4.1 The March 2012 rainfall event
Rainfall is the major triggering factor identified and monitored at the study site. Water infiltration in the shallow geological deposits has a strong mechanical effect on the dissolution and alteration of the subsurface material and can reach deeper through fractures.
It is known that high intensity rainfall are rare events which trigger landslides and induce significant morphological changes to a large area. A "rainfall event" corresponds to a group of rainy days preceded and followed by at least one (1) day with no rain. Azazga, is one of the most watered regions of Algeria with its Mediterranean climate is exposed to intense and variable meteorological episodes. The frequency and intensity of the rainfall is mainly concentrated from December to February. The average annual rainfall recorded at the “Azazga ecole” meteorological station located at a distance of about 1 km from the area affected by the movements (Fig. 2) for the period between 1950 and 2014 is 974 mm, with a minimum in 2001 (521 mm) and a maximum in 2002 (1536 mm). The highest daily value is 125 mm, which was registered on 6 December 2001. The rainfall mainly concentrates in the period from December to March and accounts for more than 50% of the annual rainfall amount. The more relevant rainfall events are measured in the fall period (December to February), and especially in December.
Analysis of these data showed clearly the great variability and irregularity of the inter-annual rains and their exceptional concentration during a given period. The most historical slope movements recorded in the city of Azazga (1952, 1955, 1973, 1974, 1984, 2003, 2004, 2012 and 2014) are linked to heavy rainfall eventS that exceeded 900 mm. These movements were particularly triggered between December and March for average rainfall exceeding 500 mm.
During the period extending between March 1st to 12th, 2012 Azazga CITY and its surrounding areas in Kabylia experienced heavy rainfall. The unusual high rainfall triggered large number of landslides. The month of March, which corresponds to the end of a winter and the beginning of the spring seasons is known, generally moderate rainfall with occasional localized heavy rainfall in the mountainous regions of the Kabylia. The high-intensity rainfall accompanied by important snow that characterized the winter of 2012 had triggered and reactivated several field instabilities in Algeria.
During the 2011 – 2012 winter season (December- February), diffuse rainfall hit the Azazga region with 564 mm of precipitation recorded in "Azazga ecole" station that represents 55% of the total rainfall of the year of 2012. Furthermore, from early spring of 2012 the same area has been affected by a series of consecutive rainfall events. On March 12th, 2012 a heavy rainstorm was recorded in Azazga region and triggered several landslides where the registered 24-hour rainfall at "Azazga ecole" rain gauge station exceeded 19.6 mm. Since January 2012 the estimated precipitations are 262 mm (January), 232 mm (February), 127 mm (March) with 56.5 mm recorded between 1 to 12 March. Cumulative peak rainfall over these three months reached 691 mm in 2012 (Fig. 5), which represents 65 % of the annual precipitations. The antecedent rainfall at that time could be the principal cause that induced the failure of the Ighil Bouzel slope (Fig. 5). Figure 5 shows the cumulative rainfall between December to March for the last ten years recorded in the city of Azazga indicating that the accumulated rainfall from May to November represents about 80% of the total rainfall. The intensive rainfalls is concentrate in the period December – March that led to the increase of groundwater level. It is obvious that a high accumulation of rainfall may be sufficient to trigger certain types of slope movements whereas for others type of landslides an exceptional climatic events are needed. The antecedent rainfall has influence the degree of saturation of the soil and consequently, plays an important role for the initiation of the landslides (Godt et al., 2006). Table 1 shows the maximum daily precipitation and antecedent rainfalls (3, 7, 10 and 15 days) registered during the period of March 2012 for the aforementioned rainstorm. The values are not very high but played an important role in the initiation of the landslides based on available rainfall records from rain stations installed close to the study area. Considering the series of rainfall data from 1 January 1950 and 12 March 2012 it is possible to know the periods during which the daily rainfall values were higher than those recorded in March 12th, 2012. These exceedance numbers could be considered as the rainfall values of the date of the beginning of the landslide.
4.2 Landslide inventory map
The inventory map of the landslides zone has been prepared at a scale of 1:10,000 based on a combination of satellite imagery, aerial photographs completed by field investigations. The resulted landslide inventory map is presented as polygons in Arc- GIS and draped over a 10-m resolution ASTER-derived digital elevation model (DEM) and geology of the study area (Fig. 6). Figure 6 shows the map of the landslides and the geology of the active zone. The Ighil- Bouzel landslide map was performed based on the results of field surveys carried out immediately after the rainfall event of March, 2012. The ground deformations such as tension cracks, fissures, scarps, minor scarps and bulges as well as disturbed morphology of the clay and scree formations have contributed to a better identification of the unstable areas (Fig. 7). However, as the area is urbanized, the limits of the upstream part of the Ighil Bouzel landslide have been delineated on the basis of the observed damage in constructions, roadways and in the ground (Figs. 8, 9 and 10).
The obtained landslides perimeter of all actived zone covers an area of 0,606 km2 which represents 6,65 % of the total urban area of Azazga (Figs. 6 and 10). The boundaries extend over approximately 1570 m length, between Tadart distict and the Iazoughen river, and a width varying from 350 to 550 m (Figs. 6 and 10). With a mean depth of 18 m, the estimated surface area of the landslide body is 852.5 × 104 m2 and the volume is around 153.45 × 105 m3.The height between the main scarp of the highest slide (Tadart market) and the lowest slide (river) is around 230 m. According to the classification of Varnes (1978), the majority of the mapped slope instabilities can be grouped into slide types. The slides are the most significant in terms of the mobilized volumes and extents (several tens of hectares). They are very widespread in the scree formations and along the area between Aboud and Iazoughen rivers (Fig. 6). They occur on 10° to 15° slopes and showed a disturbed morphology consisting of bulges, cracks, folds and slope breaks. These landslides are slow movements that can under heavy rainfall, evolve into sudden slides recognizable by the formation of head fractures (metric scale scarps) and dense traction cracks of varying depths and lengths. The landslide area is characterized by mottled topography with disturbed slope where the olive trees are inclined and with ground cracks of 1 to 3 cm.
The mapped landslides present elongated shapes and tend to be channeled towards the Iazoughen streams. Numerous mass movements display characteristics of reactivation as attested by the secondary landslide bodies existing inside them. In some cases, they are a partial reactivation of old landslides debris caused by stream erosion. These land instabilities are characterized by a main arched shape, scarp and morphological features typical of slide zones with the presence of hummocky topography and foot ridges indicating the evolution of the movement into the flow.
During the 2012 winter heavy rainfall and snowfall the accumulated precipitation reached 700 mm and many landslides have been occurred and/or reactivated. Landslide activity reaches its maximum during and just after the rainy events. Besides, runoffs related to snow melting and rain as well as infiltrations in the schistose texture marls, favored by the high permeability of fissured rocks, increased the interstitial pressure and make them less resistant to shear stresses that lead to gravitational movements. These instabilities caused damage to several public and private infrastructures. Many ground fissures were also generated and reactivated in different directions after the March 2012 rainfall event. They were quite continuously observed, along hundreds of meters, in the upstream and downstream part of the slope. It consist often on a set of an 'in echelon’-disposed fissures indicating in general a right-lateral displacement.
The Ighil Bouzel mass movements induced by the March 12th, 2012 rainfall event are located along the slope between the Iazoughen and the Aboud river. Landslide affects mainly the scree over a length of 1000 m and a width of 550 m and covers a perimeter of 0,3922 km2. With a mean depth of 18 m, the estimated surface area of the landslide body is 15.5 × 102 m2, the volume is around 27.9 × 103 m3. The unstable areas is presented in the form of an elongated catchment area where various sources of waters converged. The high humidity of the slope due to an abandoned drainage network, the discontinuity of the scree/substratum and the location of faults sometimes visible in the upstream part of the slope are probably the main factors of this landslides. Upstream, the slope is around 10° while downstream the slope becomes steeper and remains marked by traces of recent activity such as distension and compression cracks as well as lateral and transverse compression ridges. Several springs, such as the Tala Oukouchah spring, coming from the Numidian sandstone lead to water spread in the clay-sand deposits. The faults that limit the entire upstream unstable zone are the preferred location for water infiltration at the head of the landslide. The landscape also presents an often-mottled morphology linked either to the solifluction or to slower, larger-scale land movements. The shape and mode of sliding are probably controlled by the internal structure of the flysch bedrock dipping in the slope direction and the orientation of the schistose texture planes in the dip slope direction as well as the contact limit (interface) between basement and the flysch. The dense hydrographic network shows active erosion due to a relatively high rainfall (700 to 900 mm/year), that maintains the effective saturation of the soils and the fragile, clayey terrain. In addition, the down cutting erosion (regressive erosion) at the feet of slopes contribute to their destabilization.
4.3 Deformation process and characteristics of the landslide induced by the March 12th, 2012 rainfall event
4.3.1 Characteristics of landslide event
The rainfall-induced landslide is located in Ighil Bouzel village on the right bank of the Iazoughen river, an ancient active zone with an extremely wide distribution area. It represents one of the most active landslides experienced by the region during recent years that have affected highly the urbanized area and caused damage to many constructions, roads and agriculture areas. The landslide is located along a versant with an average slope of 12°, along the Aboud river that converges towards the Iazoughen river (Fig. 6). The urbanization of this slope started in early 1980 has led to an intense morphological change resulting from earthworks, constructions and roads opening. The historic inventory of these ground failures shows that rainfall is the main triggering factor. This land instability was known since February 1973, reactivated, respectively, in April 1974, in March 1985 and in March 2012 always following extended period of rainfalls that characterized the winter season. The most landslide-prone zones are located in the right bank of the Iazoughen river and in the most urbanized zone of the Azazga city, as shown in Figure 9. The landslide surface present a significant number of tension cracks, scarps, minor scarps and bulges that developed gradually at the middle of the landslide. These deformations caused damage to some residential houses, roads, and public infrastructures.
The Rainfall induced landslide of Ighil Bouzel corresponds to a large complex deep active landslide. The slope movement combine, upstream, characteristics of a rotational landslide (with two juxtaposed semi-circular scarps and two slightly down slope sectors downstream) and a characteristics of a translational landslide in the body of the landslide downstream. The slope of Ighil Bouzel has been sliding towards the Iazoughen river, with active peaks of accelerated movement. Additionally, the very altered and fractured nature of the substratum implied a main role of meteorological water infiltration (rainfall, snow melt) affecting the equilibrium state of the slope. The section of the landslide is crossed in the middle and the head by the two important national roads RN 71 and RN12 and municipality ways (Fig. 4).
4.3.2 Deformation characteristics based on field observations
The landslide presents a particularly chaotic topography aspect and a significant series of scarps (most of which are the main scarps for minor landslides - slide-flow), subsidence, transverse cracks, bulges parallel to the general slope and separated by cliffs. The landslide has experienced a long deformation history that was observed every year, particularly during the rainy season, showing a potentially highly activity area with a main scarp. Within the main landslides body have been observed arcuate main scarp (Fig. 7a) and minor scarps, fissuring, depressions, bulges, and minor landslides. Along the flank of the medium section of the landslide continuous shear surfaces with evident slickensides and lateral levees were observed (Fig. 7a, b, c, e, f and g) as well as centimeter to several meters- long fissures and ground settlement along the Aboud and Iazoughen waterways (Fig. 7c, d and e). A gentle bulge in ground surface (Fig. 7c) frequently delineates landslide toe, sometime tongue shaped in plan. Landslide toe, often tongue shaped in plan, is subsiding with high slopes (Fig. 7h). Large cracking and ground cracks were induced in the middle and the toe part of the landslide, reaching a maximum length of ~ 50 m, a width of ~ 50 cm. The cracks are observed along the main national roads of RN 12 and RN 71 (Fig. 8 c, e, f, and h).
The landslide deformation intensified in the middle with a large number of ground fissures tension cracks were found with lengths extending from 1 to 50 m, widths from 5 to 30 cm and vertical displacements of 10–60 cm. The activated landslides on March 2012 is considered as partial reactivations of pre-existing landslides deposits. Some of these landslides have experienced multiple reactivations, with also damages that manifested over the last decades.
The reactivation of the landslides was sensitive to torrential rainfall as the main triggering factor and the fragile geological structure with previous slope movement as the susceptibility factors. Under the action of rainfall, the slope gravity and the pore water pressure increased. Basal erosion by the river torrents the foot of the landslide area, markedly destabilizing the slopes. During rainstorms season, surface water penetrated into newly formed cracks, increasing the fragility of the soil. Meanwhile, the water content of materials increased, then cohesion loses and effective friction angle on sliding belt decreased noticeably on wetting state, and the stability of slopes was at limit equilibrium state. This leads to decreased resisting forces and increased sliding forces. In addition, the fragile geological structure contributed to the slide occurrence. These discontinuities favored the slide occurrence as water of rainfall penetrated into soil through these joint sets, while internal or external erosion can form new rupture surface and finally reactivated formed sliding surface. Thus, these conditions resulted in the reactivation of the Ighil Bouzel within a short time after heavy rainfall.
The in situ investigations conducted immediately following the landslide event showed that the natural surface drainage system seems to have increased the accumulation of water at the surface, sub-surface and even groundwater in some areas, as well as the collection and stagnation of large quantities of water, making the soils more susceptible to the landslides occurrence (Fig. 7 h). Moreover, rainwater infiltrated through the tension cracks and reached the internal weak layer.
Based on morphology and deformation characteristics, the Ighil Bouzel landslide can be classified into two deformation zones: (i) the rear low deformation zone in upstream showing small deformation features, with some crakes and fissures were induced during the 2012 rainfall event. The elevation of the upstream area ranges between 440 to 500 m, and the mean gradient is 10°. This area has a surface of 400 × 470 m2 and (ii) the front high deformation zone in downstream (Figs. 7, 8, 9, 10, 14 and 12) with expanded, large, heavy and abundant deformations such as transverse cracks, bulges, tensile cracks and subsidence. The elevation of the downstream zone ranges between 320 to 420 m, the mean slope is 20° and the extent is approximately 1170 × 550 m2.
4.3.3 Damagebased on field observations
After the slope failure, the landform topography had deformed significantly presenting a large number of ground fissures, cracks, a surface bulge that caused collapse of national roads, rural roads, residential houses and public infrastructures. The landslide began, with slight cracking starting to appear in the roads and around some of the houses in the head part of the landslide. Figs. 6, 7, 8 and 10 show the landslides and prepared damage map based on field surveys carried out immediately after the rainfall event of March 2012. The landslides caused damage to the provincial road, several private constructions and agriculture mainly in the upstream and middle portion of the landslide. Some residential buildings were also experiencing serious cracking, accompanied by noise, with the situation obviously deteriorating. In addition, the landslides caused indirect costs related to the many people evacuated for both short and long periods of time. Immediately after the event, 20 houses were formally evacuated, because either of structural damage suffered or as precautionary measure to avoid damage to people.
Several constructions exhibit cracks to total destabilization, two roads including the national roads RN 12 and RN 71 were deeply damaged and put out of traffic service (Fig. 8e, f, g, h). In total, ten houses were severely damaged in the villages of Tala Oukouchah, Zen and Ighil- Bouzel (Fig. 8c). The landslide induced large-scale damages with cracking of constructions and shear zones in columns, beams, load-bearing walls as well as tilting (Fig. 8c). The most significant damage affected the cultural centre (Fig. 8a), the stadium (Fig. 8b), dwellings, and water supply pipes (Fig. 7 d) that showed tilts (Fig. 8f), fissures and utility poles inclined (Fig. 8f). In addition, several other housing units suffered minor cracks.
4.3.4 Field evidences of previous landslides
The Ighil Bouzel landslide includes a series of unstable areas located along a slope with an average slope of 100, oriented North-South towards the Aboud and Iazoughen rivers. Many field observations in and around the landslides zone revealed the existence of previous evidences of instabilities and indicate that the landslides induced by the March 2012 rainfall event are partial reactivations of pre-existing landslides. Indeed, field observations showed steep to moderate slopes where the reliefs have a mottled morphology characteristic of low stability slopes with counter slopes, double crests, counterscarps, old scarps and long cliff-like scarps parallel to the general slope trend that are evidence of ancient movements as can be shown in Fig. 9. These are common forms of instabilities that have already functioned in past landslide still slowly evolving nowadays. This specific morphological feature is evidence of a deep-seated landslide affecting hundreds of meters along the Iazoughen River (Fig. 9). (Fig. 10).
The area is affected by a series of successive rotational slides where the displaced materials show chaotic morphology characterized by a strongly hummocky topography with the bulges increasing in a downstream direction. Upstream, the slopes are gentler with scarp zones consisting of multiple interacting rotational landslides (Fig. 9). The cutting zone is reshaped by a series of complex rotational and translational slides. The bodies of landslide are dotted with numerous indicators of 'fresh' instability (shear and friction cracks, fractures, bulges, drunken forest, seepages and stagnant water). Finally, in the most downstream part, the basal undercutting of the Iazoughen stream by the suppression of the abutment and the public landfill have worsened the instability. Downstream area is marked by human activities such as farming, earthworks, and housing (Fig. 10).
Various morphological and vegetation indices (chaotic topography, slope breaks, dead trees) are evidence of the landslide (Fig. 9). According to local witness the landslide occurred for the first time in February 1973 and was successively reactivated on April 1974, on March 1985 and on March 2012. This is a typical rotational landslide with a moderately sloping curved main scarp, longitudinal cracks and a rotational slide surface. The landslide material is principally made of lower Cretaceous clays. The locations of the photos taken (1-6) of figure 9 are shown on Fig. 10.
5 The landslide-inducing factors
The occurrence of the large Ighil Bouzel landslide can be attributed to two causative factors: (i) The triggering factors related to the antecedent rainfall and human activity through slope excavations embankment and deforestation; (ii) The susceptibility factors such as geology, morphology, hydrogeology and geotechnical characteristics of the soil.
5.1 Influence of rainfall on landslides: antecedent rainfall
Rainfall is largely regarded as the principal triggered factor of landslides. Prolonged and low intensity of rainfall often generates deep-seated landslide movements versus shallow soil landslides are mostly triggered by brief and intense rainfall (Hong et al. 2005). The historical landslides inventory carried out over the period of 1950-2014 shows that the main landslides in the city of Azazga are related to heavy rainfall event that exceeded 900 mm: 1952 (1190,8 mm), 1955 (992 mm), 1973 (1287,8 mm), 1974 (884 mm), 1984 (1236,1 mm), 1985 (930,7 mm), 2003 (999,6 mm), 2004 (917,6 mm) and 2012 (2086,7 mm). Landslides mainly triggered during the rainy period (October at April) during an average an average precipitations exceeding 500 mm. Such a high intensity and accumulation of rainfall can be enough to trigger certain types of landslides whereas for others type of landslides an exceptional climatic events is needed. However, the antecedent rainfall has influence the degree of saturation of the soil and consequently, plays an important role for the initiation of the landslides (Godt et al., 2006). The peaks of cumulated precipitations over these months reached 876 mm in 1952 (73% of the annual amount), 557 mm in 1955 (56 % of the annual amount), 841 mm in 1973 (65 % of the annual amount), 477 mm in 1974 (73 % of the annual amount), 852 mm in 1984 (68 % of the annual amount), 575 mm in 1985 (62 % of the annual amount), 544 mm in 2003 (54 % of the annual amount), 886 mm in 2004 (97 % of the annual amount) and 691 mm in 2012 (33 % of the annual amount). During the winter season of 2012 (December to February), the Azazga region experienced prolonged and intense rainfall with the cumulative rainfall exceeding 550 mm that represents 55 % of the total annual rainfall (Fig. 5). From January to March repeated rainfall episodes with high intensity resulted in a cumulative peak rainfall reached 700 mm which represents 65 % of the annual precipitations (Fig. 5). The first rainfall episode, from 12th to 17th January, was characterized by cumulative precipitation in excess of 80 mm. The second rainfall episode, from 19th to 28th January, was severe, and was as much as 153 mm. The third rainfall episode, from 2th to 13th February, was characterized by more severe precipitation in the range from 161 mm. The fourth episode, from 18th to 28th February, was characterized by cumulative precipitation around 71 mm. The last rainfall episode, from 1th to 14th March, was characterized by cumulative value exceeding 85 mm and the maximum daily rainfall exceeding 20 mm. The antecedent rainfall at that time could be the principal cause that induced the failure of the Ighil Bouzel slope. During the continuous and frequent heavy rainfall the infiltrated water increase the groundwater level and pore pressure in both the upper-soil slope and the lower-rock slope that contribute to the alteration of the material in interlayers and the degradation of its mechanical resistance with the loss in the shear strength, which finally influenced the slope stability. Continual rainfall may have increased the water content of the sliding mass and saturated the slip zone, leading to a major loss in the shear strength of the sliding zone. Therefore, the stability of the slope progress to the state of limit equilibrium. Still, though long-term rain infiltration exists, the groundwater level still increases considerably because of the low permeability of the slope.
5.2The human activity
The human activity represented by intense and fast-unplanned urbanization mainly since 1978, has led to significant morphological changes in the stability conditions by deforestation, excavation, embankments, land use changes, slope cutting for constructions and road increasing, hence, the frequency of Ighil Bouzel landslide. The most affected zones by landslides correspond to the badly protected one where the natural vegetation is degraded by deforestation, excavations and urbanization. These excavations degraded the quality of the soil mass, decreased the holding force of the slope and moved the lower rock slope towards a disequilibrium unstable state, which finally contributed to the slope failure. The gradual extension of the urban area with extensive land use activities in the northern parts of the city along inappropriate land use, environmental mismanagement and a lack of rules and regulations constitute the main factors of the increase of the frequency of landslides. Deforestation, with the removal of trees and the clearing of land for crops, increases the infiltration of rainwater into the soil, which has a negative impact of the slopes stability.
5.3Local geomorphology
The Ighil Bouzel landslide is located on a wide north-facing slope of 250 m to 500 m of altitude. It occurred on a steep slope with the slope dipping 10-20° northward (Fig. 11). The slope of the lower rock, with a structure of flysch bedding planes, was relatively steeper than the upper earth slope consisting of a thick layer of scree. In addition, the dip of the diaclase and the flyschs schistosity planes are dipping in the slope direction. Moreover, the morphology shows watershed with the presence of concave slope depression that favor natural accumulation of surface and subsurface runoff and likely groundwater during the storms, thereby making them more susceptible to landslides. The slope is cut by a set of valleys that forms the main artery of the Iazoughen stream. The valleys have narrow bottoms and steep slopes with dominant south-east - north-west direction. The Iazoughen network converges towards the main landslide area located between the settlements of Tala Oukouchah and Ighil Bouzel (Fig. 11). Field observations showed a relatively steep slope about 15° where the landscape show a mottled morphology and greater deformation with counter slopes, small displacement sheets, old tearing scars and small to high escarpments that characterize low unstable slopes (Figs. 7, 9 and 11). The landslide zone is characterized by a series of bulges parallel to the general slope and separated by the main scarps (Figs. 11). A series of tension cracks of varying depths and lengths were also encountered in the downstream part of the unstable slope (Fig. 11). Semi-permanent torrential-type flows have created gullies in the slope and caused shifts in the banks. Slope movements affects thick scree slopes that are widely developed on the surface and clearly visible in the landscape by their hilly shape and low vegetation cover. Human activities, especially after 1980 with large private constructions has led to a notable morphological modification by earthworks, filling and opening of access roads to the different districts. Through these geomorphic features, the entire slope presented limited equilibrium under normal conditions.
5.4Local geologyand structure
According to the field surveys, borehole data, the landslide structure is mainly represented by overlying thick Quaternary-scree deposits as upper-earth slope and the underlying bedrock of the Cretaceous flyschs as lower rock slope (Figs. 3 and 6). The analysis of lithological section across the landslide zone established from the boreholes and surface observations extending from the city to the Iazoughen river allows us to identify two lithological units (Fig. 11b): (i) The Cretaceous flyschs which forms thick series of gray to greenish clays finely bedded, folded and friable, can be cut into thin platelets, moist on the surface and compact at depth forming the substratum of the site. The borehole sections indicate thick-bedded and highly fractured hard rock belonging to the Cretaceous allochthone complex (Gelard, 1979). The flyschs deposit is intersected by most of the boreholes at depth ranging between 3 and 31 m as shown on the cross section AA' (Fig.12b). The depth decrease from the north to south from 5 m to 31 m and from the west to the east from 10 m to 25 m, (ii) The quaternary screes that cover the Cretaceous flyschs is made of sandstone blocks of different sizes embedded in a clayey to sandy-clay deposits.
The tectonic discontinuities (faults, fissures, shistosity) constitute the drain for water infiltration (Figs. 3 and 6). In addition, the lithological discontinuity between the permeable formations (scree) and the impermeable formations (flysch) constitute a major hydrological discontinuity planes.
5.5Local hydrology and hydrogeology
The study area is crossed by a drainage network essentially represented by the Iazoughen and Aboud permanent flow waterways associated with their temporary flow affluent drainage network. These watercourses with an irregular flow regime, sometimes becoming excessive during the wet season. The 'fluvial and torrential' system of Iazoughen and Aboud rivers passed through the slope toe generates a significant runoff, saturation of the soil, basal erosion on the banks, strong under cutting erosion of the slope accompanied by lateral undermining of the banks, which activates the dynamics of the slope and caused the removal of the slope toe (removal of abutment). Runoff water infiltration into the soil is possible through the presence of tectonic discontinuities and the existence of local, more permeable sandy layers (scree). Seasonal exceptional precipitation sequences combined with the processes of erosion, and infiltration increase the water content and the saturation of the soil layers that decreases the effective stress in the soil, and reduces shear strength, resulting in slope movement. The river dynamics that influence the triggering of landslides are generated by the concentrated runoff of water and are conditioned by the flow rates as well as by the lithology and slope. This process is observed in the Iazoughen and Boulina streams in 2012 and is manifested by basal erosion by the river torrents that lead to collapse at the foot of the slope. Seasonal heavy rainfall and fluvial incision by the river may induce failure of the banks due to slope undercutting and seriously influenced the slope stability. Moreover the Quaternary scree deposits that overlapping the clay flyschs is relatively permeable and can allow water flow channels and form groundwater table on the relative impermeable clay bedding surface where water accumulate within the vertical tension cracks during wet seasons. The infiltration of water into the contact zone between the scree deposits and the underlying flysch marls can decompression of the marls and causing the collapse of scree along the versants.
The analysis of the static level map of groundwater table based on the inventory of 78 water points (13 piezometers, 4 sources and 60 wells) by using Kriging interpolation (Isaaks and Srivastava, 1989) shows the presence of a continuous and generalized shallow groundwater with a static level of the groundwater table at depth ranging between 2.5 to 5 m (Fig. 12a). This groundwater is contained in the quaternary sandy-silty cover formations (scree) which are quite permeable. The values of static level of the groundwater table increase from the South to North indicating a water flow from south toward the North. Figure 12b illustrates the lithological structure of the landslide and the groundwater table in the landslide zone. Considering the lithological and the hydrogeological conditions of this landslide, the groundwater is mainly found in the cover quaternary scree, which could be contained in the contact area between the bedrock (flych) and the overlying deposits (scree). Groundwater is discharged to the ground surface in the form of a sources, springs and water wells (Figs. 4 and 13). The groundwater level is seasonally influenced by rainfall and closely associated with the amount of precipitation and may be quite high when the monthly precipitation is high as observed during heavy rainfall through sources and wells (Figs. 13c and d). The basement probably remains saturated all year. The hydrogeological conditions of the subsoil, which is often saturated, explain the low stability of the soil when the natural slopes exceed 10°. Indeed, the shallow water table in the study area increases the susceptibility of the deposits to landslide during rainfall events. The river dynamics that influence the triggering of landslides are generated by the concentrated runoff of water and are conditioned by the flow rates as well as by the lithology and slope. This process is observed in the Iazoughen stream and is manifested by water undercutting that lead to collapse at the foot of the slope (Fig. 13a and b). Seasonal heavy rainfall and fluvial incision by the river seriously influenced the slope stability and seriously damaged soil structures.
5.6 Geotechnical characteristics
Lithologicaly, the landslide area is characterized by the impermeable flyschs bedrock basement overlain by permeable Quaternary scree deposits. Comprised of a loose and porous soil layer which makes it favorable for surface water infiltration and groundwater retention. The flyschs that is not outcropping in the landslide area is mostly constituted by blue clays at the base and green at the top. The basement depth ranges between 3 to 31 m as shown on the cross section AA' (Fig. 12b). The quaternary screes deposits cover, almost the whole studied area (see the cross sections AA’) and are represented by sandstone blocks embedded in a clay-sandy deposits.
The statistical analysis of the geotechnical data allowed us to distinguish three geotechnical units (Table 2): (1) Blue and green clays form the flysch substratum with variable geotechnical characteristics that include two terms (Table 2): (i) The Weathered greenish-grey clays correspond to the upper part of the weathered flysch bedrock. They coincide with a fine soil (56<Fc<98%) that is moderately dense (1,35<γd<1,88 t/m³), saturated (Sr of 93%), plastic and very sensitive to the presence of water (LL vary between 48 and 77% and PL between 20 and 37%). They have low cohesions (Ccu between 0,15 bars and 1,3 bars) and low angles of friction (1,91°<Φcu< 22°). The oedometric tests showed a fairly to moderately compressible soil (Cc between 11% and 24%) and swellable soil (Cg between 2,8% and 9%), (ii) The Compact grey-blue clays corresponds to the compact part of the Cretaceous flysch substratum at depth. The statistical analysis of the physical parameters, showed a fine soil (55<Fc<100%), of average density that increases strongly with the depth (γd ranges between 1.52 and 2 t/m³). The degree of saturation reaches 100% in the first few meters and then Sr = 37%) at depth. The soil is classified as not very plastic (LL= 50 % and PL = 22). Shear tests showed deep and undisturbed clays (Ccu between 0,06 and 1,6 bars and φ between 0,6 and 33°). The oedometric tests classify the soil as moderately compressible (Cc around 20%), and swellable (Cg around 8%). (ii) The Sandy-clay scree corresponds to multi-sized sandstone blocks, embedded in a sandy-clay deposits. The granulometric analysis, shows sands and clays presenting 15% to 97%, respectively, of grain size. The physical characteristics showed a non- saturated to saturated soil (10 < Sr< 90 %), of low to high density (1.4 <γd<1.93 t/m3) and low to medium plasticity (LL= 23 to 35 % and PL = 7 to 35 %). Resistance tests of non-drained shear showed low to medium cohesions (Cuu between 0,03 bars to 0,95 bars) and low to medium angles of friction φ° (UU) between 1,8° and 18°. The results of the oedometer tests reveal sandy clays that are moderately compressible (average Cc between 6 and 34%) and not very swellable (Cg of 2 and 11%). Geotechnical tests results showed that the upper part of the flysch bedrock is plastic, weathered and very sensitive to water with a low mechanical strength very favorable to landslides. The significant variation in the physical and mechanical characteristics of the soils is confirmed by our field observations which showed saturated, reworked and altered clays at the surface, subjected to shrinkage and swelling traversed by a network of cracks, widely open due to the development of high pore pressures thus favoring occurrence of landslides. In addition, the flysch bedding structure of the overlying soft rock (altered) and underlying hard rock (bedrock) was extremely sensitive to water infiltration and easy formation of a shear slipping surface along this weak interlayers and consequently ultimately susceptible to landslide. The sliding zone is a continuous thick shear zone located in weathered upper part of the flysch.
6 Mechanism analysis
The Ighil Bouzel gravitational slope correspond to a complex deformation mechanisms resulting from a superimposition of a multitude of slide surfaces. The failure mechanisms of this rainfall-induced complex landslide were the result of the combination of antecedent rainfall, geological structures, geomorphologic settings and human activities. Rainfall was the principal triggering factor that reduced the mechanical strength of the soil and lead to a disequilibrium/unstable state, leading to different failure mechanisms and mass movements.
The infiltration of the antecedent rainfall through the permeable scree deposits made of sandy clay and sandstone blocks overlying the impermeable flysch bedrock allowed the retention of groundwater. In the meantime, the water content and pore water pressure reduced the mechanical strength the soil mass and resulted in the progressive slope deformation with appearance of the sliding surface. In addition, the lithological interface scree / flyschs constitute the shear zone that controlled and favored the slope movements. The bedding planes, the dip of the schistosity and the diaclase planes of the flyschs in the slope direction. More exactly, a shear-sliding surface can easily form along the bedding planes, which could finally be responsible for the formation of the slide. As well, rainwater infiltration resulted in a reduction in the shear strength of the impermeable interface stratum of the muddy schist that is often altered with low geotechnical characteristics due to long-term immersion. This interface (geotechnical interface) layer that formed a shear slip plane correspond to the upper part of the weathered greenish-grey clays of the flysch. In addition, the basal erosion of the Iazoughen and Aboud rivers torrent at the toe of the rock slope decreasing the retaining force and driving the slope towards a destabilization.
To characterize the dynamic nature of the unstable slope and particularly the Identification of the shape and depth of the failure surfaces was possible based on information gathered from field observation, analysis of the inclinometric measurements results and geophysical Electrical resistivity tomography (ERT) investigations. The field observations showed several indications of instability such as an irregular morphology resulting from the existence of several different ground failure lines, several deep and shallow landslides surfaces were recognized on the slope. At depth, the rupture zone was defined based on inclinometric monitoring and geophysical (ERT) investigation carried out along and across the largest landslide areas (Figs. 14 and 15). The location of inclinometers and ERT profiles is shown on Figure 4.
6.1 Inclinometric data interpretations
Six inclinometer tubes identified as BI1, BI2, BI3, BI4, BI5 and BI6 were installed at the Ighil- Bouzel site at depth varying from 24 to 36 m (Fig. 14). The measurements were performed by the LCTP in 2014. The objective of the inclinometric measurements is to determine the depth of the failures and the amplitude as well as the direction of the horizontal displacement of the ground as a function of time. The measurement procedure is based on the French standard NF 94 156. The resulted inclinometric measurements are presented in the form of curves indicating for each inclinometer the angular variations and displacements in the A direction and in direction B perpendicular to A. The resulting inclinometers are also correlated with the boreholes. Several curves are superimposed on each graph, showing the evolution of the angular variations or the displacement as a function of time. The curves show different deformations and their evolutions overs time. These curves clearly show the depth of the rupture surfaces (Fig. 14). Table 3 summarizes the results of these measurements. The obtained results show that (Fig. 14): (i) All the inclinometers showed a failure surface; (ii) The BI1, BI2, BI 4 and BI5 inclinometers show similar evolution speeds, higher than those observed in SI3 inclinometer; (iii) The deformation is millimeter sized in the same directions A for all inclinometers. The average displacement the lowest and the displacement concerns a subset of the large landslide which seems started from Tadart district and to develop in the north-west direction, towards the Iazoughen river; (iv) The failure surface varies from minimum 11 m to maximum depth of 28 m. The failure surface located at the interface between the scree and the flyschs layer. Precisely, it is located in the upper part of the weathered, plastic clays flysch bedrock and (v) All the inclinometers showed a fracture surface that coincides with the roof of the flysch bedrock identified by the boreholes except for the inclinometer BI3 which identifies a surface at 28 m located inside the scree deposit. (vi) The mean displacement for the period 2012–2013 is approximately 0.01 mm to 0.076 mm day−1 i.e. a magnitude of 1 to 28 cm year−1 for the average displacement (Tab. 3). The movement moving in the Northwest direction toward the Iazoughen and Aboud rivers (Fig. 10). According to the classification of Lateltin, 1997, the actual mean velocity makes the Ighil Bouzel an active to very active landslide with a fast phases.
In summary, the results of inclinometric measurements show a complex, deep-seated (11 - 28 m deep) and active moving landslide with an average velocity of 0,01 to 0,28 m yr-1. The measured fracture depths are localized, preferably at the base of the quaternary scree layer (Fig. 14). In addition, the fractures are located below the water table level. The possibility of several superimposed failure surfaces cannot be excluded according to the hydrogeological structure (Fig. 12b). Based on the inclinometer results correlated with the boreholes, the entire slip surface is located along the interface between the flyschs bedrock and the overlying scree. The sliding zone correspond to the weathered upper part of flysch bedrock (shear zone) with thickness 0,5 to 5 m. The shear resistance characteristics of the soil in the sliding zone (geotechnical interface), shows a saturated, plastic oil with low cohesions and low angles of friction.
Based on the speed activity, the Ighil Bouzel landslide can be classified into two active zones: (i) the rear low activity zone in upstream with an average velocity of 0,1 to 3,6 cm/yr. This area is characterized by a slight deformation features. It has an average depth of 14.5 m, an area of 400 × 470 m2, and an estimated volume of 27.26 × 105 m3 and (ii) the front high activity zone in downstream (Figs. 7, 8, 9, 10, 11 and 12) with extended and dense deformations. The extent is approximately 1170 × 550 m2 in area, the mean thickness is 18 m, and the volume estimated is around 115.83 × 105 m3. These inclinometer results are in concordance with the results of observed morphology and deformation.
6.2 The electrical resistivity tomography (ERT) data interpretations
The Electrical Resistivity Tomography (ERT) is the most applied geophysical methods widely used for the sub-surface investigation of landslide because of its relative simplicity and time effectivity. ERT is considered as the most adequate and efficient method for the identification of depth and internal structure of slope deformations. More precisely, it is used for the study the internal characteristics of landslides (such as main body, physical properties, water content, geometry and position of shear surface) based on the spatial distribution of electrical resistivity contrasts of the soil provide by 2D or 3D images (Hack, 2000; Havenith et al., 2000; Mahmut et al., 2006; Perrone et al., 2014; Omowumi Falae et al., 2019).
In this study, the ERT survey was conducted on the first week of January 2020 in a landslide area by CGS in order to verify the distribution of electrical resistivity in the sectors where mass movement was identified and particularly to characterize the internal structures of landslides. Therefore, a 2-D resistivity survey was carried out along four (4) profiles over the landslide by a dipole array device using a 48 electrodes cable 2.5 m apart with rolling. Profile 1 and 4 was along the axis of landslide while Profile 2 and 3 were across the landslide body (Fig. 15). The data were processed by a resistivity inversion technique using RES2DINV software to achieve electrical imaging along these lines. The ERT test was useful for defining some physical properties and the geometry of the landslide mass.
The results presented in Figure 15 shows the lateral and depth variation of the Resistivity. The simplified logs of the boreholes and inclinometers are also correlated with the obtained tomograms. The resistivity profiles enabled us with boreholes, inclinometers and geotechnical tests to differentiate the mass movements underlying the bedrock. It is clear that the electrical tomography should be combined and calibrated with the available geotechnical data in the studied site. A good concordance between resistivity and inclinometric results was also noted.
The profile 1 was in the direction of NW–SE over a length of about 355 m (Fig. 15). The Profile cross the borehole BI 2 equipped with an inclinometer at a depth of 35 m that shows a failure surface at a depth of 24 m. The resulting tomogram along this profile indicates low to high resistivity values. They are in the range of 5–300 Ωm, and there are two distinct resistivity zones along the line. The zone of low resistivity (10–20 Ωm), located in all the section, at a depth from 20 to 50 m, probably represents the landslide materials with higher water content. The zone of the high resistivity correspond to the sandstone screes. The landslide material, clay, sand and sandstone blocks units observed in the borehole are in good agreement with the results of the electrical resistivity tomography. The resistivity values strongly decrease with depth in the clay of flysch unit. Then, a conductive zone, which has a resistivity ranging from 5 to 10 Ωm, is observed, and its thickness varies between 5 and 10 m. This conductive layer might be composed of unconsolidated and water-saturated landslide material that contains clay, mud, sand and silt. The presence of water indicated by low resistivity, values is the key parameter in sliding.
The profile 2, which is almost perpendicular to the direction of the landslide movement, runs from NE to SW (Fig. 15). The profile of about 475 m length passes through the borehole BI 4 equipped with an inclinometer at a depth of 35m. The resistivity distribution are low (10–20 Ωm) in the northeastern part and in the middle of the profile (between 0 and 10 m), while they are rather high (20 and 300 Ωm) in the southwestern part of the profile (between 0 and 40 m). At the depth, the resistivity values are very low (<10 Ωm) in the northeastern and the middle parts of the profile (between 35 and 60 m). The low resistivity might be caused by the water content of the unconsolidated material (scree sandstone) while the higher values might be an indicator of the consolidated materials (sandstone blocks). The electrical tomography image is in good agreement with the results of borehole. The schist unit is characterized by very low resistivity values on the tomogram, while the clay, mud, sand and sandstone of scree unit have moderate to high resistivity values. The results show the resistivity contrast that corresponds to the geological interface between the quaternary scree deposits and the Cretaceous flysch substratum. Thus, the presence of an interface, identified by the borehole data, is clearly verified by the resistivity results.
The third profile crosses the direction of the landslide movement over about 355 m lenght (Fig. 15). The profile 3 in the direction of NE–SW is passing through the borehole BI 1 equipped with an inclinometer at a depth of 32 m. The obtained ERT result revealed the rupture surface and the subsurface geometry of the landslide. It is considered that the relatively high resistivity layer (15–250 Ωm) observed in shallow depths might be originated by the landslide material. The thickness of this layer varies between 5 and 15 m on the tomogram. Also, one can observe that this zone is related with the topographical changes and the depth of groundwater level. Furthermore, the electrical tomography image is in good agreement with the results of borehole: the resistivity tomography image shows that the depth of the landslide that confirmed by the borehole data.
The profile 4 was in the continuity of the profile 1 in the direction of the landslide over a length of about 235 m (Fig. 15). The profile passes through the borehole BI 2 equipped with an inclinometer at a depth of 35 m. The ERT result obtained along this profile provides very informative result for the determination of the geometry of the landslide and particularly the failure surface. It is considered that the scree layer (clay, sand and sandstone) with relatively high resistivity (15–250 Ωm) blocks observed at the depths ranging between 20–40 m that might be originated by the landslide material composed of unconsolidated and water-saturated landslide material. The zone of relatively very low resistivity (05–10 Ωm) between the depths of 20–40 m correspond to clays of flysch unit formed by the consolidated deposits. Thus, the roof of this layer may correspond to the failure surface of the recent landslide. The resistivity results were confirmed by the borehole data of SI 2 that show a surface of rupture at the depths of 24m.
The ERT method applied in our case landslide area study has allowed to characterize the geometry (shape of the slope, body, surface of rupture) of landslides. The electrical resistivity values along the carried out profiles reveal a structure composed of a layer of displaced scree materials that contains clay, sand and sandstone blocks with relatively high resistivity superimposed on the clayey substratum, which has low resistivity. The resistivity results were confirmed by the boreholes equipped with an inclinometers data.
According to the deformation process and mechanism modes, the Ighil Bouzal landslide presented a complex and progressive failure along the geotechnical interface located in the upper part flysch bedding rock slope. The landslide is favored and controlled essentially by: (1) the lithological discontinuity or interface in the contact zone scree / flyschs; (2) the dip of the flyschs conforming to the slope; (3) the dip of the schistosity and the diaclase planes of the flyschs in the slope direction (4) by the destabilization of the slope by the basal erosion and undercutting of the Iazoughen and Aboud rivers torrent. In addition tectonic contacts constitute major hydrological discontinuity planes between permeable formations (scree) and impermeable formations (flysch).
According to the landslide characteristics, a wide range of mitigation measures has been recommended in order to reducing the risk from existing landslides: (i) lowering the aquifers levels with two drainage techniques: deep drainage trenches or galleries combined with radiating drainage boreholes; (ii) restricting and regulations of land-use in landslide prone areas, a function assisted by mapping landslide susceptibility, take the potential geohazards into consideration in urban planning; (iii) requiring by means of codes that human activities (i.e. excavation, construction, grading, cutting slopes, landscaping, irrigation activities, vegetation clearance …) not contribute to slope instability; (iv) protecting existing developments and population by physical mitigation measures (such as slope geometry modifications, drainage, down counterfort berms that serve as buttresses, and protective barriers, landslide cracks should be covered with plastic sheets) and (v) implementing monitoring and warning systems of the landslide in long-term.