Rock avalanches and the pace of late Quaternary development
of river valleys in the Karakoram Himalaya
Kenneth Hewitt1†, John Gosse2, and John J. Clague3
Department of Geography and Environmental Studies, Wilfrid Laurier University, 100 University Avenue, Waterloo, Ontario
N2L 3C9, Canada
2
Department of Earth Sciences, Dalhousie University, Edzell Castle Circle, Halifax, Nova Scotia B3H 4J1, Canada
3
Department of Earth Sciences, Simon Fraser University, 8888 University Drive, Burnaby, British Columbia V5A 1S6, Canada
1
ABSTRACT
We discuss the implications of a set of terrestrial cosmogenic nuclide (TCN) ages on
blocky, cross-valley deposits of large rock
avalanches along upper Indus streams. The
dated deposits are key to understanding late
Quaternary events that play a major role
in landscape evolution in the Karakoram
Hima­laya. The landslides occurred between
3 and 8 ka ago, challenging existing chronologies of events along Indus streams. The TCN
ages may support a mid-Holocene climatic
role in preparing slopes for failure, but the
balance of evidence suggests that large earthquakes triggered the landslides. Each landslide dammed the Indus or a major tributary
and controlled base level and sedimentation
for millennia. They produced landforms long
regarded as characteristic of the region, including extensive lacustrine deposits, flights
of river terraces, epigenetic gorges, and sediment fans. Until the 1990s, most of the landslides were interpreted as moraines; related
lacustrine and other sediments continue to
be attributed to glacial damming, and stream
terraces to tectonic processes. Generally they
were seen to originate tens of thousands to
hundreds of thousands of years earlier than
the new ages require. Instead we argue that
they record interactions among different geomorphic processes in landslide-fragmented
valleys during the Holocene. Rather than
being geomorphic markers of tectonic and
climatic events, the landslides have buffered
or redirected climatic and tectonic forcing.
In such an active orogen, millennia-long
episodes of zero net bedrock incision at each
site are surprising. However, rates of sedimentation above landslide barriers and erosion controlled by their breaching are close
E-mail: khewitt@wlu.ca
†
to today’s high measured rates for geomorphic activity. We propose that landslidefragmented rivers may, in fact, characterize
interglaciations and future patterns of upper
Indus landscape evolution at time scales of
103 to 104 years.
masses into thoroughly broken, crushed, and
pulverized debris sheets that spread out over
areas of 6.5 km2 to 20 km2 (Table 1). The de­
posits are well preserved in spite of high rates
of tectonic uplift and erosion in the Karakoram
Himalaya. Preservation is partly due to semiarid
conditions on valley floors and partly, as we
show, to their youthfulness.
INTRODUCTION
In the past two decades, some 350 large land­
In this paper, we report new terrestrial cosmo­ slides have been identified in the upper Indus
genic nuclide (TCN) ages from six blocky, basin (Fig. 1). On average, deposits of one crosscross-valley landslide deposits in the Karakoram valley rock avalanche have been found for every
Himalaya. Our study sites have been seminal in 14 km of upper Indus River streams surveyed
interpretations of Quaternary events in the oro­ (Hewitt, 1998, 2009a; Shroder and Bishop,
gen. The upper Indus River and its tributaries 1998; Korup et al., 2010). However, until­our
have been the subject of geoscience investiga­ study, none had been numerically dated.
The Holocene TCN ages dramatically reduce
tion and debate for more than 150 years, espe­
cially in relation to tectonics, glaciation, and the time span involved in these events. Past re­
sediment yield (Kalvoda, 1992; Shroder, 1993; search placed most of the deposits and related
Korup et al., 2007; Seong et al., 2009). However, landforms into the Pleistocene or older. Al­
until the 1990s few catastrophic rock slope fail­ though too small a sample to represent the over­
ures were recognized, and their contributions to all incidence of the landslides, the sites chosen
denudation, and the extent of their disturbance and ages established help to address two distinct
of the upper Indus River and its tributaries, had problems of interpretation.
First, the causes and triggers of these massive
been largely missed.
The landslides chosen for dating have con­ rock slope failures are widely debated. It has
trolled local base levels and sediment delivery been suggested that they may represent post­
over large parts of the upper Indus basin for thou­ glacial readjustments of slopes, possibly a para­
sands of years. The deposits are products of run­ glacial phenomenon (Ballantyne, 2002; Dadson
out and emplacement of large rock avalanches and Church, 2005; Knight and Harrison, 2009).
due to sudden rock slope failures (Evans and Recently, a wetter mid-Holocene climate at­
DeGraff, 2002; Hewitt, 2006). Steep descents­ tributable to a more intense monsoon has been
were sufficient to convert competent bedrock proposed as a possible cause of large landslides.
TABLE 1. DIMENSIONS OF ROCK AVALANCHES DATED IN THIS STUDY
Landslide
Katzarah
Dhak Chauki
Gol Ghone “S”
Upper Henzul
Baltit-Sumaiyar
Ghoro Choh I
Satpara-Skardu
Volume (km3)
Original Eroded
2.1
1.1
1.1
0.4
1.4
0.65
0.8
0.4
1
0.65
0.4
0.2
1.4
0.8
Area (km2)
Original Eroded
23
20.0
15
6.5
10.5
8.0
19
8.0
11
8.5
16
14.0
24
22.0
Elevation (m)
Highest Lowest
4830
2150
3450
1375
3600
2450
2920
1530
5200
2070
3400
2630
4120
2190
Drop (m)
Total
2680
2075
1150
1390
3130
1100
1930
GSA Bulletin; September/October 2011; v. 123; no. 9/10; p. 1836–1850; doi: 10.1130/B30341.1; 9 figures; 3 tables; Data Repository item 2011115.
1836
For permission to copy, contact editing@geosociety.org
© 2011 Geological Society of America
Runout (km)
Source-base
10.8
6.5
5.4
5.0
9.5
6.5
9.0
Rock avalanches and river valley development in the Karakoram Himalaya
Afghanis
73°30′
STUDY AREA
75°00′
tan
Pakista
n
China
Afghanistan
Darkot Pass
China
36°45′
n
Pakista
Darkot
36°45′
Pakistan
Bhutan
Nepal
Bangladesh
NS
36°00′
Ishk
oma
n
I T
G
L
G I
+
++
Y
++
Gupis
+C
N
E
+
A G+ ++ ++
+
+
+
+
+
6
+
+
+
+
+
Pasu
±
Arabian
Sea
Bay
of
Bengal
Sri Lanka
76°30′
T
+
+
5
+
++
+
Haramosh
7397 m
T I S T A
N
B A L
+
+ + ++
Rondu
++
++
3 MKAskole
T
+
+ +
River
Chilas
+ +
T
M
M
+
+ +
+ +
1
+ ++ +
S
Shigar
4
+
+ +
+ + Skardu + +
++
+ +
7998 m
Nanga Parbat
50 km
36°00′
KF
+
+
+
+ ++ + +
+
+
35°15′
0
Shimshal
MK
Gilgit
+
+
N
+
+ + +
+ +
+
Indus
7
Hunz
a R.
R.
India
+
+ +
++
2
Hushe
S
Z
++
+ +++ +
++ +
+ +
Shy
ok R + + +
.
Khapalu
du
In
73°30′
35°15′
+ +
76°30′
s
75°00′
Haldi
R
.
Karakoram Batholith
High Himalaya - Haramosh
Nanga Parbat-Haramosh
-lskere gneisses
Largely plutonics (granodiorites, leucogranites,
monzonites, tonalites...)
Kohistan-Ladakh Terrane
Kohistan-Ladakh batholith
(predominantly plutonic)
Metasedimentaries
Thrust fault
Suture zone
(sedimentaries, metasedimentary
metamorphics)
MKT
NS
MMT
SSZ
KF
Karakoram Metamorphic Complex
(Karakoram Plate)
Gneisses, metasedimentaries,
marbles, amphiboles, etc.
Major, fault-defined
Northern Karakoram Terrane
(Island Arc sequences)
+
+ +
Structures
(Karakoram Plate;
mainly pre-India-Asia collision plutons)
(Indian Plate)
3
Rock avalanche deposits
Examples discussed in the text
Main Karakoram Thrust
Northern Suture
Main Mantle Thrust
Shyok Suture Zone
Karakoram Fault
Figure 1. Rock avalanches in the upper Indus River basin in the central Karakoram of northern Pakistan (after Hewitt, 2009a). Sites dated
in this study are: 1—Katzarah; 2—Gol-Ghone; 3—Ghoro Choh; 4—Satpara-Skardu; 5—Dhak Chauki; 6—Upper Henzul; and 7—BaltitSumayar. Geology after Searle (1991).
However, triggering by earthquakes seems more
likely (Keefer, 1984); some researchers have
suggested that rare, great (magnitude >8) earth­
quakes occur in the Himalaya and are responsi­
ble for large landslides (Feldl and Bilham, 2006;
Kumar et al., 2006). A significant, related issue
is the response of the landscape to interactions
of climate, seismicity, and rock slope processes
in formerly glaciated valleys (Bull, 2007). Our
new ages allow us to explore how Karakoram
landslide incidence might address this topic.
Second, the influence of landslides on Kara­
koram valley development remains an important
issue (Fig. 2). Our ages bracket the times when
they blocked streams and controlled sediment
movement. Our finding of no net bedrock inci­
sion over thousands of years seems surprising
in a setting identified as “the steepest place on
Earth” (Miller, 1984), and singled out as hav­
ing “… excessively high uplift, erosion, and ex­
humation rates” (Searle, 1991, p. 8). The Indus
Geological Society of America Bulletin, September/October 2011
1837
Hewitt et al.
Figure 2. The Dhak Chauki landslide on
Gilgit­River. (A) View southwest across
Gilgit valley, showing the rock-avalanche
debris in the middle and foreground and the
detachment scar in the middle background.
(B) Plan view of the landslide. (C) Longi­
tudinal cross section.
1838
LANDSLIDE
B
00
Detachment zone
Rock avalanches
i Original surface
ii Eroded exposures and
original extent (inferred)
Gi
iii Pressure ridges and raised
distal rims
lgi
tR
.
iv Movement (inferred)
OTHER DEPOSITS
Lacustrine and related older
impoundment sediments
ram
Karako y
a
Highw
1500
Modern flood plain alluvium
River terrace
20
00
Recent sediment fans
Highway and road
30
00
Contours (500 m intervals)
2500
Rivers and streams
N
A
B
0
3500
1
B
2 km
Landslide profiles
A
South
North
Dhak Chauki RS/RA)
A
4000
B
Opposing
Slope
Detachment
3500
Zone
Transport Slope
Rock Avalanche
3000
Elevation (m)
Seven rock avalanches (Fig. 3 and Table 1)
were selected for dating because of their rele­
vance for past investigations and issues raised
in this paper. The largest, the Katzarah event
(Fig. 4), is located at the west end of the SkarduShigar basin, which it impounds. The Katzarah
landslide dam is partly dissected, but it contin­
ues to be local base level for some 110,000 km2
of the upper Indus basin. The Gol Ghone “S”
0
200
15
STUDY SITES
A
00
25
is said to involve “… a dynamic equilibrium in
which bedrock uplift and incision are balanced
through adjustments in stream power…” (Bur­
bank et al., 1996, p. 510). Yet, over the extensive
river sections interrupted by large landslides,
erosion has been reversed, at least temporarily,
and replaced by large-scale intermontane sedi­
mentation followed by trenching of valley fill
that is as yet incomplete. This finding suggests
that we may need to modify existing views of
the relation between tectonics and stream inci­
sion in active orogens.
Many researchers have examined the con­
tribution of landslides to denudation of high
mountains, but their focus typically has been
on how much material is directly moved by
landslides and thus is available for transport
by streams (Hovius et al., 2000; Barnard et al.,
2004; Korup et al., 2007; Antinao and Gosse,
2009). Others have noted that streams can
rapidly incise landslide barriers (Pratt-Sitaula
et al., 2007). However, although exceptional
amounts of material are transported by Kara­
koram landslides, remobilization by streams
appears to be relatively inefficient. The volume
of lacustrine and fluvial sediment trapped be­
hind the barriers has commonly exceeded that
of the landslide deposits by an order of magni­
tude and as much as two orders in some cases.
Korup et al. (2010) suggest that recurrent large
landslides in the upper Indus basin may be a
factor in protecting the Tibetan Plateau from
rapid dissection; here we offer some measure
of time scales and episodes of landslide inter­
ruptions of Hima­layan streams. We emphasize
that the events we describe are, in effect, “gate
keepers” of some 90% of the upper Indus basin
and virtually all of its Karakoram drainage.
Deposit
2500
Talus
Kohistan
Arc
metasedimentaries
2000
1500
Original rock avalanche
surface with megaclasts
rock avalanche deposit
1000
C
Gilgit R.
0
1
2
River alluvium
?
3
5
4
km
Vertical exageration x 1.5
Geological Society of America Bulletin, September/October 2011
Kohistan
Batholith
plutonics
6
7
Rock avalanche sites
73°30′ E
Geological Society of America Bulletin, September/October 2011
0
it R
.
5
5. Dhak Chauki
1. Katzarah
2
3Shigar R.
4
Skardu
1
75°00′ E
R.
Drainage areas impounded
by:
Gilgit
6
za
Hun
76°30′ E
us
Ind
ok
Shy
R.
Leh
r
78°00′ E
ve
Ri
36°00′N
79°30′ E
32°00′ N
34°00′N
35°00′N
0
N
Arabian
Sea
1000 km
Pakistan
Afghanistan
81°00′ E
India
Bhutan
33°00′ N
Sri Lanka
Bay
of
Bengal
Bangladesh
Nepal
China
STUDY AREA
Figure 3. Locations of dated rock avalanches within the upper Indus basin, showing the extent of the watersheds draining to the Katzarah
and Dhak Chauki rock-avalanche barriers.
50 km
1. Katzarah
2. Gol-Ghone
3. Ghoro Choh
4. Satpara-Skardu
5. Dhak Chauki
6. Upper Henzul
7. Baltit-Sumaiyar
72°00′E
Gilg
7
37°00′N
Rock avalanches and river valley development in the Karakoram Himalaya
1839
Hewitt et al.
N35°25′
A
ll
l l
Rock avalanche
detachment zone
l l l
l
l
l l
l
l l
l l
l l
l l l l l l
l l
l l
l
l l
l
l
l
l l
l l
l ll l
l l l
l l
l l
l
l l l l
l l
l
l
l
l
l
ll
l
l
l
l
l
l
l
l
ll
l
l l
l l
l
l
l l
l l
l
Present-day river,
streams and lakes
l
l
l l
H
Lacustrine deposits
l
M
River terraces with coarse
flood deposits
l l
l l
l
l l
l l
l l
l l l l
PL
N35°25′
River alluvium
present-day floodplain
PL
l l
l
l l
l l
l l
l l l l l l l
l l
Pressure ridges
l
l
l l
l l l
l l
l l
l l
l
l l
l
l l
l l
l
l
l
l l l l l
l l
l l
l
l l
Katzarah
Bragardo
l
l l
B
Movement of rock avalanche
“dry rock streams” (Heim)
l l
l l
l l
l
l l
l l
l l
l l
l
l l
l l
M
PL
ll
l l
l l
l l
Katzarah
Lake
Existing undisturbed rock
avalanche deposit area
with megaclasts
l
l l l l
l l
l l
l
l l
l l
l l l
l l
l l
l l
l l l l l l
l l
B
l l
l l
ll
PL
M
Original rock avalanche
deposit area
l
l l l l l l l l l l l
l l
l l
l l
l l l
l l l l
l l l l
l l l l ll l
l l l
l l
l l
l
l l
l l
l l l l
l ll
l
l l
l
l l
l l l l
l
l
ll
l l
l l l l
l l
Shigarthang
Valley
N35°25′
H
us
Ind
l l
l l
ll
R.
Skardu
Basin
0
l
l l
l l
l l
River terraces
“Shangri La” Hotel
Lithology of Rock Avalanche Debris
PL
M
B
Plutonic
Metamorphic
“Brandung”
2 km
N35°25′
Figure 4 (on this and following page). Katzarah rock avalanche. (A) Map of the rock-avalanche deposit.
1840
Geological Society of America Bulletin, September/October 2011
Rock avalanches and river valley development in the Karakoram Himalaya
3500
3250
Features on opposing (south) slope
Highest run-up of
rock avalanche
Elevation (m)
3000
2750
Main body of
rock avalanche deposit
Katzarah
Lake
2500
Effective barrier
Main aggradation
terraces
2250
Present Indus Thalweg
2000
Bedrock floor
0
2
4
8
6
Along-valley length (km)
10
12
14
downstream
Figure 4 (continued). (B) Profile of Indus River along the length of the Katzarah rock avalanche, showing the height of the effective barrier
and other elements of the deposit.
rock avalanche, ~40 km upstream, also fills the
valley of the Indus; its dam was initially ~450 m
high. The deposit of the Satpara-Skardu rock
avalanche impounds a river draining Deosai
Plateau to the south, and the Ghoro Choh I rockavalanche dams the Basha-Braldu basins within
the highest, most heavily glacierized catchment
of the central Karakoram. The other three sites
are in the Gilgit-Hunza basin, the main western
tributary of the upper Indus River. The Dhak
Chauki landslide straddles Gilgit River just be­
low its junction with Hunza River. The Upper­
Henzul rock avalanche crossed and blocked
lower Gilgit River, and the Baltit-Sumaiyar rock
avalanche blocked the Hunza and Hispar rivers.
Channels above the landslide barriers are
braided; at times of high summer discharge,
valley bottoms in these areas become shallow
lakes. The rivers flow between terraces in val­
ley fill, most of which record aggradation be­
hind the landslide dams. Where they cross the
rock-avalanche barriers, rivers flow in narrow,
steep channels, heavily armored by boulders
eroded from the landslide material (Hewitt,
2002b, p. 66). Below the Katzarah barrier, the
Indus enters a steep rock gorge, of major inter­
est in studies of stream incision (Burbank et al.,
1996). In all other cases, however, the rivers
again enter braided reaches between terraces in
valley fill also associated with landslide barriers
(Hewitt, 1998).
Prior to the turn of the past century, re­searchers
had no tools at their disposal to directly date
landslide deposits in the Karakoram Himalaya.
Although many landslides are within the range
of radiocarbon dating, most deposits lack plant
or animal remains. Many studies assigned ages
to Karakoram events on the basis of relative age
relations and assumed correlation with major­
climatic events elsewhere (Haserodt, 1989;
Searle, 1991; Shroder et al., 1993; Owen, 2006).
This situation changed in the 1980s following
the first applications of TCN techniques to di­
rectly date exposed surfaces.
METHODS
Criteria for the recognition of rock-avalanche
deposits, including those of interest here, have
been discussed in detail elsewhere (see GSA
Data Repository Table DR11). Where the detach­
ment and runout zones and related morphology
are not clear, the crucial criteria are the character
and lithology of the landslide deposit. The rock
type must be identical to bedrock cropping out
in the vicinity of the deposit (within maximum
runout distances of 10–15 km). The main body
of the deposit must be monolithological in all
size fractions from boulders to matrix. If the
1
GSA Data Repository item 2011115, Table DR1:
Diagnostic morphological and sedimentological
characteristics of rock-avalanche deposits; Table DR2:
TCN sample lithologies; and Table DR3: CRONUS
calculator version 2.2 input, is available online at
www.geosociety.org/pubs/ft2009.htm, or on request
from editing@geosociety.org or Documents Secre­
tary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.
Geological Society of America Bulletin, September/October 2011
1841
Hewitt et al.
source consists of two or more lithologies, these
will be preserved in the same general positions
as in outcrops in the source zone, or as “rem­
nant stratigraphy” (Heim, 1932). In addition, all
clasts from largest to smallest must be angular
or very angular. Samples for age determination
are taken from surface blocks that are part of ex­
tensive sheets of uniform lithology.
Terrestrial cosmogenic nuclide dating (Gosse
and Phillips, 2001) is particularly well suited
to rock avalanches, which involve collapses of
steep bedrock slopes. A rock avalanche arises
from brittle fracture and crushing of the rock
mass, mainly in its rapid descent of the source
slope. The momentum imparted at travel speeds
of over 150 km hr –1 spread the material over a
much greater surface area than that of the origi­
nal mass. Hence, the exposed deposit should
consist almost entirely of material that was pre­
viously at depths beyond the influence of most,
if not all, cosmic radiation. The surface of the
deposit typically consists of angular blocks that
are interlocking and strongly packed and thus
resist postemplacement movement.
The climate in Karakoram valleys below
~3500 m asl is arid or semiarid, and extensive
remnants of rock avalanches are well preserved
on stream terraces, fans, and other surfaces away
from river channels. Physical and chemical
weathering of boulders operates slowly. Blocks
support no more than lichens and mosses, and
trees and tall shrubs are rare. Seasonal snow
cover on valley floors is typically thin and gen­
erally short lived. Sampled boulders from the
Katzarah, Gol Ghone “S,” and Satpara-Skardu
rock-avalanche deposits are quartz-rich gran­
itoid rocks of the Kohistan-Ladakh Batholith
(Searle, 1991), which include quartz monzo­
nite, granodiorite, and quartz diorite. The Dhak
Chauki samples are greenschist-facies meta­
sedi­mentary rocks containing abundant quartz
veins (Hewitt, 2009b). The Upper Henzul and
Baltit-Sumaiyar samples are granodiorite de­
rived from, respectively, the Shirot Pluton of the
Kohistan-Ladakh Batholith and Hunza plutonic
complex of the Karakoram Batholith (Petterson
et al., 1990).
Samples of quartz-rich rock (>1 kg, outer­
most 2 cm of rock) were chiseled from the
center of flat and horizontal surfaces of block
tops for 10Be analysis (Fig. 5; Table DR2 [see
footnote 1]). Sampled surfaces were at least
1.5 m above the surrounding surface of the
landslide deposit to minimize the influence of
snow, aeolian sediment, and postdepositional
movement. BeO targets were prepared at the
Dalhousie Geochronology Center. We used a
Milli-Q ­boron-specific filter to reduce 10B in the
Type 1 water; we oxidized BeO by flame, not in
a muffle furnace. These procedures were used
1842
A
B
Figure 5. Sampling for terrestrial cosmogenic nuclide age determinations. (A) Sampled
block from Satpara-Skardu landslide. (B) Sampled surface, Baltit-Sumaiyar landslide.
to reduce the potential for excessive isobaric in­
terference on young targets. On average, 1.5 ×
1019 atoms of 9Be were added gravimetrically to
~40 g of the 250–355-micron fraction of leached
quartz prior to dissolution, using a 985 mg/mL
Be carrier. The carrier was produced by J. Klein
(University of Pennsylvania) from a shielded
beryl crystal collected from the Homestake
gold mine, South Dakota, USA. 10Be/9Be was
determined at the Center for Accelerator Mass
Spectrometry (Lawrence Livermore National
Laboratory) against standard 07KNSTD3110.
10
Be/9Be values have precisions averaging 2.7%
(1s), with carrier blanks of 4–7 × 10–15. Concen­
trations of 10Be were interpreted as ages using
the CRONUS-Earth online calculator (Balco
et al., 2008, v. 2.2), which assumes a 10Be halflife of 1.38 Ma and adjusts for altitude, latitude,
and accelerator mass spectrometry (AMS)
normalization. Maximum corrections for topo­
graphic cosmic-ray shielding were 4%. Because
only samples with no evidence of erosion and
lacking local shielding were analyzed, no fur­
ther adjustments to the calculated exposure ages
were necessary. We collected additional sam­
ples, but regarded triplicate or higher replication
to be excessive when two ages were within 3%
coefficient of variation, i.e., ~1s precision of an
Geological Society of America Bulletin, September/October 2011
Rock avalanches and river valley development in the Karakoram Himalaya
individual AMS measurement (Table DR3 [see
footnote 1]).
Although Karakoram rock avalanches are well
suited for TCN dating, caution is required to ob­
tain reliable ages. Rock-avalanche deposits may
be composite, recording more than one event.
Furthermore, the possibility that many blocks
at the surface of a debris sheet have original
surfaces should be considered. Preservation of
original surfaces on blocks could particularly
be a problem in the Karakoram, where parts of
landslides commonly stall against an opposing
slope (Hewitt, 2002a; Hewitt et al., 2008). In
contrast, where the entire rock mass spreads to
a thin sheet, only a small fraction of blocks will
have original surfaces and hence are less likely
to be sampled. Given these issues, the largest
blocks that are attractive for sampling—the
typical megablocks scattered over and com­
monly concentrated near the edges of debris
sheets—should be avoided. They may be pieces
of the original rock surface. However, without
indications of proximity to the cliff face, such
as weathering, the only reliable approach to
avoid the effect of TCN inheritance is to date
several blocks produced by the same landslide.
If two or more ages are within their AMS pre­
cision, inheritance can be ignored because it
is unlikely that all samples inherited the same
concentration.
In addition, postemplacement conditions re­
quire an understanding of the geomorphic envi­
ronment. The landslides described below formed
relatively long-lived dams on rivers. Parts of the
landslide deposits that were drowned could have
been shielded under water or blanketed by sedi­
ment that has since washed or blown away. In
this dry, dusty, and windy environment, rockavalanche deposits are aeolian sediment traps,
and dunes occur on or move over some of them.
One needs to be confident that sampled boul­
ders were never buried in wind-blown sediment
or subject to prolonged shielding in lakes.
Finally, quartz exposed at the surface of
blocks in landslide deposits in this region pre­
sents a problem. It seems to break up rapidly,
the opposite of what is typically observed in gla­
cial, fluvial, or aeolian environments here. Per­
haps quartz is damaged microscopically due to
brittleness, impact, or acoustical shock as blocks
are transported, fragmented, and emplaced. In
any case, it complicates sampling of quartz-rich
rock for TCN analysis. We made a special effort
to find blocks with undamaged quartz. Where
we could find none, we sampled blocks with a
thin (1–2 cm) resistant lithology covering quartz
veins or the quartz-rich rock.
TCN AGES
All sample exposure ages are reported in
­ able 2 and shown in Figure 6. The mean ex­
T
posure age of the Katzarah avalanche (n = 2) is
7.83 ± 0.19 ka (uncertainty = standard error); Gol
Ghone (n = 2), 4.03 ± 0.17 ka; Satpara-Skardu
(n = 3), 4.07 ± 0.08 ka; Dhak Chauki (n = 2),
6.48 ± 0.16 ka; Upper Henzul (n = 3), 8.37 ±
0.25 ka (sample UHen II yielded a 15.2 ka age
and is a statistical outlier, perhaps due to TCN
inheritance; it is not included in the mean age);
and Baltit-Sumayar (n = 3), 4.36 ± 0.14 ka. The
Ghoro Choh rock avalanches also are Holocene
in age. Wood fragments in a body of organic
sediment entrained and transported by Ghoro
TABLE 2. TERRESTRIAL COSMOGENIC NUCLIDE EXPOSURE AGES
Latitude
Longitude
Elevation
Age
Uncertainty†
Mean
S.E.§
Sample*
(°)
(°)
(m)
(ka)
(ka)
(ka)
(ka)
KATZ-II
35.428
75.460
2310
7.96
0.90
7.83
0.19
KATZ-IV
35.443
75.433
2500
7.69
0.88
GG-I
35.285
75.867
2590
4.15
0.50
4.03
0.17
GG-II
35.285
75.867
2590
3.91
0.45
STSK-I
35.248
75.628
2850
4.04
0.46
4.07
0.08
STSK-II
35.247
75.628
2850
4.21
0.48
STSK-III
35.233
75.628
2850
3.95
0.45
DCh-II
35.895
74.435
1500
6.59
0.76
6.48
0.16
DCh-III
35.895
74.435
1500
6.37
0.73
Uhen-I
35.996
74.200
1800
8.55
0.97
8.37
0.25
1.7#
Uhen-II
35.996
74.200
1800
15.2#
Uhen-III
35.996
74.200
1810
8.19
0.94
BaSu-I
35.996
74.200
1800
4.56
0.52
4.36
0.14
BaSu-II
36.304
74.673
2200
4.36
0.50
BaSu-III
36.304
74.673
2200
4.16
0.48
Note: The ages given are the mean of the ages computed using the three geomagnetic field-corrected scaling
schema (Dunai, 2001; Desilets and Zreda, 2003; Lifton et al., 2005; Desilets et al., 2006).
*Sample abbreviations: KATZ—Katzarah, GG—Gol Ghone, STSK—Satpara-Skardu, DCh—Dhak Chauki,
Uhen—Upper Henzul, and BaSu—Baltit-Sumayar.
†
Uncertainty is the 1σ total uncertainly as described in Balco et al. (2008). See Table DR3 for details.
 σ 
§
S.E.—Standard error 
.
 n − 1
#
Sample Uhen-II yielded a much older age than the other two samples and is considered an outlier. The mean
age of the Uhen dates does not include this age.
Choh 1 yielded a calibrated radiocarbon age of
7.9 ± 0.16 ka (Hewitt, 1999, p. 232). This age is
a maximum for the landslide, because the dated
mass of organic sediment must predate the land­
slide. Seong et al. (2009, p. 260–261) confirm
that three rock avalanches occurred at Ghoro
Choh and consider them to be Holocene in age.
We interpret all mean ages to be times of
landslide dam emplacement and reservoir infill­
ing. They are also maximum ages for incision of
barriers by streams, although incision probably
commenced some time after emplacement. The
ages enable us to evaluate average rates of dam
incision and removal of impounded materials to
the present.
DISCUSSION
The TCN ages place all six dated landslides
in the Holocene. The TCN ages are consistent
with the observation that nearly all 340+ rock
avalanches affecting the upper Indus streams
descended­across ice-free valleys. A few de­
posits may have survived since an earlier inter­
glacial, but most occurred since the last major
glaciation. Our ages are consistent with the idea
that the last glaciation in the Himalaya correlates
with the Wurm-Weichselian glaciation in Europe
and the late Wisconsinan glaciation in North
America (Kuhle, 2006; Hewitt, 2009a, 2009b).
Our landslide sites became ice-free before 10 ka
and perhaps before 12 ka. Their geomorphic
influences appear typical of the Holocene. An
increase in damming by large landslides may ex­
plain an apparent relative decrease in Himalayan
sediment yield during the past 7–8 Ma (Burbank
et al., 1993). Punctuated sediment storage and
release in these landslide-prone reaches appear
to affect the whole of the Holo­cene at least and
may compromise measure­
ments of sediment
flux at gauging stations.
More broadly, the features seem to record
processes characteristic of interglacial condi­
tions. They help to address important questions
about the incidence and causes of massive rockslope failures—rates of incision and denudation
along the Indus and relations of landslides to
glaciation and tectonics. The interpretation pro­
posed here supports and extends recent work on
landslides in the region but challenges some pre­
vailing Quaternary chronologies, most of which
emerged from studies that consider the dated
landslide deposits to be of glacial origin.
Landslide Causes and Triggers
Our sites are key to understanding Quater­
nary landform developments, but the data set
is too small and geographically dispersed to
yield definitive information about landslide
Geological Society of America Bulletin, September/October 2011
1843
Hewitt et al.
hI
II
I
D
C
hI
C
D
III
II
Su
Ba
Su
Su
Ba
Ba
G
G
I
ST I
SK
ST I
SK
II
ST
SK
III
G
I
I
6
G
Exposure age (ka)
8
U
KA
TZ
U
II
KA
TZ
IV
he
nI
he
nI
II
10
4
2
Relative position
Figure 6. Terrestrial cosmogenic nuclide exposure ages of blocks reported in this study; see
Figure 1 for sample locations. An anomalous age for the Upper Henzul event (* in Table 2)
may represent inherited exposure of the sampled block surface and is not shown. Error bars
represent 1s total error.
triggers­or temporal distributions in the study
area. Never­the­less, our results add to the data­
base of well-dated Himalayan landslides. His­
toric examples in the upper Indus basin include
earthquake-triggered landslides in 1841 and
­
2002 and rainfall and snowmelt events in 1858
and 1986 (Shroder, 1993; Hewitt, 2006).
It may be more than coincidence that our ages
fall within or close to the Holocene climatic op­
timum (ca. 8–4 ka), suggesting a climatic role in
reducing slope stability. A case can also be made
for a paraglacial connection (Hewitt, 2009a).
Detachment zones of the dated landslides are
near the upper limit of Himalayan glaciers dur­
ing the last Pleistocene glaciation. Glacial ero­
sion and debuttressing of slopes due to glacier
retreat at the end of the Pleistocene may have
prepared the slopes for failure (Huggel, 2009).
However, our data require either that it takes
millennia for most sites to reach a critical con­
dition or that only rare triggering events cause
failure (Hewitt, 2009b). There is no discernable
age relation with detachment zone elevation or
distance from centers of glaciation.
We note that Holocene TCN ages have been
reported for six similar landslides in the Indian
Himalaya southeast of our study area (Barnard
et al., 2004; Dortch et al., 2009). Five of the six
ages are close to the oldest of our age determina­
tions—ca. 6.5–8.5 ka—and one is close to our
youngest age. Gasse et al. (1991) suggested a
connection between these events and intensified
1844
monsoonal precipitation between 8.8 and 7.1 ka,
which they inferred from lake sediments of that
age at Pangong Tso near the border between
India and China. They also report increased
monsoonal precipitation at 4.0–2.6 ka—an in­
terval that encompasses our younger dates—but
Bookhagen et al. (2006) question the reliabil­
ity of their radiocarbon ages. Moreover, this
evidence is from an area more than 500 km east
and southeast of our study area. Precipitation
in the Karakoram in recent centuries has been
dominated by winter westerlies (Hewitt, 1993);
it is not clear how periods of stronger monsoons
would affect precipitation and slope stability in
our study area.
The usefulness of our dates can be extended
with geomorphic and stratigraphic evidence that
bears on the ages of nearby, undated landslides.
In particular, there are ten other rock-avalanche
deposits within a radius of less than 20 km of
the Satpara-Skardu deposit, all emplaced at or
close to the upper levels of aggradation behind
the Katzarah landslide dam and therefore of
about the same age (Hewitt, 1999, p. 223). Like
the Satpara-Skardu landslide, they disturbed the
upper strata of the Katzarah basin fill. The TCN
age for Gol-Ghone “B” is close to that of the
Satpara-Skardu event, and these two landslides
originated on opposite flanks of the same moun­
tain ridge, consistent with an earthquake trigger.
Although a mid-Holocene wetter or warmer cli­
mate cannot be ruled out as a factor in preparing
slopes for failure, we suggest the presence of 12
large rock avalanches of demonstrably similar
age in the vicinity of the Satpara-Skardu failure
is consistent with earthquake triggering. Most
of these rock avalanches are much larger than
any in the historical period and involved deepseated failures in hard crystalline bedrock. Al­
though this region has been seismically “quiet”
during recent decades (Cronin, 1989), the
cluster of large mid-Holocene landslides sup­
ports the hypothesis of “missing” great Hima­
layan earthquakes (Feldl and Bilham, 2006).
This hypothesis is based on the discovery that
observed seismicity is insufficient to explain
the large tectonic stresses that result from col­
lision of the Indian and Eurasian plates in the
Himalaya. Intermittent great earthquakes, with
average recurrence intervals of 500–1000 years
or more, are required to accommodate the stress
imbalance (Hubert-Ferrari et al., 2005). The
northwest Hima­laya, in particular, is believed to
be overdue for one or more great earthquakes
(Hough et al., 2009). An added factor is that
seismic ground accelerations in large earth­
quakes may be intensified two to four times in
areas of high relief, and for more than 150 km
from the epicenter (Hough and Bilham, 2008).
It is reasonable to expect that sets of Holocene
rock avalanches triggered by large earthquakes
are preserved in Karakoram valleys.
Disturbance Parameters
Rock avalanches are short-lived events, but
their detachment zone scars and deposits can
remain in the landscape for millennia. More
importantly, each of the dated landslides es­
tablished a local base level tens to hundreds of
meters above the pre-landslide river channel,
creating major disturbances in valley develop­
ment. They impounded large lakes whose origi­
nal full height is recorded by remnants of former
spillways. In four cases—Katzarah, Ghoro
Choh I, Dhak Chauki, and Baltit-Sumaiyar­—
the reservoirs eventually filled completely with
sediment. In the other three cases, breaching
commenced after only partial infilling, but ag­
gradation was substantial. Impounded sediment
generally exceeded landslide volumes by an
order­of magnitude in total, and much more in
the Katzarah impoundment (Tables 1 and 3).
Each large rock avalanche initiated a twophase sequence of events—first aggradation
in the impounded upstream area; and second,
degra­
dation of the barrier and trenching of
stored sediments (Table 3). However, local
base level continues to be controlled by the
landslide barrier until the river cuts down to the
pre-landslide valley floor, and along the full ef­
fective length of the barrier. Complete trenching
Geological Society of America Bulletin, September/October 2011
Rock avalanches and river valley development in the Karakoram Himalaya
TABLE 3. DIMENSIONS OF THE INTERRUPTION EFFECTS OF THE LANDSLIDE DAMS
Impoundment
Dam height (m)*
Length
Landslide (age, ka)
Original
Incised
Remnant
(km)
Katzarah (7.6)
200
110
90
55
Dhak Chauki (6.3)
60
60
0
18
Gol Ghone “S” (3.9)
450
400
50
149
Upper Henzul (8.0)
260
210
50
33
Baltit-Sumaiyar (4.2)
270
250
20
8.5
Ghoro Choh l (7.8)
45
40
5
7
Satpara-Skardu (3.9)
340
150
190
8
*Height of original dam crest, incision to date, and remaining barrier height.
†
Maximum volume of sediment behind barrier.
§
Remaining volume of sediment behind barrier.
Original
volume
(km3)†
52
2
20
4.5
1.5
0.5
1.0
Present
volume
(km3)§
30
1.4
15
1.5
0.9
0.4
1.0
down just 90 m through that barrier. In a similar
length of time, Gilgit River, with half the dis­
charge, has cut down 210 m through the Upper
Henzul barrier, whereas Shigar River at Ghoro
Choh I, possibly the oldest of our events, has cut
down the least. Thus, in some cases, net rates
of barrier incision are higher than estimates for
bedrock incision along the upper Indus River; in
other cases, the rates are comparable; whereas at
Ghoro Choh, the rate is much lower (Hancock
et al., 1998). Dam heights obviously limit how
much downcutting occurs in landslide materials,
but they do not explain incision rates or the frac­
tion of each barrier remaining.
Bookhagen et al. (2006) relate Holocene
episodes of sedimentation, incision, and terrace
formation in the Sultej basin southeast of our
study area to monsoon fluctuations. The scale of
1200
Nomal Complex
Rondu-Mendi I
Gol Ghone “N”
Gol Ghone “S”
Baltit-Sumaiyar
Satpara
Upper Henzul
Katzarah
Dhak Chauki
Ghoro Choh I
1100
1000
900
800
Height (m)
has only happened in the case of Dhak Chauki.
Sections through this barrier reveal that the rock
avalanche was emplaced over thick, preexisting
valley fill that had accumulated behind other,
older landslide barriers downstream (Hewitt,
2009b). Moreover, the location of the pre-land­
slide river channel is not known; nor is it known
whether it is buried at a lower level than the
present one. Even here, therefore, there has been
no incision in bedrock since well before 6.5 ka.
Lacustrine deposits drape the upstream flanks
of the landslides and, in most cases, can be
traced for tens of kilometers upvalley (Hewitt,
1998). In the Skardu-Shigar basins, river in­
cision has exposed up to 70 m of these sedi­
ments, but they are much thicker, extend more
than 40 km back from the landslide barriers,
and cover more than 200 km2. The runways at
Skardu airport, for example, lie on extensive ter­
races developed in these deposits.
A simple, but useful measure of the aggra­
dation threshold related to local base levels
established by rock avalanches, is the effective
barrier profile—the vertical, along-valley en­
velope at the initial, lowest path over the crest
of the landslide deposit (Fig. 7). It may not co­
incide with the thickest or highest cross-valley
landslide debris or the pre-landslide channel.
It does, however, provide the basic parameters
for measuring the fluvial response and sediment
delivery—maximum impoundment level and
barrier length.
The difference in height between the maxi­
mum impoundment level and present river level
is a measure of net reincision at the landslide
barrier (Table 3). It is important to stress that
incision in these examples bears no relation to
landslide size, age, or stream discharge. The
highest dam and greatest incision (450 m) are
at Gol Ghone “S” (Fig. 8). In a slightly longer­
period, Hunza River, with about one-third the
discharge, has cut down 250 m through the
Baltit-Sumayar barrier; and the Satpara outlet,
with 1% of the discharge, has cut down ~150 m.
The Indus at Katzarah, the largest of the streams
considered here, has taken twice as long to cut
Area
(km2)
300
60
450
40
10
13.5
6.5
the aggradation and trenching they document is
similar to the landslide-related episodes that we
describe. However, even if changes in the mon­
soon affected the Karakoram, impacts on sedi­
ment yields and movement would be masked or
rescheduled by the landslide interruptions.
In the river reaches that we have investigated,
the landslide disturbance signal overrides cli­
matic and tectonic events. Situational factors at
impoundment sites outweigh age, discharge, or
any unique rate that may apply to stream cut­
ting in rock-avalanche debris. That breaching is
incomplete in all but one case suggests incision
rates may decline over time, perhaps because
stream profiles through the barriers undergo
shifts in slope and length as breaching pro­
gresses, effective barrier height decreases, but
also moves upvalley, and the length of stream
channel cut into the barrier increases. The point
where the river first enters and is controlled by
landslide debris at Katzarah, Gol Ghone “S,”
Upper Henzul, and Baltit-Sumayar is now sev­
eral kilometers upvalley of, and 20–30 m higher
than, the original barrier crest (Fig. 4B). As the
length of river profile cutting into the landslide
barrier increases, the pace of incision may de­
crease, in part, as a response to decreasing
stream slope.
We emphasize that, unlike some landslides
studied elsewhere, breaching of the barri­
ers in the study area was relatively slow or
episodic and not catastrophic (cf. Costa and
700
600
500
400
300
200
100
0
0
10
5
Along-valley length (km)
15
upstream
Figure 7. Profiles of effective barriers of rock avalanches along the upper Indus River
and its tributaries. All the dated rock avalanches are shown, as well as three of the highest barriers yet discovered. Gol Ghone “N” lies just below Gol Ghone “S” on the middle
Indus River; Rondu Mendhi is in the Indus gorge 40 km below the Katzarah rock avalanche; and the Nomal rock avalanche is in the Hunza Valley 10 km above the confluence of
Gilgit and Indus rivers.
Geological Society of America Bulletin, September/October 2011
1845
Hewitt et al.
Schuster­, 1988; Stedile, 2006). Major floods
have occurred repeatedly in the affected valleys
(Hewitt, 1982) and must have eroded the dams,
but floods have not destroyed them. Incision of
the dams was accompanied by the development
of a series of river terraces upstream, down­
stream, and through each of the landslides. The
terraces include both fill and fill-cut terraces,
as well as some bedrock straths. They are re­
sponses to interruption by the landslides and
are controlled by them, including “defended”
forms where resistant materials deflect stream
flow or create protected areas (Cotton, 1958,
p. 246–248). Defended forms are associated
with coarse, highly compact rock-avalanche
debris and with debris flow fans and tribu­
tary deltas entering above or below landslide
impoundments. They are also found where
streams encounter or are let down onto bedrock,
in some cases leading to the development of
epigenetic gorges that control stream incision
rates (Ouimet­et al., 2008). The history of these
events does not support the idea of a single
base level at barrier crest height, followed by
a breach that restores the pre-landslide profile.
Rather, the series of terraces record episodic
incision and, at least in these examples, succes­
sive knickpoints moving upvalley in valley fill.
A
Relations to Bedrock Incision and
Terrace Development
B
Figure 8. Breaching sequence at the Gol Ghone “S” rock avalanche. (A) View south down
Indus River, which has incised up to 400 m into the rock-avalanche deposits. The river still
flows entirely in these deposits and has not reached its pre-landslide level. (B) View north up
Indus River from the site of the photograph shown in (A).
1846
Discussions of bedrock gorges, stream thal­
wegs, and terraces in this region assume control
by tectonics (Gansser, 1983; Seeber and Gornitz­,
1983; Owen, 1988; Leland et al., 1998). Ac­
cording to these views, valley fill records devel­
opments as far back as the late Tertiary or early
Pleistocene. Some researchers ascribe a role to
mid- to late-Pleistocene glaciations but much
earlier than the last glaciation (Derbyshire and
Owen, 1990; Seong et al., 2007). All of these
chronologies imply exceptionally slow rates of
landform development and inefficient removal
of sediment (see below). The remarkably lim­
ited extent of bedrock gorges and the changing
geometry of terraces above, through, and below
landslide barriers are rarely noted.
The landslides we dated, and others like
them, suggest a framework in which high rates
of erosion and huge sediment fluxes coexist
with, and are temporarily disturbed by, episodes
of blocking of rivers by landslides and rapid
upstream aggradation. Evolution of the land­
scape involves adjustments among Earth surface
processes­on century to millennial time scales.
Further complications arise because each land­
slide interruption interacts with previous ones
and has had its impacts disturbed by later ones.
Indeed, episodic incision and sedimentation
Geological Society of America Bulletin, September/October 2011
Rock avalanches and river valley development in the Karakoram Himalaya
coexist­spatially, including tributaries or sec­
tions of the Indus with aggradation occurring
behind landslide barriers separated by reaches
in bedrock undergoing rapid incision (Shroder
and Bishop, 1998; Cornwell et al., 2003).
A major challenge is the seminal work of
Burbank et al. (1996), who used TCN dating of
rock-cut straths to bracket rates of stream inci­
sion in the Indus gorge above and through the
Nanga Parbat–Haramosh massif. They found a
close relation between stream incision in bed­
rock and rates of tectonic uplift, and their study
has been widely cited as confirming such a rela­
tionship. However, the entire river section they
investigated has been subject to multiple and
repeated episodes of landslide blockage. This
reality was better appreciated by Leland et al.
(1998). The ages of the Holocene straths re­
ported by Leland et al. (1998) and recalculated
using the Balco et al. (2008) CRONUS-Earth
calculator have a bimodal distribution: straths
61–139 m above the level of Indus River range
from 7.0 ka to 8.6 ka, whereas straths 2–61 m
above river level range from 3.0 ka to 3.2 ka.
Leland et al. (1998, p. 104) suggested that high
sediment discharge from large landslides could
account for a significant component of bedrock
incision, but so could sediment starving leading
to removal of valley fill. The close similarity in
the exposure ages of their straths and the rock
avalanches in this study is notable, although
we disagree that the breaching was necessar­
ily catastrophic. The Katzarah landslide occurs
at the upper end of their profile, where they
found one of the lowest rates of incision. Our
results suggest no incision in bedrock here for
at least 7.8 ka. Some of their highest incision
rates are found in the middle part of their pro­
file, where the 1100-m-high Rondu-Mendi “A”
landslide dam is still not completely dissected.
This landslide is likely much older than Kat­
zarah landslide (Hewitt, 1998). Everywhere,
except for a 30 km section below the mouth
of Stak River and above Shengus village, the
Indus rock gorges are superimposed from for­
mer landslides and valley fill, and are thus “epi­
genetic” gorges; an example is the 500-m-deep
gorge at Rondu-Mendi (Hewitt, 1998; Ouimet
et al., 2008). These observations indicate that
much of the bedrock incision has been repeated
at these sites, and actual rates of incision can be
much greater than the net rates established.
Higher tectonic activity could explain the
greater concentration and more rapid removal
of landslide barriers where Indus River crosses
the east flank of the Nanga Parbat–Haramosh
massif (Hewitt, 2009a). However, other factors
weaken this argument. The main bedrock gorge
has been repeatedly starved of sediment by the
Katzarah event and other landslides upstream,
and most tributaries between Skardu Basin
and Gilgit Junction have landslide blockages
(Fig. 1). Additionally, Shroder et al. (1993) have
argued that no trunk glacier could be sustained
here for any length of time, if at all, because of
the absence of large tributaries reaching high
elevations between Skardu and Sassi. Hence,
several conditions have affected fluvial activity
in this gorge: exceptional tectonic uplift and an
active fault system, landslide interruptions with
associated sediment starving and flushing epi­
sodes, and limited glaciation.
Another question that arises is how and
why the Indus, on the west flank of the Nanga
Parbat–Haramosh massif between the Gilgit
Junction and Sazin, flows entirely in valley
fill, epigenetic rock gorges, and through as
yet incompletely breached landslide barriers
(Shroder­et al., 1989; Hewitt, 2009a). This ques­
tion, and the fact that the Katzarah and Dhak
Chauki landslides indicate 6.4–7.8 ka of zero
net incision at the margins of the Nanga ­Parbat–
Haramosh massif, speaks to an alternative hy­
pothesis, that rapid gorge incision could be a
driver of, rather than a response to, lo­calized
rapid uplift (Montgomery, 1994; Simpson­,
2004; Montgomery and Stolar, 2006). If so, the
scale, frequency, and longevity of landslide in­
terruptions need to be examined, as they may
disturb and reconfigure tectonic uplift in an ac­
tive orogen (Koons et al., 2002).
Even more severe problems arise from no­
tions that the terraces along these streams,
500–1000 m above present river level near the
highest landslide barriers, mainly reflect tec­
tonic uplift. In the Nanga Parbat–Haramosh
massif, Gansser (1964, 1983) invoked tectonic
forcing beginning in the late Tertiary to explain
the “uplifted terraces” he observed. Shroder
et al. (1993) accepted a similar time frame and
tectonics, but beginning with an early Pleisto­
cene glaciation.
However, terraces along the Indus streams
in the Nanga Parbat–Haramosh massif and in
most of the Karakoram are controlled by land­
slides and are Holocene in age. Their gradients
and lengths have no clear relation to observed
climatic or tectonic variations within the moun­
tains. The terraces lack continuity over distances
of more than a few tens of kilometers, and com­
monly much less. There is no evidence to sug­
gest they are products of the internal adjustment
or autocyclic developments of stream systems
independent of landslide events. Whereas most
tectonic and climatic terraces are erosional, the
highest landslide-related terraces, such as the
main Skardu and Shigar Basin terrace, record
deposition in the aggrading phase of landslide
interruptions.
Large alluvial fans are common where tribu­
taries enter Karakoram valleys (Fig. 9; Drew,
1873). Most are partly or wholly a consequence
of sedimentation in landslide-interrupted valleys
and represent a major part of total intermontane
sedimentation. The Satpara-Skardu rock ava­
lanche is one of at least six that descended over
Figure 9. View west up Gilgit River, where it cuts through the Dhak Chauki rock avalanche.
Note the upvalley sediment fans aggraded to the level of the effective barrier, thick lacustrine sediments, and river terraces cut as the dam was breached.
Geological Society of America Bulletin, September/October 2011
1847
Hewitt et al.
sediment fans constructed behind the Katzarah
barrier in Skardu basin, near the limit of Kat­
zarah aggradation (see below).
Most sediment fans near our sites are also
terraced due to trenching by streams or debris
flows following dam breaching. First called the
“lion’s paw” landform by Rickmers (1913), the
fans resemble the “telescopic-like alluvial fans”
of Colombo (2005). In this case, landslide in­
terruption of trunk streams is responsible for
the segmented and terraced insets that formed
during successive episodes of aggradation and
trenching related to changes in local base level.
Trenching of fans and formation of stream ter­
races are closely associated; both help track the
breaching histories of impoundments.
In summary, our TCN ages bracket devel­
opments related to landslide disruptions along
the Indus streams. The conditions and features
described are typical of the fluvial zone in the
entire watershed and of the more than 340
cross-valley rock-avalanche deposits that are
currently known, some of which are older and
some younger than the ones dated here. The
conditions observed would likely be typical of
previous interglaciations in these high moun­
tain valleys.
Implications for Quaternary Developments
The sites singled out here have played a
prominent role in Quaternary investigations.
The earliest scientific observers emphasized
the great quantities of sediment in many parts
of the Indus valleys and on surrounding slopes
(Strachey, 1853; Drew, 1873). The landslides
and related forms, although not recognized as
such, were integral parts of what Dainelli (1922)
referred to as the “characteristic morphologi­
cal elements within the ancient valley floor,”
and Owen and Derbyshire’s (1993, p. 115)
“… settings of the main landform types in the
Karakoram mountains.” However, until the
1990s and, in most cases to the present time,
our landslide deposits were interpreted as gla­
cial in origin. They were treated as terminal
moraine complexes and correlated with major
glaciations and glacial sequences (Li et al.,
1984; Shroder et al., 1993; Kamp and Haserodt,
2004). The lacustrine deposits were attributed to
glacier damming (Bürgisser et al., 1982; Owen,
1996), and the terraces were thought to be longterm or postglacial responses to tectonic uplift
(Gansser, 1983). The alluvial fans were attrib­
uted to “local regional and tectonic events,” to
paraglacial sedimentation, or to “catastrophic
events” (Derby­shire and Owen, 1990, p. 52).
Dhak Chauki and other landslide deposits were
viewed as glacial sediment 45–50 ka or older
(Desio and Orombelli, 1983; Owen, 2006).
1848
Seong et al. (2007) attribute the SatparaSkardu rock avalanche and others in the same
region to a middle Holocene glacier advance,
during which glaciers advanced tens of kilome­
ters from the Deosai Mountains, reaching as
low as 2300 m asl. No valley glacier exists in
these valleys above Skardu today, and the Deo­
sai Mountains have less than 2% glacier cover
and few peaks above 5000 m asl. The ages that
Seong et al. (2007, Table 2) report on boulders
on the terrace at Skardu and on sands directly
below the terrace agree with our TCN ages. We
interpret the boulders as the outer zone of the
Satpara-Skardu rock avalanche, dated to ca.
4.07 ka. Lacustrine sediments below the Skardu
terrace have been attributed to a glacier dam
at our Katzarah landslide site (Dainelli, 1922;
Cronin­, 1989), but we think they record aggrada­
tion behind the Katzarah landslide dam ~7.83 ka.
The glacial argument stems from Dainelli’s
(1922) view of glaciations in the study area. It
ascribes massive disturbance of fan and lacus­
trine sediments around Skardu to glacitectonism
(Owen, 1988; Seong et al., 2007). However,
such disturbance is common at rock-avalanche
sites elsewhere (Krieger, 1977; Yarnold, 1993;
Blikra et al., 2006), and has been documented
at most Skardu-Shigar Basin landslide sites by
Hewitt (1999, 2002a). We suggest the landslide
interpretation offers a viable and consistent view
of relevant landforms and sediments between
Satpara and Skardu, whereas the proposed midHolocene glaciation hypothesis raises insur­
mountable climatic and geomorphic problems.
The Pleistocene chronologies discussed
above imply extraordinary slowing of landform
development within Indus valleys, quite apart
from whether glacial processes or rock slope
failures are involved. In the pivotal case of
stream activity in a major, active orogen, they
would extend episodes of zero net incision at
our sites by 103–106 years. Landslides greatly
disturb streams, but on shorter time scales and
with much shorter response times. In summary,
our observations complicate, but are not incon­
sistent with, the high rates of uplift and geo­
morphic activity in the region (Zeitler, 1985).
They are incompatible with glacial and tectonic
interpretations of the sediment records in our
study areas.
CONCLUSIONS
A series of large Karakoram rock avalanches,
chosen for their locations and relation to muchdiscussed Pleistocene developments, are actu­
ally Holocene in age. Associated lacustrine
deposits, stream terraces, and large alluvial fans
up to hundreds of meters above contemporary
valley floors are products of the recent past and
of ongoing activity by geomorphic processes
responding to landslide disturbance. Their
rates of development are much more in keep­
ing with measured rates of contemporary uplift,
sedimentation, and stream action than formerly
envisaged. Recurrent landslide damming and
subsequent incision of the dams provide the
tools to form straths and high terraces along
upper­Indus River and its tributaries.
The volume of landslide material in most
Karakoram valleys below ~3000 m asl exceeds
that of glacial deposits. Postglacial valley-fill de­
posits exceed both by two orders of magnitude
or more, but most of these sediments accumu­
lated behind former and existing landslide barri­
ers. This interpretation differs from the view that
glaciers and the glacier system exercised such
control (Owen and Derbyshire, 1993).
Hitherto, alluvial fans, terraces, and straths
along the upper Indus have been seen as what
Burbank and Anderson (2000) call “geomorphic
markers”—Quaternary legacies of tectonics or
climatic events, especially glaciations. Tecton­
ics and climate change are surely important; a
large sample of landslide ages may well reveal
trends and patterns related to them. However,
complex geomorphic adjustments have buff­
ered, altered, and rescheduled the influence of
external forcing and the legacy of glaciations.
We propose that many of the geomorphic mark­
ers along the Indus record landslide disturbance
and fragmentation of the river system.
In general, past researchers have attributed
the landslide-related features along the Indus to
climatic forcing, to spatial patterns and rates of
tectonic uplift, or to some combination of the
two. They infer systematic changes over time—
trends or periodicities with respect to tectonic
or climatic forcing, including glaciations. We
identify the roles of other, very different condi­
tions that are episodic in character, and, relative
to external forcing, chaotic. Fragmentation of
Indus streams by landslides has created unstable
and nonstationary conditions, but over a much
shorter time span than previously envisaged.
They appear as random pulses, highly variable
in time and space. Local, situational features
of the landslides—valley geometry, sediment
delivery, and the regional mix of geomorphic
processes—play a decisive role in shaping the
postglacial landscape in the Indus valleys. The
new ages suggest that post-landslide develop­
ments can mask or remove all evidence of cli­
mate or tectonic trends, leaving only a record of
interruption episodes themselves. If anything,
they disrupt and annul developments or spatial
organization related to the drainage system or
tectonic patterns.
So large are the volumes of material almost
instantaneously but repeatedly moved by land­
Geological Society of America Bulletin, September/October 2011
Rock avalanches and river valley development in the Karakoram Himalaya
slides and emplaced across valley floors that
streams never develop normally. Taken as a
whole, the landslide disturbances seem to ex­
press an entropy-related condition: a dispersive
and disorganizing response to extreme energy
events, as opposed to the unidirectional, causeeffect relations attributed to external forcing.
We emphasize again that there has been zero
net bedrock incision at the sites of the land­
slides and related deposits over their lifetime.
As episodes within the Holocene, they indicate
a hiatus in stream incision, if not overall denu­
dation, which is surprising for an area with such
extreme relief. Nevertheless, the new ages are
consistent with aggradation and trenching pro­
ceeding at globally extreme rates in each land­
slide-interruption cycle.
These conditions are exceptionally well dis­
played in the upper Indus basin, but there are
parallels elsewhere in high Asia (Abdrakhmatov
and Strom, 2006; Korup et al., 2007). In the Euro­
pean Alps, recent work reinforces the picture of
landslide-fragmented mountain rivers. The wellknown Flims and Kofels rock avalanches, which
are now assigned to the Holocene, continue to
control fluvial and hillslope developments and
sediment flux in the Upper Rhine and Oetztal
valleys, respectively (von Poschinger, 2002,
2004). Similar conditions have been described
in the New Zealand Alps and the Andes, al­
though the rock avalanches in those orogens are
mostly smaller (Whitehouse and Griffiths, 1983;
Fauque and Tchilinguirian, 2002; Larsen et al.,
2005; Antinao and Gosse, 2009).
ACKNOWLEDGMENTS
Our research was supported by Natural Sciences
and Engineering Research Council (NSERC) of Can­
ada Discovery Grants, an NSERC Major Resources
Support Grant to the three authors, and grants from
the Canadian Foundation for Innovation and the At­
lantic Innovation Fund to J.G. Thoughtful reviews from
B. Bookhagen, J. DiPietro, and An Yin improved our
discussion of the roles of tectonic processes and cli­
mate change in the study area.
REFERENCES CITED
Abdrakhmatov, K., and Strom, A.L., 2006, Dissected rock­
slide and rock avalanche deposits: Tuien Shan, Kyrgyz­
stan, in Evans, S.G., Scarascia Mugnozza, G., Strom,
A.L., and Hermanns, R.L., eds., Landslides from Mas­
sive Rock Slope Failure: Celano, Italy, Proceedings,
North Atlantic Treaty Organization Advanced Work­
shop, June 2002: Dordrecht, Netherlands, SpringerVerlag, p. 551–570.
Antinao, J.-L., and Gosse, J., 2009, Large rockslides in the
southern Central Andes of Chile (32–34.5° S): Tec­
tonic control and significance for post-Miocene land­
scape evolution: Geomorphology, v. 104, p. 117–133,
doi: 10.1016/j.geomorph.2008.08.008.
Balco, G., Stone, J.O., Lifton, N.A., and Dunai, T.J., 2008,
A complete and easily accessible means of calculating
surface exposure ages or erosion rates from 10Be and
26
Al measurements: Quaternary Geochronology, v. 3,
p. 174–195, doi: 10.1016/j.quageo.2007.12.001.
Ballantyne, C.K., 2002, Paraglacial geomorphology: Qua­
ternary Science Reviews, v. 21, p. 1935–2017, doi:
10.1016/S0277-3791(02)00005-7.
Barnard, P.L., Owen, L.A., Sharma, M.C., and Finkel, R.C.,
2004, Late Quaternary (Holocene) landscape evolution
of a monsoon-influenced high Himalayan valley, Gori
Ganga, Nanda Devi, NE Garhwal: Geomorphology,
v. 61, p. 91–110, doi: 10.1016/j.geomorph.2003.12.002.
Blikra, L.H., Longva, O., Braathen, A., Anda, E., Dehl, J.F.,
and Stalsberg, K., 2006, Rock slope failures in Norwe­
gian fjord areas: Examples, spatial distribution and tem­
poral patterns, in Evans, S.G., Scarascia Mugnozza, G.,
Strom, A.L., and Hermanns, R.L., eds., Landslides from
Massive Rock Slope Failure: Celano, Italy, Proceedings,
North Atlantic Treaty Organization, June 2002: Dor­
drecht, Netherlands, Springer-Verlag, p. 475–496.
Bookhagen, B., Fleitmann, D., Nishizumi, K., Strecker,
M.R., and Thiede, R., 2006, Holocene monsoonal dy­
namics and fluvial terrace formation in the northwest
Himalaya, India: Geology, v. 34, p. 601–604, doi:
10.1130/G22698.1.
Bull, W.B., 2007, Tectonic Geomorphology of Mountains: A
New Approach to Paleoseismology: Malden, Massachu­
setts, Blackwell.
Burbank, D.W., and Anderson, R.S., 2000, Tectonic Geomor­
phology: New York, Wiley Interscience.
Burbank, D.W., Derry, L.A., and France-Lanord, C., 1993,
Reduced Himalayan sediment production 8 Myr ago
despite an intensified monsoon: Nature, v. 364, p. 48–
50, doi: 10.1038/364048a0.
Burbank, D.W., Leland, J., Fielding, E., Anderson, R.S.,
Brozovic, N., Reid, M.R., and Duncan, C., 1996, Bed­
rock incision, rock uplift and threshold hill slopes in
the northwestern Himalayas: Nature, v. 379, p. 505–
510, doi: 10.1038/379505a0.
Bürgisser, H.M., Gansser, A., and Pika, J., 1982, Late glacial
lake sediments of the Indus valley area, northwestern
Himalayas: Eclogae Geologische Helvetica, v. 75,
p. 51–63.
Colombo, F., 2005, Quaternary telescopic-like alluvial fans,
Andean Range, Argentina, in Harvey, A.M., Mather,
A.E., and Stokes, M., eds., Alluvial Fans: Geomor­
phology, Sedimentology, Dynamics: The Geological
Society of London Special Publication 251, p. 69–84.
Cornwell, K., Norsby, D., and Marston, R., 2003, Drainage,
sediment transport, and denudation rates on the Nanga
Parbat Himalaya, Pakistan, in Butler, D.R., Walsh,
S.J., and Malanson, G.P., eds., Mountain Geomorphol­
ogy—Integrating Earth Systems: Proceedings, 32nd
Binghamton Symposium in Geomorphology, Geomor­
phology, v. 55, p. 25–43.
Costa, J.E., and Schuster, R.L., 1988, The formation and
failure of natural dams: Geological Society of America
Bulletin, v. 100, p. 1054–1068, doi: 10.1130/0016-7606
(1988)100<1054:TFAFON>2.3.CO;2.
Cotton, C.A., 1958, Geomorphology: An Introduction to
the Study of Landforms: Christchurch, New Zealand,
Whitcombe and Tombs.
Cronin, V.S., 1989, Structural setting of the Skardu inter­
montane basin, Karakoram Himalaya, Pakistan, in
Malin­conico, L.L., Jr., and Lillie, R.J., eds., Tec­tonics
of the Western Himalayas: Geological Society of
America Special Paper 232, p. 203–216.
Dadson, S.J., and Church, M., 2005, Postglacial topographic
evolution of glaciated valleys: A stochastic landscape
evolution model: Earth Surface Processes and Land­
forms, v. 30, p. 1387–1403, doi: 10.1002/esp.1199.
Dainelli, G., 1922, Studi Sul Glaciale: Relazione Scientifiche
della Spedizione Italiana de Filippi, nell’Himalaia,
Caracorum­e Turchestan Cinese (1913–1914): Bologna­,
Italy: Zanichelli, 3 v. and plates.
Derbyshire, E., and Owen, L.A., 1990, Quaternary alluvial
fans in the Karakoram Mountains, in Rachocki, A.H.,
and Church, M., eds., Alluvial Fans: A Field Approach:
Chichester, United Kingdom, John Wiley and Sons,
p. 27–53.
Desio, A., and Orombelli, G., 1983, The “Punjab erratics”
and the maximum extent of the glaciers in the middle
Indus valley (Pakistan) during the Pleistocene: Aca­
demia Nazionale die Lincei, Classe di Scienzia Fisiche­,
Matematiche e Naturali, Ser. VII, XVI, Memorie,
Sezione­IIa, p. 135–180.
Desilets, D., and Zreda, M., 2003, Spatial and temporal dis­
tribution of secondary cosmic-ray nucleon intensities
and applications to in-situ cosmogenic dating: Earth
and Planetary Science Letters, v. 206, p. 21–42.
Desilets, D., Zreda, M., and Prabu, T., 2006, Extended
scaling factors for in situ cosmogenic nuclides: New
measure­ments at low latitude: Earth and Planetary Sci­
ence Letters, v. 246, p. 265–276.
Dortch, J., Owen, L.A., Haneberg, W.C., Caffee, M.W.,
Dietsch, C., and Kamp, U., 2009, Nature and tim­
ing of mega-landslides in northern India: Quaternary
Science Reviews, v. 28, p. 1037–1054, doi: 10.1016/
j.quascirev.2008.05.002.
Drew, F., 1873, Alluvial and lacustrine deposits and glacial
records of the Upper Indus Basin: Quarterly Journal of
the Geological Society of London, v. 29, p. 441–471,
doi: 10.1144/GSL.JGS.1873.029.01-02.39.
Dunai, T.J., 2001, Influence of secular variation of the geo­
magnetic field on production rates of in situ-produced
cosmogenic nuclides: Earth and Planetary Science Let­
ters, v. 193, p. 197–212.
Evans, S.G., and DeGraff, J.V., eds., 2002, Catastrophic
Landslides: Effects, Occurrence, and Mechanisms:
Geological Society of America Reviews in Engineer­
ing Geology, v. 15.
Fauque, L., and Tchilinguirian, P., 2002, Villavil rockslides,
Catamarca Province, Argentina, in Evans, S.G., and
DeGraff, J.V., eds., Catastrophic Landslides: Effects,
Occurrence, and Mechanisms: Geological Society
of America Reviews in Engineering Geology, v. 15,
p. 303–324.
Feldl, N., and Bilham, R., 2006, Great Himalayan earth­
quakes and the Tibetan Plateau: Nature, v. 444, p. 165–
170, doi: 10.1038/nature05199.
Gabet, E.J., Burbank, D.W., Putkonen, J.K., Pratt-Sitaula,
B.A., and Tank, O., 2004, Rainfall thresholds for land­
sliding in the Himalayas of Nepal: Geomorphology,
v. 63, p. 131–143.
Gansser, A., 1964, The Geology of the Himalayas: New
York, Wiley Interscience.
Gansser, A., 1983, The morphogenetic phase of moun­
tain building, in Hsu, K.J., ed., Mountain Building
Processes­: London, United Kingdom, Academic Press,
p. 221–228.
Gasse, F., Arnold, M., Fontes, J.C., Fort, M., Gilbert, E.,
Huc, A., Bingyan, L., Yuanfang, L., Qing, L., Méllères,
F., Van Campo, E., Fubao, W., and Qingsong, Z., 1991,
A 13,000-year climate record from western Tibet: Na­
ture, v. 353, p. 742–745, doi: 10.1038/353742a0.
Gosse, J.C., and Phillips, F.M., 2001, Terrestrial in situ
cosmo­genic nuclides: Theory and application: Qua­
ternary Science Reviews, v. 20, p. 1475–1560, doi:
10.1016/S0277-3791(00)00171-2.
Hancock, G.S., Anderson, R.S., and Whipple, K.X., 1998,
Beyond power: Bedrock river incision process and
form, in Tinkler, K., and Wohl, E.E., eds., Rivers over
Rock: Fluvial Processes in Bedrock Channels: Geo­
physical Monograph 107, p. 35–60.
Haserodt, K., 1989, Zur pleistozanen und postglazialen Ver­
gletscherungzwishen Hindukutsch, Karakorum und
Westhimalaya, in Haserodt, K., ed., Beitragen und Ma­
terialen zur Regionalen Geographie, Volume 2: Berlin,
Institut fur Geographie der Technischen Universitat,
p. 182–233.
Heim, A., 1932, Bergsturz und Menschenleben: Zurich,
Fretz and Wasmuth.
Hewitt, K., 1982, Natural dams and outburst floods of the
Karakoram Himalaya, in Glen, J., ed., Hydrological
Aspects of Alpine and High Mountain Areas: Inter­
national Hydrological Association Publication 138,
p. 259–269.
Hewitt, K., 1993, Altitudinal organization of Karakoram
geomorphic processes and depositional environments,
in Shroder, J.F., Jr., ed., Himalaya to the Sea: Geology,
Geomorphology, and the Quaternary: London, United
Kingdom, Routledge, p. 159–183.
Hewitt, K., 1998, Catastrophic landslides and their effects
on the Upper Indus streams, Karakoram Himalaya,
northern Pakistan: Geomorphology, v. 26, p. 47–80,
doi: 10.1016/S0169-555X(98)00051-8.
Hewitt, K., 1999, Quaternary moraines vs. catastrophic
rock avalanches in the Karakoram Himalaya, northern
Geological Society of America Bulletin, September/October 2011
1849
Hewitt et al.
Pakistan: Quaternary Research, v. 51, p. 220–237, doi:
10.1006/qres.1999.2033.
Hewitt, K., 2002a, Styles of rock avalanche depositional
complex in very rugged terrain, Karakoram Himalaya,
Pakistan, in Evans S.G., and DeGraff, J.V., eds., Cata­
strophic Landslides: Effects, Occurrence, and Mecha­
nisms: Geological Society of America Reviews in
Engineering Geology, v. 15, p. 345–378.
Hewitt, K., 2002b, Postglacial landform and sediment asso­
ciations in a landslide-fragmented river system: The
trans-Himalayan Indus streams, Inner Asia, in Hewitt,
K., Byrne, M-L., English, M., and Young, G., eds.,
Landscapes of Transition: Landform Assemblages and
Transformations in Cold Regions: Amsterdam, Nether­
lands, Kluwer, p. 63–91.
Hewitt, K., 2006, Rock avalanches with complex run out
and emplacement, Karakoram Himalaya, Inner Asia, in
Evans, S.G., Scarascia Mugnozza, G., Strom, A.L., and
Hermanns, R.L., eds., Landslides from Massive Rock
Slope Failure: Celano, Italy, Proceedings, North Atlantic
Treaty Organization Advanced Workshop, June 2002:
Dordrecht, Netherlands, Springer-Verlag, p. 521–550.
Hewitt, K., 2009a, Catastrophic rock slope failures and
late Quaternary developments in the Nanga ­Parbat–
Haramosh Massif, Upper Indus basin, northern
Pakistan: Quaternary Science Reviews, doi: 10.1016/
j.quascirev.2008.12.019.
Hewitt, K., 2009b, Paraglacial rock slope failures, distur­
bance regimes, and transitional landscapes, Upper
Indus Basin, northern Pakistan, in Knight, J., and Har­
rison, S., eds., Periglacial and Paraglacial Processes
and Environments: The Geological Society of London
Special Publication 320, p. 235–255.
Hewitt, K., Clague, J., and Orwin, J., 2008, Legacies of cata­
strophic rock slope failures in mountain landscapes:
Earth-Science Reviews, v. 87, p. 1–38.
Hough, S.E., and Bilham, R., 2008, Site response of the
Ganges basin inferred from re-evaluated macro­
seismic observations from the 1897 Shillong, 1905
Kangra, and 1934 Nepal earthquakes: Journal of Earth
System Science, v. 117, p. 773–782, doi: 10.1007/
s12040-008-0068-0.
Hough, S., Bilham, R., and Bhat, I., 2009, Kashmir Valley
megaearthquakes: American Scientist, v. 97, p. 42–49,
doi: 10.1511/2009.76.1.
Hovius, N., Stark, C.P., Chu, H.T., and Lin, J.C., 2000, Sup­
ply and removal of sediment in a landslide-dominated
mountain belt: Central Range, Taiwan: The Journal of
Geology, v. 108, p. 73–89, doi: 10.1086/314387.
Hubert-Ferrari, A., Suppe, J., Van Der Woerd, J., Wang, X.,
and Lu, H., 2005, Irregular earthquake cycle along
the southern Tianshan front, Aksu area, China: Jour­
nal of Geophysical Research, v. 110, p. B06402, doi:
10.1029/2003JB002603.
Huggel, C., 2009, Recent extreme slope failures in glacial
environments: Effects of thermal perturbation: Qua­
ternary Science Reviews, v. 28, p. 1119–1130, doi:
10.1016/j.quascirev.2008.06.007.
Kalvoda, J., 1992, Geomorphological Record of the Qua­
ternary Orogeny in the Himalaya and the Karakoram:
Elsevier Developments in Earth Surface Processes 3.
Kamp, U., and Haserodt, K., 2004, Quaternary glaciations in
the high mountains of northern Pakistan, in Ehlers, J.,
and Gibbard, P.L., eds., Extent and Chronology of Gla­
ciation: Volume 3: South America, Asia, Africa, Aus­
tralia, Antarctica: Amsterdam, Netherlands, Elsevier­,
p. 293–311.
Keefer, D.K., 1984, Landslides caused by earthquakes:
Geological Society of America Bulletin, v. 95, p. 406–
421, doi: 10.1130/0016-7606(1984)95<406:LCBE>
2.0.CO;2.
Knight, J., and Harrison, S., eds., 2009, Periglacial and para­
glacial processes and environments: The Geological
Society of London Special Publication 320, p. 71–84.
Koons, P.O., Zeitler, P.K., Chamberlain, C.P., Craw, D., and
Meltzer, A.S., 2002, Mechanical links between erosion
and metamorphism in Nanga Parbat, Pakistan Hima­
laya: American Journal of Science, v. 302, p. 749–773,
doi: 10.2475/ajs.302.9.749.
Korup, O., Clague, J.J., Hermanns, R.L., Hewitt, K.,
Strom, A.L., and Weidinger, J.T., 2007, Giant land­
slides, topography, and erosion: Earth and Planetary
1850
Science Letters, v. 261, p. 578–589, doi: 10.1016/
j.epsl.2007.07.025.
Korup, O., Montgomery, D.R., and Hewitt, K., 2010,
Glacier and landslide feedbacks to topographic re­
lief in the Himalayan syntaxes: Proceedings, Na­
tional Academy of Sciences: www.pnas.org/cgi/
doi/10.1073/pnas.0907531107; supporting information
online awww.pnas.org/cgi/content/full/0907531107/
DCSupplemental.
Krieger, M.H., 1977, Large Landslides Composed of Mega­
breccia, Interbedded in Miocene Basin Deposits, South­
eastern Arizona: U.S. Geological Survey Professional
Paper 1008.
Kuhle, M., 2006, The past Hunza Glacier in connection
with a Pleistocene Karakoram ice stream network dur­
ing the last ice age (Würm), in Kreutzmann, H., ed.,
Karakoram in Transition: Culture, Development, and
Ecology in the Hunza Valley: Karachi, Pakistan, Ox­
ford University Press, p. 24–48.
Kumar, S., Wesnousky, S.G., Rockwell, T.K., Briggs, R.W.,
Thakur, V.C., and Jayangondaperumal, R., 2006, Paleo­
seismic evidence of great surface rupture earthquakes
along the Indian Himalaya: Journal of Geophysical Re­
search, v. 111, p. B03304, doi: 10.1029/2004JB003309.
Larsen, S.H., Davies, T.R.H., and McSaveney, M.J., 2005,
A possible coseismic landslide origin of late Holocene
moraines of the Southern Alps, New Zealand: New
Zealand Journal of Geology and Geophysics, v. 48,
p. 311–314, doi: 10.1080/00288306.2005.9515117.
Leland, J., Reid, M.R., Burbank, D., Finkel, R., and Caffee,
M., 1998, Incision and differential bedrock uplift along
the Indus River near Nanga Parbat, Pakistan Himalaya,
from 10Be and 26Al exposure age dating of bedrock
straths: Earth and Planetary Science Letters, v. 154,
p. 93–107, doi: 10.1016/S0012-821X(97)00171-4.
Li, J., Derbyshire, E., and Shuying, X., 1984, Glacial and
paraglacial sediments of the Hunza Valley, north­
west Karakoram, Pakistan: A preliminary analysis, in
Miller, K.J., ed., The International Karakoram Project:
Cambridge, United Kingdom, Cambridge University
Press, v. 2, p. 496–535.
Lifton, N., Bieber, J., Clem, J., Duldig, M., Evenson, P.,
Humble­, J., and Pyle, R., 2005, Addressing solar modu­
lation and long-term uncertainties in scaling secondary
cosmic rays for in situ cosmogenic nuclide applications:
Earth and Planetary Science Letters, v. 239, p. 140–161.
Miller, K.J., ed., 1984, The International Karakoram Project:
Cambridge, United Kingdom, Cambridge University
Press, 2 volumes.
Montgomery, D.R., 1994, Valley incision and the uplift of
mountain peaks: Journal of Geophysical Research,
v. 99, p. 13,913–13,921, doi: 10.1029/94JB00122.
Montgomery, D.R., and Stolar, D.B., 2006, Reconsidering
Himalayan river anticlines: Geomorphology, v. 82,
p. 4–15, doi: 10.1016/j.geomorph.2005.08.021.
Ouimet, W.B., Whipple, K.X., Crosby, B.T., Johnson, J.P.,
and Schildgen, T.F., 2008, Epigenetic gorges in fluvial
landscapes: Earth Surface Processes and Landforms,
v. 33, p. 1993–2009, doi: 10.1002/esp.1650.
Owen, L.A., 1988, Terraces, Uplift and Climate, Karakoram
Mountains, Northern Pakistan [Ph.D. thesis]: Leicester,
United Kingdom, University of Leicester.
Owen, L.A., 1996, Quaternary lacustrine deposits in highenergy semi-arid mountain environment, Karakoram
Mountains, northern Pakistan: Journal of Quaternary
Science, v. 11, p. 461–483, doi: 10.1002/(SICI)1099
-1417(199611/12)11:6<461::AID-JQS282>3.0.CO;2-G.
Owen, L., 2006, Quaternary glaciation, in Kreutzmann, H.,
ed., Karakoram in Transition: Culture, Development,
and Ecology in the Hunza Valley: Karachi, Pakistan,
Oxford University Press, p. 12–23.
Owen, L.A., and Derbyshire, E., 1993, Quaternary and
Holocene intermontane basin sedimentation in the
Karakoram Mountains, in Shroder, J.F., Jr., ed., Hima­
laya to the Sea: Geology, Geomorphology, and the
Quaternary: London, United Kingdom, Routledge,
p. 108–131.
Petterson, M.G., Windley, B.F., and Sullivan, M., 1990, A
petrological, chronological, structural and geochemi­
cal review of Kohistan Batholith and its relationship to
regional tectonics: Physics and Chemistry of the Earth,
v. 17, p. 47–70.
Pratt-Sitaula, B., Garde, M., Burbank, D.W., Oskin, M.,
Heimsath, A., and Gabet, E., 2007, Bedload-tosuspended­load ratio and rapid bedrock incision from
Himalayan landslide-dam lake record: Quaternary Re­
search, v. 68, p. 111–120.
Rickmers, W.R., 1913, The Duab of Turkestan: Cambridge,
Cambridge University Press.
Searle, M.P., 1991, Geology and Tectonics of the Karakoram
Mountains: New York, Wiley.
Seeber, L., and Gornitz, Z., 1983, River profiles along
the Himalayan arc as indicators of active tec­tonics:
Tectonophysics, v. 92, p. 335–367, doi: 10.1016/
0040-1951(83)90201-9.
Seong, Y.B., Owen, A., Bishop, M.P., Bush, A., Clendon,
P., Copland, L., Finkel, R.C., Kamp, U., and Shroder,
J.F., Jr., 2007, Quaternary glacial history of the cen­
tral Karakoram: Quaternary Science Reviews, v. 26,
p. 3384–3405, doi: 10.1016/j.quascirev.2007.09.015.
Seong, Y.B., Bishop, M.P., Bush, A., Clendon, P., Copland,
L., Finkel, R.C., Kamp, U., Owen, L.A., and Shroder,
J.F., 2009, Landforms and landscape evolution in the
Skardu, Shigar, and Braldu valleys, central Karakoram:
Geomorphology, v. 103, p. 251–267, doi: 10.1016/
j.geomorph.2008.04.026.
Shroder, J.F., Jr., ed., 1993, Himalaya to the Sea: Geology,
Geomorphology, and the Quaternary: London, United
Kingdom, Routledge.
Shroder, J.F., Jr., and Bishop, M.P., 1998, Mass movement
in the Himalaya: New insights and research direc­
tions: Geomorphology, v. 26, p. 13–35, doi: 10.1016/
S0169-555X(98)00049-X.
Shroder, J.F., Jr., Khan, M.S., Lawrence, R.D., Madin, I.P.,
and Higgins, S.M., 1989, Quaternary glacial chronol­
ogy and neotectonics in the Himalaya or Northern
Pakistan, in Malinconico, L.L., Jr., and Lillie, R.J.,
eds., Tectonics of the Western Himalayas: Geological
Society of America Special Paper 232, p. 275–294.
Shroder, J.F., Jr., Owen, L., and Derbyshire, E., 1993, Quater­
nary glaciation of the Karakoram mountains and Nanga
Parbat Himalaya, in Shroder, J.F., Jr., ed., Himalaya to
the Sea: Geology, Geomorphology, and the Quaternary:
London, United Kingdom, Routledge, p. 132–158.
Simpson, G., 2004, Role of river incision in enhancing
defor­mation: Geology, v. 32, p. 341–344, doi: 10.1130/
G20190.2.
Stedile, L., ed., 2006, Security of Natural and Artificial Rock­
slide Dams: Italian Journal of Engineering Geology and
Environment Special Issue I.
Strachey, H., 1853, Physical geography of western Tibet:
Journal of the Royal Geographical Society, p. 1–69.
von Poschinger, A., 2002, Large rockslides in the Alps: A
commentary on the contribution of G. Abele (1937–
1994) and a review of some recent developments,
in Evans, S.G., and DeGraff, J.V., eds., Catastrophic
Landslides: Effects, Occurrence, and Mechanisms:
Geological Society of America Reviews in Engineer­
ing Geology, v. 15, p. 237–256.
von Poschinger, A., 2004, The Flims rockslide dam, in
Abdrakhmatov, K., Evans, S.G., Hermanns, R.,
­Scarascia-Mugnozza, G., and Strom, A.L., eds., Se­
curity of Natural and Artificial Rockslide Dams: Ex­
tended abstracts volume: Bishkek, Kyrgyzstan, North
Atlantic Treaty Organization Advanced Research
Workshop Volume, p. 141–144.
Whitehouse, I.E., and Griffiths, G.A., 1983, Frequency and
hazard of large rock avalanches in the central Southern
Alps, New Zealand: Geology, v. 11, p. 331–334, doi:
10.1130/0091-7613(1983)11<331:FAHOLR>2.0.CO;2.
Yarnold, J.C., 1993, Rock-avalanche characteristics in
dry climates and the effect of flow into lakes: In­
sights from mid-Tertiary sedimentary breccias near
Artillery Peak, Arizona: Geological Society of
America Bulletin, v. 105, p. 345–360, doi: 10.1130/
0016-7606(1993)105<0345:RACIDC>2.3.CO;2.
Zeitler, P., 1985, Cooling history of the NW Himalaya,
Pakistan: Tectonics, v. 4, p. 127–151, doi: 10.1029/
TC004i001p00127.
Manuscript Received 19 May 2010
Revised Manuscript Received 24 September 2010
Manuscript Accepted 13 October 2010
Printed in the USA
Geological Society of America Bulletin, September/October 2011