Terrorism Risk Assessment for Historic Urban Open Areas

Abstract

Making cities resilient and secure remains a central goal in urban policy strategies, where established methods, technologies, and best experiences are applied or replicated when the knowledge of a threat is already well established. The scientific community and specialized bodies are invited to comprehend and evaluate disastrous events that are still not well explored to broaden the concept of resilient cities. Among these, terrorism in the European-built environment remains an underexplored topic, despite various studies assessing its economic, social, and political dimensions, exploring the radicalist matrix, or examining the post-effects of high-impact disastrous events. Within this framework, this work presents an algorithm for the risk assessment of historic urban open areas (uOAs) in Europe, combining theories of the terrorism phenomenon, the normative experiences, and the phenomenological results of violent acts in uOAs. Specifically, the algorithm is determined by studying physical qualities/properties and elements that usually feature the uOAs, using a limited set of descriptors. The descriptors and their formulation are set starting from their qualification, in compliance with the risk determinant (Hazard, Vulnerability, and Exposure), and discussed starting from participatory methods (Delphi and AHP). The algorithm is finally applied to Italian historic squares, testing the mathematical approach, verifying theories of the phenomenon, and setting up a comprehensive three-dimensional risk matrix for both soft and hard targets. This latest constitutes an operative tool to assess the investigated built environment exposed to terrorist threats aimed at developing more detailed mitigative strategies.

1. Introduction

Two main macro-dimensions connect the “Cultural Heritage” and “terrorism threat” in cities. The first is prevalently related to the ideological–political component of the extremist movement, aimed at exalting its own cultural supremacy and promoting armed conflict [1,2,3]. In that sense, the main terrorist activities aim at the intentional disruption or smuggling of Cultural Heritage. The second one describes and exhibits the operative dimension of violent acts in terms of maximization of the impact, considering the number of victims and/or the potential publicity [4]. Here, the discussion focuses on the touristic attraction of places or buildings, which increases the number of potential people involved in the attack or/and the mediatic relevance of the violent act. As is clear, the issues relate, respectively, to national and local sizes, involving legislative regulations in fighting/preventing radicalism [5,6,7] and promoting secure cities [8], extending the issue to all nations [9,10].

When the discussion on the terrorism threat focuses on the cultural/historical sites/buildings and their users, the main issues are related to secure cities and people’s safety. This trend is also a consequence of the increasing number of violent acts in less politically exposed nations, such as Europe [11]. Several national norms and guidelines have been promoted in order to manage the threat after, during, and before the attacks. The activation of national regulations has made Germany, the United Kingdom, and France represent the most prepared European states, due to several recent hazardous events. The USA and Australia are notable worldwide for their historical background on the issue [12]. Regarding the specificity of European countries’ activities, the focus is on mitigative strategies and the dissemination of good behavior for users and policy advisors in tracing suspicious individuals [13,14]. On the other hand, the UK has supported the discussion on secure cities and the invisibility of strategies as inherent argumentation for reducing the risk proneness of urban places, particularly safeguarding their image in cultural and historical sites [15,16,17]. This evidence was fully discussed and presented in recent works by Quagliarini et al. [12,18]. This study is appropriately focused on urban open areas, which constitute the most vulnerable areas in cities due to their minor attention to regulations and the exploitation of low levels of mitigation and prevention. In fact, the phenomenological analysis of previous events in Europe has highlighted how such places are prone to such events [18], as a consequence of the high level of likelihood and the low quality of security. Moreover, when these places are combined with the presence of strategic and cultural buildings or sites, the level of hazard is increased due to the nature of the phenomenon, which is mainly interested in maximizing the publicity of the violent act [4].

1.1. Previous Experience in Risk Assessment in Cities Prone to Terrorism

When the focus is on terrorist risk assessment for the promotion of secure cities, the latest scientific studies that correlate terrorism and risk assessment in cities discuss four key themes (the studies derived from the SCOPUS database focus on the keywords “risk assessment; terrorist; terrorism; city”, excluding medical and psychological research, and referring to the period 2018–2024):

• New technologies and approaches to support counterterrorism agencies by means of innovative IoT tools [19,20,21] or social and collaborative approaches [22].
• Risk assessment of terrorist threat focusing on specific infrastructures or cities and regions [21,23,24,25,26,27], with major applications referring to Asia and the USA, and very limited cases in Europe.
• Vulnerability of buildings or the security assessment of human health when exposed to specific attacks in cities [28,29,30,31].
• Risk assessment of specific classes of special cultural buildings [32,33] to protect cultural contents.

Consequently, this overview underlines the lack of a specific discussion on urban open areas; however, the importance of historical and cultural ones is emphasized by the concept that these places are not constituted only by unbuilt areas but are the result of a combination of infrastructure, buildings, and sights with a rigid and invariable shape [34,35].

At the same time, the analysis of the regulation framework determined by the European guidelines in supporting risk management and assessment [36,37,38] has highlighted the necessity to use a common method to measure terrorism risk in cities, favoring the focus on the local characterization of the phenomenon, which influences the weapon availability and the attack type [4]. As it is clear, the risk assessment of terrorism is a complex matter involving economic, political/religious conditions that change at national and local levels and affect people’s security and (psychological) health [39,40,41,42] in different ways. However, this investigation can be supported by the individuation of a limited set of conditions and well-considered aims.

Research on promoting risk disaster comprehension has highlighted the strong relevance of three main elements of risk, called “determinant” [43,44,45]: Hazard is related to the nature of the study events and their frequency; Vulnerability describes the propensity or predisposition to be adversely affected; and Exposure discusses the potential effects or damages (on people, services, buildings). For the qualification of such determinants, the phenomenon requires to be parametrized coherently with features, conditions, and properties that affect single determinants. This process is widely applied for assessing and qualifying other physical or anthropic events; moreover, the process is usually combined with participatory methods to discuss the interdependencies and single weights in the overall risk assessment [46,47,48,49,50]. Undoubtedly, the numerical assessment of risk allows for the quantification of any analyzed threat, offering the opportunity to re-iterate the procedure in assessing strategies or stressors that alter the starting state of observation of the phenomenon or conditions. However, the quantification of risk requires critical translation in terms of constraints and rules for its management, also taking into account suitable levels of acceptability of risk. In that sense, the use of qualitative methods for terrorism risk assessment offers the opportunity for easy re-iteration using a limited set of data and details, also limiting the goodness of quantitative results for specific analysis (e.g., economic damages, number of victims, behavioral quantification during the evacuation process) [51,52,53,54] or more complex goals (i.e., national strategies for counterterrorism prevention) [55].

1.2. Aim of This Study

According to the described framework featured by specific gaps in the literature and the relevance of the matter for European cities, this paper aims to present a fast formulation to support the risk assessment of historic urban outdoor areas (uOAs), such as squares, exposed to terrorist events, mainly applied to places located within historic centers of cities. This is a common thread and a progression activity related to previous studies by the authors on observing, analyzing, and assessing terrorist events in common public open spaces [18]. This is the result of recent events in Europe that have involved pubs, museums, theatres, as well as promenades and squares, affecting the security of citizens. As the phenomenological study by the authors has already highlighted [18], such public outdoor places, also combined with the inner uses of public, cultural, and strategic buildings, have a critical level of inherent hazardousness, determined by the higher frequency of events in Europe. Moreover, these scientific endeavors are related to the national project BE S²ECURe—(make) Built Environment Safer in Slow and Emergency Conditions through behavioral assessed/designed Resilient solutions. This project aims to evaluate multi-risk conditions for uOAs during common activities, addressing both sudden and slow onset disasters to human behaviors.

Specifically for the paper contents, this work is structured into the five following sections:

• Method applied to determine the risk assessment formulation (Section 2).
• Characterization of parameters affecting terrorism risk assessment, derived both from previous works and the scientific literature (Section 3), and algorithm setup (Section 4).
• Calibration and test of the formulation in a set of real Italian case studies (Section 5), characterized by cultural/historical relevance.
• Discussion of results related to matrices of terrorism risk for uOAs as both hard and soft targets (Section 6).

2. Materials and Methods

As mentioned in the previous section, this work’s aim is to set up a rapid formulation for the qualification of risk assessment in uOAs intended as soft targets within the overall urban extension. The conceptual basis of the goal relates to the necessity of discussing the uOAs as systems of buildings, infrastructures, unbuilt areas, and people, focusing on material and immaterial properties and qualities. In that sense, human behavior is ignored due to the dynamic relations among people; however, the possible interferences between physical elements of the uOAs and users are discussed, starting from the existing regulations on mitigative strategies. All the international regulations and guidelines are referenced in Section 3. However, the focus of the formulation setup is on European uOAs. For them, previous phenomenological analysis [18] is pursued, starting from the recorded violent events in Europe within the Global Terrorism Database [56], which became the basis of the frequentist details within the formulation. Finally, it is useful to remark that the formulation is centered on the built environment in an as-built approach.

The methodological process is structured into two functional phases (Figure 1) for the purpose of formulating the setup:

• The parametrization process and formulation hypothesis where factors to be considered in the terrorism risk assessment for uOAs, and their qualification in terms of association with the risk determinant (Hazard, Vulnerability, Exposure) is identified. That is, it starts from the first results discussed in [12,18] for specific boundary conditions of the matter (attack types and the relevance of building and place usage) (Section 3).
• The collaborative validation of factors and formulations is identified in the previous phase by means of the Delphi technique, while the relationships among them are quantified using the Analytic Hierarchy Process (AHP) (Section 4).

Specifically for the structured methods, the Delphi technique allows us to analyze and describe a phenomenon or a problem that is still not fully supported by argumentation, as well as where experts’ judgment is required. The proposed technique is based on delivering anonymous questionnaires through controlled feedback to a specific group of experts [57,58]. Specific notes regarding the experts and their numbers are offered in the literature. The survey may involve from 10 to more than 1000 people who have relevant knowledge and experience related to urban risk. As far as the application of the Delphi technique to this work’s goals is concerned, the Delphi questionnaire is structured into three main round surveys, mainly applied for the following:

• Validating and implementing the recognized factors (previous operative phase) in the first round of the survey, including association with risk determinants (Hazard, Vulnerability, Exposure).
• Validating the dependencies of elements and properties in the formulation during the second survey round.
• Finally, the validation of influencing factors.

Particularly for the second and third rounds, all the answers can be validated considering the Lawshe Content Validity Ratio (CVR) [59], as usually noted in Delphi applications [60,61,62]. Clearly, the involved experts were informed about boundary conditions and previous results [12,18] of this study, ensuring a well-thought analysis.

In addition to validating the concerns regarding the influence factors, the AHP was applied. Specifically, similar to the Delphi method, the AHP method is based on the multicriteria theory formulated by Saaty [63], aiming at defining priority scales among a set of elements according to individual judgment. AHP is also based on questionnaires centered on the pairwise comparison among the involved elements, administered to a set of experts. Therefore, as in the Delphi method, all the results are analyzed in terms of consistency ratio (CR): a CR equal to 0% reflects a fully consistent element, while a CR > 10% corresponds to the superior limit of acceptability of consistency. The AHP process for the presented study is applied to the class of elements derived from the third survey round of the Delphi method, deriving their specific weight in the formulation [63]. It is worth noting that both Delphi and AHP are participatory methods and they constitute valid systems in supporting and considering expert judgment in risk identification, assessment, and management [36,37,38].

At the end of its setup, the formulation is tested on a real sample of squares (Figure 1) in order to validate it and, if required, calibrate it. In addition, its testing allows to determine the reduced risk matrices to describe the uOAs. Here, the testing phase takes advantage of real case studies as part of historic squares in Italy that have already been considered for the same project BE S²ECURe.

3. Setting up the Formulation for the Terrorist Risk Assessment for uOAs

3.1. The Translation of uOA as a System of Open Space and Uses of Buildings

As described in the introduction, the main goal of the intended rapid formulation is to support the risk assessment of real uOAs exposed to terrorism events. However, as previously demonstrated in [18], a close relationship exists between open spaces and functions or uses of buildings facing the uOAs in the selection of places by perpetrators. A recent study about the issue [64] has introduced the Space of Relevance (SoR) as the interface of the internal function of buildings and external public uses of uOAs. It is borrowed from the Space of Interferences introduced by Li Piani [65] who has already introduced the concept of interferences among buildings and spaces during violent acts. Each SoR can be delimited on the floor as the external area of the building facing open places, located along the facades with access; the SoR extension is calculated [64] in the following Equation (1):

A_SoR [m²] = A_CommBuild [m²] × C_B [pp/m²]/C_OUT [pp/m²]

(1)

where

• A_CommBuild is the commercial extension of the building.
• C_B is the maximum density capacity of people in the buildings [pp/m2].
• C_OUT the maximum density of people when public activities are conducted outside, as considered for public buildings.

C_B and C_OUT follow the Italian regulations (D.M. 03/08/2015 and National Ministerial Decree 19/8/1996) but can be extended for other cases in European countries, using the local values.

SoRs identification and qualification, as for position and extension, became functional in assessing real uOAs as complex open and public spaces where their uses and distribution of people may change during the day.

3.2. Selection and Determination of Factors Influencing the Terrorist Threat in uOAs

As described in the previous sections, the terrorist threat is a controversial issue. Due to the project goals, the related risk assessment has to determine a simple formulation based on the risk triangle that is useful in describing the terrorism risk of urban Open Areas (uOAs) in European countries. Previous studies by the authors have defined two main elements to delineate the complex theme. Specifically, in Cantatore et al. [18] the phenomenological analysis of the European threat in the last 20 years has highlighted that the most dangerous and frequent events in uOAs are prevalently featured by the following:

• Attacks in squares and streets (identified as Environmental class “F”) where public activities are present (pubs, museums) (Environmental class “B”) or representative and/or strategic buildings (Environmental class “D”).
• Type of attacks (T) to maximize the relevance of the damages referring to the armed assault (identified as T2) and car-bombing (T3).

Moreover, starting from the same literature framework, the same work [18] has already discussed the set of indexes (properly codified) that describe the determinant macro-properties (see

Table 1. Summary of indexes related to Hazard, Vulnerability, and Exposure determinants, as in [18].

Determinant	Code	Index	Description of Index
Hazard	H_I.1	Target	Inherent and environmental relevance of uOAs to be attacked
	H_I.2	Uses	Typologies of uses influence the choice of a perpetrator aiming at maximizing the effect of the violent act
	H_I.3	Prevention	Qualification of uOAs and their components regarding the presence of countermeasures or mitigative solutions
Vulnerability	V_I.1	Shape of BE	The shape of open spaces influences the attack-type effects
	V_I.2	Accessibility	Geometric dimension of uOAs and their components and permeability to perpetrators and their weapons
	V_I.3	Obstacles	Presence and level of social “attractive” urban furniture in the uOAs and their components
Exposure	E_I.1	Attack type	Effect of violent event related to the weapon type
	E_I.2	Crowd level	Character that reflects the potential numerousness of involved people
	E_I.3	Attack reaction	Potential level of users’ protection or obstruction within the uOA and its components thanks to the presence of protective or blocking elements

). Their factorization is discussed in the following sections.

Starting from such classification, all the indexes have been matched again with regulations and scientific literature in order to determine their relations with specific descriptors (K-types). For each K-descriptor, a formulation of calculation or assessment is introduced to be shared and validated with the panel of experts for the Delphi and AHP methods. Moreover, the formulation aims to simplify the matter, identifying specific features that are useful for translating indexing into quantitative characteristics. Descriptors and association with indexes are schematized in Figure 2, while a summary of their formulation is provided in

Table 2. System of K-descriptors and details for their calculation and value ranges.

Index Name	K Type	Ref. for Values	Normative	General Classif.	Equation	Details of Classification	Value
Hazard
Target index	KENV	[18]			KENV = [1, … 5]	Likelihood levels	Remote	Unlikely	Possible	Likely	Very Likely
							1	2	3	4	5
	KSYMB			x	KSYMB = [1, … 5]	Symbolicity classes	negligible	low	medium	high	Very high
							1	2	3	4	5
Index of use	KTUR			x	KTUR = Tour.Int = (n.arrivals)/(n.inhab)	Classes of intensity	very low	low	medium	high	Very high
							1	2	3	4	5
	KUSE			x	KUSE = [1, … 5]	Classes of use	rarely	low	normal	high	Very high
							1	2	3	4	5
Prevention index	KCON	[12]	x		${\mathrm{K}}_{\mathrm{CON}} ={\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}(Zi/Zeff)}{N.ACCESS}}$	Eff (T2)	Remote control	Direct/local control	Video Surveillance	Innovative systems
						Eff (T3)	Innovative systems	Reinforced urban furniture	Barriers	Bollards
Vulnerability
Shape index	KSHP	[35]		x	KSHP = fEXT × fSHP fEXT = [1, 5], fSHP = f(2P/A) fSHP = [1, 1.5] fSHP = f(w/l)	Classes of fEXT	0 < 2P/A < 0.02	0.02 ≤ 2P/A < 0.03	0.03 ≤ 2P/A < 0.06	0.06 ≤ 2P/A < 0.03	2P/A ≥ 0.09
							1	2	3	4	5
						Classes of fSHP	Compact w/l ≥ 0.7		1.5 (T2)	1.0 (T3)
							elongated or very elongated fSHP < 0.7		1.0 (T2)	1.5 (T3)
Accessibility Index	KPER	[35]		x	$$\begin{gathered} {\mathrm{K}}_{\mathrm{PER}}=[1,5] \\ {\displaystyle \frac{{\sum }_{i=1}^n\left(A_{vi}\right)}{2P}} \end{gathered}$$	Classes for r	0 < r < 0.05	0.05 < r < 0.1	0.1 < r < 0.2	0.2 < r < 0.3	r > 0.3
							1	2	3	4	5
	KACC		x	x	${\mathrm{K}}_{\mathrm{ACC}}={\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}(Avi\ast facc i)}{\sum _{\mathit{i}=1}^{\mathit{n}}Avi}}$	fACC = [1, …, 5]	Not accessible	Limitedly	Moderately	Alternatively	Accessible
							1	2	3	4	5
Obstacle index	KOBST(V)			x	${\mathrm{K}}_{\mathrm{OBST}}={\sum }_{i=1}^{n}d i\ast finf i$ di =Ai/Avi	fINF = [1, 1.25, 1.5]	No influence	average increase	increasing
							1	1.25	1.5
Exposure
Index of attack type	KATT	[18]			KATT = [4,5]	Consequence levels for Katt	Minor	moderate	Medium	Major	Extreme
							1	2	3	4	5
Crowding index	KCRW			x	KCRW = [1, …5]	Occupancy classes for KCRW	negligible	low	medium	high	Very high
							1	2	3	4	5
Index of Reaction	KOBST(E)			x	${\mathrm{K}}_{\mathrm{OBST}\left(\mathrm{E}\right)} ={\sum }_{i=1}^{n}d i\ast finfi\ast fshp\mathrm{o}\mathrm{b} i$	fINF	Decreasing	average decreasing	not influential	average incremental	incremental
							0.5	0.75	1	1.25	1.5
						fSHPob	negligible	low	medium	high	Very high
							1	2	3	4	5
	KCM	[12]		x	KCM = WEFF/Wi	WEFF = 3	Alarm countermeasures	Evacuation countermeasures	Systems of physical interventions

. A detailed description of descriptors and indexes follows in subsequent sections.

3.2.1. Descriptors Affecting the Hazard Indexes

Previous studies on European terrorism events have underlined the main relevance of ideology based on the phenomenon. The concept of “target” includes the necessity for perpetrators to understand the attack objective to guarantee the goals and the significance of the attack. These results allow us to understand which elements or features may affect perpetrators in choosing a uOA or a specific component. As the main findings from the literature on the topic, the elements affecting choice are related to (i) the political, religious, and/or cultural symbolism of the place or its components in order to maximize media impact, also considering the resonance of the target type (target index); (ii) the possibility to maximize the effect in terms of the number of potential people involved as an indirect consequence of the uOA uses (use index); and (iii) the level of protection within or along the frontier of the area to attack (protection index) [18].

Target Index [H_I.1]

Considering the target index [H_I.1], the phenomenological analysis in the European continent has highlighted that an inherent level of proneness exists for each class of uses of urban spaces and buildings, particularly concerning the weapon type. In Cantatore et al. [18], the frequency of events for classes of uses and attack types was standardized focusing on “Environmental uses” classes to which a specific level of likelihood is associated. Thus, the following points are important to consider:

• K_ENV is introduced as a descriptor of the statistical relevance of attacks for each environmental class combined with the attack in equivalent levels of likelihood, as already qualified in Cantatore et al. [18], with the following five levels: remote, unlikely, possible, likely and very likely classes. On the other hand, for similar classes of uses, the target choice can be associated with different levels of significance of places due to the political, social, cultural, and/or religious relevance [66,67].
• K_SYMB, as a descriptor of such features, may describe the variation in symbolic relevance of spaces; in fact, if some uOAs have a permanent (as inherent) symbolic significance for the institution and population, for others, the symbolic significance may be considered contingent on the presence of specific events [68]. For the description of this characteristic, five main classes can be introduced, ranging from negligible to very high symbolism classes.

Index of Uses [H_I.2]

As far as the Index of Uses [H_I.2] is concerned, the required descriptors for the “Hazard” have to discuss the attractiveness of the places, leaving the feature independent of the number of people involved. In fact, if the number of people affects the dimension of “Exposure” in risk assessment, the attractiveness of the real place influences the choice of place by perpetrators for soft or hard targets (Gordon Woo defines [4] it as “Inter-dependence and replacement of targets”). Specifically, the following points are important to consider:

• The K_TUR may be discussed for risk evaluation, as a descriptor of the inherent and potential reflection of representativeness of the place and the city. The touristic inflow usually correlates inhabitants to arrivals, and it may be considered at the city scale for annual, seasonal, or daily references, according to the primary nature of tourism. This is due to previous scientific outcomes in studying the interrelation between the touristic inflow and violent acts, even if their discussion has economic and political reflections [69,70,71].
• K_USE describes the standard use of the uOA and a single SoR. In fact, alongside the external level of attractiveness of places, the inherent vulnerability of uOAs and their sub-parts (SoRs) to attacks by perpetrators should also consider their use by inhabitants. Some daily conditions of use derived by the nature of the place (e.g., rendezvous points for people) and the use of buildings that border the uOA may also alter the potential level of assault, considering the daily variation (nocturnal, diurnal, evening).

For both descriptors, any normative requirements and ranges exist at the European scale; thus, a system of classes is considered, describing qualitatively the touristic intensity and use of places (Table 2).

Prevention Index [H_I.3]

Finally, focusing on the prevention index [H_I.3], the main significance is related to the presence of prevention strategies or solutions. As it is fully argued in the literature about the theme, the choice of place also depends on the fulfillment capability, according to the meaning and differences between hard and soft targets. In addition, as Quagliarini et al. [12] recently summarized, such solutions must be related to the attack type to consider their efficaciousness. In fact, as their main results, the review of major guidelines in documenting the compatibility and efficacy of preventive solutions, such as remote control, direct/local control, video surveillance, and innovative systems (i.e., face-detecting videos [72]) are recognized as efficient control along the boundaries of uOAs for the T2 attack type; on the other hand, innovative systems, reinforced urban furniture, barriers and bollards for the T3 attack type (see Table 2) [73]. Thus, a quantitative descriptor of the prevention index, such as K_CON, should be introduced to describe the presence and the number of protective systems for each possible access point. This involves comparing Zi, the number of effective protective systems, to the maximum number of systems (Zeff = 4) for each access of uOAs and each attack type analyzed in the present study (T2 and T3) (Table 2).

3.2.2. Descriptors Affecting the Vulnerability Indexes

Three major indexes are identified for Vulnerability (Table 1). When the analysis is focused on this determinant, the discussion relates to the main elements or features, which are independent of the perpetrator’s will; however, these elements describe the inherent propensity of exposed uOAs to suffer adverse effects when impacted by hazardous events. In detail, the analysis of the Vulnerability aspects of uOAs highlights three main features: (i) the geometric shape of spaces that may affect the potential of the arms (index of shape), (ii) the inherent capacity of places to be accessed by perpetrators (accessibility index), and (iii) the presence of sub-areas or places that can locally increase the crowd levels independently from the main use of places and buildings along the borders (obstacle index).

Index of Shape [V_I.1]

Specifically regarding the shape of BE [V_I.1], the main correlation to the literature is related to the dimension of places and the means of the attack, requiring two factors to assess K_SHP.

• The first factor qualifies the extension of the uOA (f_EXT) coherently with the ratio between its perimeter (2P) and area (A) extension, which usually describes the similarity between polygons. As in the other previous cases, five ranges of values are introduced, following the results of [35] (see Table 2).
• Discussing the relationship between the shape of uOA and T2 and T3 attack types, prevalent differences are recognized in the weapon classes: centralized or in movement [74]. All the weapons in T2 (armed assault) can be categorized as “centralized” arms, where the capacity for an attack is related to the maximum achievable distance of cold steel or firearms, including gunshots/ thrown weapons within a 360° range. While considering T3, the focus is on the vehicle that moves into the uOA excluding the possible range of associated arms (e.g., for car bombing). In this case, the ability to achieve and sustain high speeds during motion is a key feature of significance for vehicles [73,75]. Due to that, the index of shape describes the geometric employing the shape factor (f_SHP) that relates the width and the length of the places according to the following ratio:

f_SHP = f(w/l)

(2)

However, three classes of geometric shapes can be determined in qualitative ranges, as properly described for open areas in Rosso et al. [35], where f_SHP < 0.7 describes elongated or very elongated shapes, while f_SHP > 0.7 I is used for compact ones.

The relationship between the value of the shape factor and the associated index is influenced by their interactions with the weapon types. Specifically, for centralized attacks, elongated and very elongated spaces reflect high levels of Vulnerability when exposed to attacks involving moving vehicles [73], while the effect is the opposite when the weapon is centralized. In that sense, for both morphology classes, K_SHP is considered either non-influential or influential in terms of Vulnerability, assuming specific coefficients (see Table 2).

Accessibility Index [V_I.2]

Following the assessment of uOA Vulnerability, the second geometric feature is place accessibility [V_I.2]. In this case, the focus is centered on the possibility of the perpetrator accessing the places, thus the index may describe the intrinsic capacity of accesses to facilitate entrance. In this case, two main factors are found to describe the property:

• The physical and geometric accessibility of the perimeter of BE as inherent features for continuous or discontinuous fronts, called K_PER (perimeter factor). It correlates specific values to the ratio (r) between the sum of the width of entrances (A_vi) and the total perimeter (2P) of squares. Specifically, r values may vary within the limits of 0 (enclosed places) and 1 (open places), even though no major classifications are found in the literature for European uOAs.
• The accessibility to uOAs considers the width of entrances, the urban mobility features, and the presence of physical elements along the entrances, described as K_ACC. In detail, it should consider the width (A_vi) and the accessibility level (f_ACC) of the i-entrance, properly assessed according to the attack types. Specifically, for T2 attack types, all the entrances can be considered always accessible due to the inherent significance of the entrance, while for T3 attacks, where perpetrators move in vehicles, ease of access is related to the possible levels of car accessibility, defined by urban regulation (e.g., traffic-restricted zone, hourly accessibility, …) or geometric restrictions.

Obstacle Index [V_I.2]

Finally, the third element assessed for the Vulnerability of uOAs is described by obstacles [V_I.3]. In detail, the attention is on the physical objects of uOAs that may influence meeting and temporal attractiveness in specific sub-areas. Sights, urban furniture, bar-covered terraces, geomorphological or physical discontinuities (i.e., stairs), and gardens may be considered. Their relevance is not related to the protection level but to the generated attractiveness, which may locally alter the general conditions of risk. For these purposes, all elements are assessed in terms of the number of obstacles, their extension, and the factor of influence as meeting points [75]. Specifically, the obstacle factor is determined considering the ratio between the extension of the i-th type of obstacle and the total surface area of obstacles in the uOA (d_i). This ratio is then multiplied by the associated relevance of each obstacle in terms of attractiveness influence (f_INF), which is described as an increasing factor, an average increasing factor, or not influential (Table 2).

3.2.3. Descriptors Affecting the Exposure Indexes

Considering the discussion of the last risk determinant, the Exposure to terrorism threat in this study is centered on victims, excluding building damages, in line with the data available from events recorded in the GTD database [56]. The qualification of elements that affect exposure in the risk assessment is carried out focusing on the properties that may affect the events before and during the event, overlooking the behavioral response of the people involved and the emergency phase (post-traumatic event). According to the summary in Table 1 and in line with the results discussed in previous studies [12,18], the number of people involved (including victims and the injured) may be considered as a consequence of (i) the attack type as the phenomenology has underlined, (ii) the number of bystanders in the place, and (iii) the presence of protective elements. Besides these elements, three descriptors have been identified and quantified, translating the reference indexes in Table 1.

Attack Index [E_I.1]

When the discussion is centered on the attack type [E_I.1], the focus is on the potential level of people in relation to recurrent boundary conditions. As it is usually reflected by risk management, Exposure includes the level of damage resulting from the phenomenological analysis of previous events. For this study, the relevance of attack typologies in the determinant is related to the phenomenological analyses conducted by the authors. Specifically, in line with the descriptor of the environmental factor, the K_ATT (attack factor) is introduced to describe the effect of the weapon type. Specifically, for each attack type, K_ATT relates the statistical relevance of the impact for each combination of environmental classes (square F, SoR of bar or special buildings F_B and F_D respectively) and attack types, with the specific details on those discussed (T2 and T3). The quantification assigns a score (4 or 5) based on the consequence levels for the selected combination in [18].

Crowd Index [E_I.2]

As far as the possible effects are concerned (E_I.2), the Exposure class is also determined by the potential number of people that could be involved in the attack. The crowding index K_CRW describes the possible impact related to the density of people in the open area or outside the public activities facing the open area, and it is properly assessed according to the technicians’ experience, in line with the qualitative ranges (see Table 2).

Index of the Attack Reaction [E_I.3]

Finally, “hide” and “run” represent two of the most recommended actions, consistent with several guidelines and norms regarding preventive actions at the national level [13,75]. The detail is on the effect of “obstacles” or “objects” that can participate in violent actions generating protective areas (positive effect) [76] or obstructing the possible evacuation (negative effect) [77]. In the first case, as reported in major national experiences, the physical object with high resistance and mass can be considered a “protective zone” for both T2 and T3 attacks. On the other hand, it was demonstrated that a system of urban furniture or objects can alter the flux, generating real obstacles during evacuation. It includes elements such as poles with chains, railings, benches, and planters, which are predominantly characterized by horizontal development. Single elements, such as trees, monuments, and bollards, may alter the flux [78]. Thus, the attack reaction [E_I.3] is focused on the potential impact on users’ reaction of the physical elements included in the uOAs from a static point of view. Instead, the impact of violent acts should consider the presence of countermeasures specifically fitted for the attack types in terms of users’ involvement, according to the results presented by Quagliarini et al. [12]. Here, the presence of coordination and evacuation plans, alarm systems, and suitable physical elements in supporting evacuation (evacuation paths that are narrow, with lighting to ensure a secure emergency exit) influence the emergency phase, limiting the final number of people involved. Thus, two main descriptors are introduced for the assessment of the related index (i_REA):

• K_OBST(E), which describes the influence of «obstacles» and «objects» on the identified aims. Specifically, K_OBST(E) has to consider their total extent in relation to that of the uOA (d), their shape or the final shape resulting from the replicability of individual elements in the spaces (vertical, horizontal, compact development) (f_SHPob) and, finally, their influence on protection or evacuation (f_inf). Thus, K_OST(E) should consider all the obstacles in the uOAs and determine a mean value for all the observed objects. Specifically,

  Table 3. Summary of the influence determined by obstacle typology on classes of shape and prevalent development.

  Icon	Examples of Shape and Prevalent Development of Obstacles	Influence of Obstacle’s Shape
  	Poles and trees— vertical development	Negligible
  	Monuments— vertical development	Low
  	Bar covered terraces— Compact development	Average
  	Benches, planters, new jersey— horizontal development	High
  	Railings, steps— horizontal development	Very high

  shows the selected influence of obstacles’ shape in the process, consistent with the state of the art [12].
• K_CM describes the positive effect on the number of people involved due to the presence of countermeasures. In this case, the factor considers strategies that may influence the preparedness for emergency activities in terms of alarms and evacuation countermeasures (both for T2 and T3). Due to that, K_CM has to consider the number of classes of countermeasures in the emergency phase present in the uOA (Wi) and the effective ones discussed in Quagliarini et al. (W_EFF).

3.2.4. Final Remark for Assessing K-Factors and Indexes

Due to the complexity of qualifying all the elements, properties, and characteristics of places, some specifications were required, and they are discussed below.

The formulations concern both uOA and SoR for all the public and symbolic or strategic buildings facing the actual open areas. However, some K-factors require to be linked to one of them in accordance with some main principles:

• Due to the inherent concept assumed in the study of uOA as a system of buildings, infrastructures, and open areas, K_TUR, K_CON, K_SHP, K_PER, K_ACC, K_CM are calculated considering the overall uOA, even when describing a SoR. This is due to the overlapping classes of qualities described by the K-factors, which are usually pre-determined by the historical evolution (e.g., the morphology of the city and district) or external strengths (e.g., tourism, local norms).
• The assessment of the SoR explicates its physical qualities when the K-factors describe properties related to their use or function (K_env, K_symb, K_USE, K_ATT, K_CRW) or link properties and qualities to their position within the overall uOA. This is the case of obstacles that are discussed in terms of efficacy, shape, or influence when they are included in the perimeter of the SoR. Thus, all the SoR graphical details, starting from their perimeter, are necessary to support the formulation.

3.3. The Mathematical Structure of the Risk Determinants

After selecting and identifying factors that influence terrorist threats in uOAs, the nature of the indexes and descriptors varies in terms of the qualitative and quantitative discussions due to the complexity of the matter. However, some common points can be recognized in order to solve the qualitative–quantitative assessment. In detail, the following points should be considered:

• Considering the relevance of the matter, some parameters cannot be related to specific ranges of values (K_SHP, K_PER). Others are related to classes of features that qualitatively describe their relevance (e.g., low, high) (K_ENV, K_SYM K_TUR, K_USE, K_ACC, K_ATT, K_CRW, K_OBST(E)). Due to that, all the descriptors are categorized into five classes of possible quantitative values, considering a range between 1 and 5, each associated with five possible qualitative classes.
• When the descriptors are related to factors that constitute corrective properties, ranges of factors are introduced in three classes in order to address the absence of influential and non-influential factors, as well as the influence of corrective factors (K_SHP, K_OBST(E)).
• All the descriptors related to the same index are considered independent and are combined for the index calculation as a product.
• Due to the large variability of the results (in terms of maximum and minimum values), all the indexes are normalized in five ranges, considering the associated final values in terms of class, from 1 to 5. This structure supports the limitation of results and their control in the overall process.
• All the values are conceived in order to exclude the zero value. This is fundamental to solving the undetermined ratios and excluding the zero value for the final risk triad (the minimum value of risk is 1).

As far as the final formulation is concerned, and consistent with the general rules, the terrorist risk for uOAs is structured as the product of three determinants. Each is calculated as the sum of all the indexes properly weighted (w), and it is evaluated for each attack type (T2 and T3), following Equation (3) and the detailed equations (Equations (4)–(6)).

R_{(T-type(F.Fb,Fd))} = H_{(T-type(F.Fb,Fd))} × V_{(T-type(F.Fb,Fd))} × E_{(T-type(F.Fb,Fd))}

(3)

H_{T-type(F.Fb,Fd)} = ((i_TRG × w_TRG) + (i_Use × w_Use) + (i_PREV × w_PREV))/w_Tot

(4)

V_{T-type(F.Fb,Fd)} = ((i_SHP × w_SHP) + (i_ACC × w_ACC) + (i_OBST × w_OBST))/w_Tot

(5)

E_{T-type(F.Fb,Fd)} = ((i_ATT × w_ATT) + (i_CRW × w_CRW) + (i_REA × w_REA))/w_Tot

(6)

Moreover, all of them are calculated for actual open areas (square or street F) and each SoR (public or special buildings, F_b and F_d, respectively). This is due to the conceptualization of uOAs as a system of buildings, infrastructures, and unbuilt areas, where the interactions among the building/infrastructure uses and the uses of squares/streets are part of the SoRs (see Section 3.1) within the same uOAs distribution. In that sense, the conceptual extension of uOAs (A_tot) has to consider the extent of all the SoRs, as well as that of the square/street (Equation (8)). Thus, the values of the determinants are calculated for both the square and each SoR (R_F = H_F × V_F × E_F; R_Fdi = H_Fdi × V_Fdi × E_Fdi; R_Fbi = H_Fbi × V_Fbi × E_Fbi), considering the ratio between single SoRs and A_tot. Moreover, for each attack type, a final triplet and risk value (R_uOAT-type) are introduced as the values of three determinants that need to be appropriately weighted to the area extension of F, SoR_FB, and SoR_FD. More specifically,

$$\begin{array}{c}{R}_{uOA\left({\displaystyle \raisebox{1ex}{$T2$}\!\left/ \!\raisebox{-1ex}{$T3$}\right.}\right)}=H_{uOA}\times V_{uOA}\times E_{uOA}=\\ \left[\right({\displaystyle \frac{H_F}{{\raisebox{1ex}{$A_F$}\!\left/ \!\raisebox{-1ex}{$A$}\right.}_{tot}}})+{\sum }_{i=1}^n{\displaystyle \frac{H_{Fbi}}{\raisebox{1ex}{$A_{Fbi}$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}}+{\sum }_{i=1}^m{\displaystyle \frac{H_{Fbi}}{\raisebox{1ex}{$A_{Fbi}$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}}]\times \left[\right({\displaystyle \frac{V_F}{\raisebox{1ex}{$A_F$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}})+{\sum }_{i=1}^n{\displaystyle \frac{V_{Fbi}}{\raisebox{1ex}{$A_{Fbi}$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}}+\\ {\sum }_{i=1}^m{\displaystyle \frac{V_{Fbi}}{\raisebox{1ex}{$A_{Fbi}$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}}]\times [\left({\displaystyle \frac{E_F}{\raisebox{1ex}{$A_F$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}}\right)+{\sum }_{i=1}^n{\displaystyle \frac{E_{Fbi}}{\raisebox{1ex}{$A_{Fbi}$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}}+{\sum }_{i=1}^m{\displaystyle \frac{E_{Fbi}}{\raisebox{1ex}{$A_{Fbi}$}\!\left/ \!\raisebox{-1ex}{$A_{tot}$}\right.}}]\end{array}$$

(7)

$$\mathrm{where} A_{tot}=A_F+{\sum }_{i=1}^n{A}_{Fbi}+{\sum }_{i=1}^m{A}_{Fdi}$$

(8)

Table 2 summarizes the K-parameters, formulations, and details used for setting up the formulation, applicable to the AHP and Delhi participatory methods. Specifically, detailed references to the literature and normative are reported when the classes or values are already present; in all the other cases, the formulations are derived; in this latest case, they are reorganized for work purposes, reporting them in the participatory validation process (Section 4).

4. Influence Factors and Weighting with Participatory Methods

The indexes and descriptors are submitted to a pool of participants which includes the following:

• Eight master’s degree students involved in studying the terroristic phenomenon in Europe.
• Seven European academics with experience in resilient and secure cities.
• Two experts in participatory methods applied to architectural built environment issues (static failure, preventive maintenance).
• Four public policymakers involved in the management of security for big events.

The participants were engaged online to explain the main goal of the questionnaire. However, they were equipped with a set of general information related to the goal of setting up the algorithm before answering the questionnaire.

4.1. Validation of Influencing Factors and Formulation

In accordance with the described method, three levels of detail are required to the participants (

Table 4. Results of the Delphi method (✓ yes, x not).

	First Round		First Round	First Round CRV	NOTE	Second Round CRV
	Consensus: the Index is Compliant with the Risk Determinant	K-Type	Consensus: the Index is Compliant with the Associated Index	Formulation and Ranges Are Acceptable?		Formulation and Ranges Are Acceptable
Target index	✓	KENV	✓	0.7143
		KSYMB	✓	1.0000
Index of use	✓	KTUR	✓	0.5714
		KUSE	✓	0.8571
Prev. index	✓	KCON	✓	1.0000
Shape index	✓	KSHP	✓	0.8571	✓	0.8571
Accessibility index	✓	KPER	✓	1.0000
		KACC	✓	0.7143
Obstacle index	✓	KOBST(V)	✓	0.7143
Index of attack type	✓	KATT	✓	1.0000
Crowding index	✓	KCRD	✓	1.0000	✓	1.0000
Index of reaction	✓	KOBST(E)	✓	0.7143
		KCM	✓	1.0000

).

• The first and second levels ensure that the indexes and K-types comply with the selected risk determinant and the associated index, respectively. In these phases, the Delphi method follows the “Consensus” goal.
• The third level aims at the acceptability of formulation and ranges for each K-type. Acceptability is measured by “Yes” or “Not” answers, and a field for comments is included.

The results of the first and second rounds of consensus met a higher level of acceptability in the first round of the Delphi surveys, both in assigning indexes to the risk determinants and in the association of K-descriptors to indexes.

The third level was addressed in two rounds due to recurrent notes about the ranges. Specifically, K_CRD, despite reaching an acceptability rate in the round (CRV > 0.5), saw more than half of the participants propose the use and association of standard ranges of values with the qualitative classification, highlighting the opportunity to resolve the issue using normative details. A new classification was determined for the selected K-type, including the opportunity to use national ranges to manage crowding events for K_CRD.

4.2. Quantification of Factor Relations in the Determinant Calculation

In accordance with the AHP procedures, the participants were invited to determine the final weight of indexes. The choice of the weight of indexes instead of K-parameters results from the necessity to provide an easier way to solve the algorithm and, above all, to understand the major results of the participatory approach. Moreover, in accordance with Equations (2)–(4), the weights are assessed considering single determinants, thus comparing three indexes for time. This is due to the necessity to consider independent Hazard, Vulnerability, and Exposure determinants in alignment with the goals of the work. The final weights for each index and the associated CR values are summarized in

Table 5. Results of the AHP method applied to the K-descriptors.

Index Name	Weight	CR (%)
Hazard
Target index	0.6	0.6%
Index of use	0.09
Prev. index	0.31
Vulnerability
Shape index	0.24	6.9%
Accessibility index	0.65
Obstacle index	0.12
Exposure
Index of attack type	0.2	3.2%
Crowding index	0.65
Index of reaction	0.17

.

Regarding the comparison of participatory methods, consistent opinions can be highlighted at the end of the procedure, supporting the coherence and the structure of the applied method:

• The concurrence between the higher values of CRV parameters for K-types and the weight of the associated index. This is the case of the target and accessibility indexes, where both the CRV values associated with K-parameters showed higher values.
• The relevance of major details about the crowding index in determining the most coherent formulation and the associated final weight of the index.

In both cases, the participants demonstrated a clear and coherent perception of key elements and features to be considered. Similarly, a good alignment with the review research conducted as the foundation of the work can be highlighted.

5. Numerical and Theoretical Test and Validation of the Algorithm

The last phase of the work involves algorithm testing to validate the participatory results and the suitability of the elements parametrized, thus testing them.

In order to exploit the calculation tool in accordance with national details, the algorithm was processed for 22 Italian cases, providing different levels of peculiarities, both in symbolicity and time of use.

Table 6. Major details about tested case studies in Italy.

Italian Case	Touristic Relevance	Symbolicity	Presence of Strategic Buildings	Principal Symbolic Buildings	Presence of Mitigative Strategy
Milano— Piazza Duomo	High, independent of season and time of day	High— Political and economic		Duomo, Galleria	yes
Napoli— Piazza del Plebiscito	High, independent of season and time of day	High— Cultural		Basilica Pontificia, Palazzo Reale	yes
Roma— Piazza San Pietro	High, independent of season and time of day	High— Religious and political		Basilica di San Pietro	yes
Venezia— Piazza San Marco	High, independent of season and time of day	High— Cultural and economic		Basilica di San Marco, Palazzo Ducale	yes
Corato (BA)— Piazza Sedile	Low, citizen uses	Low		minor churches	no
Matera— Piazza Vittorio Veneto	High, “Matera Capitale della Cultura” and “Sassi” UNESCO Site	Medium—touristic		Balcony on the “Sassi”	no
Ostuni (BR)— Piazza della Libertà	Seasonal and mainly nocturnal	Medium—touristic			no
Trani (BAT)—Piazza Duomo, Piazza Re Manfredi	Medium, presence of cultural attraction	High— Touristic and strategic	Courthouse of the province	Castello Svevo, Representative church of the Romanic style	yes
Narni (TR)—Piazza dei Priori	Medium high, independent of season	Medium—cultural	City hall	Palace “dei Priori”	no
Caldarola (MC)— Piazza Vittorio Emanuele II	Very low usage by citizens	Low	City hall	Two churches	no
Catania— Piazza Università	Medium, presence of cultural attraction and university attractiveness	Medium—touristic	University of Catania	“Machiavelli” Theatre	no
Genova— Piazza delle Vigne	Medium, presence of cultural attraction and university attractiveness	Medium—touristic		Basilica di Santa Maria delle Vigne	no
Parma— Piazza del Duomo	Medium, presence of cultural attractions	Medium—touristic		Basilica Cathedral, Baptistery, Bishop’s palace	no
Monza— Piazza Trento e Trieste	Medium, business activities	High—political and economic			no
Perugia— Piazza IV Novembre	Medium, presence of cultural attractions	Medium—touristic	Curia of Bishop		no
Pavia— Piazza del Duomo	Medium, presence of cultural and university attractiveness	Medium—touristic		Palazzo Vescovile; Cattedrale di Santo Stefano	no
Padova— Piazza delle Erbe	Medium, presence of cultural attractions	Medium—touristic		Palazzo della Ragione	no
Reggio Calabria— Piazza Duomo	Medium, presence of cultural attractions	Medium—touristic		Duomo	no
Cagliari— Piazza Palazzo	Medium, presence of cultural attractions	Medium—touristic and strategic	Prefettura; Ecclesiastic Courthouse	Palazzo Regio, Cathedral, Ancient City Hall	yes, but related to strategic buildings
L’Aquila— Piazza Duomo	Medium low	Medium—touristic		Palazzo Poste e Telegrafi, Duomo	no
Ancona— Piazza del Plebiscito	Medium, presence of cultural attractions	Medium—touristic and strategic	Prefecture; Headquarter regional Finance police	San Domenico Church	yes, and related to strategic buildings
San Gemini (TR)— Piazza San Francesco	Very low, usage by citizens	Low		San Francesco Church	no

summarizes the case studies in accordance with the main conditions and the elements for discussion. More details about F_B and F_D and SoRs extensions are detailed in Appendix A.

Data gathering involved using Google Maps details (through Google Earth Pro v.7.3.6.9796 using images ©2023Google, Airbus Maxer Technologies) in order to provide compatible and homogeneous data. Daily uses and opening times were also gathered using Google Maps details. The use of these details helps in overcoming the regional variability of the data available in vectorial maps (such as GIS data), which may determine local variability when considering the physical objects in the case studies. Finally, due to the variety of physical elements within the analyzed OAs, two main conditions were considered for the algorithm testing: with (S) and without mitigative (NS) strategies. The higher political and religious relevance of some case studies (Milano, Roma, Venezia) has already moved policymakers to introduce physical solutions aiming at increasing the local security of the places. As part of this study, the algorithm was also tested for the assessment of strategies by means of the global risk calculation, in accordance with Equation (7).

Table 7. Summary of results of applying the algorithm in all the analyzed Italian case studies, expressed in triads of values (H, V, E) for each environmental class (F, Fb, Fd).

City		Environ. Class		T2				T3
		Type	%Area	H	V	E	R	H	V	E	R
Milano	NoStr.	F	71%	3.55	2.84	3.55		3.55	2.84	3.55
		Fb	13%	0.52	0.53	0.52		0.52	0.53	0.52
		Fd	16%	0.78	0.63	0.70		0.78	0.63	0.70
		mean		5	4	5	100	5	4	5	100
	WithStr.	F	71%	3.55	2.84	3.55		3.55	1.42	3.55
		Fb	13%	0.40	0.53	0.52		0.38	0.27	0.52
		Fd	16%	0.66	0.63	0.70		0.70	0.31	0.70
		mean		5	4	5	100	5	2	5	50
Roma	NoStr.	F	71%	3.55	2.84	3.55		3.55	2.84	3.55
		Fb	1%	0.03	0.05	0.06		0.03	0.04	0.06
		Fd	28%	1.39	1.11	1.39		1.39	0.84	1.39
		mean		5	4	5	100	5	4	5	100
	WithStr.	F	71%	2.84	2.13	3.55		3.55	1.42	3.55
		Fb	1%	0.03	0.02	0.05		0.03	0.01	0.05
		Fd	28%	1.11	0.84	1.39		1.39	0.56	1.39
		mean		4	3	5	60	5	2	5	50
	Napoli	F	65%	2.58	1.94	3.23		2.58	1.94	3.23
		Fb	4%	0.10	0.14	0.16		0.11	0.09	0.15
		Fd	31%	1.28	0.94	1.19		1.32	0.94	1.19
		mean		4	3	5	60	4	3	5	60
Venezia	NoStr.	F	71%	3.54	2.83	3.54		-	-	-	-
		Fb	17%	0.49	0.44	0.46		-	-	-	-
		Fd	12%	0.78	0.57	0.74		-	-	-	-
		mean		5	4	5	100	-	-	-	-
	WithStr.	F	71%	3.54	2.83	3.54		-	-	-	-
		Fb	17%	0.37	0.44	0.46		-	-	-	-
		Fd	12%	0.61	0.57	0.74		-	-	-	-
		mean		5	4	5	100	-	-	-	-
	Matera	F	70%	2.78	2.78	3.48		2.78	2.78	2.78
		Fb	27%	0.75	0.93	0.85		0.78	0.85	0.85
		Fd	4%	0.11	0.15	0.08		0.11	0.11	0.08
		total		4	4	4	64	4	4	4	64
Ostuni (BR)	Summer	F	77%	2.31	3.08	3.85		2.31	3.08	3.85
		Fb	23%	0.69	0.91	0.86		0.69	0.87	0.86
		Fd	0%	0.00	0.00	0.00		0.00	0.00	0.00
		mean		3	4	5	60	3	4	5	60
	Winter	F	77%	2.31	2.31	3.08		2.31	2.31	3.08
		Fb	23%	0.69	0.73	0.73		0.69	0.69	0.73
		Fd	0%	0.00	0.00	0.00		0.00	0.00	0.00
		mean		3	3	4	36	3	3	4	36
	Trani	F	73%	2.19	2.19	2.92		2.19	2.19	2.92
		Fb	7%	0.22	0.21	0.23		0.22	0.22	0.23
		Fd	20%	0.70	0.51	0.59		0.70	0.59	0.59
		mean		3	3	4	36	3	3	4	36
	Corato (BA)	F	85%	1.70	3.40	1.70		1.70	2.55	1.70
		Fb	14%	0.28	0.58	0.28		0.28	0.42	0.48
		Fd	1%	0.02	0.04	0.02		0.02	0.03	0.02
		mean		2.00	4.00	2.00	16	2.00	3.00	2.00	12
	Narni (TR)	F	67%	2.69	2.69	2.69		2.69	2.69	2.69
		Fb	33%	1.23	1.23	1.23		1.23	1.23	1.23
		Fd	0%	-	-	-		-	-	-
		mean		4.00	4.00	3.00	48	4.00	3.00	3.00	36
	Caldarola (MC)	F	76%	1.52	2.28	1.52		1.52	3.04	1.52
		Fb	5%	0.15	0.20	0.10		0.15	0.21	0.10
		Fd	19%	0.37	0.67	0.37		0.37	0.67	0.37
		mean		2.00	3.00	2.00	12	2.00	4.00	2.00	16
	Catania	F	59%	1.77	2.37	2.37		1.77	1.77	2.37
		Fb	41%	1.20	1.24	1.53		1.20	1.22	1.53
		Fd	0%	-	-	-		-	-	-
		mean		3.00	4.00	4.00	48	3.00	3.00	4.00	36
	Genova	F	36%	1.08	1.08	1.08		0.72	0.72	1.08
		Fb	61%	1.69	1.82	1.70		1.21	1.21	1.70
		Fd	3%	0.07	0.10	0.07		0.07	0.07	0.07
		mean		3.00	3.00	3.00	27	2.00	2.00	3.00	12
	Parma	F	59%	1.78	2.96	1.78		1.78	1.18	1.78
		Fb	4%	0.12	0.17	0.09		0.09	0.09	0.09
		Fd	36%	1.09	1.79	1.01		1.09	1.06	1.01
		mean		3.00	5.00	3.00	45	3.00	2.00	3.00	18
	Monza	F	64%	1.93	2.58	1.93		1.93	2.58	1.93
		Fb	34%	1.05	1.40	1.02		1.05	1.40	1.02
		Fd	2%	0.01	0.02	0.01		0.01	0.02	0.01
		mean		3.00	4.00	3.00	36	3.00	4.00	3.00	36
	Perugia	F	68%	2.71	2.71	2.04		2.71	2.04	2.04
		Fb	13%	0.46	0.46	0.29		0.46	0.39	0.29
		Fd	19%	0.76	0.76	0.57		0.76	0.57	0.57
		mean		4.00	4.00	3.00	48	4.00	3.00	3.00	36
	Pavia	F	42%	1.67	1.67	1.25		1.67	1.25	1.25
		Fb	7%	0.24	0.26	0.18		0.24	0.21	0.18
		Fd	51%	2.06	2.06	1.54		2.06	1.54	1.54
		mean		4.00	4.00	3.00	48	4.00	3.00	3.00	36
	Padova	F	66%	2.64	1.98	1.98		1.98	1.98	1.98
		Fb	34%	1.12	1.02	1.00		0.90	1.02	1.00
		Fd	0%	-	-	-		-	-	-
		mean		4.00	3.00	3.00	36	3.00	3.00	3.00	27
	Reggio Calabria	F	75%	2.25	2.25	2.25		2.25	2.25	2.25
		Fb	13%	0.39	0.36	0.36		0.39	0.32	0.39
		Fd	12%	0.36	0.48	0.36		0.36	0.48	0.48
		mean		3.00	3.00	3.00	27	3.00	3.00	3.00	27
	Cagliari	F	79%	2.37	2.37	2.37		1.58	1.58	2.37
		Fb	17%	0.51	0.47	0.47		0.37	0.34	0.47
		Fd	4%	0.12	0.12	0.08		0.10	0.08	0.08
		mean		3.00	3.00	3.00	27	2.00	2.00	3.00	12
	L’Aquila	F	87%	2.62	2.62	2.62		2.62	2.62	2.62
		Fb	5%	0.16	0.11	0.16		0.16	0.15	0.13
		Fd	7%	0.22	0.29	0.22		0.22	0.29	0.22
		mean		3.00	3.00	3.00	27	3.00	3.00	3.00	27
	Ancona	F	66%	2.63	1.97	2.63		1.97	1.97	2.63
		Fb	9%	0.21	0.28	0.28		0.85	0.61	0.78
		Fd	25%	0.85	0.75	0.72		0.75	0.50	0.60
		mean		4.00	3.00	4.00	48	3.00	3.00	3.00	27
	San Gemini (TR)	F	50%	1.00	1.50	1.00		1.00	2.00	1.00
		Fb	17%	0.36	0.54	0.34		0.36	0.69	0.34
		Fd	33%	0.65	1.25	0.65		0.65	1.25	0.65
		mean		2.00	3.00	2.00	12	2.00	4.00	2.00	16

summarizes the results of all the cases, highlighting the distribution of risk determinants for the environmental classes involved (F, Fb, and Fd), in accordance with their extensions, for both attack types (T2—armed assault, T3—car bombing attack). The mathematical application across several case studies has not established limitations, it solves all the queries, and shows the expected results for most representative and symbolic Italian case studies. This is due to the relevance of both Hazard and Exposure in relation to the places, resulting from the weighted average of SoRs and square extensions. However, additional observations can be noted:

• Considering the variation in risk values for the S and NS cases, the algorithm provided sufficient variations for assessing the presence of physical mitigative or protective strategies. However, in compliance with the gathered data, all the elements aim to reduce the Vulnerability of a place to external attacks (car bombing in movement). In that sense, the reduction affects the Vulnerability values (see the case of Milano and Roma) as a consequence of the physical reduction in the accessibility of openings.
• When provincial case studies are assessed (Trani, Ostuni, Narni), the algorithm returns medium values for the actual state of the places. This is due to the inherent critical features (morphology, accessibility, lower level of protection) rather than the symbology or attractiveness of the places.
• Minor case studies, such as Corato, San Gemini, and Caldarola, reflect minor risk values, as a consequence of the combination of lower levels of attractiveness and symbology of places, combined with variable values of Vulnerability and Exposure.

In addition to the mathematical significance reached, these results also align with the main principle of the terrorism phenomenon in classifying hard and soft targets, as introduced by Gordon Woo [4]:

• Protective strategies are independent of the Hazard of places when describing hard targets. Considering the most relevant case studies (Milano, Roma, Napoli, Venezia), the algorithm describes them as hard targets where the violent acts require operative resourcefulness both in planning and executing; in fact, despite the higher symbolicity of places and the high level of protection achieved in the squares, Hazard still maintains superior values, in the range [4 or 5]. This is in accordance with the weights assigned to the target and the protection indexes (0.60 and 0.31, respectively) in the Hazard assessment. On the other hand, the strategies reflect positive effects on reducing the overall risk values, having effects on Vulnerability as the main descriptor of the inherent physical criticalities of places.
• Protective strategies affect the Hazard value of places when the soft targets are assessed. Consistent with the definition of “soft target”, the occurrence of events may increase due to the lower level of protection of places, allowing for their replicability. In that sense, the weights assigned to the target and protection indexes are well calibrated, particularly in describing soft targets. This allows for the description of the strategies’ effect on reducing both the Hazard and inherent Vulnerability of places.

6. Discussion of Results from the Risk Assessment to Its Control and Management

In accordance with the results discussed in the previous section, the 5-point scales of the Hazard, Vulnerability, and Exposure determinants are translated into qualitative meaning in order to support coherent and intelligible discussions of numerical results.

Table 8. Qualitative description of Hazard, Vulnerability, and Exposure determinants.

Hazard	Vulnerability	Exposure
1–2 unlikely	1–2 low	1–2 minor
3 probably	3 medium	3 moderate
4–5 likely	4–5 high	4–5 critical

summarizes the classes and the qualitative meaning assigned.

These classes of significance are thus merged with previous results in reading soft and hard targets, as well as with the terroristic principles in terms of result maximization, and three main recurrences can be enunciated:

• Hard targets belong to the higher class of Hazard (likely), exploiting the higher value of damage (E critical).
• Soft targets fit within all the other classes of Hazard (unlikely and probably), varying for all the other determinant classes.
• Due to the lower significance of places (H unlikely) and the lower potential effects (E minor), soft targets do not require any specific measures regarding terroristic attack risks. They are assumed to be negligible due to the lower attractiveness and efficacy of violent events, assuming that Vulnerability is an independent factor.

Due to these results, the risk management of the terrorist phenomenon can be solved by introducing and discussing three classes of risk typologies of real places as the combination of Hazard and Exposure. Moreover, considering Hazard as the determinant to describe a place’s proneness to violent acts, Vulnerability and Exposure are discussed in terms of the degree of danger, combining the potency of damage and the intrinsic vulnerabilities of places. These are thus combined to determine equivalent classes of actions to support risk reduction in real places.

Table 9. Classes of risk associated with specific combinations of H and level of danger, discussed for hard and soft targets, associating red, orange, green, and grey colors to represent high, medium, low, and negligible classes of risk.

H Values	Degree of Danger	Class of Risk
Soft target
H [1, 2] ∧ E [1, 2]	all the combinations	Negligible
V [1, 5]
H [4, 5]	VxE = [1, 9]	Medium
V [1, 5]; E [1, 3]	VxE = [9, 15]	High
H [1, 2] V [1, 5]; E [3, 5]	VxE = [3, 6]	Low
	VxE = [6, 15]	Medium
	VxE = [15, 25]	High
H [3] V [1, 5]; E [1, 5]	VxE = [1, 4] VxE = [4, 12] VxE = [12, 25]	Low
		Medium
		High
Hard target
H [4, 5] ∧ E [4, 5]	VxE = [4, 10]	Medium
V [1, 5]	VxE = [10, 25]	High

sums up the results in degrees of danger and classes of risk, combining the typologies of soft/hard targets. Figure 3 shows the expanded matrices for four combinations of the same classes of targets.

In more detail, besides the operative benefit of understanding the proneness to a single risk of uOAs, some possible goals can be reached:

• The comprehension of the possible critical points of the place can support the analysis of behavior during the emergency, matching the geometric and physical properties of the place to the users’ movements, while also assessing mitigative strategies introduced for the reduction of pre-emergency phase, toward a more comprehensive behavioral–physical-based approach for emergencies [79,80].
• Considering an asynchronous multi-risk perspective, recognizing critical points vulnerable to harm in uOAs, and designing and assessing physical transformations of the place to reduce susceptibility to terrorism offer the opportunity to evaluate the resilience of such strategies in other risk occurrences. The inherent double relationship between the protective and obstructing features of the physical objects in the area may affect emergency conditions for other sudden threats (e.g., earthquakes). In that sense, the design of mitigative and protective solutions may consider all the possible hazards in uOAs, taking advantage of behavior analysis for emergency planning [64,81].
• Considering a synchronous multi-risk perspective, the identification of critical points vulnerable to harm can be correlated with the local distribution of people when external pressures occur, such as heatwaves. Here, the slow nature of natural events may affect the local re-distribution of users within the uOAs, altering the local crowd density and shifting potential point of attacks, as well as affecting user behavior [82].

All these advantages are strictly related to the qualitative approach at the basis of the formulation, which became fundamental in describing a human-based phenomenon. At the same time, some limitations can be expressed.

First of all, it is necessary to test the results related to previous events in order to calibrate the quality of the data. This includes discussing the effects of an event before and after the application of mitigative strategies. However, this can be achieved by including a more comprehensive structure of results that includes the users’ movements. On the other hand, this requires that the location has experienced the same type of attack in the same place, which constitutes a significant limitation.

Another limitation of the analysis is related to the independence of the human dimension of the perpetrators. The proposed analysis is centered on the built environment aligning with the aim of the work and addressing the perpetrator’s actions during a violent act may change the final riskiness of the place.

The setup of the discussed algorithm for squares concerns a specific matter, compared to the general phenomenon. This is in line with most of the previous studies found in the literature that tried to solve specific problems in cities exposed to these violent acts. In accordance with European regulations [36,37,38], all these fields of application require direction toward a common method to first measure the risk and then assess the overall resilience level of cities. The application of a standardized method to assess terrorist risk and the use of a set of parameters to describe and evaluate elements in the scenarios may allow for the complexity of resolving the overall general threat, combining new parameters and re-weighting them within a broader collaborative effort.

7. Conclusions

In this work, European theories on the terrorism phenomenon, normative experiences, and phenomenological results about violent acts in urban open and public places (uOAs) are combined to determine a simplified risk matrix for qualifying this urban built environment prone to terrorism. The matrix of risk, properly defined to describe uAOs as either soft or hard targets, resulted from the assessment of the elements (objects and obstacles, buildings, accesses) and properties (functions, geometry, symbolism) that usually feature the real urban built environment, in accordance with city development and urban relevance (political, religious, …). Such properties, qualities, and deficiencies, both quantitative and qualitative, are translated into descriptors (K-descriptors) by merging scientific and normative experiences to provide a limited number of factors for the study of the real uOAs. The nine K-descriptors, properly identified in formulation and meaning, were discussed for compliance with the three determinants of risk (Hazard, Vulnerability, and Exposure) and shared with the external judges through a participatory method. All the experts involved in the AHP process were coached in the thematic study, presenting previous findings, i.e., the relevance of uOA within the European phenomenon, and the classes of mitigative and preventive strategies. Specifically, the coherence with the determinants and formulations is first assessed using the Delphi method. Then, the K-determinants were evaluated using an AHP process to determine the main relevant ones. The main result of the participatory process is the higher relevance of three main properties: the symbolism of place and its level of protection (target index), which affect the Hazard dimension of the risk, the level of accessibility (accessibility index) for the Vulnerability of the place, and the crowding level for the standard use of the space for the Exposure dimension (crowding index).

The testing phase of the mathematical algorithm on a wider set of case studies in Italy supported the validation of the tool, as well as its potentialities, both in testing possible mitigation strategies and in classifying real case studies as soft and hard targets. In fact, the determination of a formulation as a triad of determinant values allows us to control their variation as single factors and thus to understand the possible effectiveness of space transformation in terms of reduction in inherent vulnerabilities, increasing protection to reduce the likelihood of events, or increasing the users’ protection when exposed to the threat.

Finally, the use of SoRs to translate the interferences and inter-relationships between square/street—as physical open areas—and the function of buildings facing these open and public places offer the opportunity to study the actual urban open area as a system of infrastructures, buildings, and uses. This is compliant with the recent studies on the phenomenon in Europe, where the proneness of public places and public buildings to violent actions influences perpetrator choice. In that sense, the physical delimitation of SoR/s within the uOA and their qualification in terms of risk can support the identification of major critical parts of uOAs where violent actions can be perpetrated. This provides the setup for coherent and efficient evacuation plans, going beyond the assessment of the pre-emergency phase toward the resolution of the emergency one, in accordance with the BE S²ECURe project of which this study is part.