1、Abstract: The January 13, 2001 earthquake （M7.6） off the coast of El Salvador triggered widespread damaging landslides in many parts of the country. The Las Colinas landslide, located in Balsamo Ridge west of San Salvador, caused the greatest loss of life in a single location from the earthquake. Th
2、e preearthquake and post-earthquake seismic stability analysis of the Las Colinas slope is studied using the limit equilibrium and finite element methods in this paper. The safety factor of the pre-earthquake slope was likely over 1.50 based on the low groundwater level, which was noted by local res
3、idents prior to the January 13 earthquake. The slope failed as the horizontal earthquake coefficient, kc, reached between 0.18g and 0.31g depending on the failure analysis methods （i.e. circle and trial surface methods） and the low soil strength assumed. The safety factor of the post-earthquake slop
4、e falls in a range from 1.851 at low water saturation to 1.337 when fully saturated. As compared with the pre-earthquake slope, the stability is about 8% higher for the post-earthquake slope.Keywords: Earthquake, landslide, seismic, stability, limit equilibrium, finite element.1、INTRODUCTION El Salv
5、ador locates on the Pacific coast of theisthmus and bordered by Guatemala to the west and Honduras to the north and the east. The country has experienced, on average, one destructive earthquake every decade during the last hundred years. The earthquake of January 13, 2001 that struck El Salvador was
6、 the first major seismicdisaster of the third millennium and the fifth destructive earthquake to affect this small Central America republic in 50 years （Bommer and Benito, 2002）. The January 13 earthquake triggered the Las Colinas landslide off the steep northern flank of Balsamo Ridge. The landslid
7、e originated at an elevation of about 1,040m to 1,070m and traveled northward about 700m to 800m into the Las Colinas neighborhood of Santa Tecla, west of San Salvador. The vertical drop from the ridge to the neighborhood is about 160 m. The volume of the landslide material is estimated at about 250
8、,000 m3 （Jibson and Crone, 2001）.According to a prior investigation （Jibson and Crone, 2001）, the landslide material exposed in the headwall scarp appeared somewhat moist but not saturated, but once this material mobilized it behaved as a semi-liquid mass with a soupy consistency; this allowed the m
9、ass to travel anunusually long distance from the base of the slope. The transformation of the landslide material from behaving as solid to fluid in the absence of large quantities of water （such as in a heavy storm） and the long runout distance indicate an abnormal material behavior that requires de
10、tailed investigation and analysis. Landslide hazards and stability study is carried out by several organizations in the earthquakerelated areas. The pre-earthquake and postearthquake seismic stability analysis and sensitivity analysis of the Las Colinas landslide are studied using both the limit equ
11、ilibrium and the finite element methods in this paper. Safety analysis and suggestions are given in the following sections of the paper.2. GEOLOGY AND VOLCANIC SOIL The landslide scarp of the Las Colinas slope exposes the upper 25m - 30m of deposits involved in the landslide. The stratigraphy of the
12、se deposits is typical of the volcanic deposits for Balsamo Ridge and much of the surrounding Cordillera Balsamo. Based on the reconnaissance work by USGS, the locations of abundant earthquake induced landslides in the Cordillera Balsamo seem to be influenced by the presence of relatively thick depo
13、sits of TB （Tierra Blanca tephra）, which erupted from a volcanic source near Lago de Ilopango about 10,000 years ago, and the overlying fresh and weathered tephra. Tierra Blanca is a dalacitic pumice ash composed of acidic and epiclastic deposits, which covers most of the upper part of San Salvador,
14、 and is poorly consolidated and originated from multiple volcanic eruptions that can reach up to 50 m thick. Study demonstrated that one of the most important factors in terms of seismic slope stability is the relatively weak cementation that is broken when the soil is subjected to even small strain
15、 （Bommeret al., 2001）. It is most probably that the soils may experience a drastic reduction of shear strength during earthquake and hence as soon as the slope failure begins the material undergoes almost adebris flow.The strong ground shaking, along with the presence of thick, loose to poorly conso
16、lidated,young volcanic deposits, the steep topography on the northern flank of Balsamo Ridge, and possibly the presence of a relatively impermeable ancient soil at the top of the Balsamo Formation were contributing factors to the Las Colinas landslide. The cross-section of the Las Colinas slope is s
17、hown in Figure 1. The relatively thin layer of paleo-soil might have played an important role in initiating the movement. There are four main geological strata in the slope section: 1） pyroclasts; 2） brown ashes; 3） 1.0m1.5m paleo-soils; and 4） consolidated tuffs and pyroclastic flows. Figure 2 is a
18、 profile at the top of the Las Colinas landslide where landslide mitigation is under construction. The light color formation near the bottom of the profile is paleo-soil. Figure 1. Cross-section of the Las Colinas slopeshowing the geological stratum Figure 2. A profile at the top of the Las Colinas
19、landslide One of the most critical steps in slope stability analysis is to determine the shear strength parameters （c and f） along the sliding surface. Laboratory and field soil testing （e.g. Standard penetration test, downhole seismic velocity, unconsolidated triaxial test, etc.） were performed by
20、C. Lotti and Associati （2001）. The physical and mechanical properties of the soils at the site are summarized in Table 1.The available strong ground motion recordings （UCA 2001） during the earthquake are shown in Figure 3. The largest peak ground velocity （PGV） was observed at the Santa Tecla Statio
21、n （Te） with a value of approximately 57 cm/s and a peak ground acceleration （PGA） of 485 cm/s2 （around 0.5g）. This ground motion was sufficiently large to trigger large landslide in the area.Table 1. Soil properties of the Las Colinas slope（C. Lotti & Associati, 2001）Figure 3. Acceleration history f
22、rom TE station3. SLOPE STABILITY ANALYSIS The pseudo-static limit equilibrium analysis and the finite element method were employed in slope stability analysis. The primary method of evaluation was based on the Modified Bishop methods for circular and trial surface failures. For comparison, Janbu cir
23、cle and random surface and Sarma methods were also used to analyze the safety factor of the landslide. The potential sliding surface, circle or polygonal, can be pre-specified or randomly generated. The programs, PCSTABL5 （Achilleos, 1988） developed at Purdue University and Reinforced Slope Stabilit
24、y （RSS） by Geocomp Corporation （1996）, were used in the static and pseudo-static limit equilibrium analyses. ABAQUS/Standard was used in the finite element analysis.3.1 Static analysis3.1.1 Pre-earthquake The failure plane was likely restricted by the consolidated pyroclastic flows, which have much
25、higher strength properties than the other formations above. The most probable sliding plane would be along （or through） the third stratum, paleo-soils that is relatively thin and has the lowest strength. Both Bishop and Janbu circle methods employ automatic search for a failure surface. However, the
26、 trial surface method requires a pre-determined sliding surface. In the study, a slip surface following the weak paleo-soil layer was treated as a potential sliding plane. The effect of groundwater on the slope stability was assessed at different water saturation conditions. The locations of groundw
27、ater table are shown in Figure 4. Three strength groups （i.e. low, medium, and high） were considered for the sensitivity analysis of slope stability （Table 2） that includes ranges of c and f found from the field and lab testing. Table 3 summarizes the analysis results based on different methods and
28、failure modes with the assumed water table more or less following the paleo-soil layer.Table 3. The safety factor of pre-earthquakestability analysis with water table following the paleo-soil layer.Table 2. Strength parameters groups used instability analysis. The results of pre-earthquake static st
29、ability analysis indicate that the slope with the lowest strength properties and under the saturation condition noted in the field would have been stable if not because of the earthquake. Even if the slope was fully saturated, it would still be stable with safety factor larger than 1. The most proba
30、ble circular failure surface of a fully saturated slope is a circle cutting through the top three strata, following the relatively weak third stratum, and then emerging out near the boundary between thefirst and second strata. The same circular failure surface was analyzed using the Sarma method, an
31、d it gives a safety factor of 1.52. The most critical surface, however, is determined based on the trial surface that follows the actual slip surface observed in the field （Figure 4）. The minimum safety factor for a fully saturated slope with the lowest strength properties is 1.235 based on the Janb
32、u trial surface method. The Bishop Trial surface method yielded a similar Result.3.1.2 Postearthquake During earthquake, the potentially unstable soil mass detaches from the strata, and results in a more stable slope. The results of post-earthquake slope analysis （i.e. the current slope at the site） are listed in Table 4. The ranges of safety factor are higher than that of the pre-earthquake slope unde
