Experimental study on slope sliding and debris flow evolution w ith and w ithout barrier

Ji-kun Zho*，Dn Wng，Ji-hong ChenCollege of Engineering，Nnjing Agriulturl University，Nnjing 210031，PRChinJingsu Key Lortory for Intelligent Agriulturl Equipment，Nnjing Agriulturl University，Nnjing 210031，PRChinCollege of Civil Engineering，Tongji University，Shnghi 201804，PRChin Reeived 28 Ferury 2014；epted 9 Septemer 2014 Aville online 17 Jnury 2015



Experimental study on slope sliding and debris flow evolution w ith and w ithout barrier

Ji-kun Zhaoa，b，*，Dan Wanga，Jia-hong ChencaCollege of Engineering，Nanjing Agricultural University，Nanjing 210031，PRChinabJiangsu Key Laboratory for Intelligent Agricultural Equipment，Nanjing Agricultural University，Nanjing 210031，PRChinacCollege of Civil Engineering，Tongji University，Shanghai 201804，PRChina Received 28 February 2014；accepted 9 September 2014 Available online 17 January 2015

Abstract

A constitutivemodelon the evolution of debris flow w ith and w ithouta barrierwasestablished based on the theory of the Bingham model.A certain area of the Laoshan Mountain in Nanjing，Jiangsu Province，in Chinawas chosen for experimental study，and the slope sliding and debris flow detection system w as utilized.The change curve of the soilmoisture contentw as attained，demonstrating that themoisture content of the shallow soil layer increases faster than thatof the deep soil layer，and that thegrow th rateof the soilmoisture contentof the steep slope is large under the firstweak rainfall，and thatof thegentle slope issignificantly affected by the second heavy rainfall.For the steep slope，slope sliding firstoccurs on the upper slope surface under heavy rainfall and further develops along the top platform and lower slope surface，while under weak rainfall the soilmoisture content at the lower part of the slope first increases because of the high runoff velocity，meaning that failure occurring there ismore serious.W hen a barrierwas p laced ata high position on a slope，debris flow was separated and distributed early and had less ability to carry solids，and the variation of the greatest depth of erosion pits on soil slopes was not significant.

©2015 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license（http:// creativecommons.org/licenses/by-nc-nd/4.0/）.

Debris flow；Slope sliding；Geological disaster；Time domain reflectometry（TDR）technique；Detection system；Constitutivemodel

1.Introduction

Debris flow isa very destructive geological disaster.Loose materialmoves in response to debris flow's shearing force，thereby creating a secondary disaster induced by erosion. Rainfall is themain reason for slope instability，which leads to large-scale landslides.Current research on processes of debris flow always focuseson numerical simulation and experiments（Yair and Klein，1973-1974；Hottan and Ohta，2000；Magnus and Oliver，2012）.Nicholasetal.（2014）analyzed seven debris flows initiated in proglacial gullies.Gartner etal.（2014）used multiple regressions to developmodels for predicting volumes of sediment.Setting a barrier is an effectivemeasure of controlling theprocessof debris flow.Based on thesimulation and experim ents，many scholars（Mancarella etal.，2012；Brighenti et al.，2013）discussed the barrier's effect on debris flow evolution.Salciarinietal.（2010）used thediscrete elem entmethod to assess theeffectivenessof earthfillbarriers.Mancarellaetal.（2012）studied barrier effects and their possible role in infiltration processesand slope stability.They have found that debris flow was separated when itwent through a barrier，and the barrier's position and rotation angle could change the deposition areas.Time domain reflectometry（TDR）is an electricalmeasurement techniqueused to determine the spatial location and nature of various objects（Robert，2009；Suits etal.，2010；Ragnietal.，2012）.Research on theuseof the TDR detection technology inmonitoring geological disasters beganin them id-1990s（Dowding and Pierce，1994）.Pastuszka etal.（2014）determ ined the impactof the location of TDR probes in soil samples on moisture measurement.Results from some studies showed that TDR detection technology was valid for landslidemonitoring（Liang etal.，2005）.According toanalysis of laboratory tests and field data，scholars have proposed a landslide monitoring method based on this technique，for example，Chen etal.（2009）measured thedielectric constant in highly conductivesoilsbased on surface reflection coefficients.

In summary，real-time monitoring of landslides can be achieved using the TDR technology.However，results are m ostly empirical.The scope of applicability of the regression formula needs further validation.Research on debris flow is difficult due to its sophisticated composition and the variability of dynamic processes.Studies that combine the constitutive theory of the erosion processw ith laboratory tests are few.Research on technology for detection of geological disasters can help to obtain related information and shed light on the process of slope sliding and evolution of debris flow.

In order to explore the process of slope failure under the influence of rainfall，a rainfall-controlled slope model was built based on the geological data of a certain area of the Laoshan Mountain in Nanjing，Jiangsu Province，in China，and a constitutive model of evolution of debris flow under the influence of barriers based on the theory of the Bingham modelwasalso established.The rationality of the constitutive modelwas validated w ith experimental results and inversion analysis.

2.Establishment of constitutivemodel of debris flow evolution

The turbulence power of debris flow can be ignored becauseof high viscosity.Thus，thesimplified Bingham model can be adopted:

whereτis the shear strength，τBis the yield strength，ηis the coefficientof viscosity，and d v/d y is the speed gradient in the y direction（the positive direction is downward）.The Manning equation isused in the formula；the initialspeed is（Han etal.，2012）

where Svis thevolumetric concentration，d10is the lower lim it of particle size，h is the depth ofmud，andβis the gradientof the slope.This formula has been verified w ith themeasured data from the Jiangjia Gully and HunshuiGully.

2.1.Constitutivemodelofdebris flow erosionw ithoutbarrier

Debris flow is affected by friction resistance and internal viscous forces.Theslopesurface resistanceandmassof debris flow at time tican bew ritten as

where fiand miare the slope surface resistance and mass of debris flow at ti，respectively；μ0is the friction coefficient of the slope surface；d m/d t is the change ratio of the mass of debris flow；g is the acceleration of gravity；andΔt is the time interval，whereΔt=ti-ti-1.

The law of conservation of energy can be expressed as follow s:

where viis the velocity at ti，Y is the initial height（relative to the ground）of debris flow，yiis the decreasing heightof debris flow at ti，Wfiand Wsiare the amountsof energy consumed in overcoming the slope surface resistance and viscous force from ti-1to ti，respectively.

Another expression of energy（Legros，2002）at tiis

where hciis the heightof the center ofmass of debris flow at time ti.Then，the energy consumption of debris flow from ti-1to tiis

Combining Eq.（4）w ith Eq.（6）leads to the recursive expression Eq.（7）regarding vi2:

where A，B，and C are expressed as 2g2（yi-yi-1），2g2（yi-Y），and 2g（hci-hci-1），respectively.The common expression of vt2can be obtained:

where vtand ytare thevelocity and decreasing heightof debris flow at time t，respectively；C1is expressed as 2g（hct-hc0），where hc0is the initial height of the center ofmass of debris flow，and hctis the heightof the center ofmass of debris flow at time t；and m0is the initialmass of debris flow.A new equation of shear strength can be obtained by substituting the derivation of Eq.（8）into Eq.（1）:

where mtis themass of debris flow at time t.

Fig.1.Velocity diagram of debris flow with barrier on axis of slope surface.

Fig.2.Velocity diagram of debris flow with barrier rotating through an angle ofφ.

2.2.Constitutivemodelofdebrisflow erosionwithbarrier

The barrier is designed as an equilateral triangular prism in order to sim plify the calculation.It is sufficiently high and fixed on the central axis of the slope（Fig.1）.The distance from thevertex of the barrier to the top platform of theslope is x1.Theenergy consumed by the debris flow in overcoming the friction when it flows through the barrier is ignored，because the effects of the barrier on shunting and obstructing are stronger than those on buffering.

According to the decomposition principle，as shown in Fig.1，v1is decomposed into two symmetrical components at point e1at tim e t1:

whereθis the vertex angle of the barrier.

Sim ilarly，them ass of debris flow has an equal distribution at time t1:

According to the law of conservation of energy，the velocity of the left body at point e2at time t2can be expressed as follows:

where A，B，and C are expressed as2g2（y2-y1），2g2（y2-Y），and 2g（hc2-hc1），respectively.

From e2to e3，the trajectory of the debris flow is approximately a parabola.The resistance of the slope surface consists of longitudinal resistance and lateral resistance.The velocity of debris flow separating from the barrier at e3at time t3can be expressed as

where A，B，and C are expressed as2g2（y3-y2），2g2（y3-Y），and 2g（hc3-hc2），respectively.

The formula of the shear strength of the debris flow on the left sidewhen it passes through the barrier can be obtained by substituting the derivation of Eq.（13）into Eq.（1）:

We can also rotate the barrier counterclockw ise through an angleφ，as shown in Fig.2.Assum ing that the distance from the top platform of the slope to the vertex of the barrier is a fixed value x1，the velocity v1at point e1at time t1w ill be decomposed into vL1and vR1when the debris flow reaches the vertex of the barrier.According to the sine theorem，

where the subscripts L and Rmean the left-side and right-side moving bodies.Themassof themoving body on the left side is not equal to that on the right side under the influence of rotation of the barrier.Themasseson leftand rightsidesafter decomposition at point e1at time t1are

where D=ρtanφR d V，inwhichρis thedensity of thedebris flow，and V is the integration variable of volume.

According to the law of conservation of energy，thevelocity of the left-side moving body at point e3at time t3can be obtained:

where A，B，and C are expressed as 2g2（yL3-yL2），2g2（yL3-Y），and 2g（hLc3-hLc2），respectively.

An identical formula of the shear strength of the left-side m oving body at time t can be expressed as follow s: where mLtis（m0-D）/2+（t-t1/2）d m/d t.

Similarly，the shear strength of the right-sidemoving body at time t is

where mRtcan be expressed as（m1+D）/2+（t-t1）d m/d t.

3.Experimental design

3.1.Establishmentof slopemodel

Fig.3 show s the elevation contour of part of the area of the Laoshan M ountain.Them ountain consists of bedrock on the bottom，the gravel layer，and the soil layer w ith a thickness ratio of about 3:2:9.Most of the area is steep，w ith gradients from 30°to 60°.

A slopemodelw ith two platforms，4m long，2.1m wide，and 2 m high，as shown in Fig.4，was set up indoors.The modelwas a reduced-scale representation of the natural proportionsof the soil structurewith gradients from 30°to 45°.A bevel facew ith a thicknessof 30 cm and a gradientof 5°was builton the bottom，and cementm ortarwas used to level it.It was considered the bedrock of the slope.The slope model consisted of a 20 cm-deep sand layer，a 90 cm-deep clay layer above the sand layer，a flat crestat its top，and a flatbase near the slope toe.The platform of the slopewas about 1 m long.

The soilwas taken from the study area.In order to let the soil return to the pre-disturbance state，the slope model was allowed to stand full consolidation under natural conditions. Before the test，the initial values ofmonitoring indices were measured.After each index reached a relatively stable value，the experimentson slope sliding and debris flow erosion were perform ed.

Fig.3.Elevation contour of study area（units:m）.

Fig.4.Stratum s of slopemodel（units:cm）.

Fig.5.Views of monitoring points in lower part of slope model（units:cm）.

3.2.Detection system

The detection system was connected to eight TDR soil moisture probes，of the type CS635.Each probewas15.0 cm long，the diameter was 0.318 cm，and the size was 5.75 cm×4.0 cm×1.25 cm.The probe resistance changed w ith the increase of the dielectric perm ittivity of soils（Mojid and Cho，2004；W raith et al.，2005）.Fig.4 show s the stratum s of the slopemodelw ith two platforms.The lower part of the slopemodelwas taken as the research object for simulation of slope sliding and debris flow evolution because the flow of mud，the amountofwater collection，and the erosion intensity of this partwere greater than those of the other part.Fig.5 shows theside and top viewsofmonitoring points in the lower part of the slope model.The probes were embedded in two layers，10 cm and 25 cm from the soil surface.

The strain sensors were arranged along the slope on both sides of the axisof the slope surface.They were labeled A and B from right to left（Fig.6）.The sensor detected the deformation by measuring the friction when sliding occurred. Through treatment of collected data，the deformation was transformed into the strain（Cataldo etal.，2014）.

After scaling，mechanical properties of soil at corresponding locations in the study areawere tested.Fig.7 shows that shear strengths of the undisturbed soil in the upper layer are greater than those in the lower layer.Itcan also be reasonably concluded that the shear strength of experimental soil is close to that of undisturbed soil.

3.3.Rainfall generation

The rainfall control system，as shown in Fig.8，ism eant to simulate the rainfall process w ith different intensity levels. Themain system wasmade of 19 PPR pressure pipesw ith a diameter of 20 mm.They were parallel to each other w ith a uniform space of 10 cm and connected by joints-tees.Each of the pipeswasmounted w ith a spraying unitand arranged w ith a certain number of holesw ith a uniform space of 1mm.Two ball valveswere also installed on each pipe to control rainfall.

Fig.6.Positions of sensor ST350（units:cm）.

Fig.7.Comparison of shear strength of undisturbed soiland resultsat monitoring points in experiment.

Fig.8.Structure of rainfall pipes.

4.Result analysis

The experiment consisted of two phases:the firstphasewas rainfall-induced landslides and the occurrence of debris flow，and the second phasewas the evolution of the debris flow.

4.1.Analysis of soilmoisture content after rainfall

During the first phase，the comparative trialsw ith different slopeswere divided into four tests，as shown in Table 1.Two rainfall processes w ith different intensitieswere simulated in each test，and a duration of 15 m in was set between them.

4.1.1.Variation of soilmoisture contentwith time

The curvesof the soilmoisture content in Fig.9 show that，in the first stage of different rainfall intensities，the soil moisture content at points in the shallow soil layer（about 10 cm below the slope surface）increases rapidly，w ith the maximum soilmoisture contentin the rangeof 0.30-0.44，and the increase m ainly occurs during the latter segment of rainfall.Soil in the deep slope layer，25 cm below the surface，has a relatively low level of initial soilmoisture content because the infiltrated rainwater has not yet reached it.Then，affected by rainwater infiltration，the soilmoisture contentsatpoints1，2，and 3 inmost cases increase quickly，while the shallow soil moisture contentsatpoints4，5，6，and 7 decline.In thisstage，the soilmoisture content at each pointwas in a smooth transition state，and slope sliding firstoccurred in the shallow soil layer along w ith shallow landslide gullies.

In the second stage of heavy rainfall，the soil moisture contentatmostpoints demonstrates itssecond phase of grow th for the steep slope，while，the soil m oisture content at the points in the deep soil layer demonstrates that the increasing trend isevenmore significant for the gentle slope than that for the steep slope.As the rainfall continued，the rainwater gradually penetrated into the deep soil，theviolentphenomena of shock and soil slumping began to appear.The slope failure becamemore severe，and the landslide gradually evolved into debris flow.

As the rainfall stops，the soilmoisture contentsat different points tend to be constantw ith time，meaning that them oisture content of the slope soil reaches a stable state.

Table 1 Rainfall conditions of four tests.

4.1.2.Analysis of function of soilmoisture content

The numerical simulations reveal that，for the same rainfall event，the variation of soilmoisture content ismainly affected by the gradient of slope（β）and depth of soil（h）:where wtis the soilm oisture contentat time t；A1，A2，x0，and p are parameters；and f（β）and g（h）are functions ofβand h，respectively.

In the exam ination of the soilmoisture contentat different depthsof soilunder the influenceof rainfall intensity，shallow soil and deep soilwere distinguished by the sliding surface. The soil moisture content above the sliding surface was classified as the shallow soilmoisture content（wst），and the soilmoisture content under the sliding surfacewas classified as the deep soilmoisture content（wdt）.By fitting the experimental results，unified expressions were derived to forecast the soilmoisture content at different depths near the sliding surface:

Fig.10 shows the fitted resultsof the soilmoisture content. Eqs.（21）and（22）are consistentw ith the variationsof the soil m oisture content at corresponding depths.

Fig.9.Soilmoisture content curves at different points under different testing conditions.

4.1.3.Effectofgrowth rateofsoilmoisture contenton slope sliding

Based on the soil moisture content monitoring data，the grow th rate of the soilmoisture content in deep and shallow soils can be expressed w ith the same equation:

Fig.11 shows the calculated results regarding the growth rate of the soilmoisture content in each group of tests.

As shown in Fig.11（a），for the steep slope，the soilmoisture content rapidly increases at the points near the slope surface during the first heavy rainfall，and the soilmoisture content at the point close to the platform increases at an even faster rate，show ing that，under the heavy rainfall，variationsof the soilmoisture content of the steep slope began from the upper slope surface，and grew sequentially along the platform and lower slope surface.As a result，slope sliding first occurred at the upper position of the steep slope during the first heavy rainfall.For the gentle slope，as shown in Fig.11（d），during the heavy rainfall，the shallow soilmoisture content increases rapidly at first，and then the deep soil moisture content increases rapidly after the grow th rate of the shallow soilmoisture content reaches the peak value.Under such a condition，rainwater infiltrated into the shallow soil over a w ide range，and it was highly possible for the gentleslope to slide.In fact，under the heavy rainfall，channels of debris flow developed along the gentle slope，proving the conclusions stated above.

Fig.10.Variations of soilmoisture content at different depths.

Fig.11.Curves of grow th rate ofmoisture contentat different points.

Fig.11（b）and（c）show s that，after the first weak rainfall，the grow th rate of the soilmoisture content of the steep slope is larger than that of the gentle slope overall，and that of the gentle slope is significantly affected by the second heavy rainfall.According to the changes in the soil moisture content，failure modes were different for different slopes:

For a steep slope，the runoff velocity was greater than for a gentle slope during the firstweak rainfall.In the process，the soil moisture content at the lower part of the slope first increased w ith theamountofwatergathering there（see points 6 and 7），resulting in failure at the lower part of the slope surface，whichmay cause further destruction.

For a gentle slope，the runoff velocity was low during the first weak rainfall，and it was easier for the rainwater to infiltrate through the slope surface.In this case，several debris flow gullies formed.Thus，at the beginning of the second heavy rainfall，destruction occurred along thegullies.With the continuous rainfall，destruction along the gullies gradually expanded，followed by the evolution of debris flow and the phenomenon of clods slumping.

Table 2 Positions and rotation angles of barrier set in experimental study.

4.2.Analysis of debris flow evolution

The experimentwas performed on a slope w ith a gradient of 45°and soil density of 2.01 kg/m3.Themajor debris flow was considered the research subject in the experimental study. A 30 cm-high barrier was utilized，and each side of it was 30 cm long.Itwasburied 10 cm deep in the slopesoil.Table 2 shows the conditions for differentpositionsand rotation angles of the barrier in the experimental study.l is the length of the slope，and the barrier position means the distance from the barrier to the top platform along the slope.

The experimental and calculated results of the debris flow velocity under four conditions are shown in Fig.12.In the experim ent，the right side of the flow surfacewas regarded as the research object.Fig.13 show s the curve of the shear strength of the debris flow.

Fig.12 shows that，under condition 1，the debris flow is accelerated，and two seconds and four seconds are two significant turning pointsof the debris flow velocity.Thevelocity of the debris flow reaches the peak value of 1.6 m/swhen it arrivesat the slope toe at4.5 s.Loosematerialswere carried by the debris flow due to the decreasing shear strength，and the volume of the eroded slope soil increased as well.Under condition 2，the debris flow velocity dem onstrated amoderate increase when it came into contacts w ith the barrier as compared w ith condition 1 at the same tim e because the barrierweakened the capacity of the debris flow to carry loose particles.Under condition 3，the barrier diverted the debris flow，leading to a decrease of velocity of 0.1m/sat2.5 s.The mass of debris flow was greater on the right side，and the velocity on the right sidewas significantly higher than thaton the other side.As a result，through use of the barrier，the velocity，kinetic energy，and overall shearing force of the debris flow decreased as compared w ithwhat they did w ithoutuse of the barrier，and the severity of soil erosion was lessened.

As shown in Fig.12，under condition 4，when the barrier was placed at the position of l/2，the debris flow m aintained its originalmovement patterns fora relatively long period of time before making contact w ith the barrier as compared w ith condition 2.During this period of time，the velocity of the debris flow was the same as that under condition 1，and both velocities were greater than that under condition 2.The velocity of the debris flow reached 0.5 m/s at 3.5 s under condition 4 when itmade contactw ith the barrier，and then the debris flow moved at a new ly accelerated pace along the barrier.At this point，themaximum difference between debris flow velocitiesunder conditions2 and 4was0.1m/s，and，after this，the velocity under condition 4 was greater than that under condition 2.Therefore，the soil erosion in the region of l/3 to l/2 from the platform under condition 4wasmore serious than that under condition 2，which meant that the protective effectof the barrier in the downstream region under condition 4 wasweaker than that under condition 2.

Fig.12.Curves of debris flow velocity under different conditions.

Fig.13.Curves of shear strength under different conditions.

Fig.13 shows that there is a nonlinear negative correlation between the shear strength and time.The shear strength of the debris flow reaches its maximum at one second under four conditions.With the evolution of the debris flow，thenumber of loose particles increased.In spite of the increasing velocity of the debris flow，the overall shear strength of the slope soil decreased.Especially for the duration from one to three seconds，there was a significant reduction of the shear strength. The curve for condition 1 flattened after four seconds，and the head of the debris flow was close to the slope toe.Along w ith the worsening liquidity conditions，the overall shear strength wasweak and approached theyield stress.With abarrier placedon the slope surface at the position of l/3 from the platform under conditions 2 and 3，the acceleration of the debris flow decreased after 2.0 s，as compared w ith that under condition 1. The overall shearing force declined and flattened outafter five seconds and six seconds，respectively，and remained constant w ith the value close to the yield stress.The curve of the shear strength under condition 3 isgentler than thatunder condition 2 becauseof the influenceof the rotation angleof thebarrier.The overall energy consumption of the debris flow under condition 3 was larger，the overall shear strength decreased，and the ability of thedebris flow to carry loose particlesalong the slope was relatively weak，leading to the weakened erosion.Under the influence of velocity and mass，total reduction of the shearing force under condition 3 was less than under condition 2 during the evolution of debris flow.Therefore，variations of the shear strength under condition 3 weremoderate.Although there is a certain difference between the calculated resultsand measured data，theoverall trendsaresimilar，and the difference ismainly caused by the fact that the deduced formulas are suitable only for the debris flow in the fluid state.

5.Conclusions

In this study，variations of the soil moisture content and shear strength，as well as their relationships with rainfall intensity and slope gradientwere obtained，and variationsof the velocity of debris flow were also deduced during itsevolution. Themain conclusions are as follows:

（1）With the occurrence of rainfall，themoisture content of the shallow soil layer increases faster than thatof the deep soil layer.As the shallow soil reaches the saturated state，the deep soilmoisture contentw ill increase rapidly，meaning that slope sliding first occurs in the shallow soil layer w ith the occurrence of shallow landslide gullies.As the rainfall stops，soil moisture contents at different points concentrate at certain values，meaning that the moisture content of the slope soil reaches a stable state.

（2）During the process of rainfall，the variation of the soil moisture content at the same depth varies with the slope gradient and rainfall intensity.Under the first heavy rainfall，the variation of the soilmoisture content of the steep slope begins from the upper slope surface，and the soil moisture contentgrows sequentially along the platform and lower slope surface，demonstrating that slope sliding first occurs at the upper position of the steep slope under such a condition. Meanwhile，under the first weak rainfall，changes of the soil moisture contentof the steep slope are significantoverall，and the soilmoisture content at the lower part of the slope first increasesbecause of the high runoff velocity，where relatively serious failuremay occur.

（3）Debris flow acceleratesw ith an initialvelocity asit flows down along the slope，and loosematerials are carried by the debris flow because of thedecrease of the shear strength along the slope surface.The impact of the debris flow on the slope soil increasesw ith the ever-increasingmass of the debris flow.

（4）A barrier placed on the slope surface separates and blocks the debris flow.The installation of a barrier at a high position，rotating clockw ise at a certain degree，has a better effect on reducing the rainfall-induced slope soil erosion and the destruction caused by debris flow evolution.

Acknow ledgements

We are grateful to Professor Matthew Yen at California State University，Fresno，for hisadvice and selflesshelp in the modification of this paper.

References

Brighenti，R.，Segalini，A.，Ferrerob，A.M.，2013.Debrisflow hazardmitigation: a simp lified analyticalmodel for the design of flexible barriers.Comput. Geotech.54（10），1-15.http://dx.doi.org/10.1016/j.compgeo.2013.05.010.

Cataldo，A.，Persico，R.，Leucci，G.，de Benedetto，E.，Cannazza，G.，Matera，L.，de Giorgi，L.，2014.Time domain reflectometry，ground penetrating radar and electrical resistivity tomography:a comparative analysis of alternative approaches for leak detection in underground pipes.NDT E Int.62（3），14-28.http://dx.doi.org/10.1016/ j.ndteint.2013.10.007.

Chen，R.P.，Xu，W.，Chen，Y.M.，2009.Measuring dielectric constant in highly conductive soils based on surface reflection coefficients.J.Geotech. Geoenviron.Eng.135（12），1883-1891.http://dx.doi.org/10.1061/（ASCE）GT.1943-5606.0000170.

Dowding，C.H.，Pierce，C.E.，1994.Measurementof localized failure planesin soil w ith time domain reflectometry.In:Proceedings，Symposium and Workshop on Time Domain Reflectometry in Environmental，Infrastructure，and M ining Applications.US Department of Interior Bureau of M ines，M inneapolis.

Gartner，J.E.，Cannon，S.H.，Santib，P.M.，2014.Empirical models for predicting volumesof sedimentdeposited by debris flowsand sediment-laden floods in the transverse rangesof southern California.Eng.Geol.176（24），45-56.http://dx.doi.org/10.1016/j.enggeo.2014.04.008.

Han，Y.S.，Liang，C.，Han，J.，Huang，P.，Li，L.W.，Peng，S.G.，2012.Gravitation erosion in afflicted areas and its effects of sediment yield and transportation:a case study in Niujuangou Gully.J.Sichuan Univ.（Eng. Sci.Ed.）44（s2），110-116（in Chinese）.

Hottan，N.，Ohta，T.，2000.Pore-water pressure of debris flow s.Phys.Chem. Earth Ser.B 25（4），381-385.http://dx.doi.org/10.1016/S1464-1909（00）00030-7.

Legros，F.，2002.Themobility of long-runout landslides.Eng.Geol.63（3-4），301-331.http://dx.doi.org/10.1016/S0013-7952（01）00090-4.

Liang，Z.G.，Chen，Y.M.，Chen，R.P.，Chen，Y.，2005.Study on applicationsof coaxial-cable electromagnetic wave reflection technique in monitoring slope stability.Chin.J.Geotech.Eng.27（4），453-458.http://dx.doi.org/ 10.3321/j.issn.1000-4548.2005.04.018（in Chinese）.

M agnus，B.，Oliver，S.，2012.Combining airborne and terrestrial laser scanning for quantifying erosion and deposition by a debris flow event.Geomorphology 138（1），49-60.http://dx.doi.org/10.1016/j.geomorph.2011. 08.024.

M ancarella，D.，Doglioni，A.，Simeone，V.，2012.On capillary barrier effects and debris slide triggering in unsaturated layered covers.Eng.Geol. 147-148（12），14-27.http://dx.doi.org/10.1016/j.enggeo.2012.07.003.

M ojid，M.A.，Cho，H.，2004.Evaluation of the time-domain reflectometry（TDR）-measured composite dielectric constant of root-m ixed soils for estimating soil-water content and root density.J.Hydrol.295（1-4），263-275.http://dx.doi.org/10.1016/j.jhydrol.2004.03.012.

Nicholas，T.L.，Andrew，J.M.，Gordon，E.G.，Paul，K.，2014.Debris flow initiation in proglacial gullies on Mount Rainier，Washington.Geomorphology 226（1），249-260.http://dx.doi.org/10.1016/j.geomorph. 2014.08.003.

Pastuszka，T.，Krzyszczak，J.，Sław iski，C.，Lamorski，K.，2014.Effect of time-domain reflectometry probe location on soilmoisture measurement during wetting and drying processes.Measurement 49（3），182-186. http://dx.doi.org/10.1016/j.measurement.2013.11.051.

Ragni，L.，Berardinelli，A.，Cevoli，C.，Valli，E.，2012.Assessmentof thewater content in extra virgin olive oils by time domain reflectometry（TDR）and partial least squares（PLS）regression methods.J.Food Eng.111（1），66-72.http://dx.doi.org/10.1016/j.jfoodeng.2012.01.028.

Robert，C.，2009.Time-domain reflectometry method and its application for m easuring moisture content in porous materials:a review. M easurement 42（3），329-336.http://dx.doi.org/10.1016/ j.measurement.2008.08.011.

Salciarini，D.，Tamagnini，C.，Conversini，P.，2010.Discrete elementmodeling of debris-avalanche impacton earthfillbarriers.Phys.Chem.Earth A/B/C 35（3-5），172-181.http://dx.doi.org/10.1016/j.pce.2009.05.002.

Suits，L.D.，Sheahan，T.C.，Jonnalagadda，S.，Kumar，D.，Jain，P.，Gawande，N.，2010.Comparison of resistivity and time domain reflectometry sensors for assessingmoisture content in bioreactor landfills.Geotech.Test.J.33（3），183-191.http://dx.doi.org/10.1520/GTJ102326.

Wraith，J.M.，Robinson，D.A.，Jones，S.B.，Long，D.S.，2005.Spatially characterizing apparent electrical conductivity and water content of surface soilsw ith time domain reflectometry.Comput.Electron.Agric.46（1-3），239-261.http://dx.doi.org/10.1016/j.compag.2004.11.009.

Yair，A.，Klein，M.，1973-1974.The influence of surface properties on flow and erosion processes on debris covered slopes in an arid area.Catena 1，1-18.http://dx.doi.org/10.1016/S0341-8162（73）80002-5.

Thiswork was supported by the National Natural Science Foundation of China（Grant No.51275250），the Natural Science Foundation of Jiangsu Province（GrantNo.BK2010457），and the AgriculturalMachinery Foundation of Jiangsu Province（Grant No.GXZ14003）.

*Corresponding author.

E-mail address:jikunzhao_2006@163.com（Ji-kun Zhao）.

Peer review under responsibility of HohaiUniversity.

http://dx.doi.org/10.1016/j.wse.2015.01.003

1674-2370/©2015 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license（http:// creativecommons.org/licenses/by-nc-nd/4.0/）.