Seismic risk assessment for the downtown of the city of Blida, Algeria

(Mw6.6), and Hammam Melouane (Mw6.5). The sensitivity analysis demonstrated the importance of the selection of the performance point computation method (improved displacement coefficient method -IDCM, modified capacity spectrum-MADRS, and nonlinear analysis method-N2) and the choice of the ground motion prediction equation. IDCM results are less influenced by the choice of the GMPE, but they provide higher damage results expressed as a mean damage ratio. Moreover, the study estimated potential human impacts in the Blida region, highlighting varying levels of impact on different geounits under different earthquake scenarios. The study's primary findings from seismic risk assessments in the studied region highlight its high susceptibility to earthquakes and can be summarized as follows: The mean damage ratio will be 52.6% ± 1.4%, 50.9% ± 1.6%; 31.8% ± 3.4% and 21.4% ± 3.1% for the Blida, Bounaian, Mouzaia El Affroun and Hammam Melouane earthquakes respectively.


Introduction
Algeria, like many other countries, has experienced significant natural disasters, including earthquakes, leading to societal disruptions and substantial economic losses.These disasters have a particularly significant impact on urban areas with concentrated social infrastructures and underground lifeline facilities.The region of Blida in Algeria is characterized by relatively high seismic activity, with numerous destructive earthquakes recorded over the past two centuries.The strongest earthquake in the Blida region occurred on March 2, 1825, (Mw7.1),resulting in extensive damage and a death toll of approximately 7000 people [1][2][3][4][5][6][7][8][9][10].Another significant earthquake struck the region on January 2, 1867, (Mw6.6), causing substantial loss of life and property [8][9][10][11].
The downtown area of Blida (Fig. 1) is marked by a significant population and a large number of old buildings, including the historical centre named Ouled Soltane, commonly known as "Douirette".This area represents the oldest district, predominantly comprised single or two-storey buildings dating back to the colonial era.These coexist with a mix of recent and Ottoman-era buildings in an advanced state of decay [12].Consequently, it faces a substantial seismic exposure and inherent vulnerability.
To effectively assess and mitigate the seismic risks in downtown Blida, a comprehensive assessment must consider various factors.These include seismic risk elements such as the geological and tectonic characteristics of the region, the proximity to active faults, soil conditions and site amplification effects, structural vulnerability of buildings, and the population density and demographics.By considering these factors, it becomes possible to develop a thorough understanding of the potential impact of earthquakes in the city and implement appropriate measures to enhance the resilience and safety of downtown Blida.
In recent years, numerous studies conducted in Algeria have focused on predicting potential seismic damage to urban structures ( [13][14][15][16][17][18][19], among others).These studies have utilized physics-based methods, including empirical fragility analysis, to assess how buildings respond during earthquakes.These approaches involve obtaining data from imposed scenarios in various regions, such as northern Algeria, and using fragility curves or matrices to estimate the vulnerability and potential damage levels of different building types, with a specific focus on residential buildings.Despite the common use of physics-based methods by the aforementioned authors, these studies have employed varied tools and techniques for earthquake risk assessment.Often, they neglected the integration of geotechnical data for evaluating local site conditions, such as the shear wave velocity of the uppermost 30-m layer (Vs30).Therefore, it is crucial to emphasize that incorporating geotechnical data, specifically Vs30, is an effective approach for predicting local site amplification.Besides, similar studies have already been performed for urban areas outside of Algeria ( [20][21][22][23][24], amongst others).F. Bellalem et al.In this study, we aim to address the existing knowledge gap by conducting a comprehensive assessment of seismic risk in downtown Blida, Algeria.The main objective is to provide valuable insights into earthquake vulnerability and consequences in the area, contributing to a more comprehensive understanding of the seismic risks faced by downtown Blida.To achieve an accurate seismic evaluation at the urban level, establishing an appropriate taxonomy is crucial to allow the classification of different buildings in Blida into a set of model building types with their corresponding capacity and fragility.In our study focused on downtown Blida, Algeria, we have adopted the typological classes or taxonomy proposed by Lagomarsino and Giovinazzi [25].This classification assigns specific classes to buildings in the region of interest, taking into account the height of the buildings, the seismicity of the region, and the ductility classes.The chosen classification is based on the main characteristics of buildings in the Euro-Mediterranean region, making it relevant to our study.
We used fragility and capacity curves developed within the EU-FP5 RISK-UE project for the chosen taxonomies [25].These curves provide valuable information about the probability of damage for each of the four damage levels (Slight, Moderate, Extensive and Complete) for a given building type.To compute the damage, we used SELENA, a software tool specifically designed for this purpose [26].Furthermore, we compared the impact of four earthquake scenarios: Mouzaia-El Affroun Fault F1 (Mw6.6),Blida Fault F2 (Mw7.1),Bouinan Fault F3 (Mw7.1), and Hammam Melouane Fault F4 (Mw6.5)into our analysis.Integrating these scenarios with our procedures aimed to comprehensively assess the potential impact of earthquakes on Blida city and its inhabitants.This approach enabled us to evaluate the damage and losses depending of the magnitude and the distance to the city, thereby gaining a deeper understanding of the potential consequences.Ultimately, our findings provide valuable insights into the seismic resilience of downtown Blida, since it can contribute to the development of effective strategies for mitigating earthquake risks.

Seismotectonic setting
Numerous studies related to the seismicity, seismological and seismotectonic in the southern border of the Mitidja basin have been carried out to better understand the tectonic activity and assess its seismogenic potential [27][28][29][30][31].The active deformation model of the Mitidja Basin remains a matter of debate.However, the recent re-evaluation of the historical seismicity and seismological analyses of recent earthquakes occurring on its southern edge has provided a better insight into the characteristics of its active tectonics [9,27,31,32].The Blida reverse fault system represented by a more than 100 km segmented fault and its offshore continuation, may likely be the origin of the strong Zemmouri 2003, Mw6.9 earthquake.This tectonic scarp exhibits an impressive topographic offset, sometimes exceeding 1000 m along the northern base of the Blida Tell Atlas [29].Fig. 2 illustrates the four different fault segments capable of generating destructive earthquakes that could cause damage in the city of Blida.Historically, one of the most devastating earthquakes in the Blida area was that of the 1867 Mouzaia-El Affroun earthquake.This event could be linked to the E-W Mouzaia-El Affroun reverse fault segment which displays left-lateral strike-slip movement on its SW end [27].The assessed magnitude of this earthquake shows great uncertainty but could be 5.5 ≤Ms ≤ 6.5 [9].Towards the east the Blida, Bouinan and Hammam Melouane segments are also sources of the ongoing seismicity, particularly the earthquakes of 1854, 1908 and 1961 with Mw5.5 [9,11,33,34].Additionally, a more recent earthquake occurred in 2013 with a magnitude of Mw5.0 [35].The maximum magnitudes in Blida and its Fig. 2. Seismotectonic map illustration of major structural features in the Blida zone (Green) and neighbouring area.The plotted seismicity represents the most significant earthquakes that have occurred in the Mitidja basin from 1365 to the present day.Each seismic event is represented by a transparent box, indicating the year of its occurrence, the epicentral intensity I 0 , and its magnitude (M) if available; (de): destructive event.The faults MEA, Blida, Bouinan and HM (F1 to F4) are highlighted in transparent yellow boxes (modified from Refs.[9,27]).(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)F. Bellalem et al. vicinity have been evaluated using three different approaches: historical records, deterministic methods (employing Wells and Coopersmith relationships [36]), and expert opinions.Additionally, depth information derived from field observations and measurements are summarized in Table 1.The most significant earthquakes from 1760 to present day are plotted in Fig. 2, with epicentral intensities (I 0 ), and magnitudes, where available.Due to its shallow nature and proximity to the highly urbanized city of Blida, such seismicity could cause significant damage.

Soil characterization
As illustrated in Fig. 3, the study area was divided into 16 cadastral sections according to the standards established by the National Soil Conservation and Cadastre Office.These divisions facilitate administrative pocesses, enabling a comprehensive inventory of the current building stock and the incorporation of demographic data from a preliminary housing survey conducted by the office for National Statistics (ONS).To characterize the local site conditions of each cadastral section, the shear wave velocity (Vs30) of the uppermost 30-m layer was selected as the key parameter.This approach effectively predicts local site amplification, as indicated by various references [37][38][39][40][41].The seismic-refraction method was chosen to investigate the distribution of seismic velocity and changes in layer thickness within the study area.By calculating wave propagation velocities in different soil layers, seismic refraction provides a precise depiction of subsoil's structural arrangement [42][43][44].This technique is applicable to shallow survey depths (from a few meters to hundreds of meters) and is particularly useful in engineering and hydrogeology problems [45][46][47].This paper employed the seismic refraction technique along three profiles (Fig. 3), using 24 geophones to acquire seismic wave data with a dominant frequency of 10 Hz.The seismic waves were recorded using a 24-channel PASI-16SG seismograph as the data logger, enabling the data to be stacked.A standard 63.5 kg weight drop at a fixed height served as the seismic source to generate the compressional wave velocity.The intercept time method was processed using WinSism™ (Geosoft) version-13 software.Due to site constraints, the total number of geophones varied based on the available space.For seismic survey lines PS-01 and PS-02, the space between geophones was set at 4 m to achieve a profile length of 92 m (24 geophones x 4 m), while for seismic survey line PS-03, the space between geophones was 3 m, resulting in a profile length of 69 m (24 geophones x 3 m).The average shear wave velocity up to a depth of 30 m, commonly referred  to as Vs30, was then calculated.The findings, as displayed in Table 2 indicate that the distribution of Vs30 decreases gradually in an easterly direction, ranging from 555 m/s at PS-03 to 539 m/s at PS-02.According to the Algerian Building Code (RPA, 2003 [48]), these values fall under the category of stiff soil and are characterized by average Vs30 values ranging from 400 m/s to 800 m/s, corresponding to the S2 soil type classification in the Algerian building code.These findings, along with soil section maps (see Fig. 4), were used to interpolate Vs30 measurements to other parts of the study area.Consequently, for all units in the study area, an average value of Vs30 was assigned to the centroid of each cadastral section.

Ground motion models
To perform seismic hazard assessment, damage and loss evaluation, and the design of seismic-resistant buildings at a specific location, a site-specific response spectrum is necessary.This spectrum can be derived from ground motion models, also known as ground motion prediction equations (GMPEs).GMPEs describe the median and the variability of ground-motion amplitudes, depending on factors such as the magnitude, site-source distance, site conditions, and other parameters.In this study, three new generations of GM-PEs were employed to calculate Peak Ground Acceleration (PGA) and spectral acceleration values, within each cadastral section through deterministic analysis.To select the GMPEs for this study, a list of approximately 800 published seismic ground motions (estimating PGA and elastic response spectral ordinates, published between 1964 and early 2021) was compiled by Ref. [49].From this list, only the most reliable GMPEs were retained, based on strong motion data obtained from Algeria ( [50] [LA18]) and other regions worldwide with comparable seismotectonic conditions ( [51] [AK13] and [52] [BR14]).It is worth mentioning that LA18, AK13 and BR14 are all based on moment magnitude Mw, which is the preferred magnitude scale.Additionally, all three of these GMPEs provide site effect coefficients, as well as a reduced prediction standard deviation (σ).The weights allocated to each GMPE, as shown in Table 3, were determined through expert judgment, following the general recommendations outlined in Ref. [53].

Earthquake scenarios for Blida and logic tree development
To properly evaluate the impact of future earthquakes in downtown Blida city and its inhabitants, a comprehensive understanding of the seismicity due to the active faults is crucial, along which choosing corresponding earthquake scenarios (in terms of magnitude, depth and fault location).Considering the seismotectonic context explained in section 2, four different earthquake scenarios on different faults (Table 1) were chosen: Mouzaia-El Affroun Fault F1 (Mw6.6),Blida fault F2 (Mw7.1),Bouinan Fault F3 (Mw7.1) and Ham-  mam Melouane Fault F4 (Mw6.5).By running simulations on these different scenarios, we can discuss the damage and loss scenarios in the city but how these results may vary for each earthquake rupture.The results can also assist in identifying the most damaged areas within the city, aiding the development and implementation of effective emergency plans and mitigation strategies.It is important to note that the seismic damage and loss results are always affected by epistemic uncertainties, which can be considered and analyzed using a logic tree.We considered the uncertainty in the earthquake source simulation four different earthquakes and for each one we included the uncertainty in the ground motion prediction equation used (Table 3) and the performance point computation method (see section 3.7).The GMPEs were assigned different weights, with LA18 having a weight of 40% while AK13 and BR14 had a weight of 30% each.This indicates that LA18's predictions are considered more critical or reliable than those of AK13 and BR14.The performance point method was equally weighted.Only one soil model, corresponding to stiff soil, was used.Fig. 5 presents a diagram of the logic tree computation scheme used in this study, so each simulated earthquake will provide nine different results (one for each branch of the logic tree with corresponding weights).

Building and population data collection
The concept of earthquake vulnerability is crucial in predicting the average damage that buildings are likely to experience on a large scale in the event of a given magnitude.Earthquake vulnerability analysis also identifies of the buildings and structures at a regional, city, or county level that may suffer higher damage from a given ground motion.This analysis is essential for determining which buildings may need reinforcement or retrofitting, making it a crucial first step in seismic risk mitigation efforts.Vulnerability analysis includes factors at risk (physical, social and economic) and the nature of associated risk such as damage to structures and systems and human losses [54].To conduct a seismic vulnerability assessment, building inventory and population information must be included.However, other elements (social, economic, etc.) would have only a minor impact on the assessment results if they were not included.For the case study in Blida, a comprehensive building inventory survey was conducted using a rapid visual screening method across a small area (∼94 ha).This survey covered all buildings within the zone and involved in-depth investigations into various attributes, including building structural type, occupancy class, total floor area, number of stories, seismic vulnerability characteristics, construction period, and typology.This in-situ fieldwork allowed the gathering of essential data without the need for structural calculations.Upon completion, this information was integrated with GIS-based city planning database for advanced risk analysis, along with other geographical layers such as demographic data from a preliminary housing survey conducted by the Office for National Statistics (ONS) in 2020 and the location of traffic lanes.In total, 2607 buildings were surveyed across 16 geographical units, constituting the building stock.The residential general occupancy is the most significant, both in terms of exposure and building count, due to the large number of collective housing and single-family dwellings, which constitute 83.6% of all buildings.Table 4 shows the number of buildings, dwellings, and the population in each cadastral section, obtained from GIS building inventory data for the downtown Blida.

Buildings typology
To conduct an accurate seismic evaluation at the urban level, beginning with an appropriate taxonomy is crucial.Taxonomies are important as they provide a simplified way to describe the structural behaviour of existing buildings, avoiding complex structural modelling.However, it should be noted that while the grouping of model building types into taxonomies is appropriate for developing damage and losses scenarios needed for emergency management, specific infrastructures like hospitals and schools should be treated individually by structural modelling.Once a taxonomy is established, each building can be assigned to a specific typological class, a critical process for meaningful seismic evaluations at the urban scale.The expected damage during a seismic event is directly related to the building type, making this attribution process crucial.The evolution of taxonomies started from the 78 different types of structural systems proposed by ATC-13 [55] which were later named FEMA model building types [56].These US-specific buildings influenced the global building classification system (PAGER-STR, [57]).In Europe and the Mediterranean region, the RISK-UE project also defined taxonomies for residential and historical buildings [25].Additionally, the Global Earthquake Model has developed a worldwide taxonomy [58], defined by 13 attributes that describe specific characteristics of individual buildings.However, the GEM taxonomy requires very detailed knowledge of the buildings so in this study, we have adopted the typological classes proposed by RISK.UE [25].This approach offers a simplified way to assign a specific class to each building in the region of interest.We selected this classification as it is based on the main characteristics of buildings in the Euro-Mediterranean region, making it relevant to our study.Table 5 presents the building typology classification adopted in this study.The classification considers three different classes of building height: Low-Rise (_L), Mid-Rise (_M), and High-Rise (_H), defined based on the number of masonry floors (_L = 1/2, _M = 3/5, _H ≥ 6) or reinforced concrete floors (_L = 1/3, _M = 4/7, _H ≥ 8).It also takes into account the seismicity of the region, with zones I (_I), II (_II), and III (_III, as well as the ductility class (-WDC = without ductility class, -LDC = low ductility class, -MDC = medium ductility class, -HDC = high ductility class) [25].
Based on the building typology and height, using the nomenclature provided in Table 5, a total of nineteen model building types have been identified in downtown Blida.The number of buildings of each model type in the studied region is shown in Fig. 6.
The distribution of buildings in downtown Blida, as revealed in Fig. 6, shows that the majority of the buildings fall under the lowrise M3 and M2 types.These buildings typically have unreinforced masonry walls made of bricks and simple stone in various combinations.Notably, all these buildings were constructed before 1980 and were not designed according to any building code.Fig. 7 rep-   resents the distribution of the M3_L and M2_L model building types.The cadastral section 147 and 152 have the highest concentration of these model building types, respectively.

Population inventory
The concept of "population exposed to seismic hazards" refers to individuals residing in seismically dangerous areas or locations susceptible to potential property destruction of property and loss of human life resulting from seismic events [59,60].Therefore, a population inventory is a crucial tool for assessing the impact of an earthquake on society.By creating a population inventory, it is possible to identify areas of high exposure and test the impact of different earthquake scenarios, aiding disaster preparedness.
This study utilized current population data obtained from a preliminary housing survey conducted by the Office for National Statistics (ONS) in 2020.The customary census activities of the ONS typically involve conducting surveys and interviews with individuals, households, and communities to collect demographic information, including age, sex, marital status, education level, occupation, and household composition.The results of this census allowed us to determine the number of inhabitants in each district and for the different existing typologies in the region.The number of inhabitants was deduced by leveraging the information gathered in a dedicated inventory sheet, which integrated building structural data with inhabitants per building obtained from the ONS census.This comprehensive integration facilitated the analysis, enabling the determination of the number of inhabitants for different existing typologies in the region.As observed, in most of the geounits the inhabitants are living in masonry buildings.Only geounits 146, 149 and 155 show a higher population in reinforced concrete buildings.

Fragility and capacity curves
To calculate the physical damage using different analytical methods, we employed fragility and capacity curves developed within the EU-FP5 RISK-UE project for the European building taxonomy [25].In this study, the capacity curves were obtained from a mechanical model known as the DBV (Displacement Based Vulnerability) method, which is applicable to both masonry and reinforced concrete frame structures.The DBV method represents a building's response using a capacity curve, also known as a forcedisplacement curve.This curve characterizes the structural response of a building in terms of stiffness, overall strength, and ultimate displacement [61].Our methodology incorporates three user-selectable methods for computing the performance point of each model building type due to a given specific response spectrum from the simulated earthquake and the corresponding capacity curve: the Modified Acceleration-Displacement Response Spectrum Method (MADRS) [62], the Improved Displacement Coefficient Method (IDCM) [62], and the nonlinear static method (N2) developed by Ref. [63].These methods are included in the software SELENA [26].
Once the performance point is computed, the damage probability is obtained through fragility curves which express the probability of a building class sustaining a specific damage state (ds) when subjected to a certain level of ground motion intensity.Typically, these curves are represented by a lognormal cumulative distribution function, with a median value and logarithmic standard deviation or dispersion.The mathematical form of the fragility curves is as follows [64]:


where S d is the spectral displacement corresponding to the performance point and S d,ds is the median value of spectral displacement at which the building reaches the threshold of damage state ds (slight, moderate, extensive and complete).
β ds : standard deviation of the natural logarithm of spectral displacement for damage state ds, Φ(): standard normal cumulative distribution function.
A detailed description of the different damage states (ds) previously mentioned can be found in Ref. [25].Table 6 defines the median value of the damage state threshold spectral displacements and the log-normal standard deviations for each taxonomy and for damage state, i.e., Slight, Moderate, Extensive and Complete.
Fig. 9 shows the capacity curves for all masonry and reinforced concrete buildings used in the study.Additionally, we have also represented the fragility curve for two different typologies, illustrating how different damage states are reached earlier in masonry buildings than in reinforced-concrete building.

Mean damage ratio (MDR)
Quantifying earthquake-induced building damage is complex due to the variance in disaggregation of damage estimates for different scenarios or building typologies.To address this challenge, the Mean Damage Ratio (MDR) serves as a useful parameter for comparing risk estimation across different geounits within a city or between different cities or countries.MDR is defined as the cost ratio corresponding to each damage state, expressed as a ratio to the cost of new construction [14,65,66].MDR is particularly beneficial for comparing risk estimation across different test beds, cities, or countries, as it remains stable and not influenced by factors such as inflation, exchange rates, and other variables that affect reconstruction and repair costs [26].However, while MDRs have their advantages, disaggregated damage estimates are crucial for predicting social losses, such as casualties, since complete damage states (particularly collapse) result in the majority of casualties and fatalities [26,66,67].According to Ref. [26], MDR, for each geounit i, is computed using the following formula: with: DR k j : is the damage ratio of model building type j corresponding to damage state k where k = S for slight, M for moderate, E for extensive and C for complete damage, N ki j : is the damaged built area corresponding to damage state k (S, M, E, C) the model building type j in geounit i, N Ti : is the total built area in the cadastral section or geounit i over all model building types.A total MDR for the whole city can also be obtained using: with N T the total built area over all model building types j and over all geounits i.

Human casualty computation method
The number of casualties due to direct structural damage for any given structural type, level of building damage, and injury severity can be calculated [26]: CSR casualty rate of severity i for damage state k P j,k : structural damage probability of the model building type j for damage state k N j POP : number of people in model building type j Severity of injuries goes from i = 1 light injuries (slight) to i = 2 hospitalized injuries (moderate), i = 3 life threatening injuries (heavy) and i = 4 injuries causing death (death) according to HAZUS [68].
To account for varying occupancy patterns dependent on the time of day, the SELENA methodology incorporates the calculation of casualty numbers for three distinct timeframes: nightime, when most of the people are in the residential buildings; daytime and commuting time scenarios.For our study, we focused on the nighttime scenario, anticipating the highest number of casualties among the population located at home during this period.We assumed that during the night, 90% of Blida's population resides indoors, and 10% are outdoors.Therefore, these results indicate that a higher vulnerability due to the presence of existing unreinforced buildings in the city will amplify the impact of upcoming earthquakes.

Sensitivity analysis
Investigating the sensitivity of three performance point assessment procedures (IDCM, MADRS, N2) across various earthquake scenarios revealed crucial insights into the uncertainties linked with earthquake damage evaluation.Fig. 11 illustrates the sensitivity of these methods across different earthquake scenarios.
In most cases, the IDCM method computes the highest values, while the N2 method computes the lowest.The differences between the methods are usually more significant in cadastral sections with a higher MDR.Additionally, we observe that the differences between the methods increase with the distance from the earthquake.For example, the cadastral geounit 141 has a standard deviation of the MDR of 3.33% and 3.61% for BL and BN earthquakes, which increase to 10.56% and 10.85% for MA and HM, respectively.
The influence of the selection of ground motion prediction equation (GMPE) on the MDR results is also notable.Fig. 12 compares the median and percentiles obtained from the logic tree for each performance point computation method and for each earthquake.A higher or lower dispersion indicates a greater or lesser influence on the GMPE choice.
For the BL earthquake, the influence of the ground motion prediction equation is more pronounced when using the N2 method, as indicated by the larger dispersion of the results.The AK13 GMPE provides the lowest MDR, while LA18 and BR14 yield higher and closest results for all the methods.This difference in results is more pronounced with the N2 method, as the median value shifts towards the highest MDR values.Regarding the BN earthquake, with a similar Joyner-Boore distance as the BN earthquake, the results for AK13 and BR14 are the same.However, due to a slightly higher hypocentral distance, the results from LA18 are now the lowest, and those from BR14 are the highest.Again, the N2 method appears more influenced by the GMPE choice, as shown by the larger dispersion.
For the furthest earthquakes, the influence of the GMPE choice increases for all methods, with MADRS showing higher sensitivity.This trend is similar for the HM earthquake, although the dispersion between results is lower than for the MA earthquake.In this case, LA18 provides the lowest MDR values, while BR14 provides the highest.
Thus, the choice of the GMPE and the performance point computation method are both crucial when applying spectral displacement methods in the seismic risk assessment.The IDCM method appears to be less influenced by the choice of GMPE for the simulated earthquakes, although this does not necessarily indicate it provides the most accurate MDR values.The reliability and robustness of  simplified methods for estimating displacement demand can only be assessed by comparing their results with those from a more accurate analysis (e.g.response history analysis).However, this is beyond the scope of this paper.
Finally, using the logic tree, it is possible to obtain a global mean damage ratio for the entire city.The results are summarized in Table 8.
As we observe the distance to the city and the magnitude of the earthquake significantly influence the Mean Damage Ratio (MDR).Furthermore, the uncertainty of the results increases as the distance to the city increases.

Building damage assessment
In this study, we have investigated the damage percentages categorized by building taxonomy under different earthquake scenarios.Across the analyzed earthquake scenarios (Mouzaia El Affroun M W 6.6, Bouainan Mw7.1, Blida Mw7.1, Hammam Melouane Mw6.5), the masonry or unreinforced taxonomies (M2_L, M3_L, M3_M, M5_L, M5_M, M6_L and M7_L) consistently exhibit a significant percentage of complete damage (Fig. 13).This highlights how the impact of a given earthquake will be more severe in buildings with higher vulnerability.
Similarly with the MDR, the closest earthquakes (BL and BN) exhibits the highest percentage of buildings with complete damage, with values exceeding 75% for taxonomies M2_L, M3_L, M3_M, M5_L, M5_M, M6_L.Reinforced concrete buildings do not exceed the 5% of complete damage, except for the taxonomy RC1-L_DCL, which reaches 17% of complete damage.The MA earthquake causes complete damage in 76% and 66% of M2L and M3_M buildings, 9% to30% in masonry and unreinforced buildings, but no damage in reinforced concrete buildings.
The HM earthquake results in 59% and 45% of complete damage in the M2L and M3M buildings, respectively and 4% to16% in masonry and unreinforced buildings.Again, none of the reinforced concrete buildings suffer complete damage or extensive damage, except for RC1_L_LDC, which reaches 1% extensive damage.
These results underscore the importance of developing strategies to improve the seismic performance of buildings, especially the more vulnerable M2_L and M3_M taxonomies, to enhance the overall resilience of communities in earthquake-prone areas.

Human casualty assessment
This section delves into the distribution of casualties and injuries across various geounits within different earthquake scenarios, providing crucial insights into the potential consequences of seismic events on affected populations.Fig. 14 visualizes the spatial distribution of the affected population (at least severity 1 injuries) across different geounits and the histogram of total injuries in the city for each severity level (1 to 4, that is slight to death) due to each simulated earthquake.The results are also averaged using the three performance point computation methods and the three GMPEs.
The analysis of different earthquake scenarios provides valuable insights into the distribution of casualties and injuries across various geounits.
Cadastral section 148, with the highest number of inhabitants: 2167 (11% of the total populations approximately), and a significance percentage of masonry buildings, is expected to experience the highest number of affected individuals in all earthquake scenarios.The affected population (at least slight injuries) due to the BL and BN earthquakes are projected to be 351 ± 24 and 344 ± 26 inhabitants (16% of the population in the cadastral section), respectively from which with 20 ± 2 are fatalities for each scenario (around 1% of the population in the cadastral section).The MA and HM earthquakes will affect to 205 ± 58 and 123 ± 50 inhabitants, respectively (9% and 6% of the population in the cadastral section) while the fatalities will be 11 ± 3 and 7 ± 3, respectively.
As we see, the impact is similar for BL and BN due to the rupture area reaching the city, and it decreases for the HM and MA earthquakes.
Cadastral section 141 has only 487 inhabitants (2.4% of the total population) but a high mean damage ratio.Thus, its impact on the population is lower due to the reduced exposure.BL and BN earthquakes will affect 98 ± 7 and 96 ± 8 inhabitants (20% of the population in the cadastral section) and the fatalities will be 6 ± 1 and 5 ± 1 inhabitants while MA and HM affect 58 ± 16 and 33 ± 15 inhabitants (12% and 7% of the total population in the cadastral section) respectively.
We can see that although a lower exposure will imply less affected population in absolute numbers however since section 141 has a higher mean damage ratio than section 148, it will cause a higher value of affected population in terms of percentage of the total population in the section.
Local authorities should prioritize districts with higher numbers of vulnerable buildings and inhabitants for first responder rescue efforts to minimize fatalities.
While our simulations should not be seen as precise predictions, given the high uncertainty in seismic losses, they are invaluable for emergency management.Identifying the most affected cadastral sections allows emergency teams to prepare evacuation routes and allocate resources efficiently to manage the affected population.

Conclusion
A comprehensive study conducted in downtown Blida, Algeria, has provided valuable insights into the seismic risk and vulnerability of the area.This study addresses the urgent need for a comprehensive assessment of seismic risk following the devastating earthquake that occurred in 1825.To accomplish this, the study used building typology classification according to Ref. [25] and incorporated the use of SELENA [26], a powerful seismic risk assessment software.SELENA played a crucial role in estimating seismic risk by enabling a thorough analysis of earthquake scenarios and their impact on different cadastral sections or geounits.The study divided the area into 16 cadastral and conducted a detailed building inventory that included demographic data, facilitating a comprehensive assessment.
The findings of the seismic risk assessment highlighted the severity and casualties associated with earthquakes in the Blida region.Cadastral sections 148 and 147 emerged as the most affected in all the simulated earthquakes due to the low number of engineering design buildings.Therefore, local authorities should be prepared for targeted risk reduction measures in these areas.
To evaluate the earthquake damage assessment procedures, a sensitivity analysis was conducted, comparing the improved displacement coefficient method (IDCM), the modified capacity spectrum (MADRS), and the nonlinear analysis method (N2).The analysis revealed that the IDCM method exhibits a lower sensitivity to the choice of the ground motion prediction equation.On the other hand, the IDCM method provides the highest results for mean damage ratio, while the N2 method computes the lowest.Regarding the choice of GMPE, if the rupture reaches the city, AK13 provides the lowest mean damage ratio, while LA18 is used if the earthquake is furthest.BR14 consistently provides the highest mean damage ratio, regardless of the earthquake's location.This highlights the importance of using logic trees to define different input parameters with different weights in order to account for the epistemic uncertainties.
Additionally, the study estimated potential human losses in the Blida region, shedding light on the varying levels of impact for different geounits under different earthquake scenarios.This information is crucial for first responders to effectively cope with emergencies.
This comprehensive study significantly enhances our understanding of seismic risk within the urban area of Blida city.The findings not only demonstrate the importance of proactive measures to mitigate the impact of earthquakes but also inform strategies for resilience and loss reduction.The study emphasizes the necessity of implementing targeted risk reduction measures in highly vulnerable cadastral sections.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Geographical situation (a) inset indicating the global localisation (b) geomorphic setting of the Blida zone surrounding area and (c) boundaries of downtown Blida city (red delimitation).(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3 .
Fig. 3. Representation of the urban area of Blida downtown as discretized by 16 cadastral sections on orthorectified aerial photographs with a pixel resolution of 12m.The blue squares represent the seismic profiles and the yellow dots indicate the centroid of each cadastral section.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Table 2 Fig. 4 .
Fig. 4. Sketch of cross section of the soil profile along E-W.CD-01 The core (wash/rotary) drilling is used to determine the sub-surface profile, by obtaining the SPT-N value of soil.

Fig. 5 .
Fig. 5. Schematic of the logic tree computation scheme used in the study.

Fig. 6 .
Fig. 6.Number of buildings belonging to each taxonomy that exist in downtown Blida.

Fig. 7 .
Fig. 7.The spatial distribution of un-reinforced masonry buildings is characterized by a) M2_L Taxonomy building type, and b) M3_L Taxonomy building type.

Fig. 8
illustrates the ratio of the population count across different typological classes present in the studied region, namely masonry buildings (a) and reinforced concrete buildings (b) for each geounit.

F
.Bellalem et al.

Fig. 8 .
Fig. 8. Population ratio for a) masonry and b) reinforced concrete buildings in the studied region.

F
.Bellalem et al.

F
.Bellalem et al.

Fig. 10 .
Fig. 10.Distribution of the averaged mean damage ratio per geounit for four different seismic scenarios, established using SELENA.

Fig. 11 .
Fig. 11.Mean damage ratio obtained for each earthquake and performance point computation method.

Table 1
Four scenarios of fault segments in the Blida zone considered in the analysis and respective source parameters.The depth is deduced using the average depth of the significant earthquakes of the Tell Atlas.

Table 3
Ground motion prediction models (GMPEs) used in the study.

Table 4
Detailed information regarding number of buildings, number of dwellings, and the population for each cadastral section.

Table 5
Building typology classification.

Table 6
Spectral displacement-based fragility curve parameters for each damage state for a specific building type.

Table 7
Comparative analysis of averaged Mean Damage Ratios and standard deviation for geounits under different earthquake scenarios.

Table 8
Impact of the simulated earthquakes in the city of Blida as a function of the global Mean Damage Ratio.