Source characteristics of long runout rock avalanches triggered by the 2008 Wenchuan earthquake,China

The May 12, 2008 Wenchuan, China Earthquake which measured M_w = 8.3 according to Chinese Earthquake Administration – CEA (M_w = 7.9 according to the USGS) directly triggered many landslides, which caused about 20,000 deaths, a quarter of the total. Rock avalanches were among the most destructive landslides triggered by this seismic event, and have killed more people than any other type of landslide in this earthquake. The Donghekou rock avalanche, one example of a catastrophic avalanche triggered by the Wenchuan earthquake, occurred in Qingchuan and buried one primary school and 184 houses, resulting in more than 780 deaths, and in addition, caused the formation of two landslide dams, which formed barrier lakes.Combining aerial images (resolution of 0.5 m) with field investigations, this paper lists some parameters of 66 cases in one table, and details source characteristics of six typical cases. It has been found that most of the long runout rock avalanches have source areas with high relief and steep inclination, causing the debris in the travel courses to accelerate. There was also a large amount of saturated Holocene-age loose deposits formed by a river or gully that existed in the travel courses. Comparison studies indicate that saturated Holocene loose deposits in the travel courses could be the most important factor for the causes of the long runout characteristic of the rock avalanches especially when they traveled over gentle or even flat ground surfaces.Furthermore, the relationships among the relief slope gradient, runout and covered area are investigated, and a threshold line for predicting the maximum horizontal runout distance under certain change in elevation is presented.     

Slope failures on the flanks of the western Canary Islands

Landslides have been a key process in the evolution of the western Canary Islands. The younger and more volcanically active Canary Islands, El Hierro, La Palma and Tenerife, show the clearest evidence of recent landslide activity. The evidence includes landslide scars on the island flanks, debris deposits on the lower island slopes, and volcaniclastic turbidites on the floor of the adjacent ocean basins. At least 14 large landslides have occurred on the flanks of the El Hierro, La Palma and Tenerife, the majority of these in the last 1 million years, with the youngest, on the northwest flank of El Hierro, as recent as 15 thousand years in age. Older landslides undoubtedly occurred, but are difficult to quantify because the evidence is buried beneath younger volcanic rocks and sediments. Landslides on the Canary Island flanks can be categorised as debris avalanches, slumps or debris flows. Debris avalanches are long runout catastrophic failures which typically affect only the superficial part of the island volcanic sequence, up to a maximum thickness of 1 to 2 km. They are the commonest type of landslide mapped. In contrast, slumps move short distances and are deep-rooted landslides which may affect the entire thickness of the volcanic edifice. Debris flows are defined as landslides which primarily affect the sedimentary cover of the submarine island flanks. Some landslides are complex events involving more than one of the above end-member processes.Individual debris avalanches have volumes in the range of 50–500 km³, cover several thousand km² of seafloor, and have runout distances of up to 130 km from source. Overall, debris avalanche deposits account for about 10% of the total volcanic edifices of the small, relatively young islands of El Hierro and La Palma. Some parameters, such as deposit volumes and landslide ages, are difficult to quantify. The key characteristics of debris avalanches include a relatively narrow headwall and chute above 3000 m water depth on the island flanks, broadening into a depositional lobe below 3000 m. Debris avalanche deposits have a typically blocky morphology, with individual blocks up to a kilometre or more in diameter. However, considerable variation exists between different avalanche deposits. At one extreme, the El Golfo debris avalanche on El Hierro has few large blocks scattered randomly across the avalanche surface. At the other, Icod on the north flank of Tenerife has much more numerous but smaller blocks over most of its surface, with a few very large blocks confined to the margins of the deposit. Icod also exhibits flow structures (longitudinal shears and pressure ridges) that are absent in El Golfo. The primary controls on the block structure and distribution are inferred to be related to the nature of the landslide material and to flow processes. Observations in experimental debris flows show that the differences between the El Golfo and Icod landslide deposits are probably controlled by the greater proportion of fine grained material in the Icod landslide. This, in turn, relates to the nature of the failed volcanic rocks, which are almost entirely basalt on El Hierro but include a much greater proportion of pyroclastic deposits on Tenerife.Landslide occurrence appears to be primarily controlled by the locations of volcanic rift zones on the islands, with landslides propagating perpendicular to the rift orientation. However, this does not explain the uneven distribution of landslides on some islands which seems to indicate that unstable flanks are a ‘weakness’ that can be carried forward during island development. This may occur because certain island flanks are steeper, extend to greater water depths or are less buttressed by the surrounding topography, and because volcanic production following a landslide my be concentrated in the landslide scar, thus focussing subsequent landslide potential in this area. Landslides are primarily a result of volcanic construction to a point where the mass of volcanic products fails under its own weight. Although the actual triggering factors are poorly understood, they may include or be influenced by dyke intrusion, pore pressure changes related to intrusion, seismicity or sealevel/climate changes. A possible relationship between caldera collapse and landsliding on Tenerife is not, in our interpretation, supported by the available evidence.     

Characteristics and numerical runout modelling of a catastrophic rock avalanche triggered by the Wenchuan earthquake in the Wenjia valley,Mianzhu, Sichuan,China

The 2008 Wenchuan earthquake triggered more than 100 rock avalanches with volumes greater than 10 million cubic metres. The rock avalanche with the longest runout amongst these destructive landslides occurred in the Wenjia valley, Mianzhu, Sichuan, China. The landslide involved the failure of about 27.5 million cubic metres of sandstone from the source area. The displaced material travelled about 4,170 m with an elevation descent of about 1,360 m, equivalent to a fahrböschung of 16.9° and covered an area of 1.5 million square metres, with the final deposited volume of approximately 49 million cubic metres. The catastrophic event destroyed the village of Yanjing, killed 48 people and buried some houses at the mouth of the Wenjia valley. On the basis of a detailed field investigation, we introduce basic characteristics of the rock avalanche and find that the rock avalanche resulted in two run-ups and a superelevation along the runout path, and downslope enlargement due to the entrainment of path materials. A numerical model (DAN3D) is used to simulate the post-failure behaviour of the rock avalanche. By means of trial and error, a combination of the frictional model and Voellmy model is found to provide the best performance in simulating this rock avalanche. The simulation results reveal that the rock avalanche had a duration of about 240 s and an average velocity of 17.4 m/s.     

Characteristics and numerical runout modeling of the heavy rainfall-induced catastrophic landslide–debris flow at Sanxicun,Dujiangyan, China,following the Wenchuan Ms 8.0 earthquake

The 2008 Ms 8.0 Wenchuan earthquake triggered a large number of extensive landslides. It also affected geologic properties of the mountains such that large-scale landslides followed the earthquake, resulting in the formation of a disaster chain. On 10 July 2013, a catastrophic landslide–debris flow suddenly occurred in the Dujiangyan area of Sichuan Province in southeast China. This caused the deaths of 166 people and the burying or damage of 11 buildings along the runout path. The landslide involved the failure of ≈1.47 million m³, and the displaced material from the source area was ≈0.3 million m³. This landslide displayed shear failure at a high level under the effects of a rainstorm, which impacted and scraped an accumulated layer underneath and a heavily weathered rock layer during the release of potential and kinetic energies. The landslide body entrained a large volume of surface residual diluvial soil, and then moved downstream along a gully to produce a debris flow disaster. This was determined to be a typical landslide–debris flow disaster type. The runout of displaced material had a horizontal extent of 1200 m and a vertical extent of 400 m. This was equivalent to the angle of reach (fahrböschung angle) of 19° and covered an area of 0.2 km². The background and motion of the landslide are described in this study. On the basis of the above analysis, dynamic simulation software (DAN3D) and rheological models were used to simulate the runout behavior of the displaced landslide materials in order to provide information for the hazard zonation of similar types of potential landslide–debris flows in southeast China following the Wenchuan earthquake. The simulation results of the Sanxicun landslide revealed that the frictional model had the best performance for the source area, while the Voellmy model was most suitable for the scraping and accumulation areas. The simulations estimated that the motion could last for ≈70 s, with a maximum speed of 47.7 m/s.     

Morphological methods and dynamic modelling in landslide hazard assessment of the Campania Apennine carbonate slope

Landslide risk of the Campanian carbonate slopes covered by pyroclastic deposits is mainly connected with the occurrence of high-velocity debris avalanches and debris flows. Analyses show that flows initiate as small translational slides in the pyroclastics. The failure process is controlled by the interaction of both natural and human-induced factors. Geomorphological settings play a decisive role in locating the source failures. Therefore, the crucial aspects in landslide hazard and risk assessment are: (a) recognise the geomorphological control factors, (b) determine parameters defining landslide intensity (velocity, volume, depth of deposit) and (c) predict landslide runout distance. An approach combining geomorphology and numerical analysis has been adopted in the work reported here. Potential future landslide intensity scenarios are simulated predicting the runout behaviour of potential instabilities by using a dynamic model previously calibrated by back-analysing observed events of similar scale and type. The selected area is a sector of the Avella Mountains having the same geomorphological environment as the 1998 Sarno landslides (Campania, Southern Italy).     

A method to develop the input parameter database for site-specific debris flow hazard prediction under extreme rainfall

The predictive hazard analysis at a detailed scale for debris flow runout analysis can be improved significantly through reliable estimation of the input parameters. In this study, a method for database establishment of input parameters at a site-specific scale was laid out for the predictive-based debris flow hazard assessment under extreme rainfall. The adoption of the DAN-3D code necessitated the estimation of three main input parameters: initial volume, bulk basal frictional angle, and growth rate. The initial volume was assessed using a 3D coupled finite element seepage and limit equilibrium-based slope stability analysis. An artificial neural network-based model was developed using 27 debris flow events for predicting the basal bulk frictional angle and consisted of eight factors: plan curvature, profile curvature, percentage of fine content, D₅₀, initial unit weight, initial volume, relative relief ratio, and channel length. Finally, the growth rate was estimated using the previously assessed initial volume, soil depth, and the approximate runout length. The proposed method was validated by application to the Raemian slope in the Woomyeon mountain region, Seoul, for the extreme rainfall event of 27 July 2011. The analysis yielded a final volume of 53,067.9 m³, a velocity upon arrival on the road of 26.81 m/s, and an approximately 0.5-m debris thickness concentrated near the Raemian apartments. The comparison of the predicted debris flow path and debris flow velocity with the actual event demonstrates good similarity and provides a conservative estimate of the volume. This study therefore illustrates the importance of an input parameter database in providing a reliable debris flow runout hazard assessment.     

The Oeschinensee rock avalanche,Bernese Alps,Switzerland: a co-seismic failure 2300 years ago?

Landslide deposits dam Lake Oeschinen (Oeschinensee), located above Kandersteg, Switzerland. However, past confusion differentiating deposits of multiple landslide events has confounded efforts to quantify the volume, age, and failure dynamics of the Oeschinensee rock avalanche. Here we combine field and remote mapping, topographic reconstruction, cosmogenic surface exposure dating, and numerical runout modeling to quantify salient parameters of the event. Differences in boulder lithology and deposit morphology reveal that the landslide body damming Oeschinensee consists of debris from both an older rock avalanche, possibly Kandertal, as well as the Oeschinensee rock avalanche. We distinguish a source volume for the Oeschinensee event of 37 Mm³, resulting in an estimated deposit volume of 46 Mm³, smaller than previous estimates that included portions of the Kandertal mass. Runout modeling revealed peak and average rock avalanche velocities of 65 and 45 m/s, respectively, and support a single-event failure scenario. ³⁶Cl surface exposure dating of deposited boulders indicates a mean age for the rock avalanche of 2.3 ± 0.2 kyr. This age coincides with the timing of a paleo-seismic event identified from lacustrine sediments in Swiss lakes, suggesting an earthquake trigger. Our results help clarify the hazard and geomorphic effects of rare, large rock avalanches in alpine settings.     

From back-analysis to run-out prediction: a case study in the Western Italian Alps

Rock avalanches are complex phenomena that occur with a low frequency but which have a high destructive potential. As a consequence, the people who are responsible for the management of a territory are more and more interested in predicting the possible evolutions of well-known potential events. Tackling the above problems from a quantitative point of view, the RASH3D code, based on continuum mechanics concepts, has been here used to predict the evolution of a potential rock avalanche in the Western Italian Alps. A calibration-based approach, in which rheological parameters are constrained by systematic adjustment during trial-and-error back-analysis of past events similar to the landslide under investigation, is proposed to set rheological parameter values to be used for prediction purposes. The back-analysis of a 2?10⁶ m³ rock avalanche located in the Divedro Valley, close to the area of the potential event, has then been analysed using both a frictional and a Voellmy rheology. The characteristics of the slope and the dynamics of the event have made the frictional rheology more suitable to come to the correct simulation of the historical case. The back-analysis results have contributed not only in the selection of the rheological parameter values but also in the choice of the type of rheological law to use in the carried out forward-analyses.     

Linking rainfall-induced landslides with predictions of debris flow runout distances

Rapid debris flows are among the most destructive natural hazards in steep mountainous terrains. Prediction of their path and impact hinges on knowledge of initiation location and the size and constitution of the released mass. To better link mass release initiation with debris flow paths and runout lengths, we propose to capitalize on a newly developed model for rainfall-induced landslide initiation (“Catchment-scale Hydro-mechanical Landslide Triggering” CHLT model, von Ruette et al. 2013) and couple it with simple estimates of debris flow runout distances and pathways. Landslide locations and volumes provided by the CHLT model are used as inputs to simulate debris flow runout distances with two empirical- and two physically-based models. The debris flow runout models were calibrated using two landslide inventories in the Swiss Alps obtained following a large rainfall event in 2005. We first fitted and tested the models for the “Prättigau” inventory, where detailed information on runout path was available, and then applied the models to landslides inventoried from a different catchment (“Napf”). The predicted debris flow runout distances (emanating from CHLT simulated landslide positions) were well in the range of observed values for the physically-based approaches. The empirical approaches tend to overestimate runout distances relative to observations. These preliminary results demonstrate the added value of linking shallow landslide triggering models with predictions of debris flow runout pathways for a range of soil states and triggering events, thus providing a more complete hazard assessment picture for debris flow exposure at the catchment scale.     

Debris-flow event in the Frangerello Stream-Susa Valley (Italy)—calibration of numerical models for the back analysis of the 16 October, 2000 rainstorm

Three debris-flow simulation model software have been applied to the back analysis of a typical alpine debris flow that caused significant deposition on an urbanized alluvial fan. Parameters used in the models were at first retrieved from the literature and then adjusted to fit field evidence. In the case where different codes adopted the same parameters, the same input values were used, and comparable outputs were obtained. Results of the constitutive laws used (Bingham rheology, Voellmy fluid rheology and a quadratic rheology formulation which adds collisional and turbulent stresses to the Bingham law) indicate that no single rheological model appears to be valid for all debris flows. The three applied models appear to be capable of reasonable reproduction of debris-flow events, although with different levels of detail. The study shows how different software can be used to predict the debris-flow motion for various purposes from a first screening, to predict the runout distance and deposition of the solid material and to the different behaviour of the mixtures of flows with variation of maximum solid concentration.     

Statistical Analysis of Landslide Events in Central America and their Run-out Distance

Statistical analyses of landslide deposits from similar areas provide information on dynamics and rheology, and are the basis for empirical relationships for the prediction of future events. In Central America landslides represent an important threat in both volcanic and non-volcanic areas. Data, mainly from 348 landslides in Nicaragua, and 19 in other Central American countries have been analyzed to describe landslide characteristics and to search for possible correlations and empirical relationships. The mobility of a landslide, expressed as the ratio between height of fall (H) and run-out distance (L) as a function of the volume and height of fall; and the relationship between the height of fall and run-out distance were studied for rock falls, slides, debris flows and debris avalanches. The data show differences in run-out distance and landslide mobility among different types of landslides and between debris flows in volcanic and non-volcanic areas. The new Central American data add to and seem consistent with data published from other regions. Studies combining field observations and empirical relationships with laboratory studies and numerical simulations will help in the development of more reliable empirical equations for the prediction of landslide run-out, with applications to hazard zonation and design of optimal risk mitigation measures.     

Numerical modeling of debris avalanche propagation from collapse of volcanic edifices

Debris avalanches produced from the collapse of volcanic edifices are destructive events that involve volumes up to two orders of magnitude larger (cubic kilometer) than most non-volcanic rock and debris avalanches. We replicate the motion and spreading of several volcanic collapses by means of a depth-averaged quasi-3D numerical code. The model assumes a frictional internal rheology and a variable basal rheology (i.e frictional, Voellmy and plastic). We back analyzed seven case-studies against observations reported in the literature to provide a set of calibrated cases. The ASTER and SRTM satellite-derived digital elevation models were used as topographic data. The numerical model captures the main features of the propagation process, including travel distance, lateral spreading and run up. At varying triggering factors and material characteristics the best fitting parameters span in a narrow interval and differ from those typical of non-volcanic rock and debris avalanches. The bulk basal friction angles (the sole parameter required in the frictional rheology) range within 3° and 7.5° whereas typical values for non-volcanic debris avalanches vary from 11° to 31°. The consistency of the back analyzed parameters is encouraging for a possible use of the model in the perspective of hazard mapping. The reconstruction of the pre-event topography is critical, and it is associated to large uncertainty. The quality of the terrain data, more than the resolution of the DEMs used, is relevant for the modeling. Resampling the original square grid to larger cell sizes determines a low increase in the back analyzed rheological parameters, as a result of the lower roughness of the terrain.     

Catastrophic rockslide-debris avalanche at St. Bernard, Southern Leyte, Philippines

On 17 February 2006, a rockslide-debris avalanche cascaded down the steep slope of Mt. Can-abag, burying the entire village of Guinsaugon in St. Bernard, Southern Leyte, Philippines. Casualties include 139 dead with 980 still missing and presumed dead, making it perhaps the most catastrophic landslide in Philippine history. The landslide started at the ridge top along a fault plane associated with the active Philippine Fault Zone. It started as a block slide that transformed into an avalanche. The entire event lasted for only a few minutes. Estimated maximum landslide velocity is 120–130 m/s. The landslide left behind a deep, wedge-shaped scarp. The central part of the deposit exhibits a hummocky topography typical of avalanches, in contrast to the flatter surface of the debris-flow-type marginal deposit. High amounts of soil in the matrix were derived from the scouring of ancient landslide deposits and rice fields in the valley. The landslide has a total area of 3.2 km² and a runout distance of 4.1 km. Estimated volume of debris is approximately 20 Mm³. At least four streams were dammed by the landslide debris. Intense precipitation and earthquakes preceding the landslide are the potential triggers. Preliminary back analyses assuming a planar and wedge slip surface yielded very low factors of safety even under dry conditions. A more rigorous analysis of the failure mechanism of the landslide is needed.     

WCoE/IPL projects in West Africa: application of Landsat ETM+ and ASTER GDEM data in evaluating factors associated with long runout landslides in Benue hills,North-central Nigeria

Under the WCoE/IPL project, Landsat Enhanced Thematic Mapper plus (ETM+) and ASTER GDEM data were used to analyze factors associated with frequent debris flow and rock-debris avalanche on the Nigeria-Cameroon mountain range. Detailed processing of the ETM+ imageries of the vulnerable portions on the western part of Adamawa plateau was carried out to identify structural trends relatable to the slope failures. The processing identified the structural characteristics of the study area and clearly distinguished the lineaments associated with the debris flow and rock-debris avalanches. The strongly weathered and fragmented gneissic rocks, shallow groundwater table, as well as the distribution of faults and joint sets were significant factors predisposing the slopes to rapid movements. Recent landslides occurred at the locations with high lineament density and in areas close to major faults and long lineaments. The interaction among the moving mass, slope geometry, local geology, topography, and drainage then resulted in a long runout and a high degree of spreading. Analysis of the ASTER–GDEM delineated landforms, slope morphology, and drainage patterns, which was not possible with conventional mapping techniques. Integration of these results yielded reasonable interpretation of the predisposing factors for the long runout failures and provided logical basis for future landslide susceptibility analysis in the area. Accurate investigation of the predisposing factors and characteristics of landslides in environments that have experienced past failures is therefore important in understanding areas that may be susceptible to landsliding in future.     

Global landslide and avalanche hotspots

Allocating resources for natural hazard risk management has high priority in development banks and international agencies working in developing countries. Global hazard and risk maps for landslides and avalanches were developed to identify the most exposed countries. Based on the global datasets of climate, lithology, earthquake activity, and topography, areas with the highest hazard, or “hotspots”, were identified. The applied model was based on classed values of all input data. The model output is a landslide and avalanche hazard index, which is globally scaled into nine levels. The model results were calibrated and validated in selected areas where good data on slide events exist. The results from the landslide and avalanche hazard model together with global population data were then used as input for the risk assessment. Regions with the highest risk can be found in Colombia, Tajikistan, India, and Nepal where the estimated number of people killed per year per 100 km² was found to be greater than one. The model made a reasonable prediction of the landslide hazard in 240 of 249 countries. More and better input data could improve the model further. Future work will focus on selected areas to study the applicability of the model on national and regional scales.     

Dynamics of the Niumiangou Creek rock avalanche triggered by 2008 Ms 8.0 Wenchuan earthquake, Sichuan, China

The Niumiangou Creek rock avalanche was triggered by an M_s 8.0 earthquake that happened on 12 May 2008 in the Sichuan Province, China. The rock avalanche traveled a horizontal distance of 3.0 km over a vertical elevation difference of 0.89 km, equivalent to a coefficient of friction of only 0.29. The travel path of the rock avalanche can be divided into three segments: (1) failing and disintegrating, (2) flying, (3) flowing. In the failing and disintegrating segment, the rock slope failed because of the coupled action of horizontal and vertical force of the earthquake, then smashed into the opposite mountain and disintegrated. In the flying segment, the disintegrating rock mass changed direction and flew into the Lianhuaxin Creek, which was different from the previous research results that concluded rock debris flowed in Lianhuaxin Creek. A great amount of air trapped and compressed under the rock debris acted as air cushion and supported the rock debris to fly a further distance. In the flowing segment, the rock debris flowed on the ground surface in Niumiangou Creek. The flowing velocity has been estimated from the maximum elevation and runup according to the damaged trimlines of the debris. The saturated fine material in Niumiangou Creek entrained by the failed debris mass is thought to have contributed to the long runout of the debris. The Niumiangou Creek rock avalanche is one of the three longest rock avalanches triggered by Wenchuan earthquake. The conclusions of the paper have implications for hazard assessment of potential rock avalanches in the earthquake area and the other similar mountainous area in west China.     

Landslide hazards triggered by the 2008 Wenchuan earthquake, Sichuan, China

The 2008 Wenchuan earthquake (M _s = 8.0; epicenter located at 31.0° N, 103.4° E), with a focal depth of 19.0 km was triggered by the reactivation of the Longmenshan fault in Wenchuan County, Sichuan Province, China on 12 May 2008. This earthquake directly caused more than 15,000 geohazards in the form of landslides, rockfalls, and debris flows which resulted in about 20,000 deaths. It also caused more than 10,000 potential geohazard sites, especially for rockfalls, reflecting the susceptibility of high and steep slopes in mountainous areas affected by the earthquake. Landslide occurrence on mountain ridges and peaks indicated that seismic shaking was amplified by mountainous topography. Thirty-three of the high-risk landslide lakes with landslide dam heights greater than 10 m were classified into four levels: extremely high risk, high risk, medium risk, and low risk. The levels were created by comprehensively analyzing the capacity of landslide lakes, the height of landslide dams, and the composition and structure of materials that blocked rivers. In the epicenter area which was 300 km long and 10 km wide along the main seismic fault, there were lots of landslides triggered by the earthquake, and these landslides have a common characteristic of a discontinuous but flat sliding surface. The failure surfaces can be classified into the following three types based on their overall shape: concave, convex, and terraced. Field evidences illustrated that the vertical component of ground shaking had a significant effect on both building collapse and landslide generation. The ground motion records show that the vertical acceleration is greater than the horizontal, and the acceleration must be larger than 1.0 g in some parts along the main seismic fault. Two landslides are discussed as high speed and long runout cases. One is the Chengxi landslide in Beichuan County, and the other is the Donghekou landslide in Qingchuan County. In each case, the runout process and its impact on people and property were analyzed. The Chengxi landslide killed 1,600 people and destroyed numerous houses. The Donghekou landslide is a complex landslide–debris flow with a long runout. The debris flow scoured the bank of the Qingjiang River for a length of 2,400 m and subsequently formed a landslide dam. This landslide buried seven villages and killed more than 400 people.     

强震区高位滑坡远程灾害特征研究——以四川茂县新磨滑坡为例

近年来,在汶川地震等强震区常发生一种特大的高位滑坡地质灾害,它从高陡斜坡上部位置剪出并形成凌空加速坠落,具有撞击粉碎效应和动力侵蚀效应,导致滑体解体碎化,从而转化为高速远程碎屑流滑动或泥石流流动,并铲刮下部岩土体,使体积明显增加。新磨滑坡就是这种典型,它发生于2017年6月24日,滑坡后缘高程约3450m,前缘高程约2250 m,高差1200 m,水平距离2800 m,堆积体体积达1637×10~4m~3,摧毁了新磨村村庄,导致83人死亡。新磨滑坡地处叠溪较场弧形构造带前弧西翼,母岩为中三叠统中厚层变砂岩夹板岩,是1933年叠溪Ms7.5级震中区(烈度X度)和汶川Ms8.0级强震区(烈度IX度),形成震裂山体。滑源区分布多组不连续结构面,将厚层块状岩体分割成碎裂块体,在高程3150~3450 m区间形成明显的压裂鼓胀区,特别是存在2组反倾节理带,具有典型的"锁固段"失稳机理。滑坡体高位剪出滑动,连续加载并堆积于斜坡体上部,体积达390×10~4m~3,导致残坡积岩土层失稳并转化为管道型碎屑流;碎屑流高速流滑至斜坡下部老滑坡堆积体后,因前方地形开阔、坡度变缓,转化为扩散型碎屑流散落堆积,具有"高速远程"成灾模式。据此,可建立强震山区高位滑坡的早期识别方法,当陡倾山脊存在大型岩质高位滑坡时,应当考虑冲击作用带来的动力侵蚀效应和堆积加载效应,特别是沿沟谷赋存丰富的地下水时,发生高速远程滑坡的可能性将明显增加。因此,在地质灾害调查排查中,在高位岩质滑坡剪出口下方的斜坡堆积体上的聚居区等应划定为地质灾害危险区。在强震山区地质灾害研究中,不仅应采用静力学理论分析滑坡的失稳机理,而且应采用动力学方法加强运动过程的成灾模式研究。     

A test of transferability for landslides susceptibility models under extreme climatic events: application to the Messina 2009 disaster

A model building strategy is tested to assess the susceptibility for extreme climatic events driven shallow landslides. In fact, extreme climatic inputs such as storms typically are very local phenomena in the Mediterranean areas, so that with the exception of recently stricken areas, the landslide inventories which are required to train any stochastic model are actually unavailable. A solution is here proposed, consisting in training a susceptibility model in a source catchment, which was implemented by applying the binary logistic regression technique, and exporting its predicting function (selected predictors regressed coefficients) in a target catchment to predict its landslide distribution. To test the method, we exploit the disaster that occurred in the Messina area (southern Italy) on 1 October 2009 where, following a 250-mm/8-h storm, approximately two thousand debris flow/debris avalanches landslides in an area of 21 km² triggered, killing 37 people and injuring more than 100, and causing 0.5 M € worth of structural damage. The debris flows and debris avalanches phenomena involved the thin weathered mantle of the Varisican low to high-grade metamorphic rocks that outcrop in the eastern slopes of the Peloritani Mounts. Two 10-km²-wide stream catchments, which are located inside the storm core area, were exploited: susceptibility models trained in the Briga catchment were tested when exported to predict the landslides distribution in the Giampilieri catchment. The prediction performance (based on goodness of fit, prediction skill, accuracy and precision assessment) of the exported model was then compared with that of a model prepared in the Giampilieri catchment exploiting its landslide inventory. The results demonstrate that the landslide scenario observed in the Giampilieri catchment can be predicted with the same high performance without knowing its landslide distribution: we obtained, in fact, a very poor decrease in predictive performance when comparing the exported model to the native random partition-based model.