A comparative analysis of attabad landslide on january 4, 2010, using two numerical models


This study determines the runout behavior (Attabad landslide, Hunza, Pakistan) of one of the biggest landslides in Pakistan's history, along the highest and strategically most important highway of the world. On January 4, 2010, at 08:36 local time, 45 Mm3 of rock mass flowed down the hill slope for 1060 m and fell in Hunza River, thus blocking the river's flow and making an artificial dam. This paper examines the landslide's failure process based on seismic data obtained from a published paper. The average velocity of the slide was found to be 14.32 m/s. To better understand the landslide dynamics and failure phenomena, a numerical simulation was conducted using DAN3D to simulate displaced materials' runout behavior. Simulation results indicate that the slope's failure lasted for 70 s, which is in good agreement with seismic wave recordings of 70 s. The combined frictional–Voellmy model obtained the most accurate results for simulation. Further, to verify the results, another simulation was run using RAMMS debris flow software; it was found that the results of both the software's are in good agreement. It is expected that the selected model and its parameters will help understand similar kind of rock avalanches in the area, which will help concerned agencies improve landslide prediction along Karakoram Highway.


The northern region of Pakistan is considered among the world's most active tectonic areas consisting of the Himalaya, Karakoram, and Hindukush mountains. This is the region where two tectonic plates, the Indian and Eurasian Plates, meetup (Ali et al., 2019), making the whole area prone to natural hazards. The area is highly characterized by fragmented and weathered rocks, steep slopes, deep gorges, active tectonic activities, and flash flooding. This region is home to sky-touching mountains like K2, Nanga Parbat, and Rakaposhi. Figure 1 shows the longitudinal cross section indicating the area's steepness from Rakaposhi top to Nasirabad, which quickly varies in a patch of 10 km from 7788 to 2000 m (Goudie et al. 1984).

Fig. 1

Longitudinal cross section indicating steepness variation within a range of 10 km from Rakaposhi top to Nasirabad

On January 2010, at 8:36 local time, a disastrous rock avalanche hit the area of Attabad, Hunza in North of Pakistan, 14 km upstream of Karimabad at 36°20′13″N 74°52′3″E. The landslide killed 20 people and whipped off two villages Attabad and Sirat located on the Hunza river's upper slope. Rock avalanche traveled a vertical distance of 1060 m and fell into the Hunza river, which caused the Hunza river's blockage and created an artificial dammed lake over some time. Debris was composed of large boulders of granderloite of Karakoram batholite separated from bedrock along the failure slope and mixed with locally available sediments. Karakoram Highway (KKH), which is strategically important, and now part of the Chin-Pak Economic Corridor (CPEC), lies in the area, swirling through sky-touching mountains. It connects Pakistan to China at Khunjerab. The devastating landslide of Attabad also affected the Karakoram Highway as the lake formed due to the Hunza river's blockage submerged 21 km section of highway, thus seizing people's mobility shown in Fig. 2. So, it is essential to mitigate the damages caused by this kind of event and, ultimately, prevent occurrence. Therefore, it is necessary to analyze this rock avalanche's runout behavior to make predictions for similar rock avalanches in the area.

Fig. 2

Post-landslide difficulty in logistics along KKH a excavators are carrying a boat through the landslide debris, at third excavator, is cutting an emergency path on landslide debris. b Due to submergence of KKH, boats are being used for logistics

The study of slope failure and landslides is a vast field, till now a lot of work has been carried out in terms of numerical modeling and simulations of landslides. In-depth knowledge of landslide characteristics and runout behavior is required to assess the landslide hazard (Dai and Lee 2002). It can only be done by modeling the landslide and doing back calculations to verify the results. Cleary and Campbell (1993) used Compbell's 1989 model to simulate landslide and find self-lubrication for long-runout landslides.

Similarly, Hutter et al. (1995) used a theoretical model to determine the dynamic behavior of avalanches from the point of initiation to runout. Hungr (1995), for the first time, developed the most advanced model for simulation of rapidly flowing landslides named dynamic analysis (DAN). Later on, this model was widely used to simulate the landslides. Hungr and Evans (2004) used the dynamic model (DAN) combined with the frictional and Voellmy model to describe the landslide's long-runout mechanism. Evans et al. (2001) successfully analyzed the 1984 avalanche of mount Cayley using the dynamic analytical model and the results were well matched with field investigation.

Kelfoun and Druitt (2005) used a theoretical model based on a depth-averaged granular flow equation to find the Socompa rock avalanche's flow behavior. Similarly, Li et al. (2012) used discrete element modeling (DEM) to simulate the sliding process of the Donghekou landslide triggered by the Wenchuan earthquake. Chen et al. (2013) analyzed mobility and time of initiation of Tsaoling rock avalanche using seismic wave data and Newmark sliding block analysis. This time–frequency analysis was also run to identify avalanche initiation time and falling time. Several case studies have been carried out to analyze different landslides using different techniques, i.e., smoothed particle hydrodynamics (SPH), discontinuous deformation analysis (DDA), particle flow code 3D (PFC3D), two-dimensional discontinuous deformation analysis (2D DDA) (Dai et al. 2014; Jiao et al. 2014; Lin and Lin 2015; Chen et al. 2018; Bao et al. 2020). The mesh-free feature of SPH gives it a priority as compared to traditional grid-based methods while analyzing high-speed fluidized landslides; however, there is a problem with SPH known as tensile instability. Though the problem can be solved by incorporating some ad-hoc techniques, still the solution is not universal. To counter the problem, Mao et al. (2019) developed a conservative and consistent Lagrangian gradient smoothing method (L-GSM) for fluidized landslide simulations. Further, the newly developed L-GSM was calibrated by using soil column collapse standard and three landslides. Several landslides started as rock avalanches from the source area, but later they carried other materials from the erosion zone, which increased the volume of slide significantly. Xing et al. (2014) used the DAN3D model combined with frictional and Voellmy parameters to simulate the runout behavior of the Guanling landslide. The 1991 Zhaotong landslide's runout behavior was analyzed using DAN-W in-combination with three different rheological parameters (Xing et al., 2016). The study concluded that the frictional and Voellmy model combinedly provide the best simulation results. Different rock avalanches have been analyzed using DAN3D (Xing et al. 2017; Zhuang et al. 2019, 2020).

Reduction in the hazards associated with sliding masses is a challenging task that requires detailed dynamic analysis. Employing a numerical model can potentially assist in providing appropriate measures to mitigate the landslides' disastrous effect. Such dynamic numerical models can perform forward simulations for future scenarios (Luna et al., 2012). The present study involves the dynamic runout analysis of a catastrophic rock avalanche using two numerical models. Although there are many dynamic runout models commercially available such as DAN-3D, RAMMS, and FLO-2D, this study uses dynamic analysis of landslides (DAN-3D) and rapid mass movement system (RAMMS) to analyze the Attabad landslide in Pakistan.

RAMMS was developed by the WSL Institute for the Snow and Avalanche Research SFl, Switzerland (Johnson 1984). It is equipped to estimate the flow velocities, runout distance, and height of the moving mass of rapid rock slides, rock avalanches, rock falls, debris flows, and snow avalanches. It is a runout model that provides assistance to physically model the gravitationally driven moving mass in a three-dimensional environment (Christen et al. 2012). Three-dimensional runout modeling can provide in-depth delineation of the moving mass throughout the runout (Rickenmann et al. 2006; Pastor et al. 2009; Cascini et al. 2010; Hussin et al. 2012; Koo et al. 2018; Zimmermann et al., 2020). Thus, the complex behavior of the landslide process can be reflected accurately by RAMMS. The application of and rheology of both models are described in the later chapters.

A rock avalanche having a long-runout path will disintegrate into small particles during the downward motion of rock mass, thus behaving like a fluid. In this case of Attabad rock avalanche, the slide's vertical fall height is 1060 m, which means the rock boulders will be disintegrated into small particles, thus behaving as fluid, so here for simulation, DAN3D and RAMMS debris flow are used. As the name indicates, 3D, which means three-dimensional topographical analysis without mesh distortion problem. So, this is the most suitable model for the backward analysis of fluidized landslides. During a landslide, when rock boulders leave their place and hit the ground, a tremendous amount of energy is released in the form of seismic waves (Zhu et al. 2019), which can be recorded at a seismic station nearby. Unfortunately, due to lack of equipment and technology, there was no nearby seismic station to Attabad where the seismic waves could be recorded. However, a seismic wave of 5.3 magnitudes at 8:36 was recorded by an algorithm developed by (Ekström and Stark 2013); Fig. 3 indicates the landslide force history diagram of Attabad from which it can be seen that the sinusoidal wave lasted for 70 s. This force was further used to determine different parameters of the landslide, as shown in Fig. 4 (Ekström and Stark 2013).

Fig. 3

Landslide force history (Ekström and Stark 2013)

Fig. 4

source inversion model applied on Attabad landslides data extracted from a published paper of (Ekström and Stark 2013)

Landslide seismic detection and

In this study, a series of back analyses were conducted to determine the best rheological parameters for simulations using DAN3D. A comprehensive investigation was carried out to find the geological and seismic history of the area. Furthermore, the reasons for the triggering process were also evaluated. While simulating the avalanche, the most appropriate rheologies have been selected (frictional and Voellmy) and calibrated. The parameters for rheology were set by the hit and trial method until we got relevant results. It is expected that this model and selected parameter will help to understand the triggering process of similar kinds of avalanches in the area in future.

The geological and climatic condition of the area

Two significant faults pass through the Hunza valley: Main Karakorum Thrust (MKT) and Main Mental Thrust (MMT). The Main Karakoram Thrust (MKT) ranges from the West of Pakistan, from the Pakistan-Afghanistan border to western Tibet in the east. In contrast, Chitral Fault marks its northwestern limit, and the Karakoram Faults define its northeastern limit. The Hunza Valley zone north of MKT is mainly composed of brittle rocks like synmetamorphic mylonites, foliated carbonaceous, and metaconglomerates. In the area between Hini and Sarat, the Hunza valley shows a complete section of the Karakoram axial batholith as medium-grained granodiorite (Valley et al. 2010).

The geology of Attabad shows that the destroyed village was located on the slope of glacial-fluvial deposits, which was underlain by gneissic rocks of the Baltit group. Simultaneously, the Baltit group is composed of different rocks, i.e., Schist, gneisses, marble, quartzite, and dolomite limestone. Glacio-fluvial deposits of Attabad consist of boulders, cobbles, gravels along with some silt, sand, and some cementing materials. Rocks found at the crown and toe of the slide are highly fractured, sheared, jointed, and weathered. Figure 5 illustrates the geology of the Attabad region (Hayat and Engineering 2010).

Fig. 5

Geology of the Attabad region, northern Pakistan. (Hayat and Engineering 2010)

According to a survey conducted by the Geological Survey of Pakistan back in 2009, there is an active fault that crosses the river having a North–South strike. Attabad landslide is associated with this fault. According to the data obtained from the metrological department, the area of Attabad and Hunza valley is icy, winters are very harsh with an average snow depth of 0.5 m to 1.0 m, whereas summer is quite pleasant. Figure 6 shows the monthly temperature variation of the area in 2010. According to World Weather online portal, December was the coldest month of the year with an average temperature of −21˚C, whereas a maximum temperature of 10˚C was observed in August 2010. March and April are considered dangerous months because of landslide and debris flow due to snow melting.

Fig. 6

Monthly temperature (Temp. ˚C) of Hunza Valley in 2010

Triggering Phenomena of Attabad Landslide

On November 20, 2002, an earthquake of magnitude 6.3 on the Richter scale, having an epicenter 75 km south of Attabad, hit the area. The earthquake resulted in large cracks on the slopes traveling through the village and surrounding area; locals reported that these cracks developed after the earthquake (Butt et al. 2013). Over some time, these cracks on the slopes of Attabad started to lengthen and widen. After a short period of fewer than three years, on October 8, 2005, another massive earthquake of magnitude 7.6 on the Richter scale hit the area again; However, the earthquake's epicenter was far away, but still shaking was felt in the area. As a result of this earthquake, the cracks on the slopes of Attabad were significantly disturbed and gained more width (Hussain and Awan 2009). These cracks posed a severe threat to people living there. Hussain and Awan (2009) found that three prominent cracks extended over hundreds of meters with a maximum displacement of 5 m. It was recommended that the village of Attabad be evacuated with immediate effect.

Just two days before the landslide on January 02, 2010, another earthquake of magnitude 5.1, having an epicenter 450 km to NW of Attabad, hit the area again. However, this earthquake didn't cause the landslide as its epicenter was far away from Attabad but affected the opened cracks. So, we can classify this landslide as a non-seismic fatal landslide (NFL). There are four factors due to which an NFL can trigger (Petley et al. 2010).

  1. a.

    An NFL can be triggered in an area having sufficient relative relief.

  2. b.

    It can be triggered in a high population density area.

  3. c.

    In an area with seismic activity, the tectonic plates' movement causes the opening of cracks and the destabilization of slopes.

  4. d.

    The area with excessive precipitation.

According to factors mentioned above Attabad landslide concedes with factor C, as the landslide occurred in a highly active seismic area, which caused cracks in slopes of Attabad and destabilized it and over a period of time due to combined action of human activities and natural behaviors, 45Mm3 of mass slide down on January 4, 2010. Figure 7 shows the location of the remains of Attabad village. It can be seen that cracks are still present near the Attabad village that possess the threat of further slides. Further, it can be seen on the downstream side of the study area that the Karakorum Highway is blocked due to the landslide deposit.

Fig. 7

Post-landslide condition of the Attabad village; wide cracks can be observed near the village. Blocked KKH can also be seen in the background

Attabad landslide

As discussed above, the Attabad landslide was a non-seismic fatal landslide, though the cracks in the source area were developed due to the 2002 earthquake. Still, it didn't cause the mass to slide; sliding mass was a time-dependent phenomenon that took eight years to slide down 45Mm3 of mass eventually. The massive landslide blocked the passage of the Hunza river and formed an artificial lake upstream; the growing water level in the lake submerged 21 km of Karakoram Highway, growing lake also submerged a bridge upstream as shown in Fig. 8a–c. Figure 9 presents the main sliding direction of the Attabad rock avalanche and the downstream deposition area. The figure also shows people crossing landslide debris to get to the other side, as the road has been buried under the landslide deposit that has a thickness of more than 130 m.

Fig. 8

a Pre-landslide location of Attabad village, Karakoram Highway, and a bridge will be submerged as water in the lake rises. b Post-landslide location, submerging Karakoram Highway, and growing lake. c Post-landslide location in April 2011, location of submerged Karakoram Highway and flooded bridge

Fig. 9

Main sliding direction of Attabad landslide. The area between the dotted lines indicates emergency passage made over the deposition area

Figure 10a indicates the post-landslide aerial image showing the source area, erosion area, and the main sliding direction. The Attabad landslide can be divided into two zones, the source zone and erosion zone, based on the dynamics. Source area having a vertical length of 410 m. The initiation zone of rockfall ranges from 3360 to 2950 m, whereas the erosion area has a length of 650 m and the erosion zone ranges from approximately 2950 m to 2300 m. A detailed longitudinal cross section of the Attabad landslide is shown in Fig. 10b. Elevation values are approximated because the obtained line, which differentiates source and erosion area, is a polyline. After an elevation of 2300 m, mass fell into the Hunza river, thus blocking the water passage and making an artificial dam with the support of opposite mountain Fig. 11.

Fig. 10

source area along with the main sliding direction. b Longitudinal section of Attabad landslide along with line AB

a Post-landslide aerial image of the

Fig. 11

Aerial image of the deposition area of Attabad landslide

Dynamic Properties of Landslide

The landslide force inversion method developed by Ekström and Stark, 2013 was used to infer the sequence of 3D force generated by massive landslide motion. Different dynamic properties were determined, i.e., momentum, potential energy, acceleration, and runout duration. Analysis of long-period waveform data in association with the inverse model gave the best results for landslide location and direction. Force-induced by the Attabad landslide's acceleration gave a sinusoidal sequence that lasted for 70 s. The three-dimensional force vector (north, east and up) varies simultaneously, giving a consistent azimuth of acceleration and declaration. North components gave us a maximum force, as shown in Fig. 3.

Runout Modeling of Rock Avalanche

DAN3D Model

The runout behavior of the Attabad landslide was studied using DAN3D and RAMMS. Appropriate rheological parameters were adopted to simulate the Attabad landslide on DAN3D. There are five rheologies in DAN3D named frictional, plastic, Newtonian, Bingham, and Voellmy (McDougall and Hungr 2005). Among them, two rheology's frictional and Voellmy in-combination are used by most researchers as these two provide the most accurate results (Zhu et al. 2019). Frictional rheology accounts for revisiting shear force, and it considers it a function of effective normal stress σ, as expressed in Eq. 1.

$$\tau = \sigma \left( {1 - r_{u} } \right){\text{tan}}\Phi$$

where τ is shear stress, σ is effective stress, ru is pore pressure ratio, and Φ represents the dynamic friction angle, whereas dynamic friction angle and pore pressure can be represented by another variable known as bulk basal friction angle, represented by Φb. as shown in Eq. 2.

$$\Phi b = {\text{ arctan }}[\left( {{1} - ru} \right){\text{ tan}}\Phi ]$$

The second rheology is Voellmy, which says that the slide's total resistance is composed of two parameters, friction, and turbulence. Equation 3 represents the relationship between them.

$$\tau = \, \sigma f \, + \rho gv^{2} /\varepsilon$$

friction coefficient f related shear stress to normal stress, whereas ε is turbulence coefficient, which indicates that velocity-dependent components are related to normal shear stress. Rheological parameters for the purpose were obtained by back-calculation until simulation results match the landslide's actual condition. Friction angle (f) of 25˚, along with pore pressure ratio ru of 0.10, was the best fit for the frictional model. Different parameters were adopted in the Voellmy model until the desired results were obtained, as shown in Table 1.

Table 1 DAN3D and RAMMS back-calculated value of rheological models for the Attabad landslide

RAMMS Debris Flow Model

Rapid mass movement system (RAMMS) has three available modules: (1) RAMMS:: AVALANCHE used to simulate the particulate flow phenomena of snow, (2) RAMMS:: DEBRIS FLOW, which is used to observe the particulate flow phenomena of rocks, (3) RAMMS:: ROCKFALL, used in simulating the rockfall behavior. RAMMS:: DEBRIS FLOW is found appropriate to simulate the fast-moving, i.e., flow phenomena of the particulate rock debris of the Attabad landslide and have been used in this study. There is an option to set the starting condition: (1) channelized debris flow, which requires an input hydrograph of the area, or (2) un-channelized debris flow, simply requiring the initial depth of the sliding material as it uses a block release mode. Block release mode of RAMMS debris flow has been used to simulate the Attabad landslide. It involves a three-dimensional terrain for the calculation of runout movement. It employs the depth-averaged equations to predict the runout flow height and velocity parallel to the slope. RAMMS:: DEBRIS FLOW also utilizes the same Voellmy-fluid flow approach (Salm et al. 1990; Salm 1993), known as the Voellmy-fluid friction model. RAMMS:: DEBRIS FLOW represents the Voellmy model as:

$$S = \mu N + \frac{{\rho gu^{2} }}{\xi }$$
$$N = \rho hg\cos \varphi$$

where S is the frictional resistance, μ is the solid phase resistance, N is the normal force, density is presented by \(\rho\), ξ represents the resistance in the turbulent phase, h represents the flow height, g is the gravitational acceleration. Simultaneously, the downslope angle is φ, and u is the flow velocity in x- and y-directions (D'Agostino et al. 2008). For accurate results, this software uses a Voellmy friction model that has been calibrated where required using well-documented case studies (Deubelbeiss and Graf 2013). RAMMS:: DEBRIS FLOW erosion module enables it to simulate the sediment erosion in the entrainment area. It indicates the volume increase by entraining sediment due to the channel bed variation (Frank et al. 2015). Based upon field observations (Berger et al. 2011) highlighted a rapid erosion rate with a greater flow strength, leading to an increase in erosion depth (Schürch et al., 2011; Frank et al., 2015). (Berger 2010; Schürch et al. 2011) suggested that sediment erosion does not always occur for a small debris flow. Hence, the algorithm used by RAMMS:: DEBRIS FLOW computes the basal shear stress in each of the grid cells to give the value of maximum erosion depth. Back-calculated rheological parameters used in RAMMS are presented in Table 1.

Simulation results and discussions

Attabad landslide was divided into two zones, as indicated in Fig. 10. Zone 1 is a source area with a length of 410 m and ranges from 3360 to 2950 m, whereas zone 2, an erosion area, has a length of 610 m, ranging from 2950 to 2300 m. Figure 12 indicates the evolution process of the Attabad landslide simulated by DAN3D, whereas Fig. 13 presents the results simulated by RAMMS. Figures 12 and 13 portray the detached mass's deposit thickness at 10 s, 30 s, 50 s, and 70 s. The duration of landslide runout was found to be approximately 70 s from both the software, which is in good agreement with the data obtained from the analysis conducted by (Ekström and Stark 2013) shown in Fig. 4d. Results from both simulations indicate that there is a massive increase in slide thickness in the downward runout direction. The maximum deposit thickness can be observed at the end of the runout process in the 70 s after the landslide initiation. The DAN3D results in Fig. 12 show a maximum thickness ranges between 130 to 135 m after the 70 s of the landslide initiation. Moreover, in Fig. 13, the results obtained through RAMMS present a similar behavior, and maximum thickness at the 70 s is approximately the same as obtained from DAN3D.

Fig. 12

Deposit distribution at different time steps of the DAN3D simulation

Fig. 13

Deposit distribution at different time steps of the RAMMS simulation

Figures 14 and 15 indicate maximum velocity obtained from DAN3D and RAMMS, respectively, and it can be seen that landslide runout had a maximum velocity in two places. In the source area, before entering the erosion zone, runout had a maximum velocity of 50 m/s. After entering the erosion zone at an elevation of 2950 m, the slide's velocity decreased gradually. But toward the end of the slide, runout velocity increased abruptly, and this behavior can be seen in the simulation results of both the software.

Fig. 14

Maximum velocity contour of DAN3D

Fig. 15

Maximum velocity contour of RAMMS

The result for the maximum velocity obtained from DAN3D is displayed in Fig. 14. It indicates a maximum sliding velocity of 55 m/s obtained at the bottom right side of landslide runout. The results from RAMMS presented in Fig. 15 portrays the same behavior. RAMMS estimates a maximum velocity of 51 m/s in and around the same bottom right side of the landslide runout. Furthermore, the maximum velocity obtained from each software indicates the same approximate velocity of 50 m/s observed near the right bottom side of the source area. Figures 14 and 15 showcased a high-velocity variation on the right side of the landslide runout compared to the left side. On further analysis, it was found that the surface was highly fractured and sheared with a steep inclination angle on the right side of the study area compared to the overall inclination of the study region. This inclination becomes considerably steep at the bottom right side of the landslide runout. The authors believe that this is a significant reason behind the sudden increase in velocity at the end of the slide. Further, the inclination angle of this part was also checked using RAMMS software, which verified the observational results.

The deposit area of the Attabad landslide is around 400 m wide and about 1.5 km in length (Shah et al. 2013). The landslide involves a massive volume estimated to be approximately 45 Mm3, spread across the accumulation area. Initially, the sliding mass raced toward the opposite slope and accumulated in this area. The deposited material's height at the river's upstream direction is more than 130 m, as shown in Figs. 12 and 13. The deposited material formed a barrier across the Hunza River, thus resulting in a landslide dam. The newly formed landslide dam caused the water level to rise in the reservoir that resulted in a massive lake over one year, as shown in Fig. 8c, and forced the habitants of the area to relocate; also, the landslide mass destroyed the KKH directly and indirectly at multiple locations.

All the results obtained from both software are in good agreement with each other and actual conditions. The final volume is estimated to be 44.9Mm3, which is consistent with the actual landslide condition of Attabad as the detached mass moved down and deposited in the Hunza river, blocking its way and resulted in a natural dam. RAMMS was used to counter check the reliability of results obtained from DAN3D, as both software's work on the Voellmy model. It was found that the results of RAMMS and DAN3D are in good agreement with each other. A subtle variation exists in the results due to different governing equations and contact laws, which is acceptable.

As highlighted in the introduction, the Northern region of Pakistan is highly susceptible to natural hazards and landslides, so it is essential to increase the region's number of seismic and natural hazard monitoring stations. As in the Attabad landslide, no seismic station was located in the vicinity except a single local station, which hindered seismic wave data gathering. Till now, countless rock avalanches have been analyzed in China and North America, and a database of selected paraments has been established and can easily be found in published papers. With the initiation of China Pakistan Economic Corridor (CPEC), it is high time to analyze and prepare a database for regional landslides and hazards. These inventories will assist significantly in research on natural hazard mapping and slope stability measures.

Summary and conclusions

  1. (1)

    A catastrophic rock avalanche occurred on January 4, 2010, at 8:36 am in Attabda, Hunza, Pakistan. In this paper, runout behavior and triggering processes have been analyzed using DAN3D and RAMMS software; the simulations' results were verified with existing data (Ekström and Stark 2013). Based on the analysis, the following conclusions are drawn.

  2. (2)

    Attabad rock avalanche was a non-seismic fatal landslide that killed 20 people and completely demolished two villages Attabad and Sirat. Landslide traveled a distance of 1060 m from top to the deposition area's edge, including 410 m of the source area.

  3. (3)

    The simulation results indicate that the landslide lasted for the 70 s, whereas data obtained from the analysis (Ekström and Stark, 2013) confirmed it. Simultaneously, the average velocity and maximum velocity of the landslide were found to be 14.32 m/s and 55 m/s, respectively.

  4. (4)

    The runout behavior of the Attabad landslide was characterized by simulating it on DAN3D. Frictional–Voellmy model provided the best simulating results, which are in good agreement with the data obtained from literature, as mentioned above. The final volume of slide mass is also in good agreement with simulation results.

  5. (5)

    Attabad rock avalanche was quite different from conventional rock avalanches. It achieved maximum velocity at two points at the start of the slide and the end of the slide. Velocity increased toward the end of the slide due to extra mass entrainment from the bottom right side of the slide, which had a sharp inclination angle compared to the rest of the slide. The entrainment of extra mass also increased the slide's thickness toward the bottom right side, which can be seen in simulation results.

  6. (6)

    Accuracy of simulation results and selected parameters was confirmed by another simulation using RAMMS software. Results indicated that the combined frictional–Voellmy model and selected parameters could provide valuable results for predicting rock avalanches along the Karakoram Highway, especially from Hunza to Khunjerab Pass.


  1. (1)

    The study recommends that in the region from Hunza to Khunjerab, observational/monitoring points for sentinel satellite systems should be installed so that a slightest of ground movement may be recorded and precautionary measures may be taken on time.

  2. (2)

    Northern region of Pakistan has a vast history of landslides and landslides dammed lakes. There have been five massive landslides since 1858. It is recommended that an inventory of potential landslides be prepared, and further human settlements in the areas along KKH with a known history of natural hazards should be restricted.

  3. (3)

    It is also recommended that the number of seismic stations is increased in the area as the area is highly prone to landslide because of the area's active seismic nature. The seismic signal generated by a landslide has a short wavelength. It is believed that limited numbers of seismic stations cannot detect these waves, as happened in the case of Attabad, so it is recommended to densify the seismic stations in the area.

  4. (4)

    In the near future, a railway track is proposed to be laid in the area for the mass movement of goods from China to central Asia; construction of track will require cutting of slopes, which will destabilize the area and chances of land sliding will increase. While keeping the above condition in mind, a detailed study should be carried out regarding slope stability and landslides analysis before the construction of railway tracks.


  1. Ali S, Biermanns P, Haider R, Reicherter K (2019) Landslide susceptibility mapping by using a geographic information system (GIS) along the China-Pakistan Economic Corridor (Karakoram Highway). Pakistan Nat Hazards Earth Syst Sci 19:999–1022. https://doi.org/10.5194/nhess-19-999-2019

    Article  Google Scholar 

  2. Bao Y, Huang Y, Liu GR, Wang G (2020) Sph simulation of high-volume rapid landslides triggered by earthquakes based on a unified constitutive Model. Part I: Initiation Process and Slope Failure. Int. J. Comput. Methods 17. https://doi.org/10.1142/S0219876218501505

  3. Berger C (2010) Debris flow entrainment and sediment transfer processes at the Illgraben catchment. The University of Bern.

  4. Berger C, McArdell BW, Schlunegger F (2011) Direct measurement of channel erosion by debris flows. J. Geophys. Res 116. : https://doi.org/10.1029/2010JF001722

  5. Butt MJ, Umar M, Qamar R (2013) Landslide dam and subsequent dam-break flood estimation using HEC-RAS model in Northern Pakistan. Nat Hazards 65:241–254. https://doi.org/10.1007/s11069-012-0361-8

    Article  Google Scholar 

  6. Cascini L, Cuomo S, Pastor M, Sorbino G (2010) Modeling of Rainfall-Induced Shallow Landslides of the Flow-Type. J Geotech Geoenvironmental Eng 136:85–98. https://doi.org/10.1061/(ASCE)gt.1943-5606.0000182

    Article  Google Scholar 

  7. Chen K, Wu J (2018) PT. Eng. Geol. #pagerange#. https://doi.org/10.1016/j.enggeo.2018.04.002

  8. Chen TC, Lin ML, Wang KL (2013) NU SC. Geol, Eng. https://doi.org/10.1016/j.enggeo.2013.11.018

    Google Scholar 

  9. Christen M, Bühler Y, Bartelt P, Leine R, Glover J, Schweizer A, Graf C, Mcardell BW, Gerber W, Deubelbeiss Y, Feistl T (2012) Integral Hazard Management Using a Unified Software Environment Numerical Simulation Tool “RAMMS.” Congr. Interpraevent pp. 77–86.

  10. Cleary PW , Campbell CS (1993) Self-lubrication for long-runout landslides: examination by computer simulation J Geophys Reshttps://doi.org/10.1029/93jb02380

  11. D’Agostino MC, V., (2008) Comparison Between FLO-2D And RAMMS In Debris-flow Modelling: A Case Study In The Dolomites. WIT Trans Eng Sci 60:10. https://doi.org/10.2495/DEB080201

    Article  Google Scholar 

  12. Dai FC, Lee CF (2002) Landslide characteristics and slope instability modeling using GIS, Lantau Island, Hong Kong. Geomorphology 42:213–228. https://doi.org/10.1016/S0169-555X(01)00087-3

    Article  Google Scholar 

  13. Dai Z, Huang Y, Cheng H, Xu Q (2014) State Key Laboratory of Geohazard Prevention and Geoenvironment Protection. Eng. Geol, Chengdu SC. https://doi.org/10.1016/j.enggeo.2014.03.018

    Google Scholar 

  14. Deubelbeiss Y, Graf C (2013) Two different starting conditions in numerical debris flow models - Case study at Dorfbach, Randa (Valais, Switzerland), in: Graf, C. (Red.) 2013: Matter-Tal - Ein Tal in Bewegung. Publikation Zur Jahrestagung Der Schweizerischen Geomorphologi-Schen Gesellschaft 29. Juni - 1. Juli 2011. pp. 125–138.

  15. Ekström G, Stark CP (2013) Simple scaling of catastrophic landslide dynamics. Science (80-). 339:1416–1419. https://doi.org/10.1126/science.1232887

    Article  Google Scholar 

  16. Evans SG, Hungr O, Clague JJ (2001) Dynamics of the 1984 rock avalanche and associated distal debris ¯ ow on Mount Cayley, British Columbia Canada; implications for landslide hazard assessment on dissected volcanoes. Eng Geology 61:29–51

    Article  Google Scholar 

  17. Frank F, McArdell B, H.C., and V.A., (2015) The importance of entrainment and bulking on debris flow runout modeling; examples from Swiss Alps. Nat Hazards Earth Syst Sci 15:2569–2583. https://doi.org/10.5194/nhess-15-2569-2015

    Article  Google Scholar 

  18. Goudie AS, Brunsden D, Collins DN, Derbyshire E, Ferguson RI, Hashnet Z, Jones DKC, Per-Rott FA, Said M, Waters RS, Whalley WB (1984) The geomorphology of the Hunza Valley, Karakoram mountains, Pakistan. Int Karakoram-Project 2:359–411

    Google Scholar 

  19. Hayat T, Engineering N (2010) Attabad Landslide - Dam disaster in Pakistan ISSMGE Bulletin : Volume 4, Issue 3 Attabad Landslide- Dam Disaster in Pakistan 2010. ISSMGE Bull 4:20–31

    Google Scholar 

  20. Hungr O (1995) A model for the runout analysis of rapid flow slides, debris flows, and avalanches. Can Geotech J 32:610–623. https://doi.org/10.1139/t95-063

    Article  Google Scholar 

  21. Hungr O, Evans SG (2004) Entrainment of debris in rock avalanches: An analysis of a long runout mechanism. Bull Geol Soc Am 116:1240–1252. https://doi.org/10.1130/B25362.1

    Article  Google Scholar 

  22. Hussain SH, Awan AA (2009) Causative mechanisms of terrain movement in Hunza Valley.

  23. Hussin HY, Quan Luna B, Van Westen CJ, Christen M, Malet JP, Van Asch TWJ (2012) Parameterization of a numerical 2-D debris flow model with entrainment: A case study of the Faucon catchment, Southern French Alps. Nat Hazards Earth Syst Sci 12:3075–3090. https://doi.org/10.5194/nhess-12-3075-2012

    Article  Google Scholar 

  24. Hutter K, Koch T, Pluüss C, Savage SB (1995) The dynamics of avalanches of granular materials from initiation to runout. Part II Experiments Acta Mech 109:127–165. https://doi.org/10.1007/BF01176820

    Article  Google Scholar 

  25. Jiao Y, Zhang H, Tang H, Zhang X, Cof A, Tian H (2014) Simulating the process of reservoir-impoundment-induced landslide using the extended DDA method. Eng Geology 182:37–48. https://doi.org/10.1016/j.enggeo.2014.08.016

    Article  Google Scholar 

  26. Johnson AM (1984) Debris flow. In: Brunsden D, Prior DB (eds) Slope Instability. John Wiley and Sons, New York, pp 257–361

  27. Kelfoun K, Druitt TH (2005) Numerical modeling of the emplacement of Socompa rock avalanche. Chile J Geophys Res Solid Earth 110:1–13. https://doi.org/10.1029/2005JB003758

    Article  Google Scholar 

  28. Li X, He S, Luo Y, Wu Y (2012) Simulation of the sliding process of Donghekou landslide triggered by the Wenchuan earthquake using a distinct element method. Environ Earth Sci 65:1049–1054. https://doi.org/10.1007/s12665-011-0953-8

    Article  Google Scholar 

  29. Lin C, Lin M (2015) Evolution of the large landslide induced by Typhoon Morakot : A case study in the Butangbunasi River, southern Taiwan, using the discrete element method. Eng Geol 197:172–187. https://doi.org/10.1016/j.enggeo.2015.08.022

    Article  Google Scholar 

  30. Luna BQ , Cepeda J , Stumpf A, Westen CJV , Malet JP Asch TWJV (2012) Application of a Monte Carlo method for modeling debris flow runout 14 13718

  31. Mao Z, Liu G, Huang Y, Bao Y (2019) A conservative and consistent Lagrangian gradient smoothing method for earthquake-induced landslide simulation. Eng Geol 260:105226. https://doi.org/10.1016/j.enggeo.2019.105226

    Article  Google Scholar 

  32. McDougall S, Hungr O (2005) Dynamic modelling of entrainment in rapid landslides. Can Geotech J 42:1437–1448. https://doi.org/10.1139/t05-064

    Article  Google Scholar 

  33. Pastor M, Haddad B, Sorbino G, Cuomo S, Drempetic V (2009) A depth-integrated, coupled SPH model for flow-like landslides and related phenomena. Int J Numer Anal Meth Geomech 33:143–172. https://doi.org/10.1002/nag.705

    Article  Google Scholar 

  34. Petley DN, Rosser NJ, Karim D, Wali S, Ali N, Nasab N, Shaban K (2010) Non-seismic landslide hazards along the Himalayan Arc. Geol. Act. – Proceedings of 11th IAEG Congress pp 143–152.

  35. Koo RCH, Kwan JSH, Lam C, Goodwin GR, Choi CE, Ng CWW, Yiu J, Ho KKS, Pun WK (2016) Back-analysis of geophysical flows using three-dimensional runout model. Canadian Geotech J. 55(8):1081–1094. https://doi.org/10.1139/cgj-2016-0578

    Article  Google Scholar 

  36. Rickenmann D, Laigle D, McArdell BW, Hübl J (2006) Comparison of 2D debris-flow simulation models with field events. Comput Geosci 10:241–264. https://doi.org/10.1007/s10596-005-9021-3

    Article  Google Scholar 

  37. Salm B (1993) Flow, flow transition and runout distances of flowing avalanches. Ann Glaciol 18:221–226

    Article  Google Scholar 

  38. Salm BW, Burkhard A, Gubler H (1990) Berechnung von Fliesslawinen: eine Anleitung für Praktiker mit Beispielen.

  39. Schürch P, Densmore AL, Rosser NJ, McArdell BW (2011) Dynamic controls on erosion and deposition on debris-flow fans. Geology 39:827–830. https://doi.org/10.13140/RG.2.2.17528.44807

    Article  Google Scholar 

  40. Shah FH, Ali A, Baig MN (2013) Taming the Monster - Attabad Landslide Dam. J Environ Treatment Tech 1:46–55

    Google Scholar 

  41. Valley H, Calligaris C, Departement G, Comi M, Tariq S, Bashir F, Karim D, Assistance FH, Khan H (2010) Executive summary on Attabad landslide survey in Hunza 7–17 April 2010 Short introduction Background of potential glacial lake outburst floods in the Hunza Valley pp 1–20.

  42. Xing A, Wang G, Yin Y, Tang C, Xu Z, Li W (2016) Investigation and dynamic analysis of a catastrophic rock avalanche on September 23, 1991, Zhaotong, China. Landslides 13:1035–1047. https://doi.org/10.1007/s10346-015-0617-y

    Article  Google Scholar 

  43. Xing A, Yuan X, Xu Q, Zhao Q, Huang H, Cheng Q (2017) Characteristics and numerical runout modelling of a catastrophic rock avalanche triggered by the Wenchuan earthquake in the Wenjia valley, Mianzhu, Sichuan, China. Landslides 14:83–98. https://doi.org/10.1007/s10346-016-0707-5

    Article  Google Scholar 

  44. Xing AG, Wang G, Yin YP, Jiang Y, Wang GZ, Yang SY, Dai DR, Zhu YQ, Dai JA (2014) Dynamic analysis and field investigation of a fluidized landslide in Guanling, Guizhou. China Eng Geol 181:1–14. https://doi.org/10.1016/j.enggeo.2014.07.022

    Article  Google Scholar 

  45. Zhu Y, Xu S, Zhuang Y, Dai X, Lv G, Xing A (2019) Characteristics and runout behaviour of the disastrous August 28 2017 rock avalanche in Nayong, Guizhou. China Eng Geol 259:105154. https://doi.org/10.1016/j.enggeo.2019.105154

    Article  Google Scholar 

  46. Zhuang Y, Xing A, Cheng Q, Li D, Zhao C, Xu C (2019)Characteristics and numerical modeling of a catastrophic loess flow slide triggered by the 2013 Minxian-Zhangxian earthquake in Yongguang villageBull. Eng. Geol. Environ Minxian, Gansu, Chinahttps://doi.org/10.1007/s10064-019-01542-x

  47. Zhuang Y, Yin Y, Xing A, Jin K (2020) Combined numerical investigation of the yigong rock slide-debris avalanche and subsequent dam-break flood propagation in tibet, china. Landslides 17:2217–2229. https://doi.org/10.1007/s10346-020-01449-9

    Article  Google Scholar 

  48. Zimmermann F, McArdell BW, Rickli C, Scheidl C (2020) 2D Runout Modelling of Hillslope Debris Flows, Based on Well-Documented Events in Switzerland. Geosciences 10(2):70

    Article  Google Scholar 

Download references


This study was supported by the National Natural Science Foundation of China (No. 41530639) and Guizhou Science and Technology Project ([2017]5402 and [2017]2814). We are grateful to Prof. O. Hungr for supplying a copy of the DAN3D software.

Author information




Hasnain Gardezi designed the research, did the DAN3D simulation, and wrote the paper; Muhammad Bilal did the RAMMS simulation; Qiangong Cheng, Aiguo Xing, and Yu Zhuang performed the research, modified the manuscript, and analyzed the results; Tahir Masood provided the information of the Attabad Landslide.

Corresponding author

Correspondence to Aiguo Xing.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Gardezi, H., Bilal, M., Cheng, Q. et al. A comparative analysis of attabad landslide on january 4, 2010, using two numerical models. Nat Hazards (2021). https://doi.org/10.1007/s11069-021-04593-0

Download citation


  • Runout behavior
  • DAN3D
  • RAMMS debris flow
  • Rock avalanche
  • Dynamics of attabad landslide