
Monique Fort et al.80
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 81 [75–86] 26.9.2009 5:35PM
Gandaki valley (Nepal Himalaya) some 70,000 years agoand is still influencing the bed profile and planform pattern(braiding) of the river channel (Fort, 2000) (Figure 7.1).This influence is related not only to the stability of the dam,but also depends on the stream power of the river beingsufficient to breach the dam. This is a function of thelongitudinal profile, and of the liquid and solid dischargeof the river (parameters controlled by climate, autocyclicprocesses, slope and tectonics).
7.2.2 Impacts on sedimentary fluxesand budgets
In the frame of sedimentary budgets, landslide masses forma particular type of sediment store, characterized by theirvolume and calibre of materials, and by their distance fromthe channel (Slaymaker, 2006). As already mentioned,large landslides impact adjacent channels and valleys,either by accelerating, or slowing down, or even interrupt-ing, the conveyance of sediment to the downstream reser-voirs. These different impacts may occur successively intime, depending on the magnitude of the landslides andtheir capacity to act or not as an efficient barrier; they mayrandomly affect the functioning of the sediment fluxes, andcontribute to the formation of a cascading sequence ofintramontane storage units (Korup, 2005).
Local scaleAn illustration is provided by the sedimentary budget ofthe upper Cerveyrette catchment (Southern French Alps).This was directly influenced by a series of earth-flows thatdeveloped after glacial retreat, and was caused by thecombined effects of post-glacial debuttressing and lique-faction of the serpentinite bedrock, in a context of seismicactivity (Cossart and Fort, 2008) (Figure 7.3A). The larg-est slide-earth flow of the Chenaillet (c. 3 × 107m3; 3.5 kmlong) efficiently dammed the valley, so that the BourgetPlain developed in response to the impoundment.Longitudinally, the trap was infilled with alluvial gravelsand silts fed by alpine periglacial and/or still-glaciatedhillslopes. This debris accumulated as a prograding fandelta encroaching upon the lake that developed in thedistal part of the valley, closest to the dam. Lacustrinedeposits are interfingered with colluvial debris derivedfrom the steep slopes of Mount Lasseron as scree andavalanche cones.The sediment budget was estimated by combined field
surveys, DEM and GIS approaches (Figure 7.3B; Cossartand Fort, 2008). The sediment stored in the Bourget Plainrepresents approximately the same volume (22.3 × 106m3)as the Chenaillet landslide dam (21 × 106m3), whereas
debris still blankets the hillslopes (18.4 × 106m3) withoutreaching the Bourget Plain. This store developed betweenthe Late Pleistocene and c. 5,000 BP; it first interrupted,then considerably reduced the sediment fluxes downstream.As soon as the blockage of the valley occurred, the down-stream part of the Cerveyrette torrential river started adjust-ing its longitudinal profile by retrogressive erosion. Thepresent situation corresponds to a total sediment removaland export of only 106m3 (c. 1/60 of the debris). However,sporadic dissection of the landslide dam triggered byextreme meteorological events is directly threatening theCervières village, as during the 1957 >100-years recurrenceflood that destabilized the entire hydrosystem. This exam-ple shows how, in alpine headwater contexts, landslidedams are persistent features controlling the sediment fluxeslong after their occurrence.
Regional scaleLarge landslides may contribute to a significant increasein sediment yield in a very short period following failure,not only by direct massive input of landslide debris intothe fluvial system, but also by secondary hillslope pro-cesses that rework the landslide material before vegetationregrowth. Such landslide-derived sediment pulses resultin aggradation in the downstream reaches from the land-slide site, and are often accompanied by metamorphosisand/or change in the river course that may induce off-sitehazards and damaging impacts to downstream settlementsand infrastructures. For instance, Korup et al. (2004)calculated that the 1999 Mount Adams rock avalanche(New Zealand) produced a specific sediment yield inexcess of approximately 75,700 ± 4,600 t km−2 a−1, asexpressed by the massive fanhead aggradation and theopening of a major avulsion channel at the mountainrange front. This sediment yield represents a sedimentdischarge of 2.5 × 106 m3 a−1 calculated for the threeyears following failure, an amount that rapidly declinedafter the event. In the Nepal Himalayas, Fort (1987)described a giant (>4 × 109 m3) collapse of the southface of the Annapurna IV peak that occurred about 500years ago, and filled the 35 km distant Pokhara valleyunder a 60 to 100m thick gravel aggradation, whichblocked and caused the flooding of adjacent tributaryvalleys. The calculated annual contribution ofAnnapurna rockslide-derived sediment is in the order of4 × 106 m3 a−1 and represents, for the upper catchment, asediment yield of 22,860m3 a−1 km−2 (Fort, 1987; Fortand Peulvast, 1995). This is a figure averaged over a500-year period, which in fact is exceptionally high ifone considers the fact that the aggradation took place ina very short time, i.e. ‘instantaneously’ after the failure.
81Catastrophic landslides and sedimentary budgets
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 82 [75–86] 26.9.2009 5:35PM
7.3 Forecasting and preventing
Catastrophic rock failures are generally preceded by peri-ods of accelerated creep, which may result in observableslope deformation and ‘sackung’ related features: develop-ment and widening of tension cracks, buckling, increasedrockfall activity and toppling, or break-out across bedding.Measured stress drops in crustal rock, and related fracturing
could provide a physical basis for quantitatively forecastingcatastrophic slope failure (Kilburn and Petley, 2003).However, monitoring of all potential threatening slopes isnot realistic, especially in developing countries where otherpriorities (food, shelter, employment) come well ahead ofunpredictable natural hazards.Other forecasting methods are based on statistical
assessment of landslide susceptibility; they rely on a series
FIGURE 7.3. Sedimentary budget assessment in response to the Chenaillet earthflow dam (French Southern Alps). (Modified andcompleted after Cossart and Fort (2008.)
Monique Fort et al.82
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 83 [75–86] 26.9.2009 5:35PM
of variables (landslide inventory with the help of GIS,terrain predisposing factors, neo-predictive variableswith a geomorphological meaning) that are tested (simu-lations) and eventually evaluated by expert judgement(Thiery et al., 2007). Another similar approach is basedon geotechnical modelling and 3-D model calibration setup from a known and well-studied event (quantitativedata), as was done after the 1987 Val Pola landslide, oneof the most destructive and costly natural disasters that hasoccurred in Italy during recent decades (Costa, 1991;Crosta et al., 2004). In their study, Crosta et al. (2004)showed how modelling techniques help one to understandthe rheology of such a failure, and to predict its timing inhighlighting the transformation of potential energy tokinematic energy.The power law curve of landslide magnitude/frequency
can also be used for landslide hazard assessment (Evanset al., 2006), and as input into a quantitative risk calcu-lation if combined with vulnerability data, as was done toassess rockfall risk along transportation corridors (Hungret al., 1999).To prevent the failure of landslide dams and resulting
catastrophic floods, the most commonly used controlmeasure consists of the construction of spillways eitheracross the landslide crest or across the adjacent bedrock.Alternatively, drainage by siphon pipes, pump systems ordiversion tunnels (Val Pola, Mount St Helens, Bairaman;Table 7.1) are short-term measures to control lake level.More radical methods consist in large-scale blasting, aswas done across the landslide dams that were formed nearBeichuan city after the 18 May 2008 Sichuan earthquakein China. However, the disaster may occur beforeadequate control measures can be completed (more par-ticularly in remote and rugged areas that render transportof heavy equipment very difficult), or because high rapidinflow to the impoundment often exceeds general predic-tions and causes an early dam failure (Shang et al., 2003).In the case of Lake Sarez and the related Usoi landslidedam (Tadjikistan), the highest dam on Earth, and despiteextensive observations and technical studies that suggest asatisfactory stability of the dam against sliding, the possi-bility of a catastrophic outburst flood that would destroythe many villages and infrastructures of the Bartang–Pyang–Amu Darya catchment, situated between the lakeand the Aral Sea, cannot be entirely ruled out (>5 millionpeople would be affected). Combined measures are thusbeing implemented, consisting of (i) the monitoring of thestability of the dam and slopes surrounding the lake, (ii) anearly warning system to alert inhabitants of the upper AmuDarya valley, together with (iii) the modelling of a seriesof flood scenarios to determine the degree of risk and
vulnerability of downstream villages and infrastructures(Alford and Schuster, 2000). Whatever these measures, asimple doubling of the present mean streamflow volumeof 2000m3 s−1 of the Bartang River would readily destroylarge portions of existing roads, low-lying villages andagricultural land for more than 100 km downstream fromthe dam, whereas a lake outburst flood would generate aninstantaneous, catastrophic peak flow of about one millioncubic metres per second (Alford and Schuster, 2000).
7.4 Conclusions
Despite their low frequency, large landslides are naturalhazards that induce geomorphic impacts that can badlydamage human settlements (Table 7.1). They are generallyassociated with episodes of extreme rainfall and/or withearthquakes, which are the natural hazards that cause thehighest human losses on Earth. The rapid failure of a land-slide dam will cause catastrophic downstream floodingwhereas a long-lasting dam and its resulting filling by sedi-ments will mostly affect mountain valley morphology andthe sediment cascade.The development of large-scale, catastrophic landslides
should be put in the broader, long-term perspective of theevolution of a mountain range. In the context of the overallbalance between erosion and rock uplift rates (Burbanket al., 1996; Montgomery, 2001), superficial variablessuch as sediment fluxes and hillslope angle respond toboth climate forcing and sporadic, catastrophic mountainslope collapse. This results in punctuated epicycles ofaggradation and/or river incision that offset the equilibriumstate at a shorter, 104–105 year time-scale (Fort, 1988; Pratt-Sitaula et al., 2004). These high magnitude, low frequencyfailures are the very ‘formative events’ of the present mor-phology of active mountains and accomplish the main partof the denudation process of these orogens.
References
Alford, D. and Schuster, R. L. (2000). Introduction and sum-mary. In Usoi Landslide Dam and Lake Sarez: An Assessmentof Hazard and Risk in the Pamir Mountains, Tajikistan. UnitedNations, ISDR Prevention Series n°1, 1–18.
Brardinoni, F. and Church, M. (2004). Representing the land-slide magnitude-frequency relation: Capilano river basin,British Columbia. Earth Surface Processes and Landforms,29, 115–124.
Brunsden, D. and Jones, D. K. C. (1984). The geomorphology ofhigh magnitude-low frequency events in the Karakorammountains. In K. J. Miller (ed.), The InternationalKarakoram Project, vol. 1, Cambridge: CambridgeUniversity Press, pp.383–388.
83Catastrophic landslides and sedimentary budgets
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 84 [75–86] 26.9.2009 5:35PM
Burbank, D.W., Leland, J., Fielding, E. et al. (1996). Bedrockincision, rock uplift and threshold hill slopes in the northwesternHimalayas. Nature, 379, 505–510.
Cossart, E. and Fort, M. (2008). Consequences of landslide damson alpine river valleys: examples and typology from theFrench Southern Alps. Norsk Geografisk Tidsskrift(Norwegian Journal of Geography), 62, 75–88.
Costa, J. E. (1991). Nature, mechanics, and mitigation of the ValPola landslide, Valtellina, Italy 1987–1988. Zeitschrift fürGeomorphologie, 35, 15–38.
Costa, J. E. and Schuster, R. L. (1988). The formation and failureof natural dams. Geological Society of America Bulletin, 100,1054–1068.
Crosta, G. B., Chen, H. and Lee, C. F. (2004). Replay of the 1987Val Pola Landslide, Italian Alps. Geomorphology, 60,127–146.
Cruden, D.M. and Varnes, D. J. (1996). Landslide types andprocesses. In Landslides, Investigation and Mitigation.Washington, DC: National Academy Press, 36–75.
Dikau, R., Brunsden, D., Schrott, L. and Ibsen, M. (eds.) (1996).Landslide Recognition: Identification, Movement and Causes.New York; Chichester: Wiley.
Evans, S.G. and Clague, J. J. (1994). Recent climatic change andcatastrophic geomorphic processes in mountain environments.Geomorphology, 10, 107–128.
Evans, S. G., Scarascia Mugnozza, G., Strom, A. L. et al. (2006).Landslides from massive rock slope failure and associatedphenomena. In S. G. Evans et al. (eds.), LandslidesFrom Massive Rock Slope Failure, Dordrecht: Springer,pp. 3–52.
Fort, M. (1987). Sporadic morphogenesis in a continental sub-duction setting: an example from the Annapurna Range, NepalHimalaya. Zeitschrift für Geomorphologie Supplement-Band,63, 9–36.
Fort, M. (1988). Catastrophic sedimentation and morphogenesisalong the High Himalayan front: implications for palaeoenvir-onmental reconstruction. In P. Whyte, J. S. Aigner,N. G. Jablonski et al. (eds.), The Palaeoenvironment of EastAsia fromMid-Tertiary, Centre of Asian Studies, University ofHong Kong, pp.170–194.
Fort, M. (2000). Glaciers and mass wasting processes: theirinfluence on the shaping of the Kali Gandaki valley (higherHimalaya of Nepal). Quaternary International, 65/66,101–119.
Fort, M. and Peulvast, J. -P. (1995). Catastrophic mass-movements and morphogenesis in the Peri-Tibetan Ranges:examples from West Kunlun, East Pamir and Ladakh. InO. Slaymaker (ed.), Steepland Geomorphology. Chichester:Wiley, 171–198.
Hewitt, K. (1988). Catastrophic landslide deposits in theKarakoram Himalaya. Science, 242, 64–67.
Hewitt, K. (1998). Catastrophic landslides and their effects onthe Upper Indus streams, Karakoram Himalaya, northernPakistan. Geomorphology, 26, 47–80.
Hewitt, K. (2002). Postglacial landforms and sediment associa-tions in a landslide-fragmented river system: theTranshimalayan Indus stream, Central Asia. In K. Hewitt,M. L. Byrne, M. English and G. Young (eds.), Landscapes ofTransition: Landform Assemblages and Transformations inCold Regions. Dordrecht: Kluwer, pp. 63–91.
Hewitt, K. (2006). Disturbance regime landscapes: mountaindrainage systems interrupted by large rockslides. Progress inPhysical Geography, 30, 365–393.
Hewitt, K., Clague, J. J. and Orwin, J. F. (2008). Legacies ofcatastrophic rock slope failures in mountain landscapes.Earth Science Reviews, 87, 1–38.
Hovius, N., Stark, C. P. and Allen, P. A. (2007). Sediment fluxesfrom a mountain belt derived by landslide mapping. Geology,25, 231–234.
Hsü, K. J. (1975). Catastrophic debris streams (sturzstroms) gen-erated by rockfalls. Geological Society of America Bulletin,86, 129–140.
Hungr, O., Evans, S. G. and Hazzard, J. (1999). Magnitude andfrequency of rockfalls and rock slides along the main trans-portation corridors of Southern British Columbia. CanadianGeotechnical Journal, 36, 224–238.
Kilburn, C. R. J. and Petley, D. N. (2003). Forecasting giant,catastrophic slope collapse: lessons from Vajont, NorthernItaly. Geomorphology, 54, 21–32.
Koi, T., Hotta, N., Ishigaki, I. et al. (2008). Prolonged impact ofearthquake-induced landslides on sediment yield in a moun-tain watershed: the Tanzawa region, Japan. Geomorphology,101, 692–702.
Korup, O. (2005). Geomorphic imprint of landslides on alpineriver systems, southwest New Zealand. Earth SurfaceProcesses and Landforms, 30, 783–800.
Korup, O., McSaveney, M. and Davies, T. R. H. (2004).Sediment generation and delivery from large historic land-slides in the Southern Alps New Zealand. Geomorphology,61, 189–207.
Korup, O., Clague, J. J., Hermanns, R. L., et al. (2007). Giantlandslides, topography, and erosion. Earth and PlanetaryScience Letters, 261, 578–589.
Li, T., Schuster, R. L. andWu, J. (1986). Landslide dams in south-central China. In R. L. Schuster (ed.), Landslides Dams:Processes, Risk, and Mitigation. American Society of CivilEngineers, Geological Special Publication, 3, pp. 21–41.
Montgomery, D. R. (2001). Slope distributions, threshold hill-slopes, and steady-state topography. American Journal ofScience, 301, 432–452.
Plafker, G. and Ericksen, G. E. (1978). Nevados Huascaranavalanches, Peru. In B. Voight (ed.), Rockslides andAvalanches: I, Natural Phenomena. Amsterdam: Elsevier,pp. 277–314.
Pratt-Sitaula, B., Burbank, D.W., Heimsath, A. and Ohja, T.(2004). Landscape disequilibrium on 1000–10,000 yearscales, Marsyandi River, Nepal, Central Himalaya.Geomorphology, 58, 223–241.
Monique Fort et al.84
C:/ITOOLS/WMS/CUP/556791/WORKINGFOLDER/ALA/9780521769259C07.3D 85 [75–86] 26.9.2009 5:35PM
Schneider, J. L., Pollet, N., Chapron, E., Wessels, M. andWassmer, P. (2004). Signature of Rhine Valley sturzstromdam failures in Holocene sediments of Lake Constance,Germany. Sedimentary Geology, 169, 1–2, 75–91.
Shang, Y., Yang, Z., Li, L. et al. (2003). A super-large landslidein Tibet in 2000: background, occurrence, disaster and origin.Geomorphology, 54, 225–243.
Slaymaker, O. (2006). Towards the identification of scaling rela-tions in drainage basin sediment budgets. Geomorphology, 80,8–19.
Thiery, Y., Malet, J. -P., Sterlacchini, S., Puissant, A. andMaquaire, O. (2007). Landslide susceptibility assessment bybivariate methods at large scales: application to a complexmountainous environment. Geomorphology, 92, 38–59.
85Catastrophic landslides and sedimentary budgets