Journal Pre-proofs Research papers Barrier lake bursting and flood routing in the Yarlung Tsangpo Grand Canyon in October 2018 Chen Chen, Limin Zhang, Te Xiao, Jian He PII: DOI: Reference:
S0022-1694(20)30063-9 https://doi.org/10.1016/j.jhydrol.2020.124603 HYDROL 124603
To appear in:
Journal of Hydrology
Received Date: Revised Date: Accepted Date:
7 August 2019 23 November 2019 19 January 2020
Please cite this article as: Chen, C., Zhang, L., Xiao, T., He, J., Barrier lake bursting and flood routing in the Yarlung Tsangpo Grand Canyon in October 2018, Journal of Hydrology (2020), doi: https://doi.org/10.1016/j.jhydrol. 2020.124603
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier B.V.
1
Barrier lake bursting and flood routing in the Yarlung
2
Tsangpo Grand Canyon in October 2018
3 4
Chen Chena, Limin Zhanga,b,c*, Te Xiaob, Jian Heb
5 6
* Corresponding author
7 8
Email addresses:
[email protected] (C. Chen),
[email protected] (L. M. Zhang),
9
[email protected] (T. Xiao),
[email protected] (J. He).
10 11
a
12
Resource and Hydropower, Sichuan University, Chengdu, China
13
b
14
and Technology, Hong Kong
15
c
State Key Laboratory of Hydraulics and Mountain River Engineering, College of Water
Department of Civil and Environmental Engineering, The Hong Kong University of Science
HKUST Shenzhen Research Institute, Shenzhen, China
1
16
Abstract
17
Ice-soil mixture landslide dams formed frequently in the Tibetan Plateau in response to global
18
warming, which pose great threats to both upstream and downstream areas due to inundation
19
and lake bursting. On 17 October 2018, a large landslide, induced by an ice-avalanche at the
20
Sedongpu Basin of the Yarlung Tsangpo, blocked the main course of the river near Gyalha.
21
The barrier lake level rose quickly and the dam was overtopped naturally at 13:30 on 19
22
October 2018, generating a dam-breaching flood with a peak flow rate of 32,000 m3/s. This
23
paper presents a comprehensive study of the disaster chain of landslide-barrier lake-dam
24
breaching-river flooding in the Yarlung Tsangpo Grand Canyon, detailed geological and
25
hydrological characteristics of the study region, rapid prediction of the dam breaching
26
hydrograph using an erosion-based numerical model, and analysis of the flood routing in a 460
27
km canyon reach along the Yarlung Tsangpo. The simulated peak discharge at the dam site is
28
over 30,000 m3/s and the corresponding dam-breaching and flood routing hydrographs agree
29
well with the observations. Two additional scenarios with larger inflow rates are also
30
considered. Results show that with a larger inflow into the barrier lake, the erosion of the dam
31
body becomes more rapid. When the inflow rate is increased by six times, the peak dam-
32
breaching discharge can be doubled. The study serves as basis to manage the flood risks due to
33
landslide dam bursting on the Yarlung Tsangpo or similar rivers.
34 35
Keywords: Ice avalanche; Landslide; Landslide dam; Barrier lake; Dam breaching; Flood
36
routing.
2
37
1. Formation and bursting of the Gyalha barrier lake
38
The Yarlung Tsangpo is the highest large river in the world with an average elevation
39
of 4000 m. The river originates from Chemayungdung Glacier in southwestern Tibet and flows
40
eastward on the Tibetan Plateau. It bends sharply in southeastern Tibet, where it passes between
41
two peaks: the 7,782 m Namcha Barwa and the 7,294 m Gyala Peri (Fig. 1). The river then cuts
42
its way through the Yarlung Tsangpo Grand Canyon and into Arunachal, where it is called
43
Siang. Down the river from Arunachal, the river becomes broader, and after reaching Assam it
44
is also known as the Brahmaputra. The river further flows into Bangladesh from Assam, where
45
it is called Padma. It crosses China, India and Bangladesh with a length of about 3,350 km, a
46
catchment area of 66.6×104 km2, and an annual mean runoff of 66.29×1010 m3 (Liu, 1999).
47
At around 5 am, 17 October 2018, a very large landslide struck the Yarlung Tsangpo in
48
southeastern Tibet, China, 7 km downstream the Gyalha Village and about 175 km upstream
49
Medog, in the vicinity of 29°47’7.20’’ N, 94°55’24’’ E (Fig. 1). The landslide originated from
50
the left bank of the Yarlung Tsangpo inside the Sedongpu Basin, formed a landslide dam and
51
blocked the main course of the Yarlung Tsangpo (Fig. 2a). The landslide dam was composed
52
of massive ice and debris 40-60 million m3 in volume, 310-620 m in width, and 79 m in height
53
(Jin, 2019; Tong et al., 2018; CCTV News, 2018a). The Gyalha landslide was featured with
54
mixtures of ice and soil materials, which marched into the Yarlung Tsangpo and spread
55
laterally along the river by 2.3 km. The barrier lake level increased very quickly; at the noon
56
of 17 October, the lake behind the landslide dam was 15 km long, the water level rose 40 m
57
above the original water level, and the reservoir volume reached about 150 million m3. On the
58
morning of 18 October, the water level of the barrier lake was already 50 m higher than the
59
ground surface of the Gyalha Village and the lake was at risk of breaking. Figure 3 shows the
60
inundated Gyalha Village after the formation of the landslide dam, where the access road to
61
the village was completely submerged. More than 20,000 people in the Mainling County and
3
62
Medog County could be affected by the barrier lake due to either inundation or lake bursting.
63
Authorities in Tibet activated the highest level of emergency response and evacuated at least
64
6,000 residents. The breach of the landslide dam could also bring massive floods along the
65
Siang River in Arunachal, posing a great threat to the people in the region. Thus, the Chinese
66
and India governments exchanged information on 18 October to cope with this disaster. The
67
upper Siang district administration has warned residents in the Siang Valley to take
68
precautionary measures (NDTV, 2018). By 7 am, 19 October, the lake water level had risen
69
about 75 m above the original level and was still rising at an average rate of 0.61 m per hour
70
(CCTV News, 2018a). Eventually, at 13:30 pm on 19 October, the Gyalha landslide dam was
71
overtopped (Fig. 2b), with a lake capacity larger than 500 million m3 (Xinhua Net, 2018a). The
72
peak breaching flow was as large as 32,000 m3/s at the dam site and resumed its base flow on
73
20 October (Jin, 2019).
74
Landslides frequently occurred in the Sedongpu Basin due to glacial activities. The
75
Yarlung Tsangpo near Sedongpu had been partially blocked after the Nyingchi earthquake in
76
2017, thus can be blocked more easily. The mixtures of ice and debris from ice avalanches are
77
very unstable, hence the barrier dam may break quickly, releasing a large quantity of water.
78
Once a landslide dam forms and blocks a major river, it may pose a great threat to residents
79
both upstream and downstream the dam. For instance, an extremely large landslide on 9 April
80
2000 at Zhamu Creek, 48 km northeast of the Gyalha landslide (30°12’03’’ N, 94°58’03’’ E,
81
Fig. 1) blocked the Yigong River that is a tributary of the Yarlung Tsangpo (Lu et al., 2003).
82
The landslide dam breached on 10 June 2000, releasing a massive impounded water volume of
83
about 3 km3, resulting in a dam-breaching flood with a peak discharge of 120,000 m3/s (Shang
84
et al., 2003). The burst of the barrier lake affected 4,000 people in the downstream. Since most
85
landslide dams are short-lived with 51% of landslide dams breaching within one week, timely
86
evacuation of downstream people is extremely important (Peng and Zhang, 2013a,b). The
4
87
prediction of the dam breaching time, flow rate, and the downstream flood routing process is
88
important for evaluating the dam-breaching flood risks and making a successful emergency
89
management plan.
90
The focus of this study is to conduct a comprehensive analysis of the Gyalha landslide-
91
barrier lake-dam breaching-flooding disaster chain in the Yarlung Tsangpo Grand Canyon.
92
Taking the Gyalha landslide as a benchmark, the processes of dam breaching and flood routing
93
with plausible larger inflows into the barrier lake are also simulated. Results of dam breaching
94
process, flow rate, final breach size, and subsequent flood routing are analyzed and compared.
95
Detailed information of the geometric and hydraulic conditions of the dam, the dam breaching
96
simulation, the flooding routing process, and the resulting outflow hydrographs are presented
97
in this paper.
98 99 100
2. Geological and geomorphological characteristics 2.1. Topography of the studying river reach and the Sedongpu Basin
101
The studying river reach for flood routing analysis starts from the Gyalha landslide dam
102
site to the downstream Pasighat Village (Fig. 1), which is the main portion of the Yarlung
103
Tsangpo Grand Canyon. The canyon is the deepest, and longest canyon in the world, which
104
stretches 496 km with an average depth of 5,000 m, passing between the peaks of the Namcha
105
Barwa (7,782 m) and Gyala Peri (7,294 m). The elevation of the canyon entrance at Pei Village
106
in Mainling County is about 2880 m, whereas the elevation at Pasighat Village is around 150
107
m, corresponding to a water level drop of 2,730 m. The deep cut of riverbed is caused by the
108
rapid uplift of the eastern Tibet. The canyon crosses the eastern Himalayan syntaxis, where the
109
continental collision between the Eurasian plate and the India plate is at its strongest.
110
The topography for the flood routing analysis can be extracted from a digital elevation
111
model, such as ALOS World 3D-30m (AW3D30) in this study (Tadono et al., 2014; Takaku et
5
112
al., 2016, downloadable from https://www.eorc.jaxa.jp/ALOS/en/aw3d30/data/index.htm). As
113
shown in Fig. 4, the elevation at the Gyalha landslide dam site is about 2,742 m, while the
114
elevation decreases sharply to about 150 m at Pasighat. Figure 4b shows three typical cross
115
sections A-C along the river reach, representing cross sections at the upper, middle and lower
116
river reach. Slopes of riverbanks become gentler from the upstream to the downstream.
117
On 17 October 2018, the Gyalha landslide occurred on the left bank of the Yarlung
118
Tsangpo in the Sedongpu Basin (29°47’7.20’’ N, 94°55’24’’ E). The Sedongpu Basin has a
119
catchment area of 66.89 km2 and an average elevation of 4,540 m. The highest point in the
120
basin is the Gyala Peri Peak of 7,294 m, while the lowest point is 2,746 m, with a large elevation
121
difference of 4,548 m. The basin strata are mainly composed of Proterozoic Namjagbarwa
122
Group, and the primary bedrocks are Proterozoic marble and gneiss (Pt), which are interbedded
123
and can lose stability under glacier erosion and freeze-thaw weathering (Huang et al., 2007).
124
The Sedongpu Basin has a special steep-gentle-relatively steep ladder terrain from the top to
125
the bottom (Fig. 5); 50% of the basin is steeper than 30° with an average slope of 34.89° (Tong
126
et al., 2018). The middle area of the basin is gentle with slopes smaller than 15°, which is also
127
the area covered with glaciers and moraines. Due to the ladder terrain, the collapse of rocks
128
and glaciers from the upper steep part accumulated in the gentle area, providing massive loose
129
materials to move along the lower relative steep area once triggered by snow melt or intense
130
rainfall. Thus, the basin with massive loose materials is susceptible to large-scale debris flows.
131
There are 12 branches of well-developed glaciers in the upper Sedongpu Basin, and the Gyalha
132
landslide was believed to be induced by large avalanches. The runout materials were a mixture
133
of glacier and rocks, which impacted into the accumulations and moraines in the lower part,
134
formed a large debris flow and dammed the Yarlung Tsangpo (Tong et al., 2018; Liu et al.,
135
2019). The avalanches were triggered under the conditions of steep terrain, broken rock masses,
136
global warming, regional rainfall, glacier ablation, and antecedent earthquake activities.
6
137 138
2.2. Structure of the Gyalha landslide dam
139
The Gyalha landslide was triggered by avalanches from the upper glaciers of the
140
Sedongpu Basin (Fig. 6). The mixtures of ice and disintegrated soils marched approximately
141
4-7 km into Yalung Tsangpo (Hu et al., 2018). The nearby Yigong landslide in April 2000
142
showed the similar feature, which was triggered by snow melt and rainfall and the debris
143
marched about 8 km into the Yigong River (Zhou et al., 2016), entraining the colluvium with
144
snow and ice. The Yigong landslide dam consisted mainly of angular coarse clasts in a matrix
145
of gravel and finer fractions. The lithology was composed of disintegrated granite, marble and
146
gneiss (Shang et al., 2003). Figure 7 shows some cut and scour features left by mass movement
147
through Zhamu Creek Valley and the materials left in the deposition zone of the Gyalha
148
landslide. The landslide deposits at both Yigong and Gyalha are loose fine debris with high
149
erodibility. Therefore, in this study, the soil materials of the Gyalha landslide dam was
150
benchmarked against the Yigong landslide dam, which will be introduced later.
151
The size of the Gyalha landslide dam is interpreted based on satellite images and remote
152
sensing (IMHE, 2018; Hu et al., 2018). The profiles of the longitudinal section and the cross-
153
section of the dam are shown in Fig. 8. Along the river, the crest length of the landslide dam is
154
about 300 m, the upstream and downstream lengths of the dam are estimated to be 720 and 880
155
m, respectively. The elevations at the base and crest of the dam are 2,758 m and 2,837 m,
156
respectively, giving a dam height of 79 m. Other information about the dam can be referred to
157
Table 1.
158 159
3. Hydrological characteristics
160
The Yarlung Tsangpo is an important international river with several major tributaries,
161
including Lhasa River, Nyang River, Nimu Maqu River and Parlung Tsangpo. It is a major
7
162
freshwater resource for China and downstream South Asian countries (Ren et al., 2018; Sun et
163
al., 2019), having an important influence on Asian hydrology and affecting millions of people.
164 165
3.1. Glacier retreat in Tibetan Plateau
166
The Tibetan Plateau is the largest glacier area in the world with a total glacier area of
167
46,640 km2, of which 9,014 km2 is in the Yarlung Tsangpo Basin (Jia et al., 2008). Glaciers
168
are very sensitive to climate change. Figure 9 shows changes of annual mean air temperature
169
of the Yarlung Tsangpo Basin (from the source to the Pasighat) from 1957 to 2004 (Liu et al.,
170
2007), showing a gradually increasing trend. Glaciers in the Tibet-Himalaya region show
171
accelerated retreat with a substantial decrease of snow cover due to global warming
172
(Oerlemans, 2005; Yao et al., 2007; Prasad et al., 2009). Figure 10 shows significant glacier
173
reduction in the region of the Gyalha landslide from 19 Jan. 2017 to 19 Jan. 2018. Melting
174
water from glacier/snow accounts for a considerable proportion of the runoff of the Yarlung
175
Tsangpo. The annual glacier meltwater is 148.8×108 m3, accounting for about 9% of its annual
176
runoff (Jia et al., 2008). Glacier related debris flows are consequences of general glacier retreat
177
due to global warming and exposure of large quantities of unconsolidated, unvegetated, and
178
sometimes ice-cored glacial sediments (Chiarle et al., 2007). These sediments are easily
179
mobilized by floods from heavy precipitation, snowmelt, or glacial lake outbursts, which help
180
to initiate rock avalanches and landslides. Glaciological phenomena as ice/snow avalanches
181
and glacial floods can have significant impacts upon society over a very short time scale
182
(minutes-days) (Richardson and Reynolds, 2000).
183 184
3.2. Precipitation and runoff in the studying region
185
Precipitation in the Yarlung Tsangpo Basin is dominated by the Indian summer
186
monsoon circulation. The monsoonal moisture penetrates into the southern Tibetan Plateau and
8
187
moves upward along the main course of the Yarlung Tsangpo due to the block of the
188
Himalayas.
189
The long-term annual mean runoff in the downstream part of the Yarlung Tsangpo is
190
about 1,500-3,000 mm, which becomes as large as 5,000 mm at Pasighat and decreases sharply
191
at the “horse-show bend” area near the Namcha Barwa (Jia et al., 2008). Figure 11 shows the
192
annual runoff from 1970 to 2013 and the long term intra-annual distribution of the runoff from
193
1960 to 2009 at the Nuxia Hydrological Station located at 76 km upstream of the Gyalha
194
landslide dam site (Liu et al., 2011; Wang et al., 2016b). Although the annual runoff only shows
195
a slightly increasing trend, the monthly runoff within a year varies significantly, showing a
196
notable climate-dependent feature. The runoff is large in the summer and autumn. In the rainy
197
August, the monthly runoff can be as large as 23.4% of the annual runoff.
198 199
3.3. Past barrier lakes near Sedongpu Basin
200
The Sedongpu Basin is a region where rock-avalanches, landslides and debris-flows
201
occur frequently. Based on the literature (Tong et al., 2018; Liu et al., 2019) and satellite
202
images, the blockages of the Yarlung Tsangpo in the Sedongpu Basin in the history are
203
summarized in Table 2. Most of the blockages were caused by ice-avalanches. Some typical
204
satellite images of Sedongpu Basin with different remaining landslide dams are shown in Fig.
205
12. The Gyalha landslide on 17 October 2018 was believed to be affected by the Ms 6.9
206
Nyingchi earthquake at Mainling County on 18 Nov. 2017 (Hu et al., 2018; Tong et al., 2018;
207
Liu et al., 2019). The earthquake triggered at least 529 landslides, including rockslides,
208
avalanches and rock falls (Hu et al., 2018). It induced three large landslide dams, one of which
209
was located at the identical location of the Gyalha landslide dam. These landslide dams
210
breached afterward. The residuals of these dams make the river narrower (Fig. 12). It should
211
be noticed that about two weeks after the Gyalha landslide on 17 October 2018, a new landslide
9
212
occurred at the same location on 29 October, which also blocked the Yarlung Tsangpo and the
213
dam was overtopped shortly afterwards (Fig. 12d).
214 215
3.4. Characteristics of the Gyalha barrier lake in 2018
216
After the formation of the Gyalha landslide dam on 17 October 2018, a barrier lake
217
formed momentarily. The water level and flow rate at the Dexing Hydrological Station (Fig.
218
1) in Medog County, 173 km downstream the dam, dropped from 74.28 m at 8 pm on 16
219
October, to 73 m at 8 am on 17 October, and further to 71.27 m at 2 pm on 17 October. The
220
corresponding flow rate reduced from 3,430 m3/s to 2,620 m3/s from 8 pm 16 October to 8 am
221
17 October, and to 1,580 m3/s by 2 pm on 17 October (Xinhua Net, 2018b). The inundation
222
areas and the lake volumes at different lake water levels are determined using a digital elevation
223
model, AW3D30. Figure 13 shows the relation between the lake surface elevation and the lake
224
volume derived from the digital elevation model. The final lowest elevation of the crest of the
225
dam is 2,837 m, determining a maximum water depth before overtopping of 79 m and a lake
226
capacity of about 4.9×108 m3. After the formation of the barrier lake, the water volume stored
227
in the first hour was about 8×106 m3 (Xinhua Net, 2018c), indicating an initial incoming flow
228
rate of 2,222 m3/s. Thus, the time required to fill the barrier lake was about 2.6 days, implying
229
a short lifespan of the landslide dam. Figure 14 further shows the inundation area of the barrier
230
lake just before breaching. The barrier lake stretches 27 km upstream with an inundation area
231
of around 11 km2. The Gyahla Village was partially inundated, and the access road to the
232
Gyahla Village was completely submerged (Fig. 3b).
233 234
4. Modelling of barrier lake bursting and flood routing process
235
4.1. Principle of dam breaching simulation
236
The dam breaching process can be simulated as a process of erosion of the dam
10
237
materials by the overtopping flow (e.g. Jiang et al., 2017; Zhong et al., 2018; Begam et al.,
238
2018; Zhang et al., 2019). A simple linear equation is widely used to estimate the erosion rate
239
(E) of both cohesive and gravelly landslide dam materials in water flow (e.g. Hanson and
240
Simon, 2001; Briaud, 2008; Shi et al., 2015; Okeke et al., 2016; Zhong et al., 2017 etc.):
241
𝐸 = 𝐾𝑑(𝜏 ― 𝜏𝑐)
242
where Kd is the coefficient of erodibility; τ is the shear stress at the soil/water interface; and τc
243
is the critical shear stress at initiation of soil erosion. Kd and τc can be measured in-situ or
244
estimated using several empirical equations (e.g. Mitchener and Torfts, 1996; Julian and
245
Torres, 2006; Annandale, 2006; Thoman and Niezgoda, 2008; Chang et al., 2011). The shear
246
stress at the soil-water interface can be calculated as:
247
𝜏 = 𝛾𝑤𝑅ℎ𝑆
248
where γw is the unit weight of water (N/m3); Rh is the hydraulic radius (m); S is the energy
249
slope, which is equal to the slope gradient. For a trapezoid breach cross-section, Rh can be
250
obtained as:
251
𝑅ℎ =
252
where H is the elevation of the water surface (m); Z is the elevation of the breach bottom (m);
253
Wb is the breach bottom width (m); and α is the angle of the side slope.
(1)
(2)
(𝐻 ― 𝑍)cos 𝛼 + 𝑊𝑏sin 𝛼 2(𝐻 ― 𝑍) + 𝑊𝑏sin 𝛼
(3)
(𝐻 ― 𝑍)
254
The breach discharge through a trapezoid breach can be calculated based on the broad-
255
crested weir flow assumption which satisfies the condition of landslide dams with a large
256
channel length but a small slope (Singh and Scarlatos, 1988):
257
𝑄𝑏 = 1.7[𝑊𝑏 + (𝐻 ― 𝑍)tan 𝛼](𝐻 ― 𝑍)
258
32
(4)
The hydrodynamics is modelled by solving the continuity equation of the lake together
259
with the breach outflow through a broad-crested weir:
260
𝐴𝑠 𝑑𝑡 = 𝑄𝑖𝑛 ― 𝑄𝑜𝑢𝑡
𝑑𝐻
(5)
11
261
where As is the lake water surface area (m2); Qin and Qout are the inflow and outflow rates of
262
the reservoir.
263
In this study, an erosion-based numerical model DABA (Chang and Zhang, 2010;
264
Zhang et al., 2016) is used to simulate the breaching process. The model has three advantages
265
compared with other models (i.e. DAMBRK, BREACH, BEED): (1) the variations in soil
266
erodibility along depth can be considered; (2) the steepening process of the landslide dam can
267
be taken into account; and (3) the final size of the breach does not need to be assumed.
268
The breach evolution in the DABA model is divided into three phases in the cross
269
section and the longitudinal section. On the cross section (Fig. 15a), at the beginning, the soil
270
at the dam crest can erode easily when overtopped naturally. During the first phase, the width
271
of the breach will not change but the breach depth and breach bottom width increase slightly
272
until the bank slope reaches a critical angle, αc. Then, in phase II, the breach size develops
273
significantly both horizontally and vertically, while keeping the bank slope angle at αc. This
274
process continues until the shear stress by overtopped water flow cannot overcome the soil
275
erosive resistance at the channel bed or the bank side, indicating that the breach process comes
276
to the final phase. Then in phase III, when the soil at the channel bed cannot be eroded, the
277
breach only develops in the horizontal direction, the breach width increases gradually at a
278
constant slope angle αc while the breach erosion depth does not change significantly. When the
279
soil at the two side walls do not erode, the erosion process can only cut in the vertical direction;
280
the breach depth increases but the breach bottom width decreases.
281
On the longitudinal section (Fig. 15b), in the beginning, water flows over the dam crest
282
and cuts the toe of the downstream slope until it reaches a limit angle, βf. Then, the downstream
283
slope keeps eroding while maintaining the slope angle at βf in the phase II. This process
284
continues until the downstream slope eventually meets the upstream slope, which implies the
285
longitudinal breach process goes into phase III. In phase III, the breach of the downstream
12
286
slope develops dramatically with a sharp decrease of the dam crest elevation.
287 288
4.2. Erodibility of the dam materials
289
Referring to Eq. (1), soil erodibility can be described by the coefficient of erodibility
290
(Kd) and the critical shear stress at initiation of soil erosion (τc). τc indicates the potential of
291
erosion in soils, while Kd represents how fast the soil erodes. Kd and τc are functions of basic
292
soil properties and soil state, which can be obtained by laboratory tests, field tests or empirical
293
equations. Kd and τc in this study were obtained based on the empirical equations proposed by
294
Chang et al. (2011), which are particularly for broadly graded landslide deposits:
295
𝐾𝑑 = 20075𝑒4.77𝐶𝑢―0.76
(6)
296
𝜏𝑐 = 6.8(𝑃𝐼)1.68𝑃 ―1.73𝑒 ―0.97
(7)
297
where e is the void ratio; Cu is the coefficient of uniformity; PI is the plasticity index; and P is
298
the fines content with particle size smaller than 0.063 mm. When the fines content is smaller
299
than 10%, τc is recommended to be estimated as (Annandale, 2006):
300
𝜏𝑐 = 3𝑔𝑑50(𝜌𝑠 ― 𝜌𝑤)𝑡𝑎𝑛𝜙
301
where g is the gravitational acceleration; ρs is the soil mass density; ρw is the water mass
302
density; ϕ is the friction angle; and d50 is the mean particle size.
2
(8)
303
The soil materials of the Gyalha landslide are benchmarked against the Yigong
304
landslide materials, both being ice-soil mixtures in the Tibet Plateau. The materials of the
305
Yigong and Gyalha landslides travelled long distances, and were highly disintegrated. Thus,
306
the distributions of Kd and τc with depth for the upper part of the Gyalha landslide dam can be
307
referred to those for the materials of Yigong landslide dam, while the lower part that was not
308
fully disintegrated is referred to the Tangjiashan landslide dam (Shi et al., 2015). The soil grain
309
size distributions at the deposition zone of the Yigong landslide (Hu et al., 2015; Kang et al.,
310
2017; Wang et al., 2017; Wang et al., 2016a) are summarized in Fig. 16a. For the Gyalha 13
311
landslide dam, Liu et al. (2019) reported a soil-rock ratio of about 8:2, and Cai (2019) found
312
soil-rock ratio values in the range of 6:4 - 7:3, both of which are in the range of 10:0 - 3.5:6.5
313
for the benchmark Yigong materials (Fig. 16a). Here the boundary between soil and rock is
314
defined as 10 mm (Cai et al., 2019). The bulk density of the Yigong landslide deposits is 1.845
315
g/cm3, the specific gravity is 2.718, and the friction angle is about 37º (Hu et al., 2009, 2015;
316
Wang et al., 2016a). Using Eqs. (6)-(8), the variations of Kd and τc for the Gyalha landslide
317
dam with depth are shown in Figs. 16b and c.
318
Furthermore, a sensitivity analysis is performed to evaluate the influence of soil
319
erodibility on the overtopping process. The analysis is conducted by (1) changing Kd by ±10%
320
and ±20% while maintaining τc as a constant; and (2) changing τc by ±10% and ±20% while
321
keeping Kd unchanged. The hydrographs of the overtopping process are quite similar when Kd
322
is changed (Fig. 17a), with variations of the peak flow rate up to 16% and a maximum time
323
difference of 48 min. The influence of τc is very small as shown in Fig. 17b. With the
324
development of overtopping, the shear stress τ at the soil-water interface is much larger than
325
the initiation critical shear stress τc; thus the effect of τc is not significant.
326 327
4.3.
Principle of flood routing simulation
328
The unsteady flooding routing process can be simulated based on the one-dimensional
329
Saint-Venant equations for the conservation of mass and momentum (e.g. Horritt and Bates,
330
2002; Oubanas et al., 2018; Lei et al., 2019):
331
∂𝑄 ∂𝑥
+
∂𝐴 ∂𝑡
332
∂𝑄 ∂𝑡
+
∂(𝑄2/𝐴) ∂𝑥
333
where Q is the flow rate; x is the distance along the channel bed; A is the cross-sectional area;
334
t is time; g is the gravitational acceleration; h is the flow depth; S0 is the bed slope; and Sf is the
335
friction slope and can be calculated using Chezy and Manning’s formulas as:
(9)
=0 +𝑔𝐴
(∂ℎ∂𝑥-𝑆0 + 𝑆𝑓) = 0
(10)
14
𝑄2𝑛2
336
𝑆𝑓 =
337
in which n is Manning’s coefficient of the roughness of the cross-section; and R is the hydraulic
338
radius of the cross-section. For mountain streams with steep banks, if the riverbed is primarily
339
covered with cobbles with large boulders but no vegetation, Manning’s n value ranges from
340
0.04 to 0.07 (Chow, 1959). For the riverbanks with medium intensity of vegetation and
341
constructed materials, the recommended n values are between 0.06 and 0.14 (USACE, 2016).
342
In this study, the n values are assumed to be 0.06 and 0.12 for the channel bed and riverbank,
343
respectively.
𝑅
43 2
𝐴
(11)
344
The study river reach is from the Gyalha landslide dam site to the downstream Pasighat
345
Village (Fig. 1), with a river length of 460 km. Based on the aforementioned digital elevation
346
model, the river is divided into two parts according to the rive gradient, as shown in Fig. 4a. In
347
the first river reach, the water elevation changes from 2742 m to 1800 m with a river length of
348
110 km and a large water level drop of 8.6 m per kilometer. Thus, the cross sections in this
349
reach are set densely at a small interval of 0.05 km. In the second river reach, the elevation
350
changes from 1800 m to 150 m in a river length of 370 km, corresponding to a smaller water
351
level drop of 4.5 m per kilometer. The cross sections are then set at a larger interval of 1 km.
352
A total of 2764 cross sections along the river are finally determined.
353
Based on the breaching hydrograph at the dam site given by DABA and the input
354
topography information of the studying river reach, the flooding routing process in this study
355
is simulated using a one-dimensional hydrodynamic model HEC-RAS (USACE, 2016).
356 357
5. Results of barrier lake breaching and flood routing analyses
358
5.1. Barrier lake bursting on 19 October 2018
359
The Gyalha landslide dam was overtopped naturally at 13:30 on 19 October 2018. The
360
inflow rate into the barrier lake was 2222 m3/s by then. A small initial channel with a gentle 15
361
slope angle is assumed in the DABA model as shown in Table 1. The hydrograph at the dam
362
site, simulated by using DABA, is presented in Fig. 18. The simulated peak discharge at the
363
dam site is 31,685 m3/s, which is almost the same as the reported value of 32,000 m3/s (Jin,
364
2019). Based on the flood routing results, the time at the peak outflow at the Gyalha landslide
365
dam site is determined at 19:11 on 19 October. Figure 18 also illustrates the attenuation of the
366
peak discharge along the river. At Dexing Hydrological Station, the predicted peak discharge
367
is 25,741 m3/s, occurring at 23:37 on 19 October, whereas the reported maximum discharge is
368
23,400 m3/s at 23:40 on 19 October (China Daily, 2018). The simulated peak flow rate
369
decreases to about 21,216 m3/s at Tuting by 3:04 on 20 October, and to 11,235 m3/s at Pasighat
370
by 18:32 on 20 October. Moreover, at 7:00 on 20 October (i.e. about 17.5 hours after the dam
371
breach), the discharge at Dexing Hydrological Station dropped significantly to 6230 m3/s
372
(CCTV News, 2018b), while the predicted discharge at that time is 6635 m3/s, with a small
373
difference of 6.5%. By 7:00 on 20 October, the total water volume passing Dexing was 550
374
million m3. After deducting 40 million m3 from other tributaries, approximately 510 million
375
m3 of water has been drained (CCTV News, 2018b). The reservoir level was lowered by about
376
56 m and the threat of the flooding was dismissed.
377
Figure 19 further shows changes of the relative water level at Tuting. The water level
378
started to decrease at about 11:30 on 17 October, and the river flow diminished at 2:30 on 18
379
October with limited flow from other tributaries (i.e. about 21.5 hours after the formation of
380
the Gyalha landslide dam) (SANDRP, 2018). It dramatically rose up to about 15 m at 2:30 on
381
20 October and returned to the base flow on the early morning of 21 October. The predicted
382
maximum water level rise at Tuting is 16.4 m, occurring at 3:04 am on 20 October, which is
383
close to the observations.
384 385
16
386
5.2. Analysis of dam breaching scenarios with larger river inflow rates
387
The runoff of Yarlung Tsangpo shows a strong climate-dependent feature and mainly
388
concentrates in the summer (Jia et al., 2008; Liu et al., 2011). Summer is also the time for
389
glacier melting (Jansson et al., 2003). There are a large number of glaciers in the lower Yarlung
390
Tsangpo Basin where the studying area is located. Hence the meltwater from the glaciers in
391
summer is of particular importance (Yao et al., 2010; Ren et al., 2018). If ice avalanches
392
develop in the glaciers on the high mountains of Sedongpu, and the landslide debris blocks the
393
Yarlung Tsangpo at a time of large runoff, the damming may cause more catastrophic
394
consequences. Therefore, it is critical to explore impacts of landslide lake bursts under possibly
395
large runoff conditions, which directly affect the duration and development of the dam
396
breaching process, the final breach size, and the subsequent flood routing process.
397
Figure 20 presents the runoff of the latest large flood in the Yarlung Tsangpo in 1998
398
and the intra-annual distribution of runoff from 1960 to 2009 at Nuxia Hydrological Station
399
(Liu, 1999; Liu et al., 2011). The maximum flow rate in 1998 is 13,487 m3/s, which is 2.6 times
400
the maximum monthly mean flow rate of 5119 m3/s in August. In this study, the maximum
401
flow rate of 13,487 m3/s and the maximum monthly mean flow rate of 5119 m3/s are used to
402
represent the possibly large runoffs of the downstream Yarlung Tsangpo. Table 3 shows results
403
of the dam breaching simulations. Taking the Gyalha landslide dam as a benchmark, with a
404
larger inflow rate, the time of overtopping decreases dramatically. The time for evacuating
405
downstream residents also shortens from 62.5 h to 27.1 h, and even to only 10.3 h when the
406
inflow rate increases from 2222 m3/s to 5119 m3/s, and further to 13,487 m3/s. Hence the dam
407
breaching risk will intensify when the river blockage occurs in the rainy season.
408
When the dam is overtopped naturally, the subsequent peak flow increases obviously
409
with the increasing inflow rate. Comparing the conditions of the Gyalha landslide and the
410
maximum flood flow in 1998, the dam-breaching peak flow rate is nearly doubled when the
17
411
inflow is increased by five times. The total breach duration and the erosion degree of the dam
412
body also differ greatly under the three different inflow conditions. With a larger inflow rate,
413
the breach initiation time becomes shorter, the materials erode more rapidly, and the dam
414
breaching process develops very quickly. On the cross section, when the initial runoff is larger,
415
the breach process goes into phase II and phase III (Fig. 15a) more quickly, causing a
416
significant enlargement of breach size in both horizontal and vertical directions. On the
417
longitudinal section, the breaching process will go into phase III (Fig. 15b) sooner, the erosion
418
of the downstream slope will develop more dramatically, and the dam crest elevation will lower
419
sharply. It can be found that in the case of the maximum runoff of flood in 1998, the landslide
420
dam can be completely eroded in 13.8 h.
421
Figure 21 shows results of the flood routing process along the river. The trends of the
422
attenuation of peak discharge are quite similar for the three inflow conditions. However, the
423
magnitude and arrival time of the peak discharge along the river are quite different. Taking the
424
Gyalha landslide dam as a benchmark, the peak discharge along the river is 23% larger at the
425
maximum monthly runoff condition and increases dramatically by 83% at the 1998 flood flow
426
condition. The corresponding average arrival times of the peak flow at Dexing, Tuting and
427
Pasighat are about 2.8 h and 1.4 h earlier, respectively, which shortens the time for emergency
428
actions. Thus, it is suggested to install a monitoring and warning system for the Sedongpu
429
Basin to assist emergency management.
430 431
6. Summary and conclusions
432
This paper reports the formation and breach of the Gyalha landslide dam on Yarlung
433
Tsangpo in 2018, and a comprehensive analysis of a typical mountain landslide-barrier lake-
434
dam breach-flooding disaster chain. The dam breaching hydrograph is simulated using an
435
overtopping-erosion based dam breaching model, DABA, and the flood routing in the 460 km-
18
436
long Grand Canyon is analyzed using one-dimensional Saint-Venant equations. The simulated
437
peak discharges at the dam site and Dexing are 31,685 and 25,741 m3/s, respectively, which
438
agree well with observed values of 32,000 and 23,400 m3/s, respectively.
439
Investigations of the dam breach and flood routing process under two larger plausible
440
inflow conditions are further conducted: the maximum flow of the flood in 1998 and the annual
441
maximum mean monthly runoff. Under a larger inflow, the duration of dam breaching
442
decreases significantly and the scour and erosion of the dam body become more rapid. When
443
the inflow rate is increased by five times, the peak discharge can be doubled.
444
The Sedongpu Basin has 12 well developed glaciers, where landslides induced by ice
445
melting occurred frequently and blocked the Yarlung Tsangpo in recent years. Once an ice-soil
446
mixture dam breaks, the flood routing process can be very fast in the Yarlung Tsangpo Grand
447
Canyon. Thus, a long-time monitoring and warning system is suggested to assist risk
448
management in the region.
449 450
Acknowledgements
451
This research is supported by the National Key Technologies Research and
452
Development Program of the Ministry of Science and Technology of China (Project No.
453
2018YFC1508600).
454 455
CRediT author statement
456
Chen Chen: Data curation, Formal analysis, Writing- Original draft preparation Limin
457
Zhang: Conceptualization, Methodology, Funding acquisition, Supervision, Writing - review
458
& editing Te Xiao: Software, Validation, Investigation Jian He: Software, Validation,
459
Investigation.
460
19
461
Declaration of competing interest
462
None.
463 464
References
465
Annandale, G.W., 2006. Scour Technology - Mechanics and Engineering Practice. McGraw-
466 467 468
Hill, New York. Begam, S., Sen, D., Dey, S., 2018. Moraine dam breach and glacial lake outburst flood generation by physical and numerical models. J. Hydrol. 563, 694-710.
469
Briaud, J.L., 2008. Case histories in soil and rock erosion: Woodrow Wilson Bridge, Brazos
470
River Meander, Normandy Cliffs, and New Orleans Levees. J. Geotech. Geoenviron. Eng.
471
134 (10), 1425–1447.
472
Cai, Y.J., 2019. The two landslide dams in Jinshajiang and Yarlung Tsangpo 2018–the
473
formation, burst and secondary disasters. In 2019 Annual Conference of Chinese Hydraulic
474
Engineering Society. Yichang, Hubei, China.
475
Cai, Y.J., Luan, Y.S., Yang, Q.G., Xu, F.X., Zhang, S.J., Shi, Y., Yi, D.L.Z., 2019. Study on
476
structural morphology and dam-break characteristics of Baige barrier dam on Jinsha River.
477
Yangtze River 50 (3), 15-22.
478 479 480 481
CCTV News, 2018a. http://m.news.cctv.com/2018/10/19/ARTIPToqmhpUYJAmKVoKBUp U181019.shtml. CCTV News, 2018b. http://m.news.cctv.com/2018/10/20/ARTIjyNKpGkVwHdXQd5JjZSY1 81020.shtml.
482
Chang, D.S., Zhang, L.M., 2010. Simulation of the erosion process of landslide dams due to
483
overtopping considering variations in soil erodibility along depth. Nat. Hazards Earth Syst.
484
Sci. 10 (4), 933-946.
485
Chang, D.S., Zhang, L.M., Xu, Y., Huang, R.Q., 2011. Field testing of erodibility of two
20
486
landslide dams triggered by the 12 May Wenchuan earthquake. Landslides 8 (3), 321-332.
487
Chiarle, M., Iannotti, S., Mortara, G., Deline, P., 2007. Recent debris flow occurrences
488
associated with glaciers in the Alps. Global Planet. Change 56 (1-2), 123-136.
489
China Daily, 2018. http://cn.chinadaily.com.cn/2018-10/20/content_37107081.htm.
490
Chow V.T., 1959. Open Channel Hydraulics, McGraw-Hill, New York.
491
Hanson, G.J., Simon, A., 2001. Erodibility of cohesive streambeds in the loess area of the
492 493 494
midwestern USA. Hydrol. Processes 15 (1), 23-38. Horritt, M.S., Bates, P.D., 2002. Evaluation of 1D and 2D numerical models for predicting river flood inundation. J. Hydrol. 268 (1), 87-99.
495
Hu, K., Zhang, X., Tang, J., Liu, W., 2018. Potential danger of dammed lakes induced by the
496
2017 Ms 6.9 Mifflin earthquake in the Tsangpo gorge. In 5th International Conference
497
Debris Flows: Disasters, Risk, Forecast, Protection. 97-104. Tbilisi: Publishing House
498
“Universal”.
499
Hu, M.J., Cheng, Q.G., Wang, F.W., 2009. Experimental study on formation of Yigong long-
500
distance high-speed landslide. Chinese Journal of Rock Mechanics and Engineering 28 (1),
501
138-143.
502 503
Hu, M.J., Pan, H.L., Zhu, C.Q., Wang, F.W., 2015. High-speed ring shear tests to study the motion and acceleration processes of the Yigong landslide. J. Mt. Sci. 12(6), 1534-1541.
504
Huang, J.X., Xu, Z.X., Gong, T.L., 2007. Characteristics and driving factors of the runoff
505
variations in the Yarlung Zangbo River. Chinese Journal of Hydrology 27 (5), 31-35.
506
Institute of Mountain Hazards and Environment (IMHE), 2018. http://www.imde.ac.cn/yjld_2
507 508 509 510
015/201810/t20181021_5146490.html?from=single%20message&isappinstalled=0. Jansson, P., Hock, R., Schneider, T., 2003. The concept of glacier storage: a review. J. Hydrol. 282(1-4), 116-129. Jia, J.W., Lv, S.Y., Wang, Z.X., 2008. Characteristics analysis of water resources in the
21
511 512 513 514 515 516 517 518 519
Yarlung Tsangpo Basin. Yangtze River 39 (17), 71-72. Jian, X.G., Cui, P., Chen, H.Y., Guo, Y.Y., 2017. Formation conditions of outburst debris flow triggered by overtopped natural dam failure. Landslides 14 (3), 821-831. Jin, X.P., 2019. Reviews and reflections on emergency response countermeasures for barrier lakes in Jinsha River and Yarlung Zangbo River. Yangtze River 50 (3), 5-9. Julian, J.P., Torres, R., 2006. Hydraulic erosion of cohesive riverbanks. Geomorphology 76 (12), 193-206. Kang, C., Chan, D., Su, F., Cui, P., 2017. Runout and entrainment analysis of an extremely large rock avalanche—a case study of Yigong, Tibet, China. Landslides 14 (1), 123-139.
520
Lei, X., Tian, Y., Zhang, Z., Wang, L., Xiang, X., Wang, H., 2019. Correction of pumping
521
station parameters in a one-dimensional hydrodynamic model using the Ensemble Kalman
522
filter. J. Hydrol. 568, 108-118.
523
Liu, C.Z., Lu, J.T., Tong, L.Q., Chen, H.Q., Liu, Q.Q., Xiao, R.H., Tu, J.N., 2019. Research
524
on glacial/rock fall-landslide-debris flows in Sedongpu basin along Yarlung Zangbo River
525
in Tibet. Geology in China 46 (2), 219-234.
526
Liu, J., Yao, Z.J., Chen, C.Y., 2007. Evolution trend and causation analysis of the runoff
527
evolution in the Yarlung Zangbo River Basin. Journal of Natural Resources 22 (3), 471-
528
477.
529 530
Liu, T.C., 1999. Hydrological characteristics of Yalungzangbo River. Acta Geographica Sinica S1, 157-164.
531
Liu, Y.B., Liang, C., 2011. Zoning for effective utilization of agricultural water resources based
532
on principal component analysis and fuzzy C-means clustering algorithm. Chinese Journal
533
of Hydrology 31 (5), 76-79.
534 535
Lu, J.T., Wang, Z.H., Zhou, C.H., 2003. Discussion on the occurrence of Yigong Landslide in Tibet. Earth Science-Journal of China University of Geosciences 28 (1), 107-110.
22
536
Mitchener, H., Torfs, H., 1996. Erosion of mud/sand mixtures. Coastal Eng. 29 (1-2), 1-25.
537
New Delhi Television (NDTV), 2018. https://www.ndtv.com/india-news/arunachal-pradesh-
538 539 540 541
on-alert-after-tibet-landslide-creates-another-artificial-dam-1940111. Oerlemans, J., 2005. Extracting a climate signal from 169 glacier records. Science 308 (5722), 675-677. Okeke, A.C.U., Wang, F., 2016. Hydromechanical constraints on piping failure of landslide
542
dams:
an
experimental
investigation.
543
doi:10.1186/s40677-016-0038-9.
Geoenvironmental
Disasters
3(1).
544
Oubanas, H., Gejadze, I., Malaterre, P.O., Mercier, F., 2018. River discharge estimation from
545
synthetic SWOT-type observations using variational data assimilation and the full Saint-
546
Venant hydraulic model. J. Hydrol. 559, 638-647.
547
Peng, M., Zhang, L.M. 2013a. Dynamic decision making for dam-break emergency
548
management-Part 1: Theoretical framework. Nat. Hazards Earth Syst. Sci. 13(2), 425-437.
549
Peng, M., Zhang, L.M. 2013b. Dynamic decision making for dam-break emergency
550
management-Part 2: Application to Tangjiashan landslide dam failure. Nat. Hazards Earth
551
Syst. Sci. 13(2), 439-454.
552
Prasad, A.K., Yang, K. H. S., el-Askary, H., Kafatos, M., 2009. Melting of major Glaciers in
553
the western Himalayas: evidence of climatic changes from long term MSU derived
554
tropospheric temperature trend (1979-2008). Ann. Geophys. 27 (12), 4505-4519.
555
Ren, W., Yao, T., Xie, S., 2018. Stable isotopic composition reveals the spatial and temporal
556
dynamics of discharge in the large river of Yarlungzangbo in the Tibetan Plateau. Sci. Total
557
Environ. 625, 373-381.
558 559 560
Richardson, S.D., Reynolds, J.M., 2000. An overview of glacial hazards in the Himalayas. Quat. Int. 65, 31-47. South Asia Network on Dams, Rivers and People (SANDRP), 2018. https://sandrp.in/2018/1
23
561
0/19/landslide-dam-on-tsangpo-creates-flood-disaster-risk-for-siang/.
562
Shang, Y., Yang, Z., Li, L., Liu, D. A., Liao, Q., Wang, Y., 2003. A super-large landslide in
563
Tibet in 2000: background, occurrence, disaster, and origin. Geomorphology 54 (3-4), 225-
564
243.
565
Shi, Z. M., Guan, S. G., Peng, M., Zhang, L. M., Zhu, Y., Cai, Q. P., 2015. Cascading breaching
566
of the Tangjiashan landslide dam and two smaller downstream landslide dams. Eng. Geol.
567
193, 445-458.
568 569
Singh, V.P., Scarlatos, P.D., 1988. Analysis of gradual earth-dam failure. J. Hydraul. Eng. 114 (1), 21-42.
570
Sun, W.C., Wang, Y.Y., Fu, Y.H., Xue, B.L., Wang, G.Q., Yu, J.S., Zuo, D.P., Xu, Z.X., 2019.
571
Spatial heterogeneity of changes in vegetation growth and their driving forces based on
572
satellite observations of the Yarlung Zangbo River Basin in the Tibetan Plateau. J. Hydrol.
573
574, 324-332.
574
Tadono, T., Ishida, H., Oda, F., Naito, S., Minakawa, K., Iwamoto, H., 2014. Precise global
575
DEM generation by ALOS PRISM. ISPRS Annals of the Photogrammetry, Remote
576
Sensing and Spatial Information Sciences 2 (4), 71-76.
577
Takaku, J., Tadono, T., Tsutsui, K., Ichikawa, M., 2016. Validation of ‘AW3D’ global DSM
578
generated from ALOS PRISM. ISPRS Annals of the Photogrammetry, Remote Sensing
579
and Spatial Information Sciences 3 (4), 25-31.
580
Thoman, R.W., Niezgoda, S.L., 2008. Determining erodibility, critical shear stress, and
581
allowable discharge estimates for cohesive channels: case study in the Power River Basin
582
of Wyoming. J. Hydraul. Eng. 134 (12), 1677-1687.
583
Tong, L.Q., Tu, J.N., Pei, L.X., Guo, Z.C., Zheng, X.W., Fan, J.H., Zhong, X., Liu, C.L., Wang,
584
S.S., He, P., Chen, H., 2018. Preliminary discussion of the frequently debris flow events in
585
Sedongpu Basin at Gyalaperi peak, Yarlung Zangbo River. Journal of Engineering
24
586
Geology 26 (6), 1552-1561.
587
United States Army Corps of Engineers (USACE), 2016. HEC-RAS, River Analysis System -
588
Hydraulic Reference Manual (Version 5.0). Hydrologic Engineering Centre, Davis,
589
California.
590
Wang, L., Chen, Z., Wang, N., Sun, P., Yu, S., Li, S., Du, X., 2016a. Modeling lateral
591
enlargement in dam breaches using slope stability analysis based on circular slip mode.
592
Eng. Geol. 209, 70-81.
593 594
Wang, X., Qin, G.H., Li, H.X., 2016b. Analysis on characteristics and variation trend of annual runoff of mainstream of Yarlung Tsangpo River. Yangtze River 47 (1), 23-26.
595
Wang, Y.F., Dong, J.J., Cheng, Q.G., 2017. Velocity - dependent frictional weakening of large
596
rock avalanche basal facies: Implications for rock avalanche hypermobility? J. Geophys.
597
Res. B: Solid Earth 122 (3), 1648-1676.
598
Xinhua Net, 2018a. http://www.xinhuanet.com/local/2018-10/19/c_1123584572.htm.
599
Xinhua Net, 2018b. http://www.xinhuanet.com/politics/2018-10/17/c_1123573907.htm.
600
Xinhua Net, 2018c. http://tibet.news.cn/ywjj/2018-10/18/c_129974498.htm.
601
Yao, T., Pu, J., Lu, A., Wang, Y., Yu, W., 2007. Recent glacial retreat and its impact on
602
hydrological processes on the Tibetan Plateau, China, and surrounding regions. Arct.
603
Antarct. Alp. Res. 39 (4), 642-650.
604
Yao, T.D., Li, Z.G., Yang, W., Guo, X.J., Zhu, L.P., Kang, S.C., Wu, Y.H., Yu, W.S., 2010.
605
Glacial distribution and mass balance in the Yarlung Zangbo River and its influence on
606
lakes. Chin. Sci. Bull. 55 (20), 2072-2078.
607 608 609 610
Zhang, L.M., Peng, M., Chang, D.S., Xu, Y., 2016. Dam Failure Mechanisms and Risk Assessment. Singapore: John Wiley & Sons. Zhang, L.M., Xiao, T., He, J., Chen, C., 2019. Erosion-based breaching analysis of Baige landslide
dams
on
the
Jinsha
River,
China,
in
2018.
Landslides
25
611 612 613
https://doi.org/10.1007/s10346-019-01247-y. Zhong, Q.M., Chen, S.S., Mei, S.A., Cao, W., 2018. Numerical simulation of landslide dam breaching due to overtopping. Landslides 15 (6), 1183-1192.
614
Zhou, J.W., Cui, P., Hao, M. H., 2016. Comprehensive analyses of the initiation and
615
entrainment processes of the 2000 Yigong catastrophic landslide in Tibet, China.
616
Landslides 13 (1), 39-54.
617 618
List of table captions
619
Table 1 Input parameters for the analysis of dam breaching and flood routing
620
Table 2 Blockage of the Yarlung Tsangpo in the Sedongpu Basin in the recent history
621
Table 3 Results of dam breaching and flood routing analysis
622 623
List of figure captions
624
Fig. 1. Location of the Gyalha landslide dam and the river reach for flood routing simulation.
625
Fig. 2. The Gyalha landslide dam: (a) formation; and (b) after breaching (Photo credit: Xinhua
626 627 628 629 630
Net). Fig. 3. Inundation of the Gyalha village: (a) before and (b) after formation of the Gyalha landslide dam (Photo credit: Xinhua Net). Fig. 4. Profile of the studying river reach: (a) drop of water surface elevation with distance from the dam site; and (b) three typical cross-sections.
631
Fig. 5. Profile of the Sedongpu basin (modified from Tong et al., 2018).
632
Fig. 6. A planar map of the Gyalha landslide dam.
633
Fig. 7. Materials at landslide dams: (a) Yigong landslide dam (Photo credit: Xu et al., 2012);
634 635
and (b) Gyalha landslide dam (Photo credit: Xinhua Net). Fig. 8. Cross sections of the Gyalha landslide dam: (a) along the river; (b) across the river.
26
636 637 638 639 640 641 642
Fig. 9. Changes of annual mean air temperature of the Yarlung Tsangpo basin from 1957 to 2004. Fig. 10. Glacier reduction around the Gyalha landslide dam: (a) on 19 Jan. 2017; and (b) on 19 Jan. 2018 (Source: Sentinel-2). Fig. 11. Annual average flow rate from 1970 to 2013 and the long term intra-annual distribution of runoff from 1960 to 2009 at Nuxia Hydrological Station. Fig. 12. Examples of landslides at the Sedongpu Basin: (a) on 18 Feb. 2017; (b) on 05 Nov.
643
2017; (c) on 08 June 2018; and (d) on 31 Oct. 2018 (Source: Sentinel-2).
644
Fig. 13. Lake storage capacity-water surface elevation curve for the barrier lake.
645
Fig. 14. The barrier lake behind the landslide dam.
646
Fig. 15. Breach enlargement process: (a) cross section; and (b) longitudinal section.
647
Fig. 16. Soil erodibility parameters: (a) particle size distributions of materials from Gyalha
648
landslide dam and benchmark materials at Yigong landslide dam; (b) assumed Kd at
649
Gyalha landslide dam; and (c) assumed τc at Gyalha landslide dam.
650
Fig. 17. Sensitivity of soil erodibility to the overtopping process: (a) Kd is changed by ±10%
651
and ±20% while τc remains the same; (b) τc is changed by ±10% and ±20% while Kd is
652
kept the same.
653
Fig. 18. Observed and simulated hydrographs along the Yarlung Tsangpo.
654
Fig. 19. Observed and simulated rises of the relative water level at Tuting.
655
Fig. 20. The long period mean monthly runoff from 1960 to 2009 and the 1998 flood at the
656 657 658
Nuxia Hydrological Station. Fig. 21. Simulated flooding process along the Yarlung Tsangpo with different runoffs: (a) hydrographs; and (b) peak discharge with distance.
659
27
660 661 662
Fig. 1. Location of the Gyalha landslide dam and the river reach for flood routing simulation.
663
28
(a)
664 (b)
665 666 667 668
Fig. 2. The Gyalha landslide dam: (a) formation; and (b) after breaching (Photo credit: Xinhua Net).
29
(a)
669 (b)
670 671 672 673
Fig. 3. Inundation of the Gyalha Village: (a) before and (b) after formation of the Gyalha landslide dam (Photo credit: Xinhua Net).
674
30
(a)
675 676
Relative height (m)
200 0 400
B
200 0 400 200 0 -900
677
(b)
A
C -600
-300
0
300
600
900
1200
Distance from the river center (m)
678 679 680 681
Fig. 4. Profile of the studying river reach: (a) drop of water surface elevation with distance from the dam site; and (b) three typical cross-sections.
682
31
7000 6600
Rocks andofglaciers Collapse rocks and are easy to collaspe glaciers in the upper along the upper steep part steep part
Elevation (m) (m) Elevation
6200 5800 5400 5000
Steep
4600
Accumulation massive Massive looseof materials accumulat in theinmiddle loose materials the middlegentle gentlearea area
Steep
4200
Glacier-debris Glacier-debris flow flowison triggered and slide the relatively steepalong the relative steep terrain terrain
3800 3400 Proterozoic marble and gneiss (Pt)
3000
Gentle Gentle Relative-steep Relatively-steep
2600 0
683
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 Distance (m)
684 685 686
Fig. 5. Profile of the Sedongpu Basin (modified from Tong et al., 2018).
687
32
Source zone
Passage and entrainment zone
Gyalha landslide dam
688 689 690 691
Fig. 6. A planar map of the Gyalha landslide dam.
692
33
(a)
693 (b)
694 695 696 697 698
Fig. 7. Materials at landslide dams: (a) Yigong landslide dam (Photo credit: Xu et al., 2012); and (b) Gyalha landslide dam (Photo credit: Xinhua Net).
699
34
2900
(a) Elevation (m)
2850
300 m
Yarlung Tsangpo
2800
2837 m
79 m River bed
2750
Gyalha landslide dam
2758 m
1.9 km
2700 0
500
700
1000 Distance (m)
1500
2000
3200
(b) Elevation (m)
3100
Namcha Barwa
3000
Sedongpu Basin
2900
2837 m
2800
79 m
Proterozoic marble and gneiss
2700 0
701
Gyalha landslide dam
500
1000
1500
Zhibai HPGbearing gneiss 2000
2500
Distance (m)
702 703
Fig. 8. Cross sections of the Gyalha landslide dam: (a) along the river; (b) across the river.
704
35
Annual mean temperature (°C)
9
8
7
6
5 1957
705
1967
Year
Average mean temperature (°C)
1961-1970 1971-1980 1981-1990 1991-2000 2001-2004
7.2 7.3 7.44 7.78 8.02
1977
1987
1997
2007
Year
706 707 708
Fig. 9. Changes of annual mean air temperature of the Yarlung Tsangpo Basin from 1957 to 2004.
709
36
2017/01/19
(a)
Gyalha landside dam
710 711
2018/01/19
(b)
Gyalha landside dam
712 713
37
714 715
Fig. 10. Glacier reduction around the Gyalha landslide dam: (a) on 19 Jan. 2017; and (b) on 19 Jan. 2018 (Source: Sentinel-2).
38
Month 1
2
3
4
5
6
7
8
9
10
11
12 50
Annual average flow rate (m3/s)
Intra-annual distribution of runoff
2500
Annual average flow rate
40
2000 30 1500 20 1000 10
500
0
0 1970
716 717 718
Intra-annual distribution of runoff (%)
3000
1975
1980
1985
1990
1995
2000
2005
2010
Year
Fig. 11. Annual average flow rate from 1970 to 2013 and the long term intra-annual distribution of runoff from 1960 to 2009 at Nuxia Hydrological Station.
719
39
Sedongpu Valley Sedongpu Valley
(a)
(b)
720
Sedongpu Valley Sedongpu Valley
(c)
(d)
721 722 723 724 725
Fig. 12. Examples of landslides at the Sedongpu Basin: (a) on 18 Feb. 2017; (b) on 05 Nov. 2017; (c) on 08 June 2018; and (d) on 31 Oct. 2018 (Source: Sentinel-2).
726
40
Lake surface elevation (m)
2910 2880 Crest of the Gyalha landslide dam
2850 2820 2790
River bed 2760 0 727
2
4 6 Storage capacity (108 m3)
8
10
728 729
Fig. 13. Lake storage capacity-water surface elevation curve for the barrier lake.
730
41
N
94°58'0"E
29°42'0"N
29°44'0"N
Gyalha landslide dam
29°38'0"N
Service Layer Credits: Source: Esri, DigitalGlobe, GeoEye, Earthstar Geographics, CNES/Airbus DS, USDA, USGS, AeroGRID,
0 731
2.5 94°54'0"E
5
29°38'0"N
29°40'0"N
29°40'0"N
Gyalha Village
29°44'0"N
94°56'0"E
29°42'0"N
94°54'0"E
km 94°56'0"E
94°58'0"E
732 733
Fig. 14. The barrier lake behind the landslide dam.
734
42
Cross-section Phase Ⅰ
αc αc
735
αc
Cross-section Phase Ⅲ
αc
(a)
Cross-section Phase Ⅲ
Cross-section Phase Ⅱ
736
βf
737
(b)
βf
βf
Longitudinal-section Phase Ⅲ
βf
βf
βf
βf
Longitudinal-section Phase Ⅱ
Longitudinal-section Phase Ⅰ
738 739 740
Fig. 15. Breach enlargement process: (a) cross section; and (b) longitudinal section.
741
43
Percentage passing by weight (%)
100
Materials of Gyalha from Liu et al., 2019 Materials of Gylha from Cai, 2019 Materials of Yigong from Hu et al. 2015 Materials of Yigong from Wang et al. 2017 Materials of Yigong from Wang et al. 2016 Materials of Yigong from Kang et al. 2016
80
60
40
20
(a) 0 0.01
742
0.1
1 Particle diameter (mm)
10
100
743 Coefficient of erodibility, Kd (mm3/N-s) 1 10 100 1000 10000
Critical erosive shear stress, Tc (Pa) 1 10 100 1000 10000
0
0
(b)
(c) 20 Depth (m)
Depth (m)
20
40
60
80
744
100
40
60 Gyalha (field) from Cai, 2019 Gyalha (field) from Liu et al., 2019 Gyalha used in the study
80
100
Gyalha (field) from Cai, 2019 Gyalha (field) from Liu et al., 2019 Gyalha used in the study
745
44
746 747 748
Fig. 16. Soil erodibility parameters: (a) particle size distributions of materials from Gyalha landslide dam and benchmark materials at Yigong landslide dam; (b) assumed Kd at Gyalha landslide dam; and (c) assumed τc at Gyalha landslide dam. 4
The Gyalha landslide dam Kkd d increased by 10% Kkd d increased by 20% Kkd d decreased by 10% Kkd decreased by 20% d
Discharge (104 m3/s)
(a) 3
2
1
0 13:30 19 Oct
13:30 20 Oct
01:30 20 Oct
01:30 21 Oct
Time
749 4
The Gyalha landslide dam ττc c increased by 10% ττc c increased by 20% ττc c decreased by 10%
Discharge (104 m3/s)
(b) 3
ττc c decreased by 20% 2
1
0 13:30 19 Oct
750
13:30 20 Oct
01:30 20 Oct
01:30 21 Oct
Time
751 752 753 754
Fig. 17. Sensitivity of soil erodibility to the overtopping process: (a) Kd is changed by ±10% and ±20% while τc remains the same; (b) τc is changed by ±10% and ±20% while Kd is kept the same.
755
45
4
Reported peak discharges at the dam site and Dexing Another reported discharge at Dexing
Discharge (104 m3/s)
Dam site 3
Dexing Tuting
2
Pasighat 1
0 13:30 19 Oct
756
01:30 20 Oct
13:30 20 Oct
01:30 21 Oct
13:30 21 Oct
01:30 22 Oct
Time
757 758
Fig. 18. Observed and simulated hydrographs along the Yarlung Tsangpo.
759
46
20
Relative water level (m)
Observed water level Simulated water level 15
10
Gyalha landslide occurred Overtopping started
5
0 02:30 17 Oct
760
02:30 18 Oct
02:30 19 Oct
02:30 20 Oct
02:30 21 Oct
02:30 22 Oct
Time
761 762
Fig. 19. Observed and simulated rises of the relative water level at Tuting.
763
47
1
2
3
4
5
Month of a year 6 7
8
9
10
11
12
15000 Intra-annual runoff of flood in 1998 Average monthly intra-annual distribution of runoff from 1960 to 2009
Runoff (m3/s)
12000
9000
6000
3000
0 0 15th Jun.
764
15 30th
30 15th Jul.
30th 45
15th 60 Aug.
30th 75
15th 90 Sept.
30th 105
Month of 1998
765 766 767
Fig. 20. The long period mean monthly runoff from 1960 to 2009 and the 1998 flood at the Nuxia Hydrological Station.
768
48
7
Discharge (104 m3/s)
6
Maximum runoff of the 1998 flood
(a)
Maximum monthly runoff of a year
Dam site
Runoff of the Gyalha landslide
5
Dexing
4
Tuting 3
Pasighat
2 1 0
13:30 19 Oct
13:30 21 Oct
01:30 21 Oct
13:30 20 Oct
01:30 20 Oct
Time
769 7
Maximum runoff of the 1998 flood Maximum monthly runoff of a year Runoff of the Gyalha landslide
Peak discharge (104 m3/s)
(b) 6 Dexing
5
Tuting
4 3
Pasighat
2 1 0 0
770
100
200
300
400
500
Distance from the Gyalha landslide dam site (km)
771 772 773 774
Fig. 21. Simulated flooding process along the Yarlung Tsangpo with different runoffs: (a) hydrographs; and (b) peak discharge with distance.
775 776
49
777 778
Table 1 Input parameters for the analysis of dam breaching and flood routing. Analysis Dam breaching
Flood routing
Parameter Overtopping time, T0 Dam bottom elevation, Db (m) Dam crest elevation, Dt (m) Water level, Hw (m) Inflow rate, Qin (m3/s) Initial breach top width, Wt (m) Initial breach bottom width, Wb (m) Initial breach depth, Hb (m) Critical breach side slope, αc (º) Initial dam crest length, Dc (m) Initial dam upstream length, Lu (m) Initial dam downstream length, Ld (m) Initial breach bottom gradient, Bg Critical dam downstream slope, βf (º) Manning’s coefficient for channel bed Manning’s coefficient for riverbank
Value 13:30, 19 Oct. 2018 2758 2837 2836.1 2222 5 2 1 45 300 720 880 0.006 30 0.06 0.12
779
50
780 781
Table 2 Blockage of the Yarlung Tsangpo in the Sedongpu Basin in the recent history. Time (year/month/day)
Possible cause
Blocking Yarlung Tsangpo?
2018/10/29 2018/10/18 2018/10/17 2018/07/26 2018/01 2017/12/21 2017/11/18 2017/11/03 2017/10/22 2014 1968 1950
Ice-avalanche Ice-avalanche Ice-avalanche Debris-flow Debris-flow Ice-avalanche The Ms 6.9 Nyingchi earthquake Debris-flow − Ice-avalanche Ice-avalanche The Ms 8.6 Modog earthquake
Blocked Blocked Blocked Partially blocked Blocked Blocked Blocked Not blocked Blocked Blocked Blocked Blocked
782
51
783 784
Table 3 Results of dam breaching and flood routing analysis. Simulation results
Inflow rate, Qin (m3/s) Time to fill barrier lake, Tf (h) Breach initiation duration, Ti (h) Breach development duration, Td (h) Total breaching duration, Tt (h) Final breach depth, Hb (m) Final breach top width, Wt (m) Final breach bottom width, Wb (m) Peak flow rate at dam site (m3/s) Peak flow rate at Dexing (m3/s) Peak flow rate at Tuting (m3/s) Peak flow rate at Pasighat (m3/s)
Runoff conditions Actual runoff at the Gyalha landslide dam in 2018 2222 62.5 3.8 21.2 25.0 65.7 280.5 149.2 31685 25268 20848 11093
Average monthly runoff in August
Maximum runoff of the flood in 1998
5119 27.1 3.1 15.9 19.0 71.2 305.0 162.7 39005 31401 25525 13529
13487 10.3 2.2 11.6 13.8 79.0 349.8 191.8 57257 46775 37619 20763
785 786
52
787 788
Declaration of interest statement
789 790
None
791 792
53
793 794
Highlights:
which breached within 2 days with a peak flood over 30,000 m3/s.
795 796
A large landslide dam formed in October 2018 in the Yarlung Tsangpo Grand Canyon,
797
This paper presents a disaster chain of landslide-barrier lake-dam breaching-river flooding in the 460 km canyon.
798
An erosion-based numerical model was used to simulate the dam-breaching process.
799
Plausible dam breaching scenarios at larger river flows are analysed to help risk
800
management.
801 802
54