1 Introduction

Earthquakes are among the most deadly and destructive natural hazards. Seismic activity resulted in more than 2 million deaths during the last century and destroyed many cities (Duval 2007). Earthquake activity in North Algeria constitutes a constant threat to human life and property, causing major economic losses and disruption. The losses result from not only great earthquakes, such as the Ms = 7.3 El Asnam event of October 10, 1980 (Ouyed et al. 1981), but intermediate events such as the Ms = 5.9 Constantine earthquake of October 27, 1985 (Bounif et al. 1987) and the May 21, 2003, Zemmouri earthquake (Mw = 6.8) (Belazougui 2008; Meslem et al. 2012; Mébarki et al. 2013). Constantine city lies in the eastern Tellian Atlas Mountains, one of the most seismically active regions of Northeast Algeria. Assessments of seismic risk in Northeast Algeria are relatively new and started in the 1990. The current work consists of regional and local studies, mainly based on seismic risk assessment and seismic hazard (Mortgat and Shah 1978; Benouar 1996; Hamdache et al. 1998; Aoudia et al. 2000; Bouhadad and Laouami 2002; Boughacha et al. 2004; Pelàez et al. 2004, 2005, 2006; Harbi 2006; Laouami et al. 2006; Farsi and Belazougui 2007; Bensaibi et al. 2007; Boukri et al. 2012a, b). Seismic risk assessment is already well established for major urban centers, but is still of great importance in risk assessment in more sparsely populated areas. For example, it is crucial to examine the potential impact of earthquakes on Constantine city because there are few connecting roads between the city and nearby centers. Accurate assessment of the seismic risk faced by urban areas is needed by public authorities and decision makers, who are responsible for regional planning and urbanization. Geotechnicians, architects, construction supervisors, and public works institutions, among others, must have a good understanding of the nature of the potential for soil liquefaction before making decisions related to development. Here, we present the results of a seismic risk scenario developed for Constantine city. For each at-risk element, the procedure consists of linking seismic vulnerability values with hazard values. Hazard is expressed in terms of maximum horizontal soil acceleration as a first step in risk analysis. Induced phenomena (liquefaction and landslides) are evaluated using soil data, expressed in terms of maximum triggering percentage of gravitational acceleration (PGA). For each of these hazards and for each element of seismic risk, we assess potential damage to structures in three homogeneous areas (Coudia, Bellevue–Ciloc, and Old Town). We then describe a procedure for the adaptation of existing seismic risk and vulnerability assessment approaches to account for local geographic and geologic features. Our input information comprises an earthquake scenario, earthquake catalogs, local geology, soil classification, geotechnical conditions, and building structural data and typology. We present results as maps, including distributions of peak ground acceleration (PGA) on bedrock and design ground acceleration (DGA), intensity (I), damage assessment (D), and coefficients of safety (Fsliq, Fslan) for induced effects. These thematic maps constitute important basic information for sustainable socioeconomic development in Northeast Algeria.

2 Overview of Constantine city

2.1 Study area

Constantine, the second most important city in Algeria and the largest city in Northeast Algeria, is assumed to have a moderate seismic risk. It houses important economic, scientific, and cultural aspects of Algerian infrastructure. Constantine province occupies a central geographic location within the region (Fig. 1). The center of Constantine occupies an area of 232 km2, with 448,374 inhabitants, yielding a high population density of 1936 inhabitants/km2 (Boudemagh 2013). Many civilizations have influenced the development and architecture of the urban area throughout its long history, from the Phoenicians to modern times (Boussouf 2002).

Fig. 1
figure 1

Location map of Constantine city and choice reference earthquake for study area

Full size image

2.2 Seismotectonic context of Constantine region

The Tellian Atlas (Northeast Algeria) is an active collision zone between the African and Eurasian plates that is experiencing shortening of ~5–6 mm/year (Anderson and Jackson 1987; Argus et al. 1989; DeMets et al. 1990). The present study area is located around of the Constantine city, where the regional tectonics has been studied previously (Fig. 2). The study area was settled in the post-ply period, after emplacement of the Numidian layers (the uppermost structural layers in the mountain range). This area is bounded to the east by the Guelma basin (specifically, by the Temolouka Fault that separates the two basins), and to the north by an unconformity with the inner area of the Petite Kabylie, which corresponds to a major structure called the Constantine North Fault (Aicha accident M’cid-Debar), oriented N100°E to N110°E (Coiffait 1992). To the south, the area is delimited by the Batna zone, where the E–W Sigus Fault marks the edge of the basin. To the west, the Setif region marks the boundary (Durand 1969; Raoult 1974; Vila 1980; Coiffait 1992; Bougdal 2007). Within this region, three important neotectonic faults cut Pliocene-Quaternary deposits. (1) The Ain Smara Fault, recognized as active since the Constantine earthquake of October 27, 1985 (Bounif et al. 1987), comprises three segments (S1, S2, and S3) extending over almost 30 km. The fault strikes NE–SW, with a dominant stress regime determined by the focal mechanism of the Constantine earthquake (Bounif et al. 1987). In this area of the Atlas Mountains, active deformation appears to be governed by transcurrent dextral movement (Harbi 2001). (2) The Constantine North Fault (accident M’cid Aicha-Debar), oriented E–W and extending over 80 km, which marks the margin of the Jijel–Skikda area. (3) The Sigus Fault oriented E–W has a length of nearly 30 km.

Fig. 2
figure 2

Tectonic map of the study area (Constantine basin) (Harbi et al. 1999). (1) Volcanism; (2) Palcozoi; (3) Jurassic–Cretaceous and lower Cenozoic basement; (4) neogene post-nappe deposits; (5) plio-quaternary deposits; (6) quaternary deposits; (7) faults; (8) anticline; (9) reverse fault; (10) strike-slip fault; (11) normal fault

Full size image

The present study area is located in an active seismic zone within Algeria, which has been shaken by several moderate-to-strong earthquakes during the last few centuries. At least three recent earthquakes had a maximum felt intensity of at least I0 = VII–IX on the MSK scale (Bounif et al. 1987; Harbi et al. 2010), i.e., Ms = 5. 2 on August 4, 1908, Ms = 5.3 on August 6, 1947, and MS = 5.9 on October 27, 1985 (Ousadou et al. 2012) (Fig. 3). The inferred recurrence interval between damaging events is 39 years (Table 1). The historic seismicity of Algeria, including the Constantine region, has been studied extensively (e.g., Hée 1933, 1950; Rothé 1950; Grandjean 1954; Benhallou and Roussel 1971; Roussel 1973; Mokrane et al. 1994; Mezcua and Martinez 1983). More recently, reviews of historic seismicity by Harbi et al. (2003, 2010) concluded that the seismic history of the region was largely unknown before 1900 (Ousadou et al. 2012).

Fig. 3
figure 3

Spatial distribution of earthquakes in Northeast Algeria from 1357 to 2014

Full size image
Table 1 Historic seismicity in the Constantine region from Rothé (1950), Grandjean (1954). Two main historic seismic events are apparent: August 4, 1908, and August 6, 1947. The recurrence interval of the main seismic event is ~39 years
Full size table

2.3 Creation of an earthquake scenario for the Constantine city

The Constantine city and its vicinity, with an area of about 55 km2, were chosen to illustrate GIS-based risk assessment, using data on the structural vulnerabilities of different types of construction. Maps were produced from GIS databases using thematic analysis tools in MapInfo 7.0, spatial analysis tools in ArcGIS 9.3 (Christophe 2008), and other techniques, such as georeferencing and vectorization layers (Thierno 2004). For a deterministic scenario, our choice of reference earthquake in the Constantine province was the October 27, 1985, main shock, for which the epicenter was located between El Khroub and Beni Yagoub. We used a magnitude of MS = 6 and a depth of 10 km (Bounif et al. 1987). We have geologic maps (1/50,000) at this scale: the eastern part of the province, which includes the city of Constantine (Vila 1980), and the western region, which covers El Aria (Coiffait et al. 1992). The association of the two parts was used to produce a geologic map, assembled, and georeferenced (Fig. 4). We created a 1-km grid for the entire Wilaya of Constantine, a 400-m grid for the city of Constantine, a 100-m grid covering the study area, and three 20-m grids covering the buildings of the Old Town (the Rock), Coudia, and Ciloc–Bellevue (Fig. 5).

Fig. 4
figure 4

Geologic map of study area

Full size image
Fig. 5
figure 5

Different grids: a grid (1 km spacing) across the Wilaya of Constantine, b grid (400 m) across the Constantine city, c grid (100 m spacing) across the study area, d finer grid (20 m) over Bellevue–Ciloc, Coudia, and the Old Town

Full size image

3 Methodology used for assessment of seismic risk in Constantine

The main goal of this work is to evaluate seismic risk levels for buildings and roads in the Constantine city, based on the vulnerability of infrastructure to potential seismic sources (e.g., magnitudes, distances, and source mechanisms). The methodology is based on evaluating the consequences of ground motion (PGA) for building stocks and soil (Fig. 6). Risk is assigned to each area element based on field assessments of the soil amplification coefficient (coeff) due to lithological site effects, and PGA triggering thresholds for various phenomena. Field surveys allow the assessment of ground shaking vulnerability by creating a specific vulnerability index for each building. From these, a more general vulnerability index can be developed for homogenous urban zones. PGA was computed for the selected earthquake scenario using existing empirical attenuation laws summarized by Bommer et al. (2003). PGA values were then multiplied by site amplification factors to obtain maximum ground shaking values (Avib). For each hazard phenomenon, a security factor was computed by comparing Avib with a threshold Ti to estimate an event probability. Finally, a damage grade was computed by cross-referencing the vulnerability index with EMS98-intensity values derived from Avib. A low-grade approach was used to infer site effect amplifications. Soil types were assigned according to building code soil classifications (NFP06-013 1995). For each soil category, an amplification factor (coeff) is given in Table 2, based on existing tables proposed for other soil classifications (TC4 2003; EC8 1998). A PGA threshold for liquefaction was assigned based on a pioneering study of susceptibility to liquefaction as a function of the age and nature of sediment deposits (Youd and Perkins 1978). Table 3 lists the PGA values corresponding to zero probability of liquefaction for each susceptibility category. These values are based on Hazus99TM (1999). The method proposed by Keefer (1984), Wilson and Keefer (1985) is used to estimate the PGA threshold for landslides (Tls). Based on the analysis of many earthquake effects, Table 4 provides the minimum PGA for triggering landslides as a function of geologic structure, slope angle, and water saturation. For each induced liquefaction and landslides, we define a security factor (Fs liq and Fs ls) to warn if a risk is likely to be triggered during an earthquake. Values of 0–1 are obtained for each factor by comparing estimated PGA (including site effects) with the threshold for a given liquefaction and landslides and site conditions (Table 5). The method used in the present study for the assessment of building vulnerability is derived from the European Macroseismic Scale (EMS98 1998), based on correlations between macroseismic intensity and apparent (observed) damage from past earthquakes. This vulnerability model has been improved in several studies (RADIUS, GNDT, Risk-UE, and VULNERALP) (Radius 1999; Mouroux et al. 2004; GNDT 1993a, b; Risk-UE 2003; Guéguen et al. 2007). In this work, we use a vulnerability evaluation criterion partly developed in the framework of the VULNERALP project, identical to that used by Nicem France during the GEMGEP project (Bard et al. 2005). This approach, derived from that developed by the GNDT group in Italy (Benedetti and Petrini 1984; GNDT 1993b; Faccioli et al. 1999; Dolce et al. 2003), is adapted to Algerian building standards. A standardized survey and inventory form are used for data collection and the rapid determination of a vulnerability index (VI), based on building typology and aggravating factors (e.g., height, irregularities in shape, position with respect to other buildings). This makes it possible to quantitatively estimate the susceptibility to specific hazards. The vulnerability index, which ranges from zero (not vulnerable) to 100 (exceedingly vulnerable), is used to construct an empirical vulnerability (or fragility) curve for a building. This curve links a seismic event, expressed in terms of macroseismic intensity, with a damage index (d), from which we compute probabilities corresponding to different levels of damage. This distribution is estimated from a probability law, with parameters scaled based on actual observations of damage during various earthquakes, mainly in Italy and Greece. Using vulnerability indices obtained from field surveys makes it is possible to estimate the damage rate (d) for a given EMS-98 intensity using the formula of Giovinazzi and Lagomarsino (2003) (Tables 6, 7). This yields a fragility function associated with each vulnerability index. The damage rate is correlated with the damage grade, defined in the EMS-98 (Table 8), but allows building distributions to be divided into five EMS-98 damage grades using a binomial formula Giovinazzi and Lagomarsino (2003).

Fig. 6
figure 6

Global concept of the method of the study

Full size image
Table 2 Proposed amplification factors for each defined soil category (EC8 1998)
Full size table
Table 3 Proposed T liquefaction (g) and PGA thresholds for landslides, as functions of geology and age. Adapted from Wilson and Keefer (1985)
Full size table
Table 4 Proposed T landslide (g), PGA thresholds for liquefaction (g) as a function of age, and nature of the sediment deposits. Based on Youd and Perkins (1978) and Hazus (1999)
Full size table
Table 5 Security factors for inducing phenomena
Full size table
Table 6 Correspondence between vulnerability classes of the EMS 98 and scores of GNDT, according Giovinazzi and Lagomarsino (2003)
Full size table
Table 7 Equivalence between GNDT and methods VULNERALP for environmental criteria
Full size table
Table 8 EMS-98 damage grades and corresponding damage rate ranges (EMS98 1998)
Full size table

3.1 Peak ground acceleration (PGA) on bedrock

In a deterministic scenario, the PGA of bedrock is assigned to each point in the zone. To evaluate scenarios for a particular earthquake, the user can designate an epicenter location and magnitude. Induced PGA can be computed from this information using the empirical attenuation laws of Douglas (2001). However, empirical laws are developed in local contexts and are not always generalizable. Their application, even when the conditions are fulfilled, is subject to large uncertainties that can be difficult to quantify. This must be kept in mind during map analyses (Bommer et al. 2003), e.g., for laws developed from European strong motion records. For earthquakes with magnitudes 4 ≤ M ≤ 7.7, epicentral distances 10 ≤ D ≤ 200 km, and focal depths Z < 30 km, PGA is given by

$${\text{PGA}} = 10^{{ - 1.482\,+\,0.264M\,-\,0.883\log^{10} \sqrt {\left( {\frac{D}{1000}} \right)^{2}\,+\,\left( {\frac{Z}{1000}} \right)^{2}\,+\,2.473^{2} } }}$$
(1)

We construct a map grid where (D) is measured for each grid point.

For each mesh of the base map grid and each element of this study, the distance (D) to the earthquake is measured. The PGA is then computed from this distance (D, even if it does not fit exactly the law definition) and the magnitude (M).

3.2 PGA soil amplification factors (Coeff)

The aim here is to account for site effects due to soft layers above bedrock (e.g., alluvial basins), which can increase PGA and amplify damage. A low-grade approach is used. Soil type is assigned according to the French building code classification scheme (NFP06-013 1995). Based on geotechnical properties of each category, and existing tables proposed for soil classification (e.g., Borcherdt et al. 1991; TC4 2003), an amplification factor (coeff) is proposed for each soil category in Table 2. For the base map, the geologic unit crossing this study that could induce site effect is first digitalized and the most relevant coefficient of Table 2 is assigned. This step has to be clarified for this study, even if it has no influence on the resulting damage scenario map. For both base maps, the estimated PGA will be multiplied by the relative coeff to produce Avib.

3.3 PGA threshold for landslides (Tls)

Landslides are among the major consequences of earthquakes. We use the method of Keefer (1984) and Wilson and Keefer (1985) to estimate the PGA threshold for this hazard (T landslide). Table 3 gives a minimum PGA for triggering landslides as a function of deposit type and age.

3.4 PGA threshold for liquefaction (Tliq)

Liquefaction is a phenomenon that can occur when superficial sand layers lose their shear strength and act as liquids, due to rapid excitation by seismic waves. Our approach to assigning PGA liquefaction thresholds is based on the pioneering work of Youd and Perkins (1978), who investigated susceptibility as a function of age and deposit type. Table 4 shows for each of the susceptibility category, a PGA threshold corresponding to zero probability of liquefaction. These threshold values are based on Hazus99TM (1999).

3.5 PGA thresholds for induced phenomena (Ti)

To determine the soil amplification factor coeff, the PGA threshold for inducing phenomena (Ti) must be inferred from geologic maps or field surveys. For a given scenario, if Avib > Ti, a given phenomenon is likely to appear. The corresponding Ti value is fixed at a very high value that cannot be realistically achieved by Avib. For both induced phenomena, we calculate a safety factor reflecting the probability that a given phenomenon occurs after the earthquake.

3.6 The method VULNERALP level 1.0

3.6.1 Vulnerability assessment

The VULNERALP method was proposed. Since some similarity exists between the Italian old buildings (especially masonry) and Algerian, it is based entirely on the method of GNDT (1993b). The weights and Italian scores of each structural criterion were considered as no return of comprehensive experience available in Algeria. However, for a context of moderate seismicity, it is essential to have a first level of evaluation (1.0 level) that is as simple as possible, for first hierarchical “seismic” buildings. That is why the structural criteria GNDT were alleviated. This implies simplified visual auscultation, especially as the GNDT was developed on the basis of observations of damage, the situation where construction is laid bare and structural criteria more easily completed. This method, certainly the most validated in Europe, also allows the representation of the damage and their equivalence to the average damage D, according to the European macroseismic scale (EMS98 2001) (Fig. 7). The types of constructs used in VULNERALP are identical to those detailed in EMS98. As a result, the IVi function of the building material is fixed on the work presented by Giovinazzi and Lagomarsino (2003) who translated the vulnerability classes of the European Scale EMS98 in IVi (Table 6).

Fig. 7
figure 7

Equivalence between EMS98 scale damage levels and the average damage D calculated by the method of GNDT, according to Giovinazzi and Lagomarsino (2003)

Full size image

3.6.2 Vulnerability index

The 1.0 level is established on the basis of six criteria that are structural building material, irregularity in elevation and plan, the shape of the roof, its period of construction, and the nature of the foundation within the meaning of GNDT. IVi the means and the range of likely values extracted from GNDT vulnerability matrices are assigned to each criterion, for example, the GNDT four types of environment (Table 7), functions of the slope, and nature of foundation materials. These classes can be grouped in any case in rock or not, hillside or not, two levels of information that can be filled with high reliability. These two classes then have median scores and the likely intervals.

So as to be consistent with the GNDT, structural criteria not accepted by the VULNERALP method are nevertheless taken into account in calculating the IV, the associated score (and range) corresponding to the median (and range) of values of GNDT the scores. The overall IV vulnerability index is normalized to 100. Under VULNERALP, IVi the quality of construction is adjusted according to the time of construction, reflecting in France changing codes design and use of building materials. This strategy then connects the vulnerability to damage level compatible with that of the EMS98. Vulnerability matrices are then offered, with the range of likely values extracted from the criteria GNDT.

The overall IV of the structure is calculated by the following formula:

$${\text{IV}} = \mathop \sum \limits_{i = 0}^{n} {\text{IV}}_{{{\text{I}} }} W_{i}$$
(2)

The average damage D is calculated based on the intensity of IEMS98 scenario earthquake and IV, and the following relationship of GNDT:

$$D = 0.5 + 0.45\arctan\,\left( {0.55\left( {I_{{{\text{EMS}}98}} - 10.2 + 0.05 {\text{IV}}} \right)} \right)$$
(3)

The estimated intensity for a given scenario can be mapped based on earthquake intensity using the formula of (Geoter 2002):

$$I = 3.92 \times \log 10 \, \left( {\text{Avib}} \right) + 10.74$$
(4)

Avib is a lithological term.

4 Results and discussion

In this work, we have created seismic scenarios for Constantine city, using spatial analysis with layouts suitable to determine vibrational and peak accelerations and estimate potential damage. We now present the results of our deterministic approach.

4.1 Risk assessment

Figures 8 and 9 present the peak ground acceleration values, which can reach 0.10–0.13 g for an M = 6.0 local earthquake. Because soil in the valley El Rhumel is mainly composed of recent quaternary alluvial deposits from the Rhumel and Boumerzoug rivers, the largest PGA values are found here. On the road between the Old Town and Boumerzoug, vibrational accelerations are estimated with soil conditions to be 0.24 g in Mansourah and Kasteur; 0.12–0.13 g in Old Town, Sidi-Msid, and Constantine University; 0.16 g in Sidi Rached and Ben Badis; and 0.20–0.22 g in Bellevue, Benchergui, Djenane-El-Zitoune, and Bent-Elisse (Fig. 10). Using geologic data, we identify homogenous zones where induced phenomena could occur. Tow threshold values corresponding to different types of induced risks are attributed to each zone. Several landslides have been detected on the slopes that border the Rhumel valley and Ciloc. Analysis of the associated slope angles, surface geologies, and hydrological conditions yields landslide threshold values of 0.10–0.25 g; the associated triggering threshold values are 0.12–0.21 g. Finally, liquefaction occurs in the main alluvial fan of the valley at 0.15 g and in secondary alluvial areas if acceleration reaches 0.09 g. The latter value may be underestimated because river alluvium is composed of coarse gravel with a few small, isolated sand lenses. Regarding the PGA values derived in the chosen scenario, liquefaction presents a non-null security risk in Bellevue, Ciloc, Bent-Elisse, and Djenane-El-Zitoune (Figs. 11, 12).

Fig. 8
figure 8

Maximum acceleration (Amax) in the Wilaya of Constantine

Full size image
Fig. 9
figure 9

Maximum acceleration (Amax) in Constantine City

Full size image
Fig. 10
figure 10

Acceleration (Avib) in the study area of Constantine City

Full size image
Fig. 11
figure 11

Intensity (I) in the study area of Constantine City

Full size image
Fig. 12
figure 12

Factor of safety (Fs) from the effects of the study area

Full size image

4.2 Calculation of maximal acceleration (Amax)

We used the attenuation law of (Bommer et al. 2003), which estimates maximum acceleration for a rocky site. The results obtained are shown in maps (Figs. 8, 9).

4.3 Calculation of vibrational acceleration (Avib)

The same factors that amplify maximum acceleration (PGA) (e.g., lithology, EC8 1998) will change the vibration risk “Avib” significantly. The risk map in Fig. 10 shows peak ground acceleration, accounting for lithological site effects (Avib). We observe the strongest vibrations along the two rivers (Boumerzoug and Rhumel). We note that EC8 gives a good estimate of acceleration in the three study areas (Bellevue–Ciloc, Coudia, and Old Town). By integrating the acceleration with the geology, we observe an acceleration of 0.13 g in the neritic limestone (Old Town), which is similar to the value of 0.14 g (PGA) obtained without taking lithology into account. Moreover, according to the RPA (Algerian Earthquake Engineering Code), the Wilaya of Constantine is classified in Zone IIa (medium seismicity) with an expected peak acceleration rate of 0.25 g (RPA 2003).

4.4 Calculation of intensities (I)

The implementation of a damage scenario enables us to map the probable extent of damage to the Constantine city. On both maps in Fig. 10, we observe a high intensity along the two rivers (Boumerzoug and Rhumel).

4.5 Calculating the coefficients of safety (Fs) of induced effects

Figure 12 defines areas susceptible to liquefaction and landslides. From the map, we notice that liquefiable zones correspond to geologic formations (sand) in these areas, e.g., Benchergui, Bellevue, and Kef-Chadad.

4.6 Damage assessment

The use of geographic information system data has allowed us to cross-reference the vulnerability of specific neighborhoods (Bellevue–Ciloc, Coudia, and Old Town) against peak ground accelerations. Figure 13 shows estimates of damage rates for these areas: values are 0.399–0.663 for Bellevue–Ciloc, 0.501–0.738 in Coudia, and 0.634–0.793 in Old Town. We note that the highest expected rates of injury are in Old Town.

Fig. 13
figure 13

Damage rates for the three districts (Bellevue–Ciloc, Coudia, and Old Town)

Full size image

4.7 Building vulnerability estimates

Three zones were defined in Constantine city: the Coudia area, Bellevue–Ciloc, and Old Town. In all, individual vulnerabilities were assessed for 90 buildings. Old Town has the highest vulnerability index; however, indices are similar among the zones, ranging from 52 (Bellevue–Ciloc) to 69 (Old Town) (Fig. 14).

Fig. 14
figure 14

Example of the boundary of the Bellevue–Ciloc area (red-dashed line), mean vulnerability index IV = 52. The boundary of the Coudia area (green-dashed line), mean vulnerability index IV = 59 and the boundary of the Old Town area (yellow-dashed line), mean vulnerability index IV = 69

Full size image

4.8 Damage evaluation

By linking PGA to the vulnerability index that we defined for the homogenous zones, we are able to estimate the mean building damage in upper Coudia. These values range between 0.502 and 0.738 (Fig. 15). Because of lithological site effects, the greatest building damage is expected in Old Town. However, only a few buildings line the Rhumel Valley; most homes are concentrated in the center of Old Town. Bellevue and Old Town are the most populated residential areas in central Constantine; it is thus interesting to study the damage that might occur at these locations on a finer scale. We find maximum damage values in the center of the Old Town. We also note that damage decreases with increasing distance from this area, reaching values as low as 0.399 (Fig. 16). Many buildings would be either destroyed or severely damaged in Old Town (Fig. 17). Nevertheless, we notice that damage is more severe in Old Town than in Bellevue–Ciloc and Coudia. These increased damage levels do not solely reflect the vulnerability index, which is higher in Bellevue–Ciloc and Coudia.

Fig. 15
figure 15

Building damage Coudia square

Full size image
Fig. 16
figure 16

Building damage Bellevue–Ciloc square

Full size image
Fig. 17
figure 17

Building damage Old Town square

Full size image

5 Conclusions

Seismic risk assessments are priority activities in all countries that can potentially experience large earthquakes. The ultimate goal is to ensure public safety in the event of an earthquake, that is, to avoid major structural damage and loss of human lives. Damage is unavoidable but can be limited. In this work, we present a deterministic seismic risk assessment for the Constantine city, Algeria. We have assessed the seismic risk potential in detail. This work presents a methodology for assessing the likely consequences of future earthquakes in urban areas. Results are provided in the form of maps of the Constantine region. Evaluation consists of assessing the maximum acceleration of ground motion, taking site effects and area geology into account, as well as induced effects (landslides and liquefaction). Therefore, it is important that data are sufficiently well structured to be properly analyzed in GIS software, including MapInfo and ArcGIS. Of note, the direct application of EC8 yields sound estimates of alluvial PGA amplifications throughout the Constantine city. This shows that EC8 correctly accounts for lithology. Constructing such risk maps for seismically vulnerable areas permits us to (1) identify geographic areas affected by seismic risks; (2) determine more precisely whether induced effects (landslides and liquefaction) are localized on backfill, on grounds reworked by relatively old natural landslides, or on rigid bedrock affected by active deformation (e.g., seismogenic faults, broken fault, isostatic adjustment); and (3) limit the damage to buildings and to construct new buildings with seismic risk prevention in mind.