Ranking Port Criticality Under Climate Change: An Assessment of Greece

Abstract

Ports are vital components of global and regional supply chains, supporting trade, transport connectivity, and socio-economic development. However, their functionality is increasingly threatened by climatic hazards such as sea-level rise and heat stress, both of which are projected to intensify under future climate change. This study presents a comprehensive framework for assessing the criticality of ports within a national network, demonstrated through its application to the Greek port system, which encompasses a multitude of ports of all types from large international hubs to small island ones. The framework combines openly accessible geospatial and socio-economic data with projections of exposure to sea-level rise and extreme heat within a structured multi-criteria decision-making (MCDM) approach, enabling the identification of critical ports and the prioritization of adaptation needs. Results show that large mainland ports dominate in socio-economic importance and network centrality, while smaller island ports are vital locally due to limited redundancy and high exposure to climatic hazards. By 2100, nearly all ports are projected to experience freeboard reductions below operational thresholds and increased heat-related stress. These results highlight the need for targeted adaptation measures, including engineering interventions for mainland ports and redundancy-enhancing actions for island ports. The proposed framework provides a replicable, data-driven tool to guide evidence-based prioritization of adaptation investments and strengthen climate-resilient maritime transport and coastal management, thereby contributing to the achievement of Sustainable Development Goals (SDGs) 1.5, 9 and 13.

1. Introduction

Ports form crucial nodes for the global supply chains, as 80% of the traded goods by volume are transited through them [1,2]. They act as catalysts for socio-economic development, by enabling trade, supporting supply chains, and facilitating industrial and urban growth and the ‘blue economy’ at local, national, and regional/global levels [3,4]. Port development appears to have a causal effect on population growth, as ports attract businesses and employment that stimulate demographic expansion [5].

Ports form complex systems, the functionality and sustainability of which are controlled by various factors, such as the economic climate, disruptions in the supply chains, engineering and technological developments and a large array of diverse policies and legislation [6]. At the same time, ports are exposed to many climatic hazards (e.g., marine storm-induced extreme sea levels -ESLs and waves, the slow-onset rise in mean sea level driven by global warming, high mean/extreme temperatures, precipitation and winds), most of which are expected to deteriorate in the future due to Climate Variability and Change—CV&C [7,8,9,10]. A growing body of research underscores that these climatic drivers can disrupt berthing and mooring operations, impair crane and equipment efficiency, accelerate structural degradation, and reduce the functional freeboard of port infrastructure. Extreme heat has also been shown to impact labor productivity, energy demand, and cargo handling, while coastal flooding and overtopping may lead to system-wide delays with cascading effects across supply chains. These climatic hazards can cause extensive direct and indirect damages/losses to ports [2,11,12], as well as to associated coastal urban/industrial clusters, populations and socio-economic activities and the connected land transport networks (e.g., [13,14]). Functioning ports are particularly crucial for islands where they form gateways to offshore trade, connectivity, communication, tourism and Disaster Risk Reduction activities (e.g., [15]).

Assessing port exposure to current and future climatic conditions is a prerequisite for developing and implementing efficient adaptation strategies. The approaches/criteria used in such assessments depend on their spatio-temporal scales and the available information and resources, and should involve evaluations of the current environmental and socio-economic trends and risks, as well as projections under the CV&C [16]. Various approaches have been previously used to describe the vulnerabilities of coastal systems that combine elements from one or more factors into composite coastal vulnerability indexes—CVIs (e.g., [17,18,19,20]). Nevertheless, such approaches still require a more pronounced integration of environmental and socio-economic considerations under the changing climate [19,21].

In the case of ports, several efforts for the development of composite indexes that can describe port vulnerabilities (PVIS) and/or resilience (RVIs) under a changing climate have been recently made, which have been structured around technical, physical, environmental, and socio-economic sub-indices (e.g., [22,23,24,25]). For example, McIntosh & Becker [26] analyzed 22 major U.S. seaports by decomposing vulnerability into exposure, sensitivity, and adaptive capacity, using expert judgment and the Analytic Hierarchy Process (AHP) to weigh indicators. On a global scale, Izaguirre et al. [9] evaluated ports on the basis of their technological capacity, recovery potential, and overall resilience, whereas Li et al. [24] introduced a dynamic resilience-inversion index using daily shipping activity data from container ports and natural disasters. Nevertheless, cross-port comparability is still limited [27], as well as the use of integrated, cross-sectoral port vulnerability metrics [28].

At regional/national scales, the extensive scope and the high potential costs of adaptation [29,30] require fit-for-purpose frameworks to rank the significance of individual ports within a port network, in order to prioritize effective responses and plan allocation of the (mostly) limited human and economic resources. In this context, assessments of port vulnerability should be complemented by the analysis of the criticality of constituent ports within the network to identify which ports matter most for maintaining overall system functionality.

The objective of the present contribution has been to develop/apply a comprehensive framework for the assessment of the criticality of the ports within a network at the national/regional scale. The framework was designed to integrate indicators of the ports’ socio-economic significance with metrics of exposure to climatic hazards, utilizing openly accessible geospatial, environmental, and socio-economic data together with projections of exposure to sea levels and extreme heat under future climate scenarios. In doing so, the approach advances previous work and directly addresses the previously mentioned gap concerning the lack of integrated, cross-sectoral port vulnerability metrics by combining climate-hazard projections with network-level port analysis within a single, structured multi-criteria decision-making (MCDM) process, enabling the joint evaluation of operational, socio-economic, and exposure-related dimensions of port criticality. The exclusive use of openly accessible datasets further enhances transferability and reproducibility, offering a practical tool for evidence-based adaptation planning in data-limited contexts. By synthesizing socio-economic indicators with climate change exposure, the framework enables evidence-based ranking of the adaptation needs within a network, supporting the development of targeted, climate-resilient coastal management strategies. The framework is demonstrated through its application in the national port system of Greece. Greece has an extensive and diverse port network, encompassing major international hubs and numerous island ports that serve as vital lifelines for insular economies.

2. Methodology

The rank-ordering of infrastructure components in a transport network, termed transport network criticality analysis [31], has as a main goal not only to calculate vulnerability scores for the network components, but also to rank them accordingly [32]. Thus, the criticality analysis is differentiated from other types of transport network investigations, such as exposure analysis [33] and robustness analysis [34]. Network criticality does not use a single formalization and, consequently, transport experts/authorities are faced with a wide choice of potential metrics. Network criticality has been considered on the basis of the probability/consequence of component failure [31], as a probability-neutral concept [35], through threat-based investigations designed to identify/rank vulnerable port facilities under uncertainty [36], or on the basis of the trade flows in the maritime transport network [37]. In the present study, a framework was developed to assess the criticality of the Greek ports under CV&C (Figure 1).

Given its geographic setting, Greek ports are particularly exposed to two major climatic hazards: extreme sea levels (ESLs) and heatwaves [10,38]. At present, coastal flooding already represents a serious risk to transport infrastructure, which is projected to become very severe by the end of the century under both moderate and high warming scenarios [39]. Marine flooding may arise from relative sea-level rise (RSLR) as well as from more frequent, intense, and prolonged ESLs. Increases in both the mean and extreme sea levels are expected to inundate ports, damage infrastructure, and disrupt operations, with severe socio-economic impacts [2,9,40,41]. Island ports are expected to be especially vulnerable due to their geographic location and limited adaptive capacity (e.g., [8]).

Rising mean temperatures and the intensification of extreme heat events also pose major challenges to connecting transport infrastructure [4,14]. Heat stress accelerates degradation of paved port surfaces, induces asphalt rutting, compromises bridges, and triggers rail buckling. It can further lead to equipment failures, higher energy demand for cooling, and risks to the health and safety of port personnel and users, thereby reducing port productivity. The combination of extreme heat with high relative humidity magnifies risks of heat-related illnesses and is expected to present an acute hazard for workers and users [2,42]. In Europe, heat stress is already considered a critical health risk and is projected to become extremely dangerous in the future, particularly in southern regions, as reported in multiple studies [14,39,43,44].

2.1. The Framework

The proposed framework (Figure 1) integrates geo-spatial, technical, and socio-economic information—compiled for the present study in a dedicated national port database—with projections of port exposure to RSLR and extreme heat under different climate scenarios. It combines a Strengths–Weaknesses–Opportunities–Threats (SWOT) analytical structure with a set of multi-criteria decision-making (MCDM) techniques (details in the following Sections) providing a port criticality evaluation under CV&C and, thus, a prioritization of the adaptation needs in the network. In total, 136 Greek ports (32 mainland and 104 island) were included in the analysis, for which information for all selected indicators could be mined. This sample accurately represents the operational and geographical diversity of Greek ports.

Indicator selection was guided by SWOT analysis, which considered quantitative and qualitative information recorded in the database, such as port dimensions, traffic metrics and connectivity. Projections of flood and extreme heat exposure were also incorporated. The analysis was supported by (20) stakeholder inputs, collected through a structured questionnaire, ensuring that the selected indicators capture both operational and environmental pressures relevant to port management and an Analytical Hierarchy Process (AHP).

The ranking procedure was implemented in two stages. In the first stage, all (136) Greek ports were ranked using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), selected for its stability when handling large numbers of alternatives/criteria [45]. TOPSIS was performed to evaluate: (i) a Socio-Economic Index (SEI) based on socio-economic indicators; (ii) an Exposure Index (EI) based on the projected CV&C hazards; and (iii) a composite Criticality Index (CRI) that integrates all indicators. Through this procedure, the most critical ports were identified, on the basis of both socio-economic significance and exposure to CV&C. A subset of the 15 highest-ranking ports was selected to enable the second-stage, more detailed evaluation.

The second stage involved a more robust ranking of port criticality, using a different MCDM technique. A smaller number of ports was chosen for this analysis, due to the computational demands and the stability of results when applied to a large number of alternatives/criteria. Thus the 15 highest-ranking ports identified at the first stage were evaluated using the Preference Ranking Organization Method for Enrichment Evaluations (PROMETHEE II), a pairwise, preference-based method well suited for ranking a limited number of alternatives based on multiple criteria. To address its lack of a built-in weighting mechanism, AHP-derived indicator weights were integrated in the analysis.

By combining SWOT based indicator selection, expert-driven weighting through AHP, large-scale screening and detailed prioritization, the proposed framework offers a transparent, scalable, and stakeholder-informed framework for ranking the criticality of ports within a network.

In the following sections, the methods used are discussed in more detail.

2.2. Data Collation and Database

An inventory of Greek ports was compiled, containing detailed information on their spatial, technical, and socio-economic attributes. The port database integrates information from diverse sources, including existing national datasets, high-resolution satellite imagery, as well as some primary information collected through stakeholder engagement (e.g., Port Authorities (PAs), the Hellenic Coast Guard).

Structural characteristics include the delineated port area, the breakwaters’ length and elevation, the quay/dock height, and the orientation of the port entrance. These features were extracted from the Coastal Zone Land Cover/Land Use (LC/LU) dataset of the Copernicus Land Monitoring Service (https://land.copernicus.eu/en/products/coastal-zones/coastal-zones-2018, accessed on 4 April 2025). Inherent inaccuracies in the derived port polygons due to mapping limitations of the Copernicus dataset were manually corrected on recent high-resolution satellite images available via the Google Earth Pro platform, following the approach of Bove et al. [46]. To ensure consistency and minimize inter-analyst variability, digitization was performed by a single trained analyst using standardized delimitation protocols. Topographic elevation data were extracted from the LSO data series from the high-resolution (2 m × 2 m) Digital Elevation Models of the Hellenic Cadastre [47]. Land use indicators incorporated into the database include asset density within a 100 m landward buffer zone—extracted from the Coastal Zone Land Cover/Land Use (LC/LU) dataset of the Copernicus Land Monitoring Service (https://land.copernicus.eu/en/products/coastal-zones/coastal-zones-2018, accessed on 4 April 2025)—and population counts within a 500 m radius, obtained from the Global Human Settlement Layer (GHSL; https://data.jrc.ec.europa.eu/dataset/2ff68a52-5b5b-4a22-8f40-c41da8332cfe, accessed on 4 April 2025). In addition, the total length of the adjacent road network was calculated using OpenStreetMap data (https://www.openstreetmap.org/, accessed on 4 April 2025).

2.3. Port Exposure to Rising Sea Levels and Extreme Temperatures

The present analysis considered three primary climatic hazards: changes in the mean (RSLR) and extreme sea levels (ESLs), temperature extremes, and dangerous/deadly heat conditions (combinations of high temperature and humidity). Sea level information was retrieved from the LISCoAST (Littoral System Coastal Assessment) database of the Joint Research Centre of the European Commission (https://data.jrc.ec.europa.eu/collection/liscoast, accessed on 10 August 2025) and comprised estimates of the RSLR and the 100-year return period ESLs for the baseline (1986–2014) and for 2050 and 2100 under Representative Concentration Pathways (RCP4.5 and RCP8.5). ESLs result from the combined effect of mean sea-level rise, tides, storm surges, and wave setup (e.g., [48]). Relative sea-level rise projections were derived using a probabilistic, process-based approach [49]. Storm surges were simulated with Delft3D-Flow, forced by wind and atmospheric pressure fields from an ensemble of eight climate models, and validated against ERA-Interim reanalysis for 1980–2014 [50]. Wave setup was excluded from ESL estimates because its accurate estimation requires detailed site-specific modeling [51,52] of intra-port wave dynamics [9,53,54], which is beyond the scope of this study.

Temperature and humidity data were obtained from the Coordinated Regional Downscaling Experiment (CORDEX), using downscaled outputs of the CNRM-CERFACS-CM5 global model via the CNRM-ALADIN52 regional model [55]. The dataset provides daily mean and maximum temperature and relative humidity at ~12 km spatial resolution across Greece, enabling localized assessment of thermal and moisture-related hazards under the same RCP scenarios and time horizons.

For each port, the nearest LISCoAST and CORDEX projection points were identified using GIS analysis.

The assessment of exposure was carried out for both historical (baseline) and projected future climate conditions. A ‘static flood’ analysis approach was employed, whereby projected mean and extreme sea levels were compared against port quay elevations (see also [10]). Although quay structures are generally designed with uniform elevation, surrounding infrastructure—such as connecting roadways and ancillary coastal facilities—often varies in elevation, particularly in older ports that have developed within pre-existing urban settings [8]. This is particularly evident in island ports, many of which have evolved incrementally over time within densely built coastal environments, leading to elevation variability within the broader port area.

To enable a consistent, national-scale analysis, a single representative elevation value was assigned to each port. These values were extracted from the Hellenic Cadastre DEM (horizontal resolution of 2 m and vertical accuracy of about 0.4 m), which enabled more precise elevation mapping than other available DEMs [56]. For each port, the maximum quay-dock elevation along the land–sea interface was derived, after excluding outliers with unusually low values (<0.5 m) relative to the surrounding port area. These adjusted elevations were then compared against the projected RSLR and the 100-year Extreme Sea Levels (ESL₁₀₀) for the baseline period and for the years 2050 and 2100 (RCP4.5 and RCP8.5 scenarios). It is acknowledged that representing quay elevation by a single value—particularly the maximum elevation—may result in flooding underestimations, as this approach does not account for the lower-lying section of the quays that could be more susceptible to ESLs. Thus, the comparison of future sea levels with the maximum quay/dock elevations is a conservative exposure indicator.

Beyond the risk of quay flooding, sea-level rise can also impose significant operational constraints. Such constraints arise from the reduction in freeboard—the vertical distance between the quay surface and the sea level. Notably, port operations may experience disruptions even when sea does not overtop the quay, as reduced freeboard limits safe vessel berthing and passenger and cargo handling. According to established port safety guidelines [57], a minimum freeboard of approximately 1.5 m is required to ensure the safe mooring, loading, and unloading of large commercial vessels, including freight and passenger ships [57,58,59]. When this freeboard is reduced below safety thresholds, berthing becomes unsafe or unmanageable, potentially leading to operational shutdowns even in the absence of quay overtopping.

Regarding the port exposure to the extreme heat, this study builds on the material/operational and human health thresholds introduced by Monioudi et al. [10] to assess potential impacts. A daily maximum temperature threshold of 32 °C [60] was adopted as an indicator of thermal stress on port (and other coastal) transport infrastructure. Exceedance of this threshold can lead to material degradation, thereby increasing accident risks and disrupting operations [14]. To account for risks to human health, the analysis incorporates heat mortality thresholds [42], which are based on combinations of high temperature and humidity linked to human mortality due to exceedance of the human thermo-regulatory capacity. Specifically, the ‘deadly heat’ threshold associated with a 95% probability of mortality was applied to assess risks for port personnel/users. High-resolution CORDEX climate projections from the CNRM-CM5/CNRM-ALADIN52 model chain were used to estimate the frequency of threshold exceedance across three 20-year periods under the RCP4.5 and RCP8.5 climatic scenarios: baseline (1986–2005), mid-century (2041–2060), and end-century (2081–2100). These non-overlapping 20-year periods were selected for consistency with the Joint Research Centre (JRC) datasets, which provide projections for comparable mid- and end-century horizons derived from multi-decadal extreme value analyses [61]. For each period, the average number of days per year exceeding the defined thresholds is calculated.

2.4. Ranking Port Criticality Through Multicriteria Analysis

Multi-criteria analysis (MCDM) is a structured decision-making tool that facilitates optimal decision-making. In this study, it was applied to rank critical ports by integrating socio-economic and climate risk criteria. MCDM generally involves criteria assessment, weighting, and aggregation. The assessment quantifies the performance of alternatives against selected criteria, with weighting reflecting the relative importance of each criterion from the perspective of decision-makers/stakeholders. Weights can be derived through various techniques, yielding either ordinal or cardinal measures based on judgments about the relevance of different issues. Criteria aggregation then combines assessment outcomes and weights, typically through algebraic rules, to evaluate overall performance and the results are commonly presented as rankings of alternatives [62].

Many MCDM techniques exist, each with its own strengths and weaknesses [63]. In order to select the most suitable techniques for ranking port criticality, these techniques were carefully reviewed. The Analytical Hierarchy Process (AHP) [64,65] simplifies decision problems into hierarchies, offering intuitive structure and consistency checks, although it is prone to ranking reversals and requires numerous pairwise comparisons. The Analytical Network Procedure (ANP) [66] relaxes the assumption of independence among criteria, improving accuracy, yet its complexity and demanding comparison requirements limit usability. Data Envelopment Analysis (DEA) [67] is valuable when criteria have different units, but it is highly sensitive to measurement error and unsuitable for large datasets. Simpler approaches, such as the Weighted Sum Model (WSM/SAW) and Weighted Product Model (WPM) [67], are intuitive and computationally efficient, but WSM requires criteria in identical units and risks oversimplification, while WPM cannot accommodate criteria with zero weights. More advanced methods address uncertainty and ranking complexity. Goal Programming (GP) [68] can handle multiple goals and constraints, but requires external techniques to determine reliable weights. Elimination and choice translating reality (ELECTRE) [69] accounts for uncertainty and supports veto options, but is difficult to communicate to non-specialists. Multi-Attribute Utility Theory (MAUT) [70] incorporates uncertainty but requires precise utility functions. The simple multi-attribute rating technique (SMART) [71] reduces complexity, although it may not suit all contexts. PROMETHEE (Preference ranking organization method for enrichment of evaluations) methods provide robust preference-based rankings [72], but lack a standardized method for weight assignment.

TOPSIS (Technique for order preferences by similarity to ideal solutions) [45] evaluates alternatives relative to both best and worst possible scenarios, supporting diverse distance metrics, but neglects correlations amongst variables. Simulated Uncertainty Range Evaluations (SURE) [73] provides a simple and understandable visualization of strength/uncertainty across alternatives, but presents challenges for decision-makers when they choose between alternatives with overlapping uncertainties. VIKOR (VlseKriterijumska Optimizacija I Kompromisno Resenje) identifies solutions, when trade-offs exist, by balancing ‘maximum group utility’ and ‘individual regret’; however, it depends on subjective weight assignments [74]. COPRAS (Complex Proportional Assessment) uses proportional assessment to rank alternatives while considering interdependencies, but requires complex calculations and subjective comparisons [74]. MOORA (Multi-objective optimization on the basis of ratio analysis) simplifies decisions by converting multiple criteria into a single ratio, offering computational efficiency and reliability when criteria are independent. Its main drawback is poor handling of dependent criteria, as well as qualitative ambiguity without hybrid or fuzzy extensions [75,76,77,78]. Finally, SWARA (Step-wise Weight Assessment Ratio Analysis) provides structured, expert-driven weighting and captures preference uncertainty. However, it cannot assign equal values to criteria, often produces large disparities in importance, and reduces the influence of lower-ranked criteria in complex decisions [79,80,81].

Based on an analytical overview of the available MCDM techniques, a fit-for-purpose combination of AHP, TOPSIS, and PROMETHEE II techniques was selected to systematically assess and rank ports. Each technique was selected for its complementarity and ability to overcome the shortcomings of others.

2.4.1. AHP Implementation

Indicators were weighted through a relevance analysis which was based on (20) stakeholder inputs, collected through a structured questionnaire. The relative importance (weights) of indicators was determined through pairwise comparisons using the Analytical Hierarchy Process (AHP) [65]. Following Saaty’s fundamental scale, ranging from 1 (equal importance) to 9 (extreme importance) [64], the stakeholders/experts provided judgments on indicator significance. These results were compiled into matrices, and their consistency was tested through eigenanalysis. A consistency ratio (CR) below 0.1 was considered acceptable; for values greater than 0.1, experts adjusted their judgments to improve consistency. Comparisons were conducted within the questionnaire, with adjustments being made by the stakeholders/experts themselves and the CR values automatically calculated to ensure consistency. This procedure provided a sound basis for integrating subjective expertise into the framework. This weighting process relies on expert judgment and adapts to geographic and contextual conditions. In this study, AHP was employed exclusively for weight assignment, allowing the integration of expert input while minimizing concerns such as ranking reversals.

The pairwise comparison matrix A = (a_ij) expresses the relative importance of element i compared to j. Consistency is tested through eigenvalue analysis:

$$Aw=\lambda w$$

(1)

where w is the eigenvector (weights) and λ is the maximum eigenvalue of A. If the comparisons are perfectly consistent, ${\lambda }_{max}=n$ (n: number of criteria being compared, i.e., the dimension of matrix $A$). The consistency index (CI) and consistency ratio (CR) are calculated as:

$$CI={\displaystyle \frac{({\lambda }_{max}-n)}{n-1}}$$

(2)

$$CR={\displaystyle \frac{CI}{RI}}$$

(3)

where RI is the random index from simulated matrices. A CR ≤ 0.1 indicates acceptable consistency; otherwise, judgments should be revised [82]. Final weights are synthesized up the hierarchy to produce overall priorities.

2.4.2. TOPSIS Implementation

The TOPSIS method was applied in the first stage of the port ranking analysis (Figure 1) due to its robustness and computational efficiency, regardless of the number of criteria or alternatives [45]. TOPSIS can identify the alternative most closely to the ideal best solution and farthest from the ideal worst solution [82].

The normalized decision matrix for i = 1, 2, …, m alternatives and j = 1, 2, …, n criteria is computed as:

$$r_{ij}={\displaystyle \frac{a_{ij}}{\sqrt{{\Sigma }_{j=1}^m{a}_{ij}^2}}}$$

(4)

Weighted normalization incorporates criterion weights:

$$p_{ij}=w_i{r}_{ij}$$

(5)

where $w_i$ is the weight of the criterion i and ${\sum }_{i=1}^n{w}_i=1$

The Euclidean distances from the ideal best (A⁺) and worst (A⁻) are then calculated:

$$s_i^{+}=\sqrt{{\sum }_{j=1}^n{(p_{ij}-p_j^{+})}^2}$$

(6)

$$s_i^{-}=\sqrt{{\sum }_{j=1}^n{(p_{ij}-p_j^{-})}^2}$$

(7)

The relative closeness to the ideal solution is:

$$C_i={\displaystyle \frac{s_i^{-}}{(s_i^{+}+s_i^{-})}}$$

(8)

Alternatives are ranked according to C_i, with higher values indicating higher priority.

2.4.3. PROMETHEE II Implementation

The PROMETHEE II method was applied in the second stage of the port ranking analysis (Figure 1) to refine and validate the results obtained from TOPSIS. It relies on pairwise comparisons of alternatives under each criterion, allowing decision-makers to express preference intensities. The decision matrix is first normalized, with beneficial and non-beneficial criteria treated separately to ensure comparability across metrics. The relative performance of alternatives is established through pairwise differences. Outranking relations are formalized via a preference function. Although several generalized functions exist [83], their reliance on thresholds often complicates real-time applications; to overcome this, a simplified function was adopted [84]:

$$P_j\left(a,b\right)=0,R_{aj}\le R_{bj}$$

(9)

$$P_j\left(a,b\right)=R_{aj}-R_{bj},R_{aj}>R_{bj}$$

(10)

where R_aj, R_bj are the normalized values of alternatives a and b under the criterion j.

The aggregated preference function of alternative a over b is given by:

$$\pi (a,b)={\displaystyle \frac{{\sum }_{j=1}^n{w}_j{P}_j(a,b)}{{\sum }_{j=1}^n{w}_j}}$$

(11)

The positive, negative, and net outranking flows are then determined as follows:

$${\phi }^{+\left(a\right)={\displaystyle \frac{1}{m-1}}{\sum }_{b=1}^m\pi \left(a,b\right)}$$

(12)

$${\phi }^{-(a)={\displaystyle \frac{1}{m-1}}{\sum }_{b=1}^m\pi (b,\alpha )}$$

(13)

$$\phi (a)={\phi }^{+\left(a\right)-{\phi }^{-\left(a\right)}}$$

(14)

The positive outranking flow indicates the extent to which an alternative a outranks all other alternatives: higher values of φ⁺(a) correspond to better performance. Conversely, the negative outranking flow reflects the extent to which an alternative a is outranked by the others; lower values of φ⁻(a) signify a more favorable alternative. The alternative with the highest net flow φ(a) is considered the most preferred. PROMETHEE II provides a complete ranking of alternatives and is particularly effective when integrated with AHP-derived weights to address its lack of a weight assignment mechanism.

3. Results

3.1. SWOT Analysis and Selection of Indicators

SWOT analysis was carried out to identify the main indicators that can describe the factors that may affect port vulnerabilities, prospects for resilience and sustainability, and their criticality within the port network.

Strengths lie primarily in the Greek ports’ strategic position at the crossroads of Europe, Asia, and Africa, which makes them transit hubs in European and global trade networks [85]; major ports such as Piraeus and Thessaloniki (Figure 2) already serve as international logistics hubs in the Mediterranean [86,87,88]. There are 25 ports within the European Union (EU) Trans-European Transport Network (TEN-T): 5 Core Network ports (Piraeus, Thessaloniki, Patra, Igoumenitsa, and Heraklion), and twenty Comprehensive Network ports. These constitute integral nodes in the European transport infrastructure and the ‘Orient/East Med Corridor’ [85]. Beyond trade, Greek ports are enablers of passenger transport and tourism, serving as lifelines for island economies.

Weaknesses include the already high exposure of Greek ports to climate-related hazards [10]. In addition, port governance is fragmented, with overlapping competencies between central and local public bodies and port authorities, slow bureaucracy, and the lack of a coherent framework for adaptation and risk management [89,90]. Development is uneven, with large ports showing advances in modernization/digitalization in contrast to smaller or island ports [91]. Financial constraints also present challenges, affecting monitoring procedures, digital infrastructure development, and the integration of sustainable technologies, particularly in the smaller ports. Given the lack of comparable governance and policy data for all ports, these aspects are reflected in the qualitative assessment rather than included directly as indicators.

Opportunities arise from the significant sustainability potential through the adoption of green shipping corridors, port electrification, renewable energy integration and circular economy practices [92,93,94]. The European and International climate/sustainability policies can provide funding pathways [95,96], whereas EU legislation (e.g., the Flood Risk (2007/60/EC), the Environmental Impact Assessment (2014/52/EU) and the Critical Entities Resilience ((EU) 2022/2557)) Directives strengthen the regulatory environment.

Threats are dominated by the intensification of climatic hazards, such as rising sea levels, storms, and heat waves which could disrupt operations and challenge the adaptation capacity [25]. Global and regional geopolitical instabilities/disruptions also threaten trade and passenger flows, energy security, investment and tourism [6,97,98,99,100]. Competition from other Mediterranean ports presents further challenges, whereas ports depending heavily on tourism (such as island ports) are exposed to disruptions in travel demand and shipping cycles.

It appears that Greek ports combine unique strengths and opportunities with serious weaknesses and threats. Their strategic role in regional trade and European transport corridors, together with emerging financial and technological opportunities, creates strong prospects for sustainable transformation. However, governance fragmentation, uneven development, climate risks, and geopolitical instability continue to pose significant challenges that require coordinated action and investment to ensure resilience and long-term sustainability. These governance and policy-related elements are therefore acknowledged as contextual factors and identified as candidates for future indicator extensions when suitable data become available.

The SWOT analysis guided the selection of a set of indicators, which were subsequently employed in the criticality ranking. The set includes socio-economic attributes that describe the port size/usage and the surrounding populations and urban development, the inland and offshore connectivity, and the existence of alternatives, as well as climatic stressors (

Table 1. The list of the selected indicators.

Indicators	Description	Type of Data/Effect
1. Port area (m2)	Calculated from digitized polygons derived from the Coastal Zone Land Cover Dataset of the Copernicus Land Monitoring Service, with manual corrections where necessary	Quantitative/Beneficial 1
2. Population	Defined as the permanent inhabitants within a 500 m radius from the port; (3) the length of road network, i.e., the total length of roads within a 1000 m radius around the port	Quantitative/Beneficial
3. Road network length	The total length of roads within a 1000 m radius around the port	Quantitative/Beneficial
4. Passenger throughput	Based on the number of boardings/disembarkations	Quantitative/Beneficial
5. Port connectivity	Expressed as the number of domestic and international maritime connections	Quantitative/Beneficial
6. Urban development	Measured as the building density behind the port, expressed as a percentage of the area of a 100 m buffer zone	Quantitative/Beneficial
7. Redundancy	Reflects the dependence of an area on a single port, with higher values assigned to well-connected mainland ports and lower values assigned to island ports with limited alternatives (port grading scheme: mainland Greece = 5, islands with 5–6 ports = 4, medium-sized islands with >3 ports = 3, islands with 2 ports = 2, islands with 1 port = 1)	Qualitative/non-Beneficial 2
8. Flood exposure	The available freeboard threshold, derived from the comparison of quay elevation with the projected relative sea-level rise (RSLR)	Quantitative/Beneficial
9. Deadly heat exposure	Calculated as the number of days when the combined effects of temperature and relative humidity are projected to exceed thresholds creating hazardous/lethal conditions for port personnel/passengers	Quantitative/Beneficial
10. Heat risk to infrastructure	Defined as the number of days when extreme temperatures surpass thresholds, affecting operational functionality, safety and port infrastructure resilience	Quantitative/Beneficial

¹ higher values correspond to higher criticality; ² higher values correspond to lower criticality.

).

Since the indicators are expressed in different units, normalization procedures were applied to ensure uniformity and comparability. Indicators (1)–(7) were used to construct a Socio-economic Index, reflecting the port’s socio-economic significance, while indicators (8)–(10) informed an Exposure Index, reflecting the climatic stressors. The entire set was then integrated into the Criticality Index, which formed the basis for the port’s final ranking.

3.2. Socio-Economic Indicators

The spatial distribution of the socio-economic indicators reflects both the geography of Greece and the functional characteristics of its ports. The indicator port area/size (Figure 2) shows a concentration of large-scale facilities in the mainland and few major islands, where port infrastructure has historically expanded to support both passenger flows and cargo shipping. By comparison, smaller islands tend to exhibit ports of limited area, reflecting their lower traffic. The population within a 500 m radius of the ports displays a complementary pattern, with dense concentrations around mainland hubs such as Piraeus, Thessaloniki, and other large urban coastal centers, while many smaller islands show lower values (Figure 3a). This relationship also manifests the link between urban growth and port activity.

The connected road network length within 1000 m further amplifies this mainland–island divide (Figure 3b). Ports located in the mainland benefit from extensive road connectivity, which strengthens their integration into broader transport systems. Island ports, particularly those on smaller islands, show sparse or even minimal land connectivity, underscoring their reliance on maritime transport. Passenger throughput (Figure 3c) mirrors these patterns, with exceptionally high values at the mainland’s major gateways and few island hubs and/or large touristic destinations (e.g., Heraklion, Rhodes) and limited flows at small island ports. It is noted that this indicator was selected instead of the passenger/cargo throughput, due to the lack of accurate cargo information for the large majority of Greek ports. Nevertheless, passenger flows in most island ports could also serve as a proxy for freight activity, as most cargo reaches islands via multipurpose ferry services, underscoring their dual logistical and social importance. Moreover, passenger throughput is also used as a proxy for social cohesion, since the majority of Greek ports are located on islands and, in many cases, ferry connections represent the only infrastructure facilitating transport to and from mainland Greece.

Port connectivity/centrality (Figure 3d) highlights the strategic importance within the maritime and transport network [101,102]. Ports with high connectivity/centrality (e.g., Piraeus, Thessaloniki) act as major nodes for mobility, trade and logistics. Conversely, limited inland connectivity (and reduced service frequency) can constrain a port’s integration into the network and undermine its criticality within the network [103,104].

Urban development within a 100 m radius (Figure 3e) follows a similar pattern to population. Densely built-up areas around the mainland and large island ports greatly exceed the modest development clusters of the smaller islands. Finally, redundancy (Figure 3f), i.e., a metric of available alternatives, provides a distinct layer of differentiation. Mainland ports achieve the highest values because of the several alternatives available. Larger islands with several ports also show good metrics, but as island size and port numbers diminish, redundancy declines, culminating in single-port islands where accessibility becomes very fragile.

When comparing across indicators, there are clear overlaps: large port areas, high population densities, extensive road connections, and large passenger throughputs consistently reinforce one another in the major mainland hubs and a few large island ports. Likewise, low values in population, connectivity, and redundancy feature in smaller islands. However, there are some exceptions. For example, certain island ports (e.g., Paros, Thira) may have small port areas and low populations, but relatively higher passenger throughputs and connectivity due to their location as touristic destinations and ferry-boat hubs, which can elevate their criticality ranking.

Piraeus port appears to dominate nearly every metric. It has the largest port area (815,000 m²), the highest population within a 500 m radius (44,473 residents), the longest connected road network within a 1000 m radius (≈260 km), the highest throughput (7.9 million passengers), the greatest number of connections (62), and dense surrounding urban development (96%). With a redundancy score of 5, Piraeus is the central mainland hub. By comparison, small island ports show minimal values. For example, Agia Marina port (island of Aegina) has a small port area (5066 m²), a low surrounding permanent population (295), and only one connection, making it much less critical despite its importance in serving local needs.

Generally, Piraeus and a few other major mainland ports dominate all indicators. Large islands with multiple ports form an intermediate, relatively resilient group, where small, single-port islands show consistently low values and minimal redundancy, underscoring their vulnerability.

3.3. Port Exposure to Climatic Factors

3.3.1. Rising Sea Level Exposure

This assessment highlights the increasing port exposure to flooding and diminishing freeboard under the projected RSLR and the 100-year ESLs, with impacts intensifying toward the end of the century. Currently, nearly half of the ports (46%) already exhibit a freeboard of less than 1.5 m—the minimum requirement for commercial vessel operations—while 1% of ports fall below the threshold for fishing boats (0.5 m) (Figure 4a and

Table 2. Flood assessment. Differences between maximum port quay elevation and projected RSLR and ESLs₁₀₀ (excluding wave setup) under various dates and climate scenarios. The table reports the number and percentage of Greek ports with available freeboard below 1.5 m, 0.5 m, and 0.15 m—thresholds corresponding to minimum safety requirements for commercial vessels (1.5 m), fishing boats (0.5 m), and leisure crafts (0.15 m) as specified by port safety guidelines [57]. It also indicates the number of ports projected to experience flooding (freeboard < 0 m).

				Number (and Percentage) of Ports
Sea-Level Rise				Freeboard (Based on the Max. Quay Elevation)
	Year	RCP	(m)	<1.5 m	<0.5 m	<0.15 m	<0 m
RSLR	Baseline		0	63 (46%)	2 (1%)	0	0
	2050	4.5	0.13–0.15	94 (69%)	4 (3%)	1 (0.7%)	0
		8.5	0.17–0.21	94 (69%)	4 (3%)	1 (0.7%)	0
	2100	4.5	0.46–0.53	121 (89%)	26 (19%)	4 (3%)	2 (1.5%)
		8.5	0.74–0.84	129 (95%)	52 (38%)	19 (14%)	7 (5%)
ESL100	Baseline		0.42–0.81	125 (92%)	29 (21%)	7 (5%)	3 (2%)
	2050	4.5	0.55–0.92	129 (95%)	36 (26%)	13 (10%)	7 (5%)
		8.5	0.60–0.99	129 (95%)	45 (33%)	16 (12%)	9 (7%)
	2100	4.5	0.86–1.35	133 (98%)	80 (59%)	38 (28%)	28 (21%)
		8.5	1.16–1.56	134 (99%)	108 (79%)	67 (49%)	56 (41%)

). By mid-century (2050), however, nearly 70% of ports are projected to fall below the 1.5 m freeboard threshold due to RSLR, under both RCP4.5 and RCP8.5 scenarios (Figure 4b and Table 2). Then, only 3% and <1% of ports are expected to breach the thresholds for fishing boats (0.5 m) and leisure craft (0.15 m), respectively, with no quay inundation expected (Table 2).

By 2100, the risks escalate considerably. Under RCP4.5, almost 90% of ports are projected to fail the commercial vessel freeboard requirement, 19% the fishing vessel threshold, and 3% the leisure craft limit, while 1.5% are expected to also experience inundation. Under RCP8.5, these figures increase to 95%, 38%, 14%, and 5%, respectively, rendering infrastructure inoperative without adaptation (Table 2).

There are differences between mainland and island ports. On the mainland, 75% of the ports will be below the commercial threshold by 2050, rising to 94% by 2100 under RCP8.5, with nearly half of these ports being unsuitable even for fishing vessels and 6% facing inundation. Island ports follow a similar but slightly less severe future, with 67% below the commercial vessel freeboard threshold by 2050 and 95% by 2100 under RCP8.5.

The situation is more difficult when ESLs₁₀₀ is considered, with large implications for port operability and safety. Even under baseline conditions, with extreme sea levels (ESLs₁₀₀) ranging between 0.42 m and 0.81 m, 92% of ports already fail to meet the minimum freeboard threshold of 1.5 m. Furthermore, 21% fall below the 0.5 m threshold set for fishing boats, 5% fall below the 0.15 m threshold for leisure crafts, and 2% experience quay inundation (freeboard < 0 m) (Figure 4c and Table 2). These results show that many Greek ports operate today under conditions that are below the internationally accepted safety guidelines.

By mid-century, under RCP4.5, the proportion of ports with insufficient freeboard will increase slightly (95% < 1.5 m; 26% < 0.5 m; 10% < 0.15 m; 5% inundated). Under RCP8.5, the situation will worsen, with one-third of ports showing freeboards < 0.5 m, 12% < 0.15 m, and 7% experiencing quay overtopping (Figure 4d and Table 2). This trend will intensify by the end of the century, when the projected relative sea level rise could drive catastrophic increases in flood exposure. Under RCP4.5, the freeboard of nearly all ports (98%) will fall below 1.5 m and 59% will be left with freeboards of less than 0.5 m. Alarmingly, 21% of ports are expected to also experience inundation. Under RCP8.5, the projections are even more severe (99% < 1.5 m; 80% < 0.5 m; 49% < 0.15 m) since quay inundation/flooding will occur at 41% of ports during ESL₁₀₀ events (Table 2). Mainland ports exhibit slightly higher exposure. It is noted that in most ports, particularly in island settings, large ferries, which are projected to face problems due to decreasing freeboard, form the main mode of transport for passengers but also for cargo.

Overall, the results demonstrate that there are already widespread deficiencies in the available freeboard across the Greek port network, which will largely deteriorate under climate change. While quay/port inundation will remain relatively limited in the near term, by the end of the century, it emerges as a significant threat, particularly under the high-emission scenario. These findings highlight an urgent need for targeted adaptation measures to preserve the functionality and safety of port infrastructure under the present and, particularly, the future sea levels.

The above projections, particularly for the extreme events, might likely be underestimations as they do not consider wave and wind dynamics within port basins [10,59]. Accurate port-level assessments would require site-specific, high-resolution modeling that includes current and future wave dynamics, port geometry/bathymetry, and infrastructure vulnerabilities, an approach far more computationally intensive. Moreover, using a single elevation value (especially the maximum elevation) to represent quay/dock height might overlook more flood-prone areas, further contributing to flood exposure underestimations.

Finally, the available freeboard, under the projected RSLR, was selected as an indicator for the port ranking analysis, since it represents long-term and continuous impacts of sea-level rise, whereas the ESL₁₀₀ represents episodic/transient events. Therefore, this metric provides a consistent basis for evaluating the port operability and safety.

3.3.2. Heat Exposure

Climate projections for Greek ports indicate a consistent warming trend throughout the 21st century, with both mean and maximum temperatures increasing steadily across mainland and island locations. Compared to the historical period (1986–2005), mean annual temperatures are projected to rise by more than 2 °C by mid-century and up to 4 °C by the end of the century under the high-emission scenario. Extreme heat events will intensify markedly, with late-century maxima exceeding 47 °C under RCP8.5. In contrast, relative humidity will remain largely stable, with only minor decreases projected across regions and scenarios (

Table 3. Projected average annual number of days with heat conditions exceeding critical thresholds at Greek ports. Values are shown for the threshold for potential harmful impacts on transport infrastructure and operations (the maximum daily temperature exceeds 32 °C) and for the 95% probability threshold for lethal heat conditions (combination of temperature and humidity) for outdoor workers.

				Number (and Percentage) of Ports
				Average Days per Year of Deadly Heat
Year	RCP	Tmean (°C)	RH	>5	>10	>30	>50
Historical		12–19	70–83	0	0	0	0
2050	4.5	13–20	69–82	46 (34%)	12 (9%)	1 (0.7%)	0
	8.5	14–21	68–82	67 (49%)	20 (15%)	1 (0.7%)	0
2100	4.5	14–21	69–82	79 (58%)	64 (47%)	3 (2.2%)	0
	8.5	16–23	68–82	127 (93%)	115 (85%)	77 (57%)	27 (20%)
				Number (and Percentage) of Ports
				Average Days per Year of Tmax ≥ 32 °C
Year	RCP	Tmax (°C)		>5	>20	>40	>70
Historical		28–43		48 (35%)	25 (18%)	11 (8%)	0
2050	4.5	29–44		69 (51%)	47 (35%)	30 (22%)	6 (4%)
	8.5	30–46		74 (54%)	49 (36%)	32 (24%)	9 (7%)
2100	4.5	30–45		90 (66%)	59 (43%)	36 (26%)	13 (10%)
	8.5	31–48		114 (84%)	84 (62%)	54 (40%)	24 (18%)

). Island ports consistently are projected to exhibit higher mean daily temperatures due to their maritime climate and reduced diurnal variability, whereas mainland ports will experience higher maximum daily temperatures.

The analysis of heat-related thresholds for Greece shows a substantial increase in exposure to extreme thermal conditions under future climate scenarios, with potentially severe implications for human safety and infrastructure functionality. Historically, the occurrence of deadly heat conditions, defined as temperature–humidity combinations exceeding the 95% probability threshold for (outdoor) lethal outcomes [10,42], has been negligible across all ports (Figure 5a and Table 3). However, projections indicate that such conditions will become increasingly common over the coming decades. By mid-century, under RCP4.5, more than 34% of the ports (16% of mainland and 39% of island ports) are projected to experience at least five days of deadly heat annually, with nearly 10% facing more than ten days. Under RCP8.5, the exposure will be more pronounced, with nearly half of ports (49%, 19% of mainland and 59% of island ports) exceeding five days and 15% exceeding ten days per year (Figure 5b and Table 3). By the end of the century, the situation will become severe. Under RCP4.5, almost 60% of ports (31% of mainland and 66% of island ports) will face lethal conditions for five days annually. Under RCP8.5, nearly all ports (93%, 84% of mainland and 96% of island) will experience such conditions, with more than half (57%, 28% of mainland and 65% of island) facing lethal conditions for over 30 days and 20% for more than 50 days (Table 3). A differentiation is also evident: island ports, with persistently higher mean temperatures and humidity, are more exposed to deadly heat conditions, while mainland ports, with higher maxima, face greater risks from acute short-term extremes.

Exposure to high maximum daily temperatures (Tmax ≥ 32 °C)—a recognized threshold for transport infrastructure thermal stress—will also rise significantly. Historically, 35% of ports have recorded more than five days per year above this threshold, but none experienced more than 70 days (Figure 5c and Table 3). By mid-century, under RCP4.5, over half of ports (51%) are projected to experience such thermal stress for more than five days annually, 35% more than 20 days, and few ports (4%) more than 70 days; the RCP8.5 pathway produces even higher projections (Figure 5d and Table 3). By the end of the century, projections become much more severe. Under RCP4.5, the infrastructure of 66% of the ports will face thermal stress for more than five days annually, and 26% more than 40 days. Under RCP8.5, the projections show that 84% of the ports (97% of mainland and 80% of island ports) will face thermal stress for more than five days annually, 62% (94% of mainland and 52% of island ports) for more than 20 days, 40% (75% of mainland and 29% of island ports) for more than 40 days, whereas 18% (41% of mainland and 11% of island ports) will be exposed to infrastructure thermal stress (Tmax ≥ 32 °C) for more than 70 days annually (Table 3).

Together, these projections highlight the high exposure of Greek ports to the rising temperatures: increasing frequency of extreme temperature days that undermine infrastructure resilience, and the growing incidence of deadly heat conditions that endanger port personnel and passengers, as well as the surrounding populations. It appears that there is an urgent need for climate adaptation strategies that can safeguard both port operations and human health.

3.4. Port Criticality Ranking

Weights were derived using the Analytic Hierarchy Process (AHP), based on the judgments of 20 port stakeholders and experts, which were compiled into pairwise comparison matrices, and their consistency was tested (Section 2.4.1). Separate weights were calculated for socio-economic indicators, exposure indicators, and for all indicators combined. The analysis showed that the socio-economic factors (weight: 0.694) were prioritized over exposure factors (0.306) in determining port criticality. Among socio-economic indicators, redundancy, port area (0.1226), and passenger throughput were weighted higher, while urban development and nearby population received lower weights. In the exposure category, sea-level exposure received the highest weight, highlighting stakeholder concern over the impacts of sea-level rise and port flooding compared to the future heat exposure (

Table 4. The AHP weights assigned to the indicators. The table presents the local weights, which represent both the relative importance of each sub-category and the relative importance of individual indicators within their respective sub-categories (e.g., “Socio-economic” and “Exposure”). It also includes the global weights, which reflect the overall contribution of each indicator to the main objective of the analysis—the composite index of port criticality.

Sub Categories	Local Weights	Indicators	Local Weights	Global Weights
Socio- economic	0.6940	1. Port area (m2)	0.1767	0.1227
		2. Population (500 m radius)	0.1123	0.0779
		3. Road network length (1000 m radius)	0.1427	0.0991
		4. Passenger throughput	0.1539	0.1068
		5. Port connectivity	0.1418	0.0984
		6. Urban development (100 m radius)	0.0827	0.0574
		7. Redundancy	0.1899	0.1318
Exposure	0.3060	8. Flood exposure	0.4806	0.1470
		9. Deadly heat exposure	0.2322	0.0710
		10. Heat risk to infrastructure	0.2872	0.0879

).

The Socio-economic Index (SEI) showed the highest values at major mainland and large-island ports with dense urban surroundings and extensive transport integration (Figure 6a and Table S2). The Piraeus cluster (Perama (ID 92), Piraeus (ID 1)) and Thessaloniki (ID 44) ports dominated, reflecting their high surrounding populations, infrastructure size, and traffic flows. Conversely, ports on smaller islands generally scored at the lower end of the index.

The Exposure Index (EI), comprising the indicators of sea level and heat exposure, showed a different pattern (Figure 6b,c and Table S3). The highest exposure scores were found in island ports, with the southeastern Aegean ports of Rhodes (ID 108), Kos (ID 66), Leros (ID 67) and Symi (ID 121) found as highly exposed to projected sea-level rise and extreme temperature conditions. Notably, exposure indexes appeared to increase throughout the century under both scenarios, with certain small-island ports (e.g., Pisaetos, Ithaca (ID 98), Mesta, Chios (ID 80) and Agios Efstratios (ID 7)) moving into the most exposed group.

The Criticality Index (CRI) according to the TOPSIS integrated both socio-economic importance and environmental stressors (Figure 6d,e and Table S4). Perama, Piraeus, and Thessaloniki ports consistently occupied the top ranks, reflecting their very high socio-economic significance. Paloukia (Salamina) (ID 94) and Rhodes (ID 108) also scored high due to their combination of significant functionalities with elevated exposure. Ports such as Volos (ID 25), Heraklion (ID 41), Patra (ID 91), and Corfu (ID 54) formed a second tier, whereas most small-island ports remained at the bottom of the criticality ranking. Nevertheless, small ports with stronger network linkages, such as Paros (ID 89), demonstrated higher criticality mostly due to their increased port connectivity. It appears that connectivity can function as both a resilience amplifier and a potential vulnerability factor, depending on the extent of redundancy and the integration available within the port network. At the same time, in islands depending on a single maritime gateway, port connectivity becomes essential not only for their socio-economic development, but also for their resilience under adverse climatic or operational conditions (e.g., [8]).

The second stage of the prioritization (PROMETHEE II method) refined the ranking of the 15 ports identified as most critical in the first-stage (TOPSIS) screening (Figure 6). Here, the AHP weights were also integrated, ensuring that expert judgments were also incorporated into the final port criticality ranking.

The results revealed a persistent ranking across all climate scenarios (Figure 7,

Table 5. Ranking scores (PROMETHEE II analysis) of the 15 selected ports, according to their socio-economic significance and climatic hazard exposure.

Port Name	Current	RCP4.5, 2050	RCP8.5, 2050	RCP4.5, 2100	RCP8.5, 2100
1-Piraeus	0.384	0.376	0.375	0.371	0.363
25-Volos	−0.108	−0.105	−0.106	−0.109	−0.122
41-Heraklion	−0.018	−0.011	−0.007	−0.011	−0.012
43-Thira	−0.090	−0.093	−0.092	−0.088	−0.077
44-Thessaloniki	0.059	0.053	0.048	0.048	0.039
54-Corfu	0.023	0.017	0.017	0.022	0.025
56-Kavala	−0.061	−0.055	−0.058	−0.059	−0.074
66-Kos	−0.078	−0.056	−0.052	−0.048	−0.039
68-Lavrio	−0.117	−0.115	−0.118	−0.117	−0.114
81-Mytilene	−0.059	−0.043	−0.037	−0.027	−0.015
89-Paros	−0.077	−0.084	−0.084	−0.083	−0.073
91-Patra	−0.163	−0.159	−0.158	−0.157	−0.147
92-Perama	0.203	0.194	0.193	0.188	0.180
94-Paloukia Salamina	−0.026	−0.040	−0.041	−0.046	−0.050
108-Rhodes	0.129	0.121	0.121	0.117	0.118

and Table S5). Piraeus (ID: 1) emerged as the most critical port, both under the baseline and the future climatic conditions, reflecting the combined effect of its socio-economic significance (large port area, high throughput, and dense urban integration) with a rising exposure to climatic stressors. Perama (ID: 92), the other port of the Piraeus port cluster, followed in second place under all scenarios. Rhodes (ID: 108) consistently occupied the third place (CRI between 0.129 and 0.118), with its criticality shaped by its high exposure to climatic hazards as well as by its role as a southeastern Aegean port hub. Thessaloniki (ID: 44) ranked fourth: its high CRI was due primarily to its socio-economic significance, as it showed lower exposure under CV&C compared to the list’s island ports. Corfu port (ID: 54) showed modest socio-economic significance, but growing exposure to climate change. Heraklion port (ID: 41) showed a balance between socio-economic significance and manageable climatic stress, whereas Mytilene (ID: 81) port’s CRI reflected more its CV&C exposure than its socio-economic significance. Finally, the lowest positions were consistently held by Kos (ID: 66), Kavala (ID: 56), Paros (ID: 89), Thira (ID: 43), Volos (ID: 25), Lavrio (ID: 68), and Patra (ID: 91), all of which registered negative scores across the tested CV&C scenarios (Figure 7, Table 5 and Table S5), demonstrating limited capacity to outrank the other selected ports under pairwise evaluation.

Generally, the PROMETHEE II results confirmed the stability of rankings across CV&C scenarios, with only minor shifts in relative scores. By integrating expert-based weighting in this MCDM technique, the second stage evaluation provided a robust criticality analysis framework.

4. Discussion

The study introduced a national/regional scale framework for ranking the criticality of ports across the extensive Greek port network, based on a combination of indicators that reflect the ports’ socio-economic significance as well as their exposure to CV&C. Unlike vulnerability and resilience assessments which focus on exposure, sensitivity, and adaptive capacity, this analysis emphasizes the concept of criticality, that is, the relative importance of each port for maintaining the functionality of the network. Hence, while vulnerability and resilience indicate the propensity to suffer damage and the capacity to recover [25,105,106,107,108], criticality assesses which elements are most essential to the system operation [32,37] and, thus, provides evidence for prioritizing adaptation investment where disruptions might have the highest consequences.

In the Greek port networks, the socio-economic indicators used revealed a geographical control. Mainland ports were found to consistently dominate across all (7) indicators, due to their larger facilities, high passenger traffic, and better integration into the national transport network. They also showed much higher redundancy due to their multiple alternatives, which enhances systemic resilience under localized disruptions [14]. In contrast, most island ports scored low in all indicators, particularly in redundancy (Figure 3f). Although this was expected given their geographical particulars, it still shows that island ports, especially the smaller ones, face significant challenges from disruptions. Resilience requires both strong external linkages and large inland connections, which are indispensable for network efficiency (e.g., [109]). Nevertheless, in some cases (e.g., Paros, ID 89), enhanced port connectivity enables certain island ports to function as regional hubs and increase their role as important network nodes [101].

The analysis of climatic exposure of the 32 mainland and 104 island ports revealed increasing exposure from both rising sea levels and extreme heat. Nearly half of the tested ports were found to already operate below the 1.5 m freeboard limit required for large commercial vessels. The situation will deteriorate markedly under CV&C: by mid-century, the freeboard of nearly 70% of Greek ports is projected to fall permanently below the 1.5 m threshold due to RSLR; whereas by 2100, up to 95% of the ports will fail this standard and about 40% will have difficulties serving even medium-sized fishing boats (requisite freeboard of 0.5 m). These projections point to major challenges for individual ports and the port network as a whole, the facing of which will require the design/implementation of effective engineering solutions as well as very large investments [110,111,112]. In addition to mean sea level rise, extreme sea level events will also pose severe risks. By mid-century, 5–7% of Greek ports are projected to face inundation under the ESL₁₀₀, whereas by 2100 21–41% of ports will be flooded depending on the scenario. These projections are also alarming. Although port flooding under extreme events is a transient incident, damages/losses might be very severe indeed [2,12,41,113]. It should be noted here that the above flood exposure projections should be regarded as conservative: not only were the estimations based on the maximum quay elevations and, as such, most probably underestimate the overall flood exposure, but also wave dynamics were excluded in the estimation of the port extreme sea levels (e.g., [59]). Moreover, due to the large scope of the exercise, estimates of port inundation were based on a simple, static comparison of the sea level with the maximum quay elevation and not on port-specific dynamic simulations, an approach that tends to underestimate port flooding (e.g., [10]).

Heat exposure is also expected to worsen markedly. By mid-century, half of all ports will experience at least 5 deadly heat days annually, whereas by the end of the century, almost all ports (93%) are projected to be affected, with 20% of them facing more than 50 such days per year. Infrastructure is also projected to be impacted by maximum temperatures of more than 32 °C, affecting up to 84% of the ports, especially on the mainland; island ports will also face higher humidity and prolonged heat stress. Without targeted adaptation, these combined hazards could severely strain port infrastructure integrity, as well as port personnel/user health and safety, and thus undermine port functionality.

In terms of port criticality, our analysis showed that the socio-economic factors (weight: 0.694) were prioritized over exposure factors (0.306) in determining port criticality. Thus, as expected, the large mainland ports (Piraeus ports, Thessaloniki) emerged as the most critical ports for the network, due mainly to their dominant socio-economic significance; Piraeus ports have been previously identified as key nodes in the European port network, the disruption of which might cause large declines in European transport network efficiency and cohesion [114]. Smaller island ports generally scored low. However, despite their low scores, their minimal redundancy makes them critical for local communities. Port disruptions on islands with a single port would sever vital connections for transport, communication, and supply chains, affecting local economies and disaster response capacities. This discrepancy—i.e., low port network criticality but high local significance—highlights the significant adaptation challenges for island ports.

These findings contribute to a broader understanding of port criticality. As noted previously [37], criticality reflects a port’s role within the maritime network and its contribution to trade and economic continuity. The large mainland ports of Greece are critical at both national and transnational scales, sustaining trade flows across the country and the wider region. Island ports, despite their mostly limited scale, are locally indispensable, maintaining insular economies and social cohesion. Recognizing this dichotomy between the network and local criticality is vital for prioritizing adaptation investment; differentiated strategies are required that can enhance both overall network strength and local resilience. For mainland ports, priority should be given to engineering measures (e.g., quay elevation upgrades, drainage enhancements, and thermal protection for critical infrastructure and the personnel/passengers) (e.g., [115,116]), whereas for island ports enhancing redundancy through secondary berths, inter-island linkages, and flexible vessel scheduling should be planned/implemented in addition to requisite engineering measures (e.g., [8]).

The regulatory framework for climate change adaptation can assist towards these objectives [2]. Key international strategies/policies, such as the 2030 Agenda for Sustainable Development Goals (SDGs)— particularly Goals 1.5 (building resilience of vulnerable communities), 9 (developing resilient infrastructure and fostering innovation), and 13 (strengthening climate action)—and the Sendai Framework for Disaster Risk Reduction 2015–2030-SFDRR [117], and legal instruments (e.g., the 1992 United Nations Framework Convention on Climate Change—UNFCCC [118] are of particular relevance in this context, as they can help create effective port adaptation plans to reduce exposure and vulnerability to flooding and other CV&C hazards (e.g., [119]).

At the EU level, relevant policies include: the 2021 EU Climate Change Adaptation Strategy [120] that aims to ensure a climate-resilient EU by 2050, and the EU Action Plan on SFDRR 2015–2030 [121] which envisages specific actions and measures to reduce disaster risk. There are also several EU legal instruments which explicitly, or implicitly, address CV&C port adaptation, resilience-building and DRR (Disaster Risk Reduction). The Floods Directive [122] imposes a general duty on EU Member States to assess and map the coastal flood risk, affected areas, assets and humans at risk and take the necessary management measures. The amended EIA (Environmental Impact Assessment) Directive [123] envisages that for all (public and private) projects falling within its scope, including ports, EIAs should be carried out, considering risks and vulnerabilities associated with CV&C. Several recent legislative developments are also of special relevance to CV&C adaptation for ports, including: the Climate Law Regulation [124] which mandates strong action on adaptation and resilience-building (Art. 5); the Union Civil Protection Mechanism Regulation (EU) 2021/836; and the TEN-T Regulation (EU) 2024/1679) [125] that prescribes climate proofing of infrastructure included in the development of the trans-European transport network, to which several Greek ports are part of (see Section 3.1).

The EU Directive on Resilience of Critical Entities [126] is also most relevant. The Directive, which had to be transposed into the national legislation of EU Member States by 17 October 2024, aims to ensure the resilience of public and private entities that provide services essential to the maintenance of functions vital to society, the economy, the environment, and the public health/safety, (‘critical entities’) for which an incident would have significant disruptive effects. EU Member States must identify critical entities by 17 July 2026, while entities must conduct their own risk assessments to recognize risks that could disrupt their essential services and take technical and organizational measures to enhance their resilience; port facilities are explicitly referred to in the Annex of the Directive (Sector Transport, subsector Water). Furthermore, Member States shall consider vulnerabilities linked to the isolation of certain areas such as islands (Article 7), and the importance of each entity in maintaining an adequate level of the essential service, also taking into account the availability of alternative means for its provision. These criteria clearly indicate that all Greek ports should be considered as critical entities, yet the redundancy deficit of the island ports requires focused attention. A main challenge lies in the implementation: the large number of ports requiring adaptation actions will likely strain the administrative and technical capacity of the State as well as its financial resources. Moreover, effective port adaptation requires interdisciplinary collaboration among engineers, coastal scientists, and planners, as well as continued innovation in design and management practices [13].

The results of the present study can be useful inputs for Port Authorities, Municipal Port Funds, and Government Institutions responsible for port governance and planning. The Greek port system is characterized by multi-layered governance structures, similar to other Mediterranean and European systems, where responsibilities and resources are unevenly distributed across port tiers [127]. The major ports (e.g., Piraeus ports and Thessaloniki) possess the financial and organizational capacity to undertake substantial adaptation measures, including capital-intensive infrastructure upgrades and operational reforms. In comparison, smaller ports typically operate with limited financial autonomy and may require targeted public support to address climate-related risks [90]. The smallest ports, most of which are located on islands, face the most significant constraints: they have minimal revenue streams and limited institutional capacity, despite being critical for local economies, tourism, social cohesion, and emergency response, consistent with findings for other island port systems globally [13]. Therefore, the Greek State should prioritize dedicated funding schemes for these ports, such as resilience funds or climate-adaptation grants, in line with EU recommendations for supporting vulnerable critical infrastructures [120]. Finally, as the Greek port industry is under ongoing privatization and concession processes, our results can provide insights regarding the scale and urgency of future capital needs for climate adaptation, a point increasingly emphasized in global port investment analyses [6].

Finally, it is noted that the study, which focused on criticality rather than vulnerability, can provide an evidence-based tool to support investment prioritization in the face of escalating climate-related hazards. The proposed framework enables a systematic ranking of ports in maritime networks, thereby guiding the assessment of adaptation needs and urgency. Future iterations should incorporate additional socio-economic metrics and climatic hazard indicators as these become available, including governance and policy dimensions, as well as processes to evaluate uncertainties (e.g., [36,128]).

5. Conclusions

The present work provides the first national-scale assessment of port criticality in Greece under climate variability and change. By integrating socio-economic indicators with climatic exposure metrics, it establishes a systematic and replicable method for identifying and prioritizing critical ports.

The analysis showed a large and increasing climatic exposure of all Greek ports. It also confirmed that the large mainland ports constitute primary nodes of the Greek port system, whereas major island ports are also functionally significant within their regional contexts. Nevertheless, small island ports, although scoring low in the network criticality, are both fragile and greatly important at the local level, due to their low redundancy.

It is noted that the analysis emphasized criticality—the importance of ports for network functionality and socio-economic continuity—rather than vulnerability, which reflects susceptibility and adaptive capacity. This distinction enables prioritization of ports whose disruption would most affect national and regional resilience. Addressing vulnerability concerns will require detailed studies at the port level, integrating adaptive capacity, institutional readiness, and detailed hazard and exposure modeling.

Coordinated national/regional action is required to enhance the resilience of port networks, which can integrate socio-economic significance and dependencies, climatic hazard and exposure projections, and financial and governance capacity. The framework proposed in this study provides a first attempt for a decision-support tool to guide such efforts which will contribute to the sustainable transformation of the maritime transport system under a changing climate.