NHESSDNatural Hazards and Earth System Sciences DiscussionsNHESSDNat. Hazards Earth Syst. Sci. Discuss.2195-9269Copernicus GmbHGöttingen, Germany10.5194/nhessd-3-675-2015Formation time and mean movement velocities of the 7 August Zhouqu debris flows extracted from
broadband seismic recordsLiZ.HuangX.huangxh19850216@gmail.comXuQ.FanJ.YuD.HaoZ.QiaoX.China Earthquake Networks Center,
Beijing, 100045, ChinaState Key Laboratory of Geohazard Prevention and Geoenvironment
Protection, Chengdu University of Technology, Chengdu, 610059, ChinaLanzhou Institute of Seismology, China Earthquake
Administration, Lanzhou, 730000, ChinaInstitute of Seismology, China Earthquake
Administration, Wuhan, 430071, ChinaX. Huang (huangxh19850216@gmail.com)22January20153167569515December201431December2014This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://nhess.copernicus.org/preprints/3/675/2015/nhessd-3-675-2015.htmlThe full text article is available as a PDF file from https://nhess.copernicus.org/preprints/3/675/2015/nhessd-3-675-2015.pdf
The catastrophic Zhouqu debris flows, which were induced by heavy
rainfall, occurred at approximately midnight of 7 August 2010
(Beijing time, UTC + 8) and claimed 1765 lives. Broadband
seismic signals recorded by the Zhouqu seismic station nearby are
acquired and analyzed in this paper. The seismic signals are divided
into two separate parts for the first time using the crucial time of
23:33:10 (Bejing time, UTC +8), with distinctly different frequency characteristics on
time-by-time normalized spectrograms and amplitude increasing
patterns on smoothed envelopes. They are considered to be generated
by the development stage and the maturity stage of the Sanyanyu
debris flow respectively. Seismic signals corresponding to the
development stage have a broader main frequency band of
approximately 0–15 Hz than that of the maturity stage,
which is around 1–10 Hz. The N–S component can detect the
development stage of the debris flow about 3 min earlier than other
components due to its southward flow direction. Two sub-stages
within the maturity stage are recognized from best-fitted amplitude
increasing velocities and the satellite image of the Sanyanyu flow
path and the mean movement velocities of the Sanyanyu debris flow
during these two sub-stages are estimated to be 9.2 and
9.7 ms-1 respectively.
Introduction
Zhouqu County is located in Longnan Prefecture and belongs to Gansu
Province in the northwestern part of China. It is geologically part of
the conjunction area of tectonic blocks in the E–W direction and in
the middle of the N–S seismotectonic zone. This area, which is
largely influenced by tectonisms that result from the western
Tibetan Plateau, suffers from high tectonic activities and destructive
earthquakes frequently. As a consequence, fold belts and faults are
widely distributed; and loose geological structures and materials can
be easily detected everywhere. Geological hazards, such as rock
collapses and debris flows have been big threats to local residents
ever since ancient times. The topography in this area varies
dramatically due to fold belts and faults, and the maximum altitude
difference is up to 2488 m. Most of its gullies have steep
slopes, more than half of which are steeper than
25 ∘. Sufficient available loose materials, unstable
geological structures and steep slopes on the catchment are the
necessary conditions for giant debris flows occurring in this area
(Takahashi, 1981).
At approximately midnight of 7 August 2010 (Beijing time,
UTC + 8), two giant debris flows induced by heavy rainfall hit
Zhouqu city and claimed 1765 lives (Tang et al., 2011; Wang et al.,
2013b; Yu et al., 2010). A total volume of 2.2 millionm3
was transported and deposited on an existing debris fan. The heads of
debris flows rushed into the Bailongjiang River and formed
a 3 km long barrier lake (Fang et al., 2010). Water ran into
Zhouqu city and caused further damages after being stopped by the
newly formed dam (Yu et al., 2010). Most previous studies about these
debris flows mainly focused on the geological and precipitation
perspectives (Liu et al., 2011; Zhang et al., 2012; Wang et al.,
2013a). However, at least three questions are still left unresolved:
first, we only know the time that the debris flows rushed into the
Zhouqu city, but what is their formation time? Second, what kind of
frequency characteristics do seismic signals have for the development
stage and the maturity stage of the debris flows? Are there any
differences? Third, could mean movement velocities of debris flows be
estimated for different stages? The questions above are what we
attempt to resolve in this paper.
Digital broadband seismic observation networks have been gradually
established and completed in China. These networks can help to acquire
new information and explain new phenomena in seismic
observations. Broadband seismic signals have also been widely used in
geological hazard researches, such as landslides (Brodsky et al.,
2003; Chen et al., 2013; Feng, 2011; Kao et al., 2012; Lin et al.,
2010; Yamada et al., 2012) rockfalls (De Angelis et al., 2007; Norris,
1994; Vilajosana et al., 2008) avalanches, debris flows (Arattano,
1999; Burtin et al., 2009) and other block movements (Deparis et al.,
2008; Suriñach et al., 2005). These studies are receiving growing
attentions. Recent studies show that the quantitative extraction and
analysis of seismic networks data can explain the key processes, as
well as landslide and debris flow mechanisms, especially their
geological characteristics (Chen et al., 2013; Petley, 2013;
Ekström and Stark, 2013). In the present study, broadband seismic
records are used to research the giant Zhouqu debris flows occurred in
7 August 2010. The formation time of Sanyanyu debris flow is revealed
to be 23:33:10 (Bejing time, UTC +8) for the first time. Combined with the satellite image
of the Sanyanyu flow path, the mean movement velocities of Sanyanyu
debris flow during two sub-stages are estimated to be 9.2 and
9.7 ms-1 respectively. This study validates that
broadband seismic signals recorded by seismic stations deployed at
proper positions can be used to extract key parameters of debris
flows, such as formation time and mean movement velocity.
Seismic data
The giant 7 August Zhouqu debris flows occurred in the catchments of
the Sanyanyu and Luojiayu torrents. A total volume of
2.2 millionm3 was transported and deposited on an
existing debris fan and into a river, and about 64 % of them
originated from the Sanyanyu drainage area (Tang et al., 2011).
Compared with Luojiayu debris flow, the Sanyanyu debris flow has
a longer path and a larger drainage area (Tang et al., 2011). The
Zhouqu seismic station is located downstream of Sanyanyu gully and
positioned only 150 m away from the exit (Fig. 1). Thus,
seismic records mainly represent energy released in Sanyanyu
gully. The longitude, latitude, and altitude of the station are
104.4∘ E, 33.8∘ N, and 1460 m
respectively. The bedrock where the seismometer is deployed is
limestone, which guarantees the high quality of seismic records used
in this research. The giant Sanyanyu debris flow hit the Zhouqu
seismic station and destroyed its power system at approximately 23:40 LT
according to monitoring logs. This time can also be confirmed by the
end time of seismic records. The debris flow then rushed into Zhouqu
city only approximately 2.1 min later (Liu et al., 2011), killed many
people and caused large economic losses. The seismic signals recorded
by a CMG-3ESPC broadband seismic seismometer with the sample rate of
100 Hz exhibited continuous variations in time and frequency
domains tens of minutes before the records ended. The amplitude of
seismic signals increases in approximately an exponential way with
time, which is partly because the actual increase of the debris flow's
kinematic energy and partly because the approaching of the debris flow
with time; and the maximum amplitude at the end of the signal is
approximately 200 times larger than that of the background
signals. Seismic signals about 15 min before the end of records
(i.e. from 23:25 LT) are selected to reveal intrinsic processes of this
debris flow, and their waveforms are shown in Fig. 2.
Methodology and resultsDevelopment stage and maturity stage of the
giant Sanyanyu debris flow
Signal characteristics are acquired and analyzed in time-frequency
domain in the first place. Time-frequency analysis is very effective
in analyzing non-stationary signals and widely used in signals
processes. Short-time Fourier Transform, wavelet transform and
S-transform are among the most frequently used methods. In this study,
short-time Fourier transform is applied to the selected seismic
signals with a moving time window of 10.24 s. Spectrogram has
the advantage of showing global energy variation patterns temporally
and spatially in the same image. But, when dealing with peculiar
signals with imbalanced energy distribution in time domain as in this
case, the frequency distribution could be misleading. Figure 2 reveals
that the amplitude of selected seismic signals increases in
approximately an exponential way with time. In spectrograms of this
kind of signals, energy distributions for different frequency
components in low amplitude regions are usually shadowed by that of
high amplitude regions, which implies that the main frequency band
cannot be correctly observed for low amplitude regions. Given that we
are more interested in the energy distributions than the total energy,
amplitude influences are eliminated for spectrograms by normalizing
each time component to 0–1. They are called time-by-time normalized
spectrograms and shown in Fig. 3. The envelopes of selected seismic
signals are also acquired using Hilbert Transform and shown in Fig. 3
after being smoothed using the time window of 10.24 s, the
same as that used in spectrogram calculations.
Generally speaking, background signals are mainly composed of linear
drifting and low-frequency oscillations with periods of several
seconds, their peak frequency is no more than 0.5 Hz. Signals
with a broader main frequency band as large as 0 to 10 Hz and
beyond are most probably generated by the debris flows and related
events. The most obvious feature of spectrograms in Fig. 3 is that
they can be divided into two separate parts with distinctly different
frequency distribution characteristics using the time of 23:33:10.
For the left part, the effective signals are widely distributed with
the main frequency band of as broad as 0 Hz to approximately
15 Hz. The horizontal components (i.e. E–W component and N–S
component) contain more effective signals than the vertical component
(i.e. U-D component). The energy distribution in the frequency domain
for the right part is relatively regular. The main frequency band of
the U-D component shifts from approximately 1.5–8.6 Hz at
the separation time to 3.5-10.6 Hz at the end of the
signal in a near-linear manner without any expansion. As for the
horizontal components, main frequency band changed from 1.8–7.1 Hz to 2.8–9.3 Hz for the E–W component
and from 2.3–7.2 Hz to 3.3–11.2 Hz for the
N–S component, probably due to the Doppler Effect. The Doppler Effect
affects the horizontal components much stronger than the vertical
component because the Sanyanyu debris flow is mainly in the horizontal
direction.
Figure 3 also reveals that the smoothed envelopes of seismic signals
are more than 10 times larger than that of background signals at the
time of 23:33:10 and rapidly increase afterward in an
unprecedented velocity. Based on the spectrograms and smoothed
envelopes, we claim that seismic signals generated by the giant
Sanyanyu debris flow can be divided into two separate parts with
distinctly different frequency characteristics corresponding to the
development stage and maturity stage of the debris flow
respectively. Unlike earthquakes and landslides, debris flows do not
break out suddenly. It takes some time to accumulate water and loose
materials before a large and destructive energy release is fully
formed. Seismic signals generated by these two stages before and after
the formation of a debris flow are quite different, especially in the
frequency domain. We define the crucial time of 23:33:10 as the
formation time (FT) of the giant Sanyanyu debris flow, after which the
debris flow reached its maturity.
It is almost impossible to determine when a debris flow starts
developing. Even from the first stream of water running, the first
pile of loose materials moving and the first boulder falling, the
development stage of a debris flow starts. As loose materials
gathering and their kinematic energy gradually increasing, seismic
signals generated by them become detectable by seismometers. However,
a rough start time can be revealed under a given criterion. Here, the
short-term average/long-term average (STA/LTA) approach (Chen et al.,
2013; Kao et al., 2012) is employed to determine the start time (ST)
of the development stage of the Sanyanyu debris flow. Threshold and
time window of 1.6 and 15 s, respectively, are adopted in the
present study and the procedure is described as follows. First, the
mean absolute values of background signals are calculated for each
component. And then we move a time window with the given width step by
step to calculate mean absolute values inside. If a calculated value
exceeds the threshold, we assign the start time. Results of three
components are 23:30:16.48 (U–D), 23:30:12.56 (E–W), and 23:27:19.48
(N–S) respectively. It can be observed that the ST of the N–S
component is approximately 3 min earlier than that of other
components. Spectrograms also reveal that the horizontal components
(E–W and N–S) can detect effective signals much earlier than the
vertical component (U-D); and within horizontal components, the N–S
component is earlier than the E–W component. Given that debris flows
mainly move in the horizontal direction, the energy amplification is
much faster for the horizontal direction than the vertical direction
and can be detected sooner by horizontal components (E–W and
N–S). The direction of the main debris flow path in this study is
mainly north–south (Fig. 1). Therefore, the calculated ST of the N–S
direction is much earlier than that of the E–W and U-D components.
Estimation of the Sanyanyu debris flow
movement velocities
To reveal the amplitude variation patterns of seismic signals with
time, firstly the Hilbert Transform is adopted to get the envelopes of
signals; and subsequently, the envelopes are linearly best-fitted to
get the mean amplitude increasing velocities using a moving time
window of 1 min. To eliminate influences from source approaching with
time, we multiply the results by the function scale(t),
which is defined as:
scale(t)=e-t/120
Where t is time in second and starts from 23:30. The scaled
best-fitted amplitude increasing velocities for three components and
scale function are shown in Fig. 4.
Amplitude of seismic signals can roughly represent the kinematic
energy released by the debris flow. Figure 4 shows that there are at
least four magnificent peaks in the maturity stage corresponding to
a sequence of fast energy accumulation stages of the debris flow,
which could be caused by the increase of movement velocity and/or
materials. These two manners have different characteristics in
amplitude increase patterns. Material increase caused energy increase
corresponds to the synchronous increase of amplitude for the three
components; while movement velocity increase caused energy increase
corresponds to more significant increase in the component of the
movement direction than other components. Four key time points (i.e.
KT1, KT2, KT3 and KT4) are depicted in Fig. 4 to help carrying out the
following discussions. Satellite image of the Sanyanyu flow path
snapshotted from Google Earth is shown in Fig. 5 to analyze different
spatial stages of the giant Sanyanyu debris flow. The Sanyanyu flow
path is depicted using color solid lines and the altitude profile
along the path is given sideward as well. A colluvial deposit area
described as a significant energy accumulation region (Tang et al.,
2011) is also sketched in Fig. 5 using a gray solid line.
Because the Sanyanyu debris flow moved mainly in the horizontal
direction, the amplitude increase of the U-D component is mostly from
the material increase instead of the movement velocity increase. This
can explain that the scaled best-fitted amplitude increasing
velocities of the horizontal components are larger than that of the
vertical component almost in the whole maturity stage; and that the
Doppler Effect affects the horizontal components much stronger. As for
the horizontal components, the amplitude increase could be from the
increase of materials and movement velocities. In the first case, the
scaled best-fitted amplitude increasing velocities of horizontal
components vary with that of the vertical component. In the second
case, the amplitude of the component in the acceleration direction
increase faster than other components. Figure 4 reveals that from KT2
to KT3, the amplitude of the N–S component increases much faster than
that of the E–W and U-D components; from KT3 to KT4, the amplitude of
the E–W component increases rapidly. This phenomenon corresponds to
a path shape of southward/northward in the beginning and turn to
eastward/westward in the end, which is in consistent with the path
section depicted using orange solid line in Fig. 5. This path section
is southward in the beginning and then turns to eastward in the end.
Therefore, we believe that seismic signals between K2 and K4 were
generated when the Sanyanyu debris flow moved on this path
section. Before KT2 in Fig. 4, there is a significant peak section
with an approximately synchronous increase pattern for three
components. Meanwhile, the path section before the orange solid line
in Fig. 5 is within the colluvial deposit area. When the Sanyanyu
debris flow moved on this path section, seismic signals with amplitude
increase patterns between KT1 and KT2 could be generated. Based on the
above information, the seismic signals corresponding to the maturity
stage is further divided into four sub-stages, with the first one from
FT to KT1, the second one from KT1 to KT2, the third one from KT2 to
KT4, and the last one from KT4 to the end. Four corresponding spatial
stages are depicted in Fig. 5 using blue, yellow, orange and red solid
lines respectively. The time intervals and length of path sections for
different stages are listed in Table 1. The mean movement velocities
of the Sanyanyu debris flow during stage 2 and 3 are estimated to be
9.2 and 9.7 ms-1 respectively.
Discussions
It is necessary to discriminate effective signals from background
signals before further processes. Normally, this can be achieved by
recognizing different amplitude characteristics in time domain and
energy distribution characteristics in frequency domain. Amplitude
differences between effective signals and background signals in time
domain can be easily recognized; but for the frequency domain it is
not that obvious. Spectrograms are frequently used to characterize
frequency distributions along time; but they contain not only
frequency information but also amplitude information in a global
way. They are merged together. When the amplitude is imbalanced
globally, signals that contain effective information in frequency
domain but with relatively low amplitude are covered up like the
situation in this case. Time-by-time normalized spectrogram is used
in signal processing for the first time and proves a good tool. It
normalizes each time component of a spectrogram to eliminate
influences from the imbalanced amplitude distribution and can reveal
true frequency distributions. By using this method, development stage
and maturity stage of the giant Sanyanyu debris flow are distinctly
separated for the first time. This method may also be used to detect
effective information of other kinds of event such as landslides,
earthquakes and volcano eruptions.
The development of a debris flow starts from collecting loose
materials and water. But, from the development stage to the maturity
stage of a debris flow, it is not a simply quantitative change;
instead, a qualitative change happens. Before and after the formation
time, the differences of seismic signals in amplitude and frequency
are quite clear. The main frequency band of seismic signals
corresponding to the development stage is broader than that of the
maturity stage. Yet, the seismic signals corresponding to the maturity
stage is more regular. The component in the direction of a debris flow
movement can detect the development of the debris flow earlier, which
can be used in alert systems. The main frequency band of seismic
signals corresponding to the maturity stage of the Sanyanyu debris
flow expanded 0, 22.64 and 61.22 % for U-D, E–W and N–S
components respectively, which implies that the Doppler Effect affects
seismic signals in the component of the movement direction the
strongest.
Mean movement velocities of the Sanyanyu debris flow during sub-stages
2 and 3 are estimated to be 9.2 and 9.7 ms-1
respectively. These results are in consistence with the previous
estimation (Tang et al., 2011). The estimation result that the mean
movement velocity of the debris flow in sub-stage 3 is larger than
that in sub-stage 2 is consistence with our previous judgment that
a significant portion of the energy increase in the sub-stage 3 is
from the movement velocity increase. As for the sub-stages 1 and 4,
the exact formation location of the Sanyanyu debris flow is very hard to
determine from the satellite image. Also, the Sanyanyu debris flow
lasted several minutes. When the head of the debris flow destroyed the
power system of the Zhouqu seismic station, the seismic source was not
necessarily reached the station. Therefore, the mean movement
velocities in these two sub-stages are left unresolved. The length of
the time window used in calculation of the best-fitting amplitude
increasing velocity is 60 s. Given the movement velocity of
the debris flow of approximately 10 ms-1, the spatial
resolution is about 600 m. Sub-stages have a smaller spatial
scale cannot be rightly resolved using this data.
Conclusions
The 7 August Zhouqu debris flows, which were induced by very heavy
rainfall killed 1765 people and caused great economic losses. Zhouqu
seismic station located on the path to Zhouqu city was destroyed by
the debris flows at approximately 23:40 LT on the same date, which marks
the time when the head of the debris flows reached the station. The
seismic signals recorded by the Zhouqu seismic station before being
destroyed are analyzed in this paper. The signals are divided into two
separate parts for the first time using the crucial time of 23:33:10,
with distinctly different frequency characteristics on time-by-time
normalized spectrograms and amplitude increasing patterns on smoothed
envelopes. They are considered to be generated by the development
stage and the maturity stage of the Sanyanyu debris flow
respectively. Seismic signals corresponding to the development stage
have a broader main frequency band of approximately 0–15 Hz
than that of the maturity stage, which is around 1–10 Hz. The
main frequency band shifted to the high frequency direction with the
seismic source approaching for all of the three components; and it
also expanded 22.64 and 61.22 % for E–W and N–S components due
to the Doppler Effect. The N–S component can detect the development
stage of the debris flow about 3 min earlier than other components
due to its southward flow direction. The seismic signals corresponding
to the maturity stage is further divided into four sub-stages
according to peaks from scaled best-fitted amplitude increasing
velocities. Combined with the satellite image of the Sanyanyu flow
path, the mean movement velocities of the Sanyanyu debris flow in
sub-stages 2 and 3 are estimated to be about 9.2 and
9.7 ms-1 respectively. Our results show that broadband
seismic signals recorded by seismic stations deployed at proper
positions can be used to extract key parameters of debris flows and
research their formation mechanisms.
Z. Li, X. Huang and Q. Xu discussed and determined
the overall framework of this study; J. Fan, D. Yu, Z. Hao and
X. Qiao prepared the seismic data and did field investigations;
X. Huang analyzed the seismic data and prepared the manuscript with
contributions from all co-authors.
Acknowledgements
We would like to thank Liu Ruifeng, Huang Zhibin, and Zhao Yong from
the China Earthquake Networks Center for their helpful comments on
seismic wave recognition. We also thank Zeng Wenhao and Niu Yanping
from the Lanzhou Institute of Seismology for their kind help in the
field investigation. Liu Zhumei from the Institute of Seismology
also helped us in preparing Fig. 1. Satellite image of Sanyanyu
flow path in Fig. 5 is snapshotted form Google Earth. This research
is financially supported by the 973 Program 2013CB733200 and 863
Program 2012AA121302.
References Arattano, M.: On the use of seismic detectors as
monitoring and warning systems for debris flows, Nat. Hazards, 20,
197–213, 1999.Brodsky, E. E., Gordeev, E., and Kanamori, H.: Landslide
basal friction as measured by seismic waves, Geophys. Res. Lett.,
30, 2236, 10.1029/2003GL018485, 2003.Burtin, A., Bollinger, L., Cattin, R., Vergne, J., and
Nábělek, J. L.: Spatiotemporal sequence of Himalayan debris
flow from analysis of high-frequency seismic noise,
J. Geophys. Res.-Earth, 114, F04009, 10.1029/2008JF001198, 2009. Chen, C.-H., Chao, W.-A., Wu, Y.-M., Zhao, L.,
Chen, Y.-G., Ho, W.-Y., Lin, T.-L., Kuo, K.-H., and Chang, J.-M.:
A seismological study of landquakes using a real-time broad-band
seismic network, Geophys. J. Int., 194, 885–898, 2013.De Angelis, S., Bass, V., Hards, V., and Ryan, G.: Seismic
characterization of pyroclastic flow activity at Soufrière Hills
Volcano, Montserrat, 8 January 2007, Nat. Hazards Earth Syst. Sci.,
7, 467–472,
doi:10.5194/nhess-7-467-2007,
2007. Deparis, J., Jongmans, D., Cotton, F., Baillet, L.,
Thouvenot, F., and Hantz, D.: Analysis of rock-fall and rock-fall
avalanche seismograms in the French Alps, B. Seismol. Soc. Am., 98,
1781–1796, 2008.
Ekström, G. and Stark, C. P.: Simple scaling of
catastrophic landslide dynamics, Science, 339, 1416–1419,
2013. Fang, H. Y., Cai, Q. G., Li, Q. Y., Sun, L. Y., and
He, J. J.: Causes and countermeasures of giant flash flood and
debris flow disaster in Zhouqu County in Gansu Province on
7 August 2010, Science of Soil and Water
Conservation, 8,
14–18, 2010 (in Chinese).Feng, Z.: The seismic signatures of the 2009 Shiaolin
landslide in Taiwan, Nat. Hazards Earth Syst. Sci., 11, 1559–1569,
doi:10.5194/nhess-11-1559-2011, 2011. Kao, H., Kan, C.-W., Chen, R.-Y., Chang, C.-H.,
Rosenberger, A., Shin, T.-C., Leu, P.-L., Kuo, K.-W., and
Liang, W.-T.: Locating, monitoring, and characterizing
typhoon-linduced landslides with real-time seismic signals,
Landslides, 9, 557–563, 2012.Lin, C. H., Kumagai, H., Ando, M., and Shin, T. C.:
Detection of landslides and submarine slumps using broadband seismic
networks, Geophys. Res. Lett., 37, L22309, 10.1029/2010GL044685, 2010.Liu, C. Z., Miao, T. B., Chen, H. Q., Dong, K. J.,
Li, Z. H., and Li, H. J.: Basic feature and origin of the
“8 × 8” mountain torrent-debris flow disaster happened in
Zhouqu County, Gansu, China, Aug. 8, 2010, Geological Bulletin of
China, 30, 141–150,
2011 (in Chinese). Norris, R. D.: Seismicity of rockfalls and avalanches at
three Cascade Range volcanoes: implications for seismic detection of
hazardous mass movements, B. Seismol. Soc. Am., 84, 1925–1939,
1994. Petley, D. N.: Characterizing giant landslides, Science,
339, 1395–1396, 2013.Suriñach, E., Vilajosana, I., Khazaradze, G.,
Biescas, B., Furdada, G., and Vilaplana, J. M.: Seismic detection
and characterization of landslides and other mass movements,
Nat. Hazards Earth Syst. Sci., 5, 791–798,
doi:10.5194/nhess-5-791-2005, 2005. Takahashi, T.: Estimation of potential debris flows and
their hazardous zones: soft countermeasures for a disaster, Journal
of Natural Disaster Science, 3, 57–89, 1981.Tang, C., Rengers, N., van Asch, Th. W. J., Yang, Y. H.,
and Wang, G. F.: Triggering conditions and depositional
characteristics of a disastrous debris flow event in Zhouqu city,
Gansu Province, northwestern China, Nat. Hazards Earth Syst. Sci.,
11, 2903–2912,
doi:10.5194/nhess-11-2903-2011,
2011.Vilajosana, I., Suriñach, E., Abellán, A.,
Khazaradze, G., Garcia, D., and Llosa, J.: Rockfall induced seismic
signals: case study in Montserrat, Catalonia, Nat. Hazards Earth
Syst. Sci., 8, 805–812,
doi:10.5194/nhess-8-805-2008,
2008. Wang, G. L., Zhang, M. S., Yu, G. Q., and Ye, W. J.:
Factor analysis for catastrophic debris flows on August 8, 2010 in
Zhouqu City of Gansu, China, J. Mt. Sci., 31, 349–355, 2013a (in
Chinese). Wang, L. M., Wu, Z. J., Wang, P., and Chen, T.:
Characteristics, causation, and rehabilitation of Zhouqu
extraordinarily serious debris flows in 2010, China, J. Cent South
Univ. T., 20, 2342–2348, 2013b.Yamada, M., Matsushi, Y., Chigira, M., and Mori, J.:
Seismic recordings of landslides caused by typhoon Talas (2011),
Japan, Geophys. Res. Lett., 39, L13301, 10.1029/2012GL052174, 2012. Yu, B., Yang, Y. H., Su, Y. C., Huang, W. J., and
Wang, G. F.: Research on the giant debris flow hazards in Zhouqu
County, Gansu Province on August 7, 2010, J. Eng. Geol., 18,
437–444, 2010 (in Chinese). Zhang, Z. X., Zhang, Q., Tao, J. C., Sun, Y., and
Zhao, Q. Y.: Climatic and geological environmental characteristics
of the exceptional debris flow outburst in Zhouqu, Gansu Province,
on 8 August, 2010, J. Glaciol. Geocryol., 34, 898–905, 2012 (in
Chinese).
Time intervals of sub-stages 1–4 and distances of sub-stages 2
and 3. Mean movement velocities of the Sanyanyu debris flow in
sub-stages 2 and 3 are also provided.
Locations of the Zhouqu seismic station (red solid triangle),
Sanyanyu Gully, and Luojiayu Gully.
Seismic waveforms of three components. The top to bottom
images are U–D, E–W, and N–S components respectively. The
formation time is marked by using black solid lines.
Smoothed envelopes (upper panel) and time-by-time normalized
spectrograms (lower panel) of three components. The formation time
is marked by using black solid lines. The smoothed envelopes in the
shadow regions are enlarged and superimposed on the spectrograms
using white solid lines. The main frequency bands of the three
components are roughly depicted between the red dashed lines.
Scale function (upper panel) and scaled best-fitted amplitude
increasing velocities (lower panel). The formation time and key time
points (KT1, KT2, KT3 and KT4) are also depicted. The maturity stage
is further divided into 4 sub-stages and distinguished using
different background colors.
Satellite image of the Sanyanyu flow path, depicted using
color solid lines with different colors corresponding to different
sub-stages. The altitude profile is provided
sideward. A significant colluvial deposit area is also sketched
using a gray solid line.