Liquefaction hazard mapping in the city of Boumerdès, Northern Algeria

Earthquake-induced geological/geomorphological hazards in northern Algeria include several phenomena such as landslides, rock falls, and soil liquefaction (Bouhadad et al. 2010; Bourenane et al. 2014; Bourenane et al. 2016). Soil liquefaction is among the main causes of structures damage in saturated loose sand deposits during earthquakes. Indeed, loose sand tends to compress under cyclic loading imposed by earthquake shaking, which can transfer normal stress from the sand matrix onto the pore water, if the soil is saturated and largely unable to drain during cyclic loading. Hence, a reduction in the effective confining stress within the soil and an associated loss of strength and stiffness occur contributing, consequently, to soil deposit deformation. Past earthquakes showed how soil liquefaction may be damaging for buildings, bridges, railways, harbors, and other infrastructures. Soil liquefaction has been, historically, observed during moderate-sized and strong earthquakes (Ambraseys 1988). However, in Algeria, liquefaction is only associated to strong earthquakes of El Asnam (Ms = 7.3, 1980) and Zemmouri (Mw = 6.8, 2003) (Philip and Meghraoui 1983; Bouhadad et al. 2004; Machane et al. 2004). Investigation in contemporary documents suggests that the 22nd, August 1856 Jijelli earthquake (I₀ = VIII- IX) triggered tsunamis and liquefaction (Harbi 2001). However, no liquefaction was reported during the several earthquakes of magnitude M_s. < 6.0 that occurred during this century. Nevertheless, paleoliquefaction is also described in quaternary geological layers triggered likely by ancient pre-historic seismic events (Bouhadad et al. 2008; Benhammouche et al. 2013). During the 2003 Zemmouri earthquake (M_w = 6.8) liquefaction induced ground deformation has been extensively observed in the most susceptible areas around Boumerdès city, mainly along riversides and along the beach land, where, the geological, geotechnical, and hydrogeological conditions are most favorable (Bouhadad et al. 2004; Machane et al. 2004). It caused damage to roads and settlements of houses and bridges in many places. The described liquefaction features include centimeter to several meters- long fissures, ground settlement and lateral spreading along rivers sides as well as sand boils that constitute a typical feature of liquefaction during this earthquake (Fig. 1). The May 21, 2003 (Mw 6.8) Zemmouri earthquake caused 2384 deaths and widespread damage of buildings. The observed severe ground motion amplitude was, partly, caused by the lithographical and topographical amplification effects of seismic motion (DLEP 2004; Hellel et al. 2010; Dunand et al. 2004; Laouami et al. 2003; Laouami et al. 2006; Meslem et al. 2012; Bensalem et al. 2010). Worldwide, during the last few decades, liquefaction hazard zonation became among the first steps in each seismic risk analysis and mitigation strategy in urban areas. It may constitute a useful guide for a future urbanism development planning in liquefaction prone areas and an important challenge to develop proper disaster management policies. Liquefaction hazard corresponds to the probability of occurrence of liquefaction in a given area within a reference period of time (Varnes 1984), that means identification of areas with high likelihood of liquefaction-induced ground deformation which is among causes of damage during earthquakes. Liquefaction hazard maps are also useful to identify areas where specific investigations for liquefaction hazard are needed or should be required prior to any project development; however, in general such maps should not be used for site-specific engineering design but for land planning only. The aim of this work is to prepare liquefaction hazard maps of the urban area of Boumerdès by using GIS technology. The liquefaction assessment has been conducted, by using field data obtained from boreholes, down-holes, and standard penetration tests, in terms of factor of safety against liquefaction (F_S) and liquefaction potential index (LPI). For this purpose two main steps have been followed: (i) Geological, hydrogeological, geotechnical, geophysical, and seismic hazard data gathering (ii) assessment and mapping of liquefaction potential. The outcome of this study is presented, herein, as liquefaction hazard maps indicating the quantitative characteristics of the liquefiable layers and the area where there was a probability of surface liquefaction. GIS technology is used in this research in order to analysis, assess, and build a liquefaction hazard zonation database for the studied area.

Boumerdès city, the capital of the Boumerdès prefecture, is located in north central Algeria at about 50 km east of the capital Algiers and it is a coastal Mediterranean city (Fig. 2). Due to rapid development during the last century, Boumerdès city has become one of the largest urban areas and an important commercial distribution center. The studied area is characterized by a Mediterranean climate, known with cold and rainy winters, with flash floods, from November to March and a hot and dryn season between May to September. The precipitations are irregular and high in winter, averaging 600 mm/year. The average annual temperature varies between 10 °C and 27 °C. Geomorphologically, the city of Boumerdès altitude ranges between 30 m and 50 m above sea level. The city is crossed by three rivers which are, respectively, Oued Boumerdès river in the east, Oued Tatareg river in the center, and the Oued Corso in the West (Fig. 2).

The adopted methodology for evaluating and mapping the liquefaction potential in Boumerdès city includes (05) five steps: (i) Collection and analysis of geological, hydrogeological, geotechnical, and geophysical data of the study area including field observations and elaboration of data base entitled (Boumerdès_Data_Based Géo_Map) using GIS system (Ms Access, ArcGIS and MapInfo); (ii) Perform seismotectonic, seismicity, and probabilistic seismic hazard analysis for the study area; (iii) Preparation of representative geological and geotechnical profiles based on the geological and geotechnical data; (iv) Evaluation of the liquefaction potential index (LPI) of the soil layers using geotechnical parameters provided by borings and SPT taking into consideration a seismic hazard scenario corresponding to the calculated seismic hazard for a return period of 500 years which is a_max = 0.48 g and finally, (v) Preparation of the liquefaction hazard maps for the city of Boumerdès by correlation between results of the liquefaction index (LPI) in the framework of a geographical information system (GIS). In addition, having computed the value of probability based on the LPI, a liquefaction appearance probability map was compiled for the considered seismic scenario.

The Boumerdès region belongs to the central part of the Tell Atlas chain of northern Algeria characterized by a NW-SE convergence of the African and the Eurasian tectonic plates. The amount of crustal shortening ranges from 4 to 8 mm/year (McKenzie 1972; Argus et al. 1989; Nocquet and Calais 2004). As a consequence of this geodynamic process a series of active faults trending NE-SW affects the Tell Atlas chain (Meghraoui et al. 1988; Meghraoui 1991; Aoudia and Meghraoui 1995; Bouhadad 2001; Meghraoui et al. 2004; Maouche et al. 2011, Bouhadad 2013; Maouche et al. 2013). These multi-segment faults were responsible for several strong earthquakes in the past (Philip and Meghraoui 1983; Meghraoui et al. 2004) and are likely to produce strong seismic events in the future (Bouhadad and Laouami 2002). Several historical strong earthquakes have been mentioned in the available seismicity catalogs of Algeria (Benouar 1994; CRAAG 1994; Ayadi et Bezzeghoud, 2015). The city of Boumerdès considered, before the May 21, 2003 (Mw = 6.8) earthquake, as a seismically quite and safe area, has to consider now the impact of major active faults recently revealed and/or identified both offshore and onshore (Bouhadad et al. 2004; Meghraoui et al. 2004; Deverchère et al. 2005; Benfedda et al. 2017). Indeed, the Zemmouri 2003 earthquake revealed a 50 km long active reverse fault, located offshore, oriented NE-SW and dipping 50° to the SW (Bounif et al. 2004; Meghraoui et al. 2004; Belabbes et al. 2009). After this seismic event Boumerdès city is now reclassified in zone III, corresponding to the highest hazard zone in the new Algerian seismic code classification (RPA 2003). The recorded peak ground accelerations during the May 21, 2003 (Mw = 6.8) Zemmouri earthquake reached 0.58 g (amplified value by site conditions) and 0.34 g (non amplified value) at 20 km far from the epicenter (Laouami et al. 2003; Laouami et al. 2006) (Fig. 3).

In order to provide the PGA value to use in this study we perform a seismic hazard analysis by using a probabilistic approach (Cornell 1968; Mc Guire 1978; NRC 1988; Geomatrix 1993; Sadigh et al. 1993). For that three steps are required: (i) Definition of a seismic source model, (ii) Determination of source parameters, (iii) Choice of attenuations relationship and, finally, computation of seismic hazard. The seismic source model and source parameters have been determined based on the most recent seismotectonic data of the region, as shown, respectively, on Fig. 4 and Table 1 and by using the probabilistic approach. Concerning the attenuation of ground motion we used the most recent available attenuation relationships of Ambraseys et al. (2005), which is an update of the Ambraseys and Bommer (1991) relationship and the Akkar and Bommer (2010). The obtained result is shown in Fig. 5. The used earthquake scenario in this study is 0.48 g which is the peak ground acceleration obtained for a return period of 500 years by combining the used attenuation relationships mentioned above. In the probabilistic approach of seismic hazard analysis used in this study, it is assumed that the occurrence of earthquakes in a seismic source results from a Poisson process. Then, the probability that at a given site a ground motion parameter, Z, will exceed a specified level, z, during a specified time, T, is represented by the expression:

Where ν(z) is the average frequency during time period T at which the level of ground motion parameter, Z, exceeds level z at a given site. The function ν(z) incorporates the uncertainty in time, size, and location of future earthquakes and uncertainty in the level of ground motion they produce at the site (Geomatrix 1993).

In order to assess liquefaction hazard, the available in situ investigations geological, geotechnical, and hydrogeological data, carried out by several companies (L.C.T.P 1988–2003; L.N.H.C 1980–2010; GEOMICA and ANRH 2007), in the framework of several construction project studies, have been compiled, analyzed, and processed. It consists of 10 down-hole, 154 boreholes and hydraulic drilling of depth ranging between 10 to 35 m, geotechnical profiles containing data about the N values of SPT, grain size distribution, liquidity limit (LL), and plasticity limit (PL) of selected samples in subsoil layers as well as 35 level checks (water table depth) of water points. The data from boreholes were used to prepare simplified cross-sections illustrating the lithological structure of the studied area. The locations of boreholes, SPT, down-hole, water points, and the position of cross-sections are shown in Fig. 6. The data analysis required a coding in a data base (Boumerdès_Data_Base and Geo_Map) under GIS framework that allows easy handling, processing, managing, exploiting, treating, and interpreting the large amount of data. It allows the needed capacity of spatial analysis, data management, and updates as well as mapping (Carrara et al. 1995; Ayalew et al. 2005). For this purpose we perform a geological, geotechnical, and hydrogeological data base using MapInfo and ArcGis softwares and Microsoft access. All the maps with database were geo-referenced at the same local projection system of Algeria (UTM zone 31 North, WGS 84—Geodetic Reference System).

Liquefaction susceptibility is a first step in the liquefaction hazard assessment of a given area. Soil liquefaction susceptibility depends on geological, hydrogeological, and geotechnical site conditions (Youd and Perkins 1978; AFPS 1995; Youd 1998; CDMG 1999; Seed et al. 2003).

In order to evaluate liquefaction occurrence in a given deposit the widely used methodology is that proposed by Iwasaki et al. (1982). The objective consists of the assessment of liquefaction hazard and the compilation of a map showing the spatially distributed probability of liquefaction-induced surface disruption in the studied area. This method is an improved version of the prior method developed by Seed et al. (1971) entitled “simplified procedure”, that was modified by Seed et al. (1985) and Youd et al. (2001). It consists of calculating the safety factor per layer (Fs), as the ratio of cyclic resistance ratio (CRR) to the cyclic stress ratio (CSR).

The CRR is computed by the following equation:

Where (N₁)₆₀ is the penetration resistance corrected (N₁)₆₀ = N_m C_n C_eC_b C_r C_s) and standardized relative to the overburden pressure factor (Cn), the correction for hammer energy ratio (Ce), the correction for borehole diameter (Cb), the correction factor for rod length (Cr), and the correction for samplers with or without liners (C_s). The (Cn) was calculated according to the equation proposed by Liao and Whitman (1986), Cn = (Pa/σ’v)^0.5 in function with (Pa) (atmospheric pressure) and the σ’v (effective vertical stress). Additionally, a “fine content” correction was applied to the calculated N1(60) value in order to obtain an equivalent clean sand value N1(60)cs given by the equations proposed by Youd et al. (2001).

The CSR defines the seismic demand and is expressed as:

Where σv is the total vertical stress at depth z, σ’v is the effective vertical stress at the same depth, a_max is the peak horizontal ground acceleration, (g) is the acceleration due to gravity, and (rd) is the stress reduction factor. The term (rd) was estimated using the Liao and Whitman (1986) equation:

The CSR values have been divided by the magnitude scaling factor (MSF), which is calculated by the following equation of Youd et al. (2001):

Afterward, the values of liquefaction potential index (LPI) of the soil columns for a given site at a depth of 20 m and for boreholes were computed using the following equation of Iwasaki et al. (1982):

Where (z) is the depth below the ground surface in meters and is calculated as w(z) = 10–0.5z; F(z) is a function of the safety factor against liquefaction, (Fs), where F(z) = 1 – Fs when Fs < 1 and F (z) = 0 when Fs >1.0).

Iwasaki et al. (1982) calibrated the severity of liquefaction induced damages with the (LPI) values. According to this calibration, no evidence of liquefaction phenomena are expected where 0 ≤ LPI, while moderate and high degree of liquefaction surface deformations are expected where (LPI) values range between 5 and 15 and LPI > 15. The advantage of (LPI) is that it quantifies the likelihood of liquefaction at the site by providing a unique value for the entire soil column instead of factors of safety for each of the layers. Consequently, the (LPI) values were used for the compilation of liquefaction hazard maps which can be used by planners as a tool for the preliminary assessment of the liquefaction potential. The liquefaction assessments in Boumerdès city has been performed by using data from geotechnical profiles of 154 boreholes with 56 SPT tests and 35 water points leveling and a seismic scenario of Mw = 7.3, taken from the seismic hazard study. By computing the CRR of the soil layers and the CSR (Fig. 14) regarding the earthquake “scenario”, the safety factor against liquefaction (Fs) was calculated (Fig. 15). Then, the depth and the thickness of the liquefied layers, per borehole, were defined based on the computed values of (Fs) of the soil layers. The depth of the liquefiable layer is evaluated by considering the depth of the upper subsoil layer susceptible to liquefaction, for which the safety factor was estimated less than one (Fs < 1). Figure 15 summarizes results of the safety factor (Fs) calculation as a function of the depth for the different lithological units in the studied area. The obtained results indicate that the Pleistocene coarse sands are liquefiable at shallow depth (1-5 m) for the accelerations a_max = 0.48 g, while the recent dune and sand Holocene alluviums are liquefiable at a depth of 1-10 m.

After computing the safety factor (Fs), the values of liquefaction potential index (LPI) of the soil columns for a given site at a depth of 20 m and for boreholes have been calculated using Eq. 8 cited above. The results from liquefaction potential analysis show that the (LPI) index, varying from 0 to 32, fluctuates horizontally according to the accelerations values (Fig. 16). The obtained results suggest that the moderate and high liquefaction area have been observed in the sand layer characterized by moderate and high susceptibility to liquefaction, particularly, along the Corso waterway at the west of Boumerdès city. These results also show a good agreement with the observations made during the May 21, 2003 (Mw = 6.8) Zemmouri earthquake (Bouhadad et al. 2004; Machane et al. 2004; Bouhadad et al. 2008). The obtained results are also compatible with the geological, geotechnical, and hydrogeological susceptibility to liquefaction of the area (Youd and Perkins 1978).

Liquefaction potential mapping of the study area is obtained by correlation (e.g., krigging) of the liquefaction potential index (LPI) obtained for each borehole and for the above mentioned scenario (Fig. 17 ). The map shows that liquefaction is low in the center and in the east of the city corresponding to clay soils, bedrock, and dense sand with a deep water table (generally higher than 12 m). The liquefaction potential is moderate to high in the West of the city corresponding to areas formed by sand deposits, fairly dense and relatively loose, of the alluvial plain, characterized by a shallow water table (ranging between 0 to 10 m). Liquefaction is quite apparent in the West of the city (Fig. 17 ), in the alluvial plain of the Corso waterway, where the geological and the hydrogeological conditions are favorable, a shallow water table, and thick loose sandy layers. This observation has been confirmed by the observations of liquefaction during the May 21, 2003 (Mw = 6.8) earthquake (Fig. 10). The susceptibility to liquefaction decreases with the distance from waterways where the water table is shallow (6 to 8 m). Moderate susceptibility is observed where the water table depth is less than 10 m. Non susceptible area corresponds to the ante-Holocene deposits where the water table depth is more than 10 m. Liquefaction impact depends on the level of susceptibility. In the high susceptibility area the consequences are represented by lateral spreading near free surfaces and settlement. In the moderately susceptible area the settlement is less while in weak susceptibility the impact is minor. Consequently, before any project of construction in liquefiable soils we need necessarily either to strengthen the site or to avoid it.

The computed values of (LPI) at each location of each borehole were used in the equation proposed by Papathanassiou (2008) to evaluate the probability of liquefaction-induced ground disruption:

The aim of the present study was, therefore, the assessment of liquefaction hazard and the compilation of a map showing spatially distributed probability of liquefaction-induced surface disruption. In this model, the (LPI) value is the independent variable and the occurrence or not of liquefaction phenomena is the dependent one. At sites where the probability (P) ≥ 0.5 liquefaction phenomena are expected to be triggered while non-occurrence of liquefaction is predicted where P < 0.5 (Fig. 17). In our study, as shown in Fig. 18 for a_max = 0.48 g, the probability of liquefaction-induced ground failures is very low at the city of Boumerdès, with the computed value of probability >50%, and concerns only the sites close to the Corso River plain. For the other sites of Boumerdès city, liquefaction phenomena is not expected at the surface under a seismic loading a_max = 0.48 g. The defined liquefaction model is almost the same as the defined one by the (LPI) values (Fig. 16).

The city of Boumerdès is located in the seismically active zone of Northern Algeria. Indeed, the peak ground acceleration (PGA) provided by the Seismic hazard study is 0.48 g for a return period of 500 years. During the May 21, 2003 (Mw = 6.8) Zemmouri earthquake liquefaction was extensively observed in the most susceptible areas around Boumerdès city, where the geological, geotechnical, and hydrogeological conditions are most favorable. In this study, we used geological, geotechnical, and hydrogeological data, gathered from boreholes and standard penetrations tests (SPT) to assess the liquefaction potential of the subsoil layers by applying the methodology of liquefaction potential index (LPI) and by considering the seismic scenario, represented by the maximum peak ground accelerations (PGA). The scenario corresponds to the a_max = 0.48 g provided by the seismic hazard study of the area. Results are shown as liquefaction hazard maps by using the classification of (LPI) values that provide the severity of liquefaction surface evidence and the computed probability of liquefaction-induced surface disruption. The obtained maps show the quantitative characteristics of the liquefiable layers and the area where there is a probability of surface evidence of liquefaction. Our results conclude that the urban area of Boumerdès city exhibits a low potential of liquefaction, except for the Corso waterway valley located at the west of the city, where the liquefaction potential is moderate to high. Our results also show that the observed liquefaction during the May 21, 2003 (Mw = 6.8) earthquake fall into the area of high and moderate liquefaction potential. The obtained liquefaction hazard maps may serve as useful tools for land management and planning in the city and as a study-example to follow in other populated cities of northern Algeria located in earthquake and liquefaction prone areas.