Through investigation and analysis of geological conditions and mechanical parameters of the Tazihping landslide, finite-volume method coupling with Voellmy model is used to simulate the landslide mass movement process. The present paper adopts the numerical approach of the RAMMS software program and the GIS platform to simulate the mass movement process before and after engineering treatment. This paper also provides the conditions and characteristic variables of flow-type landslide in terms of flow height, velocity, and stresses. The 3-D division of hazard zones before and after engineering treatment was also mapped. The results indicate that the scope of hazard zones decreased after engineering treatment of the landslide. Compared with the case of before engineering treatment, the extent of high-hazard zones was reduced by about two-thirds, and the characteristic variables of the mass movement in the case of after treatment decreased to one-third of those in the case of before treatment. Despite having engineering treatment, the Tazhiping landslide still poses significant potential threat to the nearby residences. Therefore, it suggests that the houses located in high-hazard zones should be relocated or reinforced for protection.
The hazards of a landslide include scope of influence (i.e., source area, possible path area, and backward and lateral expansion area) and secondary disasters (i.e., reservoir surge, blast, and landslide-induced barrier lake). A typical landslide hazard assessment aims to propose a systematic hazard assessment method with regard to a given position or a potential landslide. Current research on typical landslide hazard assessment remains immature, and there are multiple methods for interpreting landslide hazards. To be specific, the scope of influence prediction of a landslide refers to deformation and instability characteristics such as sliding distance, movement speed, and bulking thickness range. The movement behavior of a landslide mass is related to its occurrence, sliding mechanisms, mass characteristics, sliding path, and many other factors. Current landslide movement prediction methods include empirical prediction and numerical simulation.
The empirical prediction method involves analyzing landslide flow through the collection of landslide parameters in the field. It further consists of the geomorphologic method (Costa, 1984; Jackson et al., 1987; Scott and Vallance, 1993), the geometric change method (Finlay et al., 1999; Michael-Leiba et al., 2003), and the volume change method (Fannin and Wise, 2001). Empirical models are commonly simple and easy to apply, and the required data are easy to obtain as well. Numerical simulation methods are further divided into the continuous deformation analysis method (Hungr, 1995; Evans et al., 2009; Wang, et al., 2016), the discontinuous deformation analysis method (Shi, 1988), and the simplified analytical simulation method (Christen et al., 2010a; Sassa et al., 2010; Bartelt et al., 2012; Du et al., 2015). The numerical simulation method expresses continuous physical variables using the original spatial and temporal coordinates with geometric values of discrete points. Numerical simulations follow certain rules to establish an algebraic equation set in order to obtain approximate solutions for physical variables.
Empirical prediction models only provide a simple prediction of the sliding path. Due to the differences in geological environments, empirical prediction models commonly have low generality. Landslides move downslope in many different ways (Varnes, 1978). In addition, landslides can evolve into rapidly traveling flows, which exhibit characteristics of debris flows on un-channelized or only weakly channelized hillslopes. The geomorphic heterogeneity of rapid shallow landslides, such as hillslope debris flows, is larger than observed in channelized debris flows; however many of these flows can be successfully modeled using the Voellmy fluid friction model (Christen et al., 2012). The selection of model parameters remains one of the fundamental challenges for numerical calculations of natural hazards.
The continuous deformation method has the advantage of an extremely strong replication capability, but it is not recommended when analyzing flow-type landslides, lahars, or debris flows because of complicated rheological behaviors (Iverson et al., 1997; Iverson and Vallance, 2001; Hungr et al., 2001; Glade, 2005; Portilla et al., 2010; Chen et al., 2014). The fluid mechanics-based discontinuous deformation method has several shortcomings such as great computational burden, difficult parameter selection, and difficult 3-D implementation. The simplified analytical simulation method fully takes into account the flow state properties of landslides before introducing a rheological model and can easily realize 3-D implementation on the GIS platform. On that account, this paper adopted the continuous fluid mechanics-based finite-volume method (simplified analytical simulation method). We introduce a rheological model on the basis of using mass as well as momentum and energy conservation to describe the movement of landslides. We also employed GIS analysis to simulate the entire movement process of Tazhiping landslide and map the 2-D division of hazard zones.
Adopting the continuous fluid mechanics-based finite-volume method, this paper took into account erosion action on the lower surface of the sliding mass and the change in frictional resistance within the landslide debris flow in order to establish a computational model. The basic idea is to divide the calculation area into a series of non-repetitive control volumes, ensuring that there is a control volume around each grid point. Each control volume is then integrated by the unresolved differential equation in order to obtain a set of discrete equations. The unknown variable is the numerical value of the dependent variable at each grid point. To solve the integral of a control volume, we make a hypothesis about the change rule of values among grid points, that is, about their piecewise distribution profile. The finite-volume method can satisfactorily overcome the finite-element method's weakness of slow calculation, and solve the problem of complex region processing. Thus, we adopted the finite-volume method to establish the kinematic model for the landslide flow process.
The core of the finite-volume method is domain discretization. The finite-volume method uses discrete points as a substitute for continuous space. The physical meaning of the discrete equation is the conservation of the dependent variable in a finite control volume. Establishment of the conservation equation is based on the continuous movement model, that is, the continuity hypothesis about landslide substances. We divided the landslide mass into a series of units and made the hypothesis that each unit has consistent kinematic parameters (speed at a depth, density, etc.), and physical parameters (Fig. 1). We also established an Eulerian coordinate system-based conservation equation with regard to each control volume.
Schematic diagram of finite-volume discretization (Christen et al., 2010a).
The computational domain is defined as directions
Thus, the mass balance equation becomes
Assuming that
The momentum balance equation is
The kinetic energy balance equation is
The improved Voellmy rheological model is applied in the computational
simulation of the landslide. See the computational formula below:
Synthesizing control Eqs. (1), (3), (4), and (5), we can obtain the
simplified form of the nonlinear hyperbola equation:
In this paper a numerical solver within RAMMS is used, which was specifically designed to provide landslide (avalanche) engineers with a tool that can analyze problems with two-dimensional depth-averaged mass and momentum equations on three-dimensional terrain using both first- and second-order finite-volume methods (Christen et al., 2010b). Therefore, the finite-volume method is adopted to analyze the flow type (high mobility, high velocity, large scope of risks, etc.) of the landslide mass movement process. The present paper adopts the numerical approach of RAMMS and the GIS platform to simulate the mass movement process before and after treatment. The landslide depositional characteristics and the mass movement conditions can be combined to provide a scientific basis for engineering prevention , control, and forecast risk assessments for these kinds of disasters.
The Tazhiping landslide is located southeast of the Hongse village, Hongkou
town, Dujiangyan city of Sichuan Province. The site is located at
(103
Location of Tazhiping landslide, Baisha river basin, Dujiangyan city
(the landslide was triggered by Wenchuan
Tazhiping landslide.
Plane sketch of the Tazhiping landslide.
After the Wenchuan earthquake, the massive colluvial deposits covered the mountain slope. The colluvium is 0.5–5.0 m thick at the top of the slide and is composed of rubble and gravel. The mass consists of a small amount of fine gravel, which is composed of gray or grayish-green andesite with a clast of 20–150 cm. Field surveys indicate that the rubble in the surface layer has a maximum diameter exceeding 2 m, and that fine gravel is loosely intercalated with the rubble. A small amount of yellowish-brown and gray-brown silty clay mixed with 5–40 % of non-uniformly distributed rubble composed the first 5–10 m of the slide. From 10 to 25 m deep, there is a wide distribution of gravelly soil. The soil is grayish-green or variegated in color, is slightly compact and non-uniform, and has a rock fragment content of about 50 %. The parent rock of the rock fragments is andesite, filled with silty clay or silt (Fig. 5). Table 1 shows the parameters of the surface gravelly soil of the landslide mass based on the field sampling.
Parameters of surface soil of Tazhiping landslide.
Geological profile of the Tazhiping landslide.
The landslide is an unconsolidated mass containing relatively large amounts of crushed stones and silty clay (Fig. 6). Its loose structure and strong permeability facilitate infiltration of surface water. The Wenchuan earthquake aggravated the deformation of the landslide making deposits more unconsolidated, further reducing the stability of the landslide mass. During persistent rainfall, surface water infiltrates the landslide slope resulting in increased water pressure within the landslide mass and reduced shear strength on the sliding surface. Thus, rainfall constitutes the primary inducing factor of the upper Tazhiping landslide. After infiltrating the loose layer, water saturates the slope increasing the dead weight of the sliding mass and reducing the shear strength of soil in the sliding zone. Infiltration into the landslide mass also increases the infiltration pressure of perched water, drives deformation, and poses a great threat to villages located at the front of the landslide. Slide-resistant piles and backfill were place at the toe of the slope in order to reduce the hazards of future slides. The slide-resistant piles have enhanced the overall stability of the slope; however, under heavy rainfall the upper unconsolidated landslide deposits may cut out from the top of the slide-resistant piles.
Therefore we simulate possible movement states of the Tazhiping landslide before and after treatment with slide-resistant piles, comparatively analyzed the kinetic parameters in the movement process, and mapped the 2-D division of hazard zones.
It was assumed that the landslide was damaged before engineering treatment.
According to field investigation, the sliding mass had an estimated starting
volume of about 600 000 m
Model calculation parameters.
See the kinematic characteristic parameters of the landslide deposits in
Fig. 7. The colored bar shows the maximum values of the kinematic process for
a given time step. As shown by the calculation results, deposits accumulated
during the landslide movement process had a maximum flow height of 23.85 m,
located around the surface gully of the middle and upper slope. The middle
and lower section of the landslide deposit had a flow height of about
5–10 m; the middle and lower movement velocity of the landslide ranged from
3 and 7 m s
After fully accounting for the slide-resistant piles and mounds, we introduced the Morgenstern–Price method (Morgenstern and Price, 1965) to calculate the stability coefficient of Tazhiping landslide after treatment. The method was determined with an iterative approach by changing the position of the sliding surface until failure of the dump site (Fig. 8). The physico-mechanical parameters under a saturated state (Hydrologic Engineering and Geological Survey Institute of Hebei Province, 2010) were adopted to search for the sliding plane of the landslide.
Search for the sliding plane of the Tazhiping landslide (before treatment).
Based on numerical analysis, the Tazhiping landslide stability coefficient is
0.998. Under rainfall conditions, the middle area of the Tazhiping landslide
was unstable. Loose deposits in the middle part of the landslide might
convert into a high-water landslide and cut out from the top of the
slide-resistant piles. In the damaged area, the slope had a rear edge wall
elevation of about 1170 m. Its front edge was located on the south side of
the mountain road, with an elevation of 1070–1072 m and a length of 182 m.
Thus, the scale of the rainfall-damaged area is estimated to be about
250 000 m
Provided in Fig. 9 are the kinematic characteristics of the landslide
deposit. The colored bar shows the maximum values of the kinematic process
for a given time step. Deposits accumulated during the landslide movement
process had a maximum flow height of 18.37 m, located around the surface
gully of the middle and upper slope. The middle and lower portions of the
landslide deposit had a flow height of approximately 3–5 m. The middle and
lower movement velocity of the landslide deposits ranged between 3 and
5 m s
After treatment, the accumulation flow height and pressure of the deposits
were reduced by about one-half, and the kinematic speed is reduced by about
Landslides reflect landscape instability that evolves over meteorological and geological timescales, and they also pose threats to people, property, and the environment. The severity of these threats depends largely on landslide speed and travel distance. There may be examples where entire houses on a landslide mass are moved but not destroyed because of stable base plates. In any case, velocity plays a more important role regarding kinetic energy acting on an obstacle. However, the Miaoba residential area of Red Village is located at the frontal part of Tazhiping landslide. During landslide movement, the spatial scale indexes of a landslide mass include area, volume, and thickness. The maximum thickness of the landslide is one of the direct factors influencing the building's deformation failure status. A large landslide displacement may lead to burial, collapse, or deformation failure of the building, and thus influences its safety and stability. Thus, landslide thickness constitutes an important index for assessing the hazards of a landslide disaster, and for influencing the consequences faced by disaster-affected bodies (Fell et al., 2008; DZ/T 0286-2015, 2015). Provided in Table 3 is a landslide thickness-based division of the predicted hazard zones of Tazhiping landslide, in which the thickness of the landslide mass correlates with the ability of a building to withstand a landslide disaster (Hungr et al., 1984; Petrazzuoli et al., 2004; Glade et al., 2006; GB 50010-2010, 2010; Hu et al., 2012; Zeng et al., 2015). After treatment with slide-resistant piles, the hazard of a future slide was reduced by about one-third overall and by two-thirds in high-hazard zones.
Division table of the predicted hazards of Tazhiping landslide (unit:
m
The hazard zones of Tazhiping landslide was given by 2-D divisions before and after engineering treatment (Fig. 10). The size of the hazard zones changed after engineering treatment, particularly in the high-hazard zones. Before treatment with slide-resistant piles, the landslide posed a great hazard to eight houses on the left side of the upper Miaoba residential area, with a high-hazard zone associated with landslide mass height over 5 m and a red zone. After treatment, the number of effected houses was reduced to four. We defined outside the colored area as hazard-free.
The hazard assessment of landslides using numerical models is becoming more and more popular as new models are developed and become available for both scientific research and practical applications. There is some confusion about the mass movement process that is discussed by the rheological model presented in this contribution.
Landslides move downslope in many different ways (Varnes, 1978). In addition, landslides can evolve into rapidly traveling flows, which exhibit characteristics of debris flows on un-channelized or only weakly channelized hillslopes. The geomorphic heterogeneity of rapid shallow landslides, such as hillslope debris flows, is larger than observed in channelized debris flows; however many of these flows can be successfully modeled using the Voellmy fluid friction model (Christen et al., 2012). Results presented in this paper support the conclusion that Voellmy fluid rheological model can be used to simulate flow-type landslides.
The selection of model parameters remains one of the fundamental challenges
for numerical calculations of natural hazards. At present, there are numerous
empirical parameters obtained from 30 years of monitoring data. Such as in
RAMMS, we can automatically generate the friction coefficient of an avalanche
for our calculation domain based on topographic data analysis, forest
information, and global parameters (WSL, 2013). The friction parameters for
debris flows can found in some literature (Fannin and Wise, 2001; Iovine et
al., 2003; Hürlimann et al., 2008; Scheidl and Rickenmann, 2010; Huang et
al., 2015). However, there is little research regarding friction parameters
of flow-type landslides. Therefore, we tested different coulomb friction
coefficient
Based on the finite-volume method and the RAMMS program, simulation results of Tazhiping landslide were consistent with the sliding path predicted by the field investigation. This correlation indicates that numerical simulation is an effective method for studying the movement processes of flow-type landslides. The accumulation flow height and pressure of landslide deposits were reduced by about one-half, and the kinematic speed was reduced by about one-third after treatment. However, the Miaoba residential area of Red Village is still partially at risk. Considering that houses of two stories or less within the deposition range might be buried, it was further suggested that the design strength of the gable walls of houses on the middle and upper parts of the deposit be increased above 150 kPa.
By utilizing a GIS platform in combination with landslide hazard assessment indexes, we mapped the 2-D division of the Tazhiping landslide hazard zones before and after engineering treatment. The results indicated that overall hazard zones contracted after engineering treatment and, the area of high-hazard zones was reduced by about two-thirds. After engineering treatment, the number of at-risk houses on the left side of the upper Miaoba residential area was reduced from eight to four. It was thus clear that some zones are still at high hazard despite engineering treatment. Therefore, it was proposed that houses located in high-hazard zones be relocated or reinforced for protection.
The data and software used in this work are available in publicly accessible repositories.
The authors declare that they have no conflict of interest.
The authors sincerely appreciate the CAS Pioneer Hundred 432 Talents Program for the completion of this research. This work was supported by National Natural Science Foundation of China (grant no. 41301009 41301592) and the Hundred Young Talents Program of IMHE (SDSQB-2016-01), the International Cooperation Program of the Ministry of Science and Technology of China (grant no. 2013DFA21720). The authors express their deepest gratitude for this aid and assistance. The authors also extend their gratitude to editor and the two anonymous reviewers for their helpful suggestions and insightful comments, which have contributed greatly in improving the quality of the paper. Edited by: Thomas Glade Reviewed by: two anonymous referees