Climate Change Impact on Cultural Heritage: A Methodological Framework for Driven Reasoning in Risk Assessment

1. Introduction

Cultural Heritage (CH) is a vector of social and economic improvement. The composite ecosystem of their tangible and intangible components, made up of rituals and practices along with hand-craft products, buildings and natural elements, offers the opportunity to develop innovative forms of rural, creative and slow tourism experiences, pushing toward a positive job creation trend. However, the current Climate Change (CC) process puts at risk the fragile equilibrium that interlinks natural and human environments in CH. Long-term changes in weather patterns and temperatures, known as CC, are considered one of the most serious threats of the twenty-first century [1]. This natural process, which has been occurring for millions of years due to variations in the solar cycle, has been accelerated in the last two centuries as a result of excessive anthropization and greenhouse gas concentration, upsetting the natural balance and increasing global surface temperature [2]. Every decade since the 1980s has experienced rising temperatures, with the 2011-2020 period being the hottest on record, reaching in 2019 a global average temperature 1.1°C higher than pre-industrial levels [3]. These phenomena emerge as an increase in the frequency and intensity of extreme weather events such as heat waves, droughts, and heavy rains, increasing the risks for ecosystems (1). To mitigate the severe impacts of CC, all sectors of society must collaborate, mobilizing all fields, including education, research, and innovation, and openly encouraging shared responsibility for action [4]. According to the UNESCO Climate Action Policy Paper, CC has become one of the most significant threats to natural and cultural World Heritage sites, with the potential impact on their Outstanding Universal Value (OUV), including their authenticity, integrity, and capacity for economic and social development at the local level, as well as the quality of life of communities linked to World Heritage sites [5]. If sites are not adequately prepared or radical climate actions are not taken at the international, national, and local level, World Heritage sites may be irremediably damaged by extreme weather events [6]. As a result, ICOMOS, a UNESCO advisory body, emphasizes the importance of adequately responding to, and preparing for, CC risks for Cultural Heritage (CH). It highlights the need for effective conservation, use, and management of tangible and intangible cultural resources, considering heritage as a source of resilience and climate mitigation [7]. Not only climate change, the OUV of CH is put at risk by anthropic action which manifests itself in a series of direct and indirect actions capable of altering the ecosystem in which the protected asset is located. Anthropic risks represent a further level of complexity in the evaluation of prevention and mitigation strategies as the evaluation of specific risks is in most cases qualitative and not quantitative. The variety of risks with a potential impact on CH and in particular on Architectural Heritage requires a multidisciplinary approach and a preliminary overview of the existing methods for risk analysis and assessment in different fields. CH is permanently exposed to natural and anthropic hazards due to the impact of climate change, which accelerates the effects of natural hazards as well as those caused by human actions [8]. The traditional distinction between natural and anthropic hazards is challenged by the combination and interaction of different hazards and causes, as natural disasters also cause migration and social conflicts, and human behavior accelerates natural processes. Phenomena related to climate change and natural/anthropic risks raise new research questions and open new challenges for modern societies, requiring a broad multidisciplinary approach towards problem solving. A multi-scale approach should be applied to encompass different scenarios and to address both general and specific issues. Overall objective of the PNRR Research Project (2) is delivering innovative solutions to mitigate the effects of climate change, natural and anthropic risks on cultural heritage, based on the following methodological steps:

• identification of knowledge gaps and development of an appropriate shared paradigm (new concepts) and coordination of data acquisition and integration;
• evaluation of resilience and adaptation models of the past, re-appropriation of historical memory;
• comprehensive risk analysis on the multivariate effect of climate change and the interaction of different risks;
• development of a shared framework for modeling, simulation, and computerized data drive monitoring;
• integration of knowledge fields to support the Multi-Criteria Decision Method (MCDM);
• development of an interdisciplinary framework for a Decision Support System (DSS) aimed at the redevelopment and design of architectural heritage and historical landscape.

2. Materials and methods

2.1. Risk assessment process

In the process of risk assessment it is crucial to explore the interconnected dimensions of risk, focusing on the interplay between hazard, vulnerability, and exposure [9]. Hazards, as defined by the Intergovernmental Panel of Climate Change (IPCC), are potential sources of harm, encompassing both natural phenomena and anthropogenic events “that may cause loss of life, injury, or other health impacts, as well as damage and loss to property, infrastructure, livelihoods, service provision, and environmental resources” [10]. The dynamic interaction among hazards, vulnerability, and exposure underscores the complexity of risk, necessitating a holistic approach to mitigate its impacts. Hazards are inherently diverse and dynamic, spanning geological, meteorological, and anthropogenic domains. Natural hazards, driven by Earth's processes, include earthquakes, floods, and wildfires [11], anthropogenic hazards, stemming from human activities, encompass industrial accidents, chemical spills, and pollution [12]. Understanding the frequency, intensity, and spatial distribution of hazards is crucial for effective risk assessment [13]. Vulnerability, as defined by Adger [14], is the susceptibility of a system to harm, influenced by physical, social, economic, and environmental factors. Social vulnerability often arises from disparities in wealth, education, and healthcare access as stated by IPCC while physical vulnerability may result from inadequate infrastructure or inappropriate land use planning. A comprehensive vulnerability assessment is pivotal for identifying risk-prone areas and populations. Exposure refers to the degree to which elements at risk, such as populations, assets, or ecosystems, are subject to a hazard [15] and spatial and temporal dimensions of exposure play a crucial role in determining risk levels. Mapping exposure helps identify high-risk areas, enabling targeted interventions and resource allocation. Thus, risk assessment integrates hazard analysis, vulnerability assessment, and exposure mapping to quantify the likelihood and consequences of adverse events [16]. Effective risk management strategies involve interdisciplinary collaboration, community engagement, and policy interventions that are fundamental to activate early warning systems, build resilient infrastructure and adaptive strategies to reduce overall risk [17]. A nuanced understanding of the interconnected dimensions of risk is imperative for sustainable development and resilience. The United Nations Office for Disaster Risk Reduction (UNDRR) define resilience as: “the ability of a system, community or society exposed to hazards to resist, absorb, accommodate to and recover from the effects of a hazard in a timely and efficient manner, including through the preservation and restoration of its essential basic structures and functions.” [18]. Various methodologies and strategies are employed in the risk assessment processes [19]:

• Quantitative Risk Assessment: this method takes a numerical approach, utilizing hazard scenarios and the valuation of at-risk elements;
• Event Tree Analysis: employing a quantitative perspective, this method involves defining trees to establish relationships between diverse hazards and events;
• Risk Matrix Approach: this approach tackles risk qualitatively, allowing for the categorization of risks based on expert knowledge, particularly in situations where quantitative data is either lacking or limited;
• Indicator-Based Approach: this semi-quantitative method involves the use of indicators associated with each risk determinant or component (such as hazard, exposure, and vulnerability). These indicators are then normalized, weighted, and aggregated to derive a comprehensive risk score.

Most of the methodologies introduced above takes a quantitative approach that requires a clear overview of hazard taxonomy such as indicators, indexes and thresholds.

2.2. Methodological framework for hazards taxonomy identification

Thus, the starting point of the research consists in a systematic literature review of risk assessment approaches, methods and indicators. This process represents the base of a multidisciplinary framework consisting of the definition of risk, its categorization, its metrics represented by a measurable index and its main target, according to the classification of CH based on the harmonization of European vocabularies definitions and protocols. The framework will be also developed as a result of a co-creation process based on the collaboration between professionals and academics. This process, integrating different indicators in a synthetic evaluation model, emphasizes the multidisciplinary and multi-objective nature of the preservation of CH. Overcoming this diversity and thus barriers between disciplines has been a research priority, for which the sharing of expertise and data has been necessary. The methodological framework is defined as follow:

• Analysis of the main International and National risk assessment protocols;
• Definition of the first risk classification based on the two main classes such as natural and anthropic risks;
• Identification of the main hazard components for quantitative or semi-quantitative risks assessment.

The second stage of the research is devoted to risk's taxonomy development through a series of strategies:

• Analysis of EU Research Project completed or underdevelopment;
• Use of AI chatbot;
• Systematic Literature review and co-creation process.

As final outcome a multidisciplinary framework will be developed. The database will allow the quantitative or semi-quantitative assessment of natural and anthropic hazards that will be described taking in consideration indicators, indexes and threshold.

2.3. Hazard analysis and taxonomy

To establish a comprehensive understanding of research on hazards, the research gathered data from European and national Agencies, identified the most pertinent and comprehensive reports related to the analyzed topic, and assessed relevant online platforms for task development. Nevertheless, the in-depth overview of risk analysis at both national and European levels relies also on the findings from ongoing and completed EU projects as well as scientific publications. Presented below are the primary selections that contribute to the Task's perspective.

• 3 International references with specific focus on CH and risk assessment:

  Table 1. International references with specific focus on CH and risk assessment.

  UNESCO World Heritage Convention	https://whc.unesco.org/en/factors/
  International Centre for the Study of the Preservation and Restoration of Cultural Property ICCROM	https://www.iccrom.org/publication/guide-risk-management
  ICOMOS – ICORP International Scientific Committee on Risk Preparedness	https://icorp.icomos.org/

  Table 1. International references with specific focus on CH and risk assessment.

  UNESCO World Heritage Convention	https://whc.unesco.org/en/factors/
  International Centre for the Study of the Preservation and Restoration of Cultural Property ICCROM	https://www.iccrom.org/publication/guide-risk-management
  ICOMOS – ICORP International Scientific Committee on Risk Preparedness	https://icorp.icomos.org/

• 2 International references on risk assessment:

  Table 2. International references on risk assessment.

  DRMKC - Disaster Risk Management Knowledge Centre	https://drmkc.jrc.ec.europa.eu/risk-data-hub/#/
  UN Office for Disaster Risk Reduction	https://www.undrr.org/implementing-sendai-framework

  Table 2. International references on risk assessment.

  DRMKC - Disaster Risk Management Knowledge Centre	https://drmkc.jrc.ec.europa.eu/risk-data-hub/#/
  UN Office for Disaster Risk Reduction	https://www.undrr.org/implementing-sendai-framework

• 3 National references on risk assessment:

  Table 3. National references on risk assessment.

  Department of Civil Protection Presidency of the Council of Ministers	https://www.protezionecivile.gov.it/it/
  ISPRA - ISTITUTO SUPERIORE per LA PROTEZIONE E LA RICERCA AMBIENTALE	https://www.isprambiente.gov.it/it
  INGV - ISTITUTO NAZIONALE GEO-VULCANOLOGIA	https://www.ingv.it/

  Table 3. National references on risk assessment.

  Department of Civil Protection Presidency of the Council of Ministers	https://www.protezionecivile.gov.it/it/
  ISPRA - ISTITUTO SUPERIORE per LA PROTEZIONE E LA RICERCA AMBIENTALE	https://www.isprambiente.gov.it/it
  INGV - ISTITUTO NAZIONALE GEO-VULCANOLOGIA	https://www.ingv.it/

The analysis of the main national and international protocols on risk identification was carried out on both a general and specific level for the cultural heritage in order to identify risk clusters that may be considered invariant. During the first phase, the investigation was aimed at defining the risk class – natural, anthropic, and anthropic/natural – of the corresponding type and of the specific risk in relation to the different identified classes.

The database’s structure was articulated as follows:

• Risk class;
• Risk type;
• Specific risk;
• Probability classes (in the event of qualitative assessment);
• Indicators;
• Indices;
• Metrics (in the event of quantitative assessment);
• Bibliographical references and sources.
• Glossary.

The first protocol analyzed, the UNESCO World Heritage Convention, proposes a comprehensive classification of primary and secondary risk factors that can have a negative influence on the conservation of the “Outstanding Universal Value” (OUV), in terms of the historical, artistic, scientific, aesthetic, ethnological, or anthropological level of the world cultural heritage. The primary risk factors were defined by establishing a broad spectrum of application in a variety of natural, anthropic, and historical/cultural settings. The latter were summarized in thirteen distinct classifications within the framework, with the possibility of including others if the supplied risk is not comprehensive or if further risk elements yet to be identified should arise.

The primary factors identified at the present time are the following: Buildings and Development, Transportation and Infrastructure, Utilities and Service Infrastructure, Pollution, Biological Resource Use/Modification, Physical Resource Extraction, Local conditions affecting physical fabric, Social/cultural uses of heritage, Other human activities, Climate change and severe weather events, Sudden ecological or geological events, Invasive/alien species or hyper-abundant species, and Management and institutional factors. Although the primary factors are not classified as natural, anthropic, and anthropic/natural risks, the description of the secondary factors allows the classification within the framework to be easily refined.

Based on the identified primary and secondary factors, UNESCO proposes a list of 56 protected assets that are potentially at risk, with the consequent loss of OUV, providing a purely qualitative description of the phenomenon.

As relates to the cultural heritage, an additional classification of potential risk factors is provided by the International Centre for the Study of the Preservation and Restoration of Cultural Property, ICCROM. The protocol defines a hierarchy of elements that work together to identify the specific risks. The analysis starts from the context within which the cultural heritage, be it a building, a monument, a project, or a scenic site, is placed, identifying a series of factors that can contribute towards determining the risks, identifying them, and managing them. These include environmental, political, sociocultural, administrative financial, and legal aspects, as well as potential stakeholders. The context analysis phase is preparatory for identifying the risks, which calls for defining:

• Ten agents of decay;
• Six layers of “enclosure;
• Three risk categories in relation to their likelihood of occurrence.

Agents of decay are factors that can cause damage and therefore result in loss of value of the asset being analyzed in a specific place and in a given setting. The ten identified agents are: Physical forces, Dissociation, Incorrect RH, Incorrect temperature, Light and UV, Pollutants, Pests, Water, Fire, and Criminals. For each of these, the protocol provides specific, multi-scale examples on possible risk sources and the potential damage they can cause to the cultural heritage. In this regard, Physical Forces include as risk sources extreme events like earthquakes, wind, and erosion, as well as the improper handling of artistic assets, while the possible effects include the structures’ collapse, damage, and abrasion of their surfaces. The illustration of the agents of decay is accompanied by concrete examples that help clearly identify their cause and their effects upon the cultural heritage. To contribute towards identifying the potential hazards, the concept of “Layers of Enclosure” is introduced, through which the architectural heritage being studied is placed in relation with the corresponding ecosystem that may be considered as a potential source of damage and as an element of protection from it. If the object for which a risk analysis is performed is an archaeological find, its “Layer of Enclosure” will be the museum display case, or the container in which it is conserved, while the ecosystem with which it will relate will be the room in which it is placed. Following a bottom-up logic, the layers are defined as follows: Support, Fitting, Room, Building, Site, Region.

The final component of the analysis is the assessment of the likelihood that the risk will occur. The protocol proposes three categories:

• Rare events that take place once every 100 years (e.g. floods, earthquakes, and destructive fires);
• Common events that take place several times over the course of 100 years (e.g. earthquakes and fires of low/medium intensity);
• Cumulative processes that can take place continuously or intermittently (e.g. corrosion of metals, erosion of stone).

An additional general and trans-scalar reading on the risks is provided by two international agencies: DRMKC - Disaster Risk Management Knowledge Centre and UN – Office for Disaster Risk Reduction. The former offers a classification of the risks based on eight classes, each of which presenting risk types and specific risks. The identified classes are as follows: Geophysical, Hydrological, Meteorological, and Climatological, as natural risks, Technological and Transportation for anthropic ones, and Biological as events born from the interaction between natural and anthropic phenomena. To support the analysis, the protocol offers a “digital vocabulary” based on ShowVoc datasets, aimed at outlining the taxonomy of the risk, while identifying its main components.

The database is structured as follows:

• Risk class;
• Risk type;
• Specific risk;
• Description of the risk;
• Metrics (in the event of quantitative assessment);
• Bibliographical references and sources.

The instrument came into being with the intent to organize the base of existing knowledge in a complete taxonomy. This taxonomy aims to cover not only the components of the risk (potential damage, exposure, and vulnerability/resilience), but also the management processes and phases (for example, risk assessment and inventory of data on losses due to catastrophes). An additional tool offered to support the analysis of the risk in all its components, damage, exposure, and vulnerability, is the RISK DATA HUB through which a WEB GIS application allows the potential damage of a particular asset (building, population, economy) for a geographical area to be displayed, taking a specific risk into consideration from time to time.

Lastly, on the international level, consideration was also made of the UN - Office for Disaster Risk Reduction classification that is articulated in clusters based on seven risk classes: Meteorological and Hydrological, Geohazard, Environmental, Chemical, Biological, Technological, and Societal. Each class has different risk types corresponding to the specific risks. Supporting the classification is a webpage with a “Knowledge Base” section where the taxonomy of the risks, hazard, exposure, and vulnerability may be examined in greater depth, with content updated daily.

On the national level, as an initial instrument, that of the Department of Civil Protection, Presidency of the Council of Ministers was analyzed, which provides a holistic vision of the issue of risks, identifying nine types of risk: Seismic, Meteo/Hydrological, Volcanic, Seaquake, Forest fires, Health, Environmental, Nuclear, and Industrial. For these, a purely qualitative description of the phenomenon is provided with reference to the national risk prevention Plan and the risk prevention strategies. No indications are provided as to the risk classes and the specific risks. A broader discussion is referred to such national agencies as ISPRA – Istituto Superiore per la Protezione e la Ricerca Ambientale – and INGV – Istituto Nazionale GEO-VULCANOLOGIA – as concerns discussion of the specific risks set in relation to hydrogeological and seismic events. In this case, the description of the specific risk is broadened with indices, indicators (where present), and metrics.

Most of the analyzed protocols, except for the DRMKC, provide no classification of the risks based on the definition of the class, type, and specific risk. The DRMKC provides a “digital vocabulary” for which a clear taxonomy of the risks is outlined that facilitates understanding of these risks, also for users with no specialistic know-how. The tool also provides metrics for the quantitative assessment of the risks even if, in many cases, the datum is no longer available. In no case is information provided for indices and indicators of the specific risks.

The analysis of the general classification proposed by the main national and international protocols was preparatory for the definition of the two main risk classes, natural and anthropic, and for the selection of the various types of risk to be associated with the respective classes within the comprehensive research framework.

First, an initial partial clusterization of the risks was performed by analyzing the results in progress of the European 4CH Project (2021-2024) (3). 4CH provides a general classification of the risks based on the two main macro areas of natural and anthropic risks. As concerns natural phenomena, an additional classification was done for risks derived from cumulative processes, which is to say all the forms of deterioration that gradually accumulate over time, or any process or intermittent and fluctuating event that takes place more than once a year, and risks due to catastrophic events that are often beyond human control. The risks belonging to the first sub-category (cumulative processes) are classified as natural and biological. The former comprise the following types of risk: see level rise, glaciation, erosion, silting, desertification, ground-water, deposition, and vibration. The biological ones comprise: animal migration, vegetation, pest, decay and degradation. For the second sub-category, the classification proposed in the context of the 4CH Project is as follows: invasive species and extreme climate and geological events. The risk types underlying the invasive species include fauna and flora, while the extreme climate events are fire, downpours, squalls, floods, and hail. To conclude, extreme geological events are tsunami, earthquake, landslide, and volcano. For the anthropic class, a subdivision is proposed based on intentionality: intentional and indirect risks of causing damage to the cultural heritage. The intentional risks are in turn classified between management and crimes against the cultural heritage. As concerns the former, we find the following risk types: modern re-use, corruption, quarrying, and political; and, in the latter, vandalism, arson, theft, illegal excavations, illicit trafficking, and collectors. The family of indirect anthropic risks comprises: building/infrastructure/industry, land conversion, Heritage management, Socio-cultural, and other. Building/infrastructure/industry comprises: industrial activity, constructions, transportation, pollution, and mining. In land conversion: agriculture and forestation; Heritage management: negligence, neglect, restoration, tourism industry, visitors, and handling; Socio-cultural: change in value, veneration, loss of traditional knowledge, and performance.

Through the analysis of the deliverables dedicated to the mapping of the risks and to the development of the case studies, the types belonging to the natural and anthropic risks could be identified, for which to complete the taxonomy by providing the missing information relating to indices, indicators, and metrics of use for the purpose of the quantitative assessment of the risks. The classification proposed by the 4CH project was implemented with respect to the one proposed by the main international protocols that are the object of study, and in relation to the specific research purposes. In particular, the final classification is configured as the result of the following operations:

• homologation to the taxonomy used in the 4CH project with reference to the main international classifications;
• implementation of the natural risk types closely correlated with climate change (e.g. heat wave e cold wave);
• introduction of risk types resulting from interaction between natural and anthropic phenomena that can impact the conservation protection of the CH;
• specific classification of certain risk types that make reference to generic phenomena (e.g. Pollution, Flood);
• selection of anthropic risks in relation to the specific purposes of the research.

The following is the classification that was developed (Table 4):

Having completed the classification and identified the risk types, the second phase of the research related to the definition, if possible, of the associated specific risks, the probability classes, the indicators, the indices, and the metrics (Table 5) of use for the purpose of the quantitative or semi-quantitative assessment of the risk in relation to the cultural heritage.

In order to complete the taxonomy and provide a framework that is as comprehensive as possible, the work proceeded as follows:

• Assessment of the Projects financed by the EU framework programmes concluded or in the completion phase, considering the timeframe of the past ten 10 years, 2013-2023;
• Use of Artificial Intelligence AI, chatbot;
• Literature review and co-creation of the database with the support of specialists in the sector;
• Analysis of reports of national and international agencies specialized in managing and assessing specific risks (e.g. The World Meteorological Organization);
• Consultation of specific databases on the taxonomy of risks.

2.4. Selection of European Project on CH multi-risk assessment

Analysis of some of the main European projects on the topic was done using CORDIS, Community Research and Development Information Service, a database that collects the results of the Projects financed by the EU framework programmes for research and innovation (from FP1 to H 2020). A set of keywords was used to scan the database in relation to the specific purpose of the research, considering the timeframe of the past ten 10 years (2013 to 2023). The analysis was immediately restricted to the projects with multi-risk analysis and assessment processes. An additional investigation was done on the INTERREG CENTRAL EUROPE platform which has a database of projects financed by the specific EU 2014-2020 fund, subdivided into four categories: INNOVATION, LOW-CARBON, ENVIRONMENT/CULTURE, TRANSPORT. The screening focused on the ENVIRONMENT/CULTURE category. Within this category, the choice was made to more deeply analyze two projects which, by analyzing the Deliverables, made it possible to obtain elements of use for further refining the risk taxonomy.

The first of these projects is ProteCHt2save (2016-2019) (Table 6). In particular, the main focus of the research pertaining to the mitigation of the natural risks derived from CC, and in particular the floods and fires derived from periods of serious drought, was analyzed with the intent to preserve CH through a multi-scale approach, from the artefact to the landscape, using tailor-made solutions marked by relative simplicity of application. Analysis of the deliverables made it possible to track a partial classification of the natural risks derived from CC with some indices and indicators of use for the purpose of a quantitative assessment of the associated risks.

Other data for the completion of the framework were extrapolated from the Strench (2020-2022) project, using the associated digital platform. The portal provides a list of natural risks derived from CC, such as: heavy rain, flood, drought, and extreme heat. Moreover, some climate variables are defined, like minimum temperature, maximum temperature, and precipitation, which may be considered as determinant factors in the definition of the risks. For the specific risks cited above, various indices and indicators of use for the quantitative and semi-quantitative assessment are reported, in addition to a description to facilitate understanding of the associated phenomenon (Table 7).

In the context of the Projects financed by the EU framework programmes, Prothego - PROTection of European Cultural HEritage from GeO-hazards (H2020, 2015-2018) (Table 8) deals with the issue of protection from CH with respect to the risks derived from such geomorphological events as landslides, earthquakes, and phenomena associated with volcanic activity. Analysis of the project’s deliverables allowed partial data on the taxonomy of geomorphological risks, like threshold values and indicators for a qualitative or semi-quantitative assessment, to be collected. No indices are present.

An additional research of the potential natural risks derived from CC that may affect CH is performed in the setting of the Heracles Project (H2020, 2016-2019). Analysis of the deliverables made it possible to identify the specific risks selected for the qualitative assessment. It was not possible to arrive at the indicators and indices used for the quantitative assessment of the risks (Table 9).

The work undertaken in the STORM project (H2020, 2016-2019) underlies the development of decision-making tools aimed at facilitating the preservation of CH in the face of the challenges raised by climate change. Analysis of the deliverables made it possible to achieve the taxonomy of the main natural risks associated with climate change, with clear indication of indices and indicators (Table 10).

The RESIN project (H2020, 2015-2018) studies standardized procedures to assess the vulnerability of CH, evaluating the effectiveness of adaptation and mitigation measures. These procedures serve as a guide for decisions and the creation of mitigation strategies. For the purpose of the assessment, 31 natural risks were selected, and for 18 of them the indices and indicators for a quantitative assessment are reported (Table 12).

A holistic vision based on a data-driven process regarding the impact of the natural risks associated with climate change on the cultural heritage is provided by the European SHELTER Project (H2020, 2019-2023). Analysis of the deliverables and milestones yields a clear classification of the natural risks, developed through dedicated analysis summary sheets where essential information can be retrieved for outlining the damage component associated with the following risk types: Geophysical, Meteorological, Climatological, and Hydrological. For the following types, the main biophysical and climatic factors that may be considered as “determinant,” and the specific, derived risks, are identified. Completing the analysis are summary sheets for each risk that cast light on the following: risk type and class; possible receivers or exposed asset; analysis scale; timeframe (forecast); threshold value with respect to which the potential risk is to be determined (Table 13).

Analysis of these projects described above and of others among the major financed European Projects taken into consideration (15 in all) shows how the setting of the risks associated with natural phenomena, for the most part linked to climate change, is the one most investigated, and for which there is a solid classification of the risks and a complete description of the associated phenomena, in both qualitative and quantitative terms. For the latter, the taxonomy relating to indicators and indices described above were employed as a starting point for developing the framework. Lastly, in light of the investigation that was performed, there are no financed projects aimed at more deeply analyzing the anthropic risks associated with CH, also as pertains to natural risks, in order to asses possible interactions between the two components.

2.5. Artificial Intelligence as research assistant: using chatbot

To complete the framework’s taxonomy, use was made of Artificial Intelligence (AI) tools, particularly CHATGPT 3.5, OPENAI chatbot, an automatic learning model that uses deep learning techniques to generate a text [REFERENCE]. These machines are trained by human beings and by other machines using an enormous dataset updated to 2022. In order to determine indicators, indices, and metrics of the specific risks belonging to the two risk classes (natural and anthropic), the following queries were submitted to the chatbot:

• “(SPECIFIC RISK) indices and metrics”;
• “Can you specify the unity of measure of indices?”
• “(SPECIFIC RISK) indices and metrics scientific references”.

The following is a research example on a specific risk based on the queries (Table 14):

Although the operation is handy in methods and timing, three actions are needed to verify the scientific reliability of the results:

• Analysis of indicators, indices, and metrics provided by the chatbot through a comparison with specific scientific publications and with data contained in European Projects that have been concluded or are in their performance phase;
• Verification of the references provided by the chatbot on specific databases (e.g. Scopus);
• Selection of risk indicators and indices, excluding those that refer to exposure to vulnerability.

The verification of the references provided by the chatbot that was conducted in the experiment confirmed their reliability with respect to the topic discussed, but also highlighted many possibilities for use and many problems yet to be resolved. In particular, it is emphasized that this latter phase makes the methodology unsuitable for searching the parameters being studied, given the vastness of the identified specific risks. Therefore, the use of this tool is still to be confined to the preliminary research phase in order to restrict the field of investigation especially as relates to the types of risks that refer to the anthropic risks class. An additional innovation and possibility is outlined by ChatGPT 4’s version 4 Turbo, in which a specific chatbot can be created, able to develop and analyze a set of scientific publications in order to provide specific data required by the user. This operation might facilitate the research by considerably reducing the time dedicated to literature review.

2.6. Report analysis and co-creation through expert-based knowledge

An additional tool used to compile the database was that of seeking the parameters that describe the specific risks in reports of national and international Agencies and through the consultation of industry experts able to provide expert-based knowledge on given specific risks. Consulting the reports of International Agencies made it possible to complete or implement the taxonomy of the following specific risks:

• Anthropic risk, Air pollution [20];
• Anthropic risk, Soil pollution in agriculture [21];
• Anthropic risk, Overtourism [22];
• Natural risk, Extreme Weather and Climate Events [23,24];
• Natural risk, Fire [25];
• Natural risk, Avalanche [26];
• Natural risk, Flash Flood [27];
• Natural risk, Soil Erosion [28];
• Natural risk, Drought [29].

A multidisciplinary and opensource framework having been prepared, certain risks were defined thanks to a process of co-creation exploiting the expert-based knowledge of the Sapienza University of Rome research unit composed of researchers from various disciplinary sectors. Through the co-creation process, the taxonomy of the following risks was defined:

• Anthropic risk, Carbonation and CO2 uptake of concrete [31];
• Anthropic risk, Salt Crystallization [32];
• Anthropic risk, Corrosion [33,34];
• Natural risk, Invasive species e Biodiversity Loss [35];
• Natural risk, Plant Pest [36];
• Natural risk, Desertification [37].

2.7. Systematic literature review (SLR)

The SLR process was adopted, within the setting of this research, above all to identify the components missing within the framework relating to the natural and anthropic risks (Table 15) of use for a semi-quantitative and quantitative assessment and, in the absence of these, for identifying the methodologies and criteria employed in the assessment processes. In particular, work was done with respect to two research queries (RQs):

• ₋ (RQ1) Are there indexes and metrics to be applied to anthropic or natural risks for quantitative assessment?
• ₋ (RQ2) What criteria (indicators) these research articles employ for anthropic risks assessment?

In relation to the research queries, the SLR process followed these steps:

• ₋ Determining the keywords for building an effective research string in which the first term relates to the “Risk assessment” and the second term relates to the specific risk, e.g. “Air Pollution,” or, conversely, while a possible third term, as well as synonyms, might be employed to reduce the research field by identifying their specific setting, e.g. “Cultural Heritage,” “Indicators and indices,” and “Hazard modelling";
• ₋ Defining the list of inclusion and exclusion criteria. (Table 16);
• ₋ Selecting and analyzing the relevant research.

In order to select the database most efficient for the purposes of the research, three of the leading web search engines were compared (Figure 1):

• Rome Digital Library System of Sapienza University- SBS (Discovery Sapienza) powered by EBSCO host (https://web.uniroma1.it/sbs/discoverysapienza);
• SCOPUS peer review database (https://www.scopus.com);
• GOOGLE SCHOOLAR free Web search engine that specifically searches scholarly literature and academic resources (https://scholar.google.com/).

For the research in question, the decision was made to exclude the use of Google Scholar, as it has no peer review filters in the initial phases of the search, and Discovery Sapienza for problems relating to the host and the accesses, privileging the use of SCOPUS (SC). The initial results of the test were obtained by applying the TITLE-ABS-KEY codes to all the “query string” fields. The SC database does not automatically detect the duplicated elements. Starting from the initial query, the research was conducted for each specific risk indicated in the table (Table 12). Due to the specific nature of each of the risks and in order to achieve the best results, the initial query was modified (again within the search field) by changing the order of its terms or eliminating some of them until obtaining the query suitable to yield the results (Figure 2 and Figure 3).

The selection was further restricted by applying the “Study Area” filter on SC in order to refine the research by excluding those disciplinary areas that lie outside the research setting. The selected studies, divided into folders for each specific risk, were exported into the format (RIS) using the Rayyan online platform (htps://www.rayyan.ai) for the screening. The software in question allows users to analyze each article by displaying the title, abstract, keywords, publisher, and authors to determine the product’s inclusion, exclusion, or uncertainty in order to respond to the research queries.

2.7.1. Water pollution

The SLR process for the anthropic risk relating to Water Pollution employed the following research query (RQ): "Risk assessment" AND “Water pollution” AND “Water Quality Index.” The RQ yielded 88 results as research products within the previously defined inclusion and exclusion criteria (Table 17).

An initial analysis of the articles was carried out using the pertinent abstracts in order to identify the presence of indicators and indices of use for a quantitative or semi-quantitative assessment of the specific risk. As a result of this second phase, 28 papers were selected, with respect to which 6 scientific articles were analyzed in the portion relating to the description and assessment of the risk. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Water Pollution [38,43] to be implemented in the prepared framework.

2.7.2. Hail

The SLR process for the natural risk, Severe Weather, relating to Hail, employed the following RQ: "Risk assessment" AND “Hail”. This research yielded 22 results as research products within the previously defined inclusion and exclusion criteria (Table 18).

An initial analysis of the articles was carried out using the pertinent abstracts in order to identify the presence of indicators and indices of use for a quantitative or semi-quantitative assessment of the specific risk. As a result of this second phase, 2 papers were selected, for which an analytic study was done of the part relating to the definition and assessment of the risk. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Hail [44,45] to be implemented in the prepared framework.

2.7.3. Coastal erosion

The SLR process for the natural risk, Severe Weather, relating to Coastal Erosion employed the following RQ: "Risk assessment" AND “Coastal erosion”. The RQ yielded 120 results as research products within the previously defined inclusion and exclusion criteria (Table 19).

An initial analysis of the articles was carried out using the pertinent abstracts in order to identify the presence of indicators and indices of use for a quantitative or semi-quantitative assessment of the specific risk. Given the vastness of the results, the keyword “coastal erosion” was employed to carry out a focus on the specific risk. As a result of this additional screening, 7 papers were selected, for which an analytic study was done of the part relating to the definition of the risk. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Coastal erosion [46,52] to be implemented in the prepared framework.

2.7.4. Siltation

The SLR process for the natural risk relating to Siltation employed the following RQ: "Risk assessment" AND “Siltation”. The RQ yielded 12 results as research products within the previously defined inclusion and exclusion criteria (Table 20)

An initial analysis of the articles was carried out using the pertinent abstracts in order to identify the presence of indicators and indices of use for a quantitative or semi-quantitative assessment of the specific risk. As a result of this additional screening, 4 papers were selected, for which an analytic study was done of the part relating to the definition of the risk. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Siltation [53,56] to be implemented in the prepared framework.

2.7.5. Frost ground

The SLR process for the natural risk relating to the Frost phenomenon employed the following RQ "Risk assessment" AND “Frost,” which yielded 74 results as research products within the previously defined inclusion and exclusion criteria (Table 21).

An initial analysis of the articles was carried out using the pertinent abstracts in order to identify the presence of indicators and indices of use for a quantitative or semi-quantitative assessment of the specific risk. As a result of this additional screening, 14 papers were selected, for which an analytic study was done of the part relating to the definition of the risk. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Frost [57,69] to be implemented in the prepared framework.

2.7.6. Sea level rise

The SLR process for the natural risk relating to the phenomenon Sea level rise employed the following RQ: "Risk assessment" AND “Sea level rise”. This research yielded 449 results as research products within the previously defined inclusion and exclusion criteria. To perform an additional screening of the results, an additional keyword, “Index”, was used, for which 48 scientific products were obtained (Table 22).

An initial analysis was performed using the pertinent abstracts in order to identify the presence of indicators and indices of use for a quantitative or semi-quantitative assessment of the specific risk. As a result of this additional screening, 7 papers were selected, for which an analytic study was done of the part relating to the definition of the risk. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Sea level rise [70,77] to be implemented in the prepared framework.

2.7.7. Mining

The SLR process for the natural risk relating to the phenomenon del Mining employed the following RQ: "Risk assessment" AND “Mining hazard”. The RQ yielded 11 results as research products within the previously defined inclusion and exclusion criteria (Table 23).

An initial analysis was performed using the pertinent abstracts in order to identify the presence of indicators and indices of use for a quantitative or semi-quantitative assessment of the specific risk. As a result of this additional screening, 2 papers were selected, for which an analytic study was done of the part relating to the definition of the risk. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Mining [78,79] to be implemented in the prepared framework.

2.7.8. Deforestation/land conversion

The SLR process for the natural risk relating to the phenomenon del Deforestation/land conversion employed the following RQ: “land OR agricultural” AND “expansion” OR “land” AND “cover” OR “use” AND “changes” OR “deforestation” AND “quantitative” AND "risk assessment.” The RQ yielded 56 results as research products within the previously defined inclusion and exclusion criteria (Table 24).

An initial analysis was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage. As a result of this screening, 27 papers were discarded while for the remaining 19 an analytic study of the portion relating to the definition of the risk was performed an analytic study of the portion relating to the definition of the risk was performed. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Deforestation/land conversion [80,98] to be implemented in the prepared framework. Although climate change is indicated as the main cause, the scientific community is increasingly aware of the role played by anthropic pressures on the natural ecosystems. The indicators and indices identified to measure the deforestation risk are linked mainly to the characteristics of the soil and vegetation.

2.7.9. Vandalism

The SLR process for the natural risk relating to the Vandalism phenomenon employed the following RQ: “vandalism” AND “cultural heritage,” which yielded 21 results as research products within the previously defined inclusion and exclusion criteria (Table 25).

An initial screening was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage. As a result of this screening, 9 papers were discarded while for the remaining 6 an analytic study of the portion relating to the definition of the risk was performed. The SLR made it possible to define indicators and indices of use for a quantitative assessment of the specific risk Vandalism [99,105] to be implemented in the prepared framework. Evidence of the classification of various types of vandalism that can strike the cultural heritage (e.g. graffiti, ideological vandalism) was found. The only code found in the SLR relates to the Security Rating Index (SRI) while many of the indicators can be used to comprehend the level of vulnerability of the cultural heritage in the event of hazard.

2.7.10. Illicit trafficking

The SLR process for the natural risk relating to the Illicit trafficking phenomenon employed the following RQ: "illicit trafficking” AND "cultural heritage". The RQ yielded 23 results as research products within the previously defined inclusion and exclusion criteria (Table 26).

An initial screening was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage. As a result of this screening, 8 papers were selected, for which an analytic study was done of the part relating to the definition of the risk [106,119]. The SLR made it possible to define indicators to be implemented in the framework of use for measuring the probability that this risk might take place, most of which linked to policies for managing the cultural heritage and to natural and anthropic events (e.g. wars) that influence the management of CH. Analysis of the articles shows a growing sensitivity to the illegal traffic, as demonstrated by recent initiatives by UNESCO or ICOMOS to increase the level of protection against illegal trafficking also through international cooperation policies in order to make the specific legislation uniform. No evidence was found on the use of indices for the purposes of the quantitative assessment of the specific risk.

2.7.11. Corruption

The SLR process for the natural risk relating to the Corruption phenomenon employed the following RQs: “corruption” AND “quantitative” AND “risk assessment,” “corruption” AND "risk assessment," “corruption” AND “quantitative” AND “risk assessment.” The RQs respectively yielded 5, 56, and 14 results as research products within the previously defined inclusion and exclusion criteria (Table 27).

An initial screening was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage. As a result of this screening, 42 articles relating to the used RQs were excluded while 32 were subject to analysis. The SLR made it possible to define indicators and indices to be implemented in the proposed framework [120,134]. The corruption risk is distinguished into various settings all directly correlated to political instability and to the ideological differences that create religious and political tensions, generating a high degree of instability. The main consequence of corruption in CH is illegal trafficking.

2.7.12. Adaptive reuse

The SLR process for the natural risk relating to the Adaptive reuse phenomenon employed the following RQ: “adaptive reuse” AND “cultural heritage”. The RQ yielded 112 results as research products within the previously defined inclusion and exclusion criteria (Table 28).

An initial screening was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage, through which 90 articles were excluded. For the 22 selected papers, an analytic study of the portion relating to the definition of the risk was performed. The SLR made it possible to define indicators to be implemented in the framework of use for measuring the probability that this risk might take place, most of which linked to policies for managing the cultural heritage [135,149]. During this period of growing urbanization, CH can play a key role in achieving the goals for sustainable development, as widely recognized by such international institutions as the United Nations (UN), the United Nations Educational, Scientific and Cultural Organization (UNESCO), and the International Council on Monuments and Sites (ICOMOS). On the other hand, the impact of the conservation projects and the reuse of CH can cause it damage and loss of identity. The assessment of the risks related to the reuse projects are indispensable for assessing the projects’ feasibility. No evidence was found on the use of indices for the purposes of the quantitative assessment of the specific risk.

2.7.13. Traditional Knowledge losses

The SLR process for the natural risk relating to the Traditional Knowledge losses phenomenon employed the following RQ: “losses” OR “losing” AND “traditional knowledge” AND “cultural heritage.” The research yielded 15 results as research products within the previously defined inclusion and exclusion criteria (Table 29).

An initial screening was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage through which 9 articles were excluded. For the 6 selected papers, an analytic study of the portion relating to the definition of the risk was performed. The SLR made it possible to define indicators to be implemented in the framework of use for measuring the probability that this risk might take place [150,155]. Traditional culture is linked to the nature and customs of small communities that have maintained an intimate bond with CH over the years. One of the main risk factors is globalization, which comprises a series of transformations in the economy, in society, and in the use of the territory, and that today, along with the effects of climate change, has led to the loss of a large amount of knowledge connected to the tangible (including buildings and landscapes) and intangible cultural heritage. Indices have been identified to measure the risks of loss of ethnobotanical knowledge in areas of high naturalistic interest.

2.7.14. Political instability

The SLR process for the natural risk relating to the Political instability phenomenon employed the following RQs: “political instability” AND “risk assessment” and “political instability” AND “cultural heritage.” The RQs respectively yielded 31 and 9 research products within the previously defined inclusion and exclusion criteria (Table 30).

An initial screening was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage through which 24 articles were excluded for the first RQ and 5 for the second. For the selected papers, an analytic study of the portion relating to the definition of the risk was performed. The SLR made it possible to define indicators to be implemented in the framework of use for measuring the probability that this risk might take place [156,163]. Political instability is the origin of such other possible risks as wars, corruption, and vandalism that that can directly strike CH in the form of illegal traffic of assets or destruction in the most extreme cases. Political instability is listed in many articles as one of the main risks for the conservation of CH. No specific indicators to quantitatively measure the specific risk were identified, except for the Energy Security Index (ESI) that is placed in direct correlation with political stability.

2.7.15. War

The SLR process for the natural risk relating to the War phenomenon employed the following RQs: “war” AND “risk assessment” AND “cultural heritage” and “war” AND “quantitative” AND “risk assessment.” The first RQ yielded 6 results within the previously defined inclusion and exclusion criteria while the second one yielded 16 (Table 31).

An initial screening was performed using the following search filters: quantitative, risk, risk assessment, index, indicators, and cultural heritage. The screening process excluded 3 articles from the first query and 9 articles from the second, that could not be correlated with the queries to be searched. For the selected papers, an analytic study of the portion relating to the definition of the risk was performed. The SLR made it possible to define indicators to be implemented in the framework of use for measuring the probability that this risk might take place, while there is no evidence of the use of quantitative indices [164,170]. The analysis shows that war has a devastating impact: on CH, becoming a triggering factor for such other risks as soil and water pollution, on natural ecosystems, on the economy, and on a country’s political stability.

3. Results and discussion

3.1. Opensource framework for driven reasoning in risk assessment

Using the methodology described above made it possible to develop an opensource framework based on a reasoned classification of risks, that presents data relating to indicators, indices, and metrics whose consultation facilitates the quantitative or semi-quantitative assessment of the natural and anthropic risks that may have repercussions on CH, from the architectural heritage to the natural landscape, in a multi-scale dimension. Publicly available datasets were created in this study. This data can be found here: [https://docs.google.com/spreadsheets/d/1bg9UyY8tctCj3eFlao3GGAzHk31tiLhTpfVFSl_3EFs/edit?usp=sharing]. The framework collects, in a single repository, the main natural and anthropic risks, with the related taxonomy, representing right from the beginning an open document implemented over the course of time through knowledge-based expertise that can be consulted by researchers, professionals, and public administrations readying assessments of risks in the CH sector. The proposed model will be implemented with the Exposure and Vulnerability risk components, which characterize the various risks in relation to the applicative context (site-based evaluation process) in order to provide a complete assessment framework. The framework’s structure permits additional specializations in greater thematic depth should it be necessary to update the list of specific risks.

4. Conclusion

The research output is configured as an essential tool for completing the first task of the National Recovery and Resilience Plan (PNRR) research Project that calls for a complete quantitative analysis of the risks on the multivariate effects of climate change and on the interaction of the various risks. The adoption of specific selection and data collection methods, the definition of semantic models (ontologies and vocabularies), and the design of a scalable and interoperable logical infrastructure, are key strategies for the development of an information system that supports the entire knowledge process related to the cultural heritage. These models consider and represent, in a format that can be shared and computed, the distinctive traits, the visible and invisible uses of the past and present, and the risks that the cultural heritage is subject to, with the related impact assessments and forecast scenarios, and material and immaterial values, caused for the territory by their presence. In this direction, the definition and the monitoring and control tests that will be done on the two selected case studies will make it possible to validate the new methods for aggregating and organizing the data and metadata derived from different possible disciplinary paths converging upon the reading of the natural and building historical heritage, holding together the complexity of the processes and visions connected to knowledge of them. The overcoming of the barriers between the disciplines is in fact a priority of the project, which makes it necessary to share knowledge and data in order to achieve processes of sustainable transformation of the heritage, to be implemented through Decision Making (MCDM) opening towards new design scenarios. These planning systems use an evidence-based approach that allows data originating from the various scientific and humanistic sectors involved to be integrated.