Risk Assessment of Dams and Reservoirs to Climate Change in the Mediterranean Region: The Case of Almopeos Dam in Northern Greece

Abstract

Climate change poses significant challenges to the operation and safety of dam and reservoir (D&R) systems, particularly in regions characterized by water scarcity and high climate variability. This study presents a structured methodology for climate risk assessment that integrates regional climate projections, system-specific thresholds, and a semi-quantitative risk matrix approach. A key innovation is the explicit linkage between climate indicators and system performance through physically based thresholds, combined with empirically derived exceedance probabilities from high-resolution climate projections. The methodology is applied to the Almopeos D&R system in northern Greece, using an ensemble of statistically downscaled CMIP6 simulations under two emission scenarios (SSP2-4.5 and SSP5-8.5) and two future periods (2041–2060 and 2081–2100). Three climate indicators are analyzed: TX35 (temperature extremes), CDD (consecutive dry days), and Rx1day (extreme precipitation). Results indicate that temperature increase is the dominant climate risk hazard, leading to increased irrigation demand and reduced system reliability, with risks classified as high to very high. Drought conditions represent a secondary but important risk, becoming critical during prolonged dry periods affecting reservoir storage, while extreme precipitation events exhibit low likelihood but potentially high consequences for dam safety. Adaptation measures are prioritized using a qualitative multi-criteria approach, highlighting the effectiveness of operational measures, while structural and monitoring interventions remain essential for ensuring system safety. The proposed methodology provides a transparent and transferable framework for climate-resilient planning of water infrastructure systems.

1. Introduction

Dam and reservoir (D&R) systems play a critical role in water resources management by supporting hydropower production, irrigation, flood control, water supply, and ecosystem regulation, particularly in regions with pronounced hydroclimatic variability such as the Mediterranean [1]. Despite their importance, numerous dam failures have occurred worldwide, sometimes resulting in severe human, environmental, and economic losses [2,3]. Regardless of the underlying causes of failure, such as structural, hydrological, geotechnical, or operational factors, the consequences are often catastrophic, including loss of life, environmental degradation, and major economic damage [4,5]. A recent example highlighting the vulnerability of D&R systems under extreme climatic conditions is storm Daniel, which occurred on 10–11 September 2023 [6,7]. The event produced unprecedented precipitation over northeastern Libya, triggering catastrophic flash floods that led to the failure of the Bu Mansour and Al-Bilad dams and severe impacts in the coastal city of Derna [8,9]. Therefore, effective risk management and analysis methodologies are essential for ensuring dam safety and supporting informed decision-making throughout the life cycle of D&R systems [10,11].

Risk assessment approaches applied to D&R systems are generally classified into qualitative, semi-quantitative, and quantitative methods according to their level of complexity and data requirements [12]. Among these, semi-quantitative approaches are widely used in engineering practice, because they provide a balance between methodological rigor and practical applicability when detailed probabilistic data are unavailable.

One of the most commonly applied tools in semi-quantitative risk analysis is the risk matrix approach. Risk matrices are widely used in dam safety assessments, particularly for preliminary or screening-level analyses, due to their transparency, simplicity, and ability to incorporate expert judgement. They are included in several dam safety guidelines and national risk management frameworks [13,14,15] and support decision-making processes by helping to prioritize further investigations and mitigation measures. Applications of risk matrices in dam safety analyses are reported in studies such as Sohler et al. [16] and Xie et al. [17], while similar approaches have been applied to other water infrastructure systems, including water distribution networks [18]. Lane and Hrudey [19] provide a detailed guide for the development and application of risk matrices in water infrastructure management.

Several additional methodologies have been applied to dam safety and infrastructure risk analysis. Failure Mode and Effects Analysis (FMEA) and its extensions, such as PFMA and FMECA, are widely used for the systematic identification and evaluation of failure mechanisms [20,21,22]. Event Tree Analysis (ETA) has been applied to assess the progression of failure scenarios and their consequences [23,24,25]. Fault Tree Analysis (FTA) is commonly used to analyse the logical relationships between system failures and their causes [26,27,28]. Probabilistic approaches, such as Monte Carlo simulation, have been used to quantify uncertainties in dam performance and failure processes [29,30]. Bayesian networks provide a flexible framework for modelling dependencies and updating risk estimates under uncertainty [31,32,33]. In addition, multi-criteria decision analysis methods, including the Analytic Hierarchy Process (AHP), have been applied to support decision-making and prioritization of risk mitigation measures [34,35,36,37]. These approaches provide more detailed probabilistic or system-based analyses but often require extensive data and computational effort.

Climate change introduces additional challenges for the design, operation, and safety of D&R systems, as it affects temperature, precipitation patterns, and the frequency and magnitude of extreme hydrological events. The Mediterranean region is widely recognized as one of the most vulnerable areas to climate change, often referred to as a climate change hotspot [38], where temperature increases exceed the global average and are accompanied by a reduction in precipitation and an increase in the frequency and intensity of extreme events [39,40]. These changes are expected to affect significantly hydrological regimes, leading to increased water scarcity, prolonged drought periods, and higher variability in water availability.

In this context, D&R systems play a critical role in ensuring water security, particularly for irrigation, which represents the dominant water use in many Mediterranean regions. However, the performance and safety of these systems are increasingly challenged by climate-driven pressures, including increased evaporation losses, higher water demand, reduced inflows, and the potential for extreme hydrological events. Overtopping and internal erosion/seepage are widely recognized as among the most important dam-failure mechanisms, especially for embankment dams [3,41], while Mediterranean water systems are increasingly exposed to prolonged droughts, reduced water availability, and irrigation-related water stress under climate change [40,42,43]. Existing knowledge on dam failures is largely derived from aggregated global datasets and international inventories [44,45,46], which consistently identify overtopping as the dominant failure mode (approximately 30–40% of cases), followed by internal erosion and seepage processes (20–30%). While these datasets provide valuable insights into general failure mechanisms, they are inherently heterogeneous and primarily suited for generalized or screening-level analyses, offering limited resolution for region-specific assessments [11,47].

Under these conditions, the dominant failure modes of D&R systems are strongly influenced by region-specific hydroclimatic processes. Extreme precipitation events and flash floods, which exhibit high spatial variability in Mediterranean catchments, increase the likelihood of overtopping and spillway exceedance [48,49]. Conversely, prolonged droughts and declining inflows promote desiccation and cracking of embankment materials, leading to increased seepage and piping susceptibility, particularly in earthfill dams [47,50]. In addition, drought-induced water shortages represent a critical operational failure mode, especially in irrigation-dominated systems, where the combined effects of reduced water availability and increased demand can lead to significant service disruption and economic losses [51], especially in Mediterranean regions where irrigation demand increases during heatwaves and reduces reservoir reliability [52,53]. These characteristics make Mediterranean D&R systems particularly vulnerable to climate change and highlight the need for integrated risk assessment and adaptation planning approaches tailored to regional conditions.

Recent studies have also highlighted the importance of evaluating the performance of CMIP6 simulations for temperature and precipitation before their use in climate-impact assessments, as well as the added value of bias-adjusted and statistically downscaled datasets for regional and local applications. In particular, Varotsos et al. [54] demonstrated the value of high-resolution gridded climate data for Greece, while Daher and Kirtman [55] and Enyew et al. [56] emphasized the importance of evaluating and downscaling CMIP6 projections for robust regional hydroclimatic assessments. Therefore, case studies from Mediterranean environments provide valuable insights into the development of transferable methodologies for climate-resilient water infrastructure.

In response to these challenges, Stamou et al. [57] developed a Climate Risk and Vulnerability Assessment (CRVA) methodology within the CLIMPACT research project, based on an extensive literature review and the technical guidance of the European Commission for climate proofing of infrastructure [58]. The methodology consists of five steps implemented in two phases: a screening phase, including system description (step 1), climate change assessment (step 2), and vulnerability assessment (step 3), and a detailed analysis phase, including risk assessment (step 4), and the identification, appraisal, and prioritization of adaptation measures (step 5). The preferred measures should then be integrated into the project design and/or its operation to improve climate resilience [59]. Stamou et al. [60] applied the first three steps of this methodology to the Almopeos D&R system in northern Greece and identified three main groups of climate hazards affecting the system: (i) increasing temperatures and extreme heat, (ii) decreasing precipitation and drought conditions, and (iii) extreme precipitation and flooding events.

Despite the growing literature on climate-change impacts on water infrastructure, important gaps remain in the climate risk assessment of dams and reservoirs. In particular, many studies focus either on broad vulnerability screening without explicit linkage to system-specific performance thresholds, or on individual hazards without integrating consequence analysis, likelihood estimation, and adaptation prioritization within a single framework. This limitation is especially relevant in Mediterranean systems, where dam-safety hazards such as overtopping and seepage must be assessed together with drought stress, increasing irrigation demand, and reduced water availability [40,42,43]. Building on previous work on CRVA for D&R systems [57,60], the present study advances the methodology by explicitly linking climate indicators to system performance using system-specific thresholds. In particular, thresholds are derived based on irrigation demand, reservoir storage characteristics, and hydrological response, allowing a more physically meaningful interpretation of climate change impacts. Furthermore, the study introduces the use of empirically derived exceedance probabilities from high-resolution climate projections for the estimation of hazard likelihood. This approach enables a transparent and data-driven representation of climate uncertainty without relying on assumed statistical distributions. The proposed framework is applied to the Almopeos D&R system in northern Greece, with the objective of demonstrating how climate-informed risk assessment, combined with system-based thresholds and a semi-quantitative risk matrix, can support practical decision-making and climate-proofing of D&R systems in the Mediterranean region.

2. Materials and Methods

2.1. Methodology

The methodology adopted in this study follows the CRVA framework developed by Stamou et al. [57] within the CLIMPACT project and is consistent with the European Commission guidance on climate proofing of infrastructure [58]. The framework shown in Figure 1 comprises five steps implemented in two phases: (i) a screening phase, including system description, climate change assessment, and vulnerability assessment, and (ii) a detailed analysis phase, including risk assessment and identification of adaptation measures.

This work focuses on the detailed analysis (phase 2), which follows the screening phase (phase 1), in which the following groups of climate hazards were identified as potentially significant: (i) increasing temperatures and extreme heat, (ii) decreasing precipitation and drought conditions, and (iii) extreme precipitation and flooding events [60]. In the present detailed analysis, the impacts of these hazards are classified according to the dominant climate driver using the typologization of Stamou et al. [61] as (i) temperature-related impacts (TIMs), (ii) drought-related impacts (DIMs), and (iii) flood-related impacts (FIMs). These impacts are then analysed using an impact chain approach, which represents the causal sequence linking climate hazards to physical processes and their consequences on the performance of the D&R system.

Each impact chain therefore represents a structured pathway from climate forcing to infrastructure response and operational consequences. For each impact chain, consequences are evaluated across a set of risk areas relevant to D&R systems, whose identification is based on engineering judgement, available literature, and operational knowledge [10,13,14,58]. In this study, the following five risk areas are considered, that are consistent with EC [58]: asset damage (CA), safety and health (CH), environmental impacts (CE), service disruption and social impacts (CS), and financial and reputational impacts (CF & CR). These risk areas and indicative consequences are shown in

Table 1. Risk areas considered in the climate risk assessment of D&R systems.

Risk Area	Description	Indicative Consequences
Asset Damage (CA)	Damage to physical components of the dam and associated infrastructure.	Spillway overtopping, erosion of embankments, malfunction of outlet works.
Safety and Health (CH)	Impacts on human life and public safety caused by dam malfunction or failure.	Downstream flooding, emergency evacuations, injuries or fatalities.
Environmental Impacts (CE)	Adverse effects on aquatic and terrestrial ecosystems.	Habitat degradation, sediment transport changes, water quality deterioration.
Service Disruption & Social Impacts (CS)	Interruption of water services affecting communities or users.	Irrigation supply interruption, restrictions on water use.
Financial and reputational impacts (CF & CR).	Economic consequences related to infrastructure damage or service disruption.	Repair costs, loss of hydropower production, irrigation supply interruption.

.

In

Table 2. Impacts of mean air temperature increase and extreme heat, relevant impacts chains and affected risk areas.

Symbol	Climate Indicator	Impact	Impact Chain (Simplified)	CA	CH	CE	CS	CF & CR
TIM1	TXm, HD	Obscuring of monitoring sites due to algae growth	Increased water temperature → T-P1 increased algae growth → obstruction of monitoring sites → increased maintenance	X				X
TIM1	TXm, TR	Degradation of water quality due to increased water temperature	Increase in air temperature → T-I increase in river and reservoir water temperature → T-P1 increased biological activity and degraded water quality → reduced suitability of water for irrigation (T-O1) → environmental impacts (CE) and increased monitoring or treatment costs (CA, CF & CR)	X		X		X
TIM2	TXm, HD	Reduction in reservoir storage due to increased evaporation	Higher temperature and heat waves → T-P2 increased evaporation from the reservoir surface → reduction in effective reservoir storage → reduced water availability for irrigation (T-O1) → service disruption (CS) and economic losses (CF & CR)	X			X	X
TIM3	TXm, TX35	Increased irrigation water demand during heat waves	Increased temperature and evapotranspiration → T-O1 increased irrigation water demand → higher withdrawals from the reservoir → reduced reliability of irrigation supply (CS) → financial losses in agriculture (CF & CR)				X	X
TIM4	TXm, TX35, TR	Desiccation and cracking of embankment materials	Prolonged heat and drought conditions → T-A1 desiccation and shrinkage of clayey materials in the embankment → formation of cracks and increased seepage susceptibility → increased inspection and maintenance requirements (CA) and higher repair costs (CF & CR)	X				X
TIM5	TXm, HD	Thermal deterioration of spillway and auxiliary structures	High temperature and solar radiation → T-A2 thermal expansion and cracking of spillway concrete structures and T-A3 deformation of metallic auxiliary components → reduced structural reliability → increased maintenance and repair needs (CA, CF & CR)	X				X
TIM6	HD, TX35, TR	More difficult working conditions for personnel	Heat waves and tropical nights → T-S4 increased thermal stress for personnel → difficult outdoor working conditions and reduced operational efficiency → occupational health risks (CH) and operational disruptions (CS)		X		X	X

,

Table 3. Impacts of decreased mean precipitation, aridity and droughts, relevant impacts chains and affected risk areas.

Symbol	Climate Indicator	Impact	Impact Chain	CA	CH	CE	CS	CF & CR
DIM1	PRCPTOT	Reduced reservoir storage due to reduced inflows	Reduced precipitation → D-I reduced inflows to the reservoir → D-P1 reduced reservoir volumes and water levels → reduced water supply potential for irrigation (D-O1) → service disruption (CS) and economic losses (CF & CR)	X			X	X
DIM2	PRCPTOT, CDD	Degradation of water quality due to low reservoir levels	Reduced inflows and prolonged dry periods → D-P1 reduced reservoir volumes → increased concentration of pollutants and degraded water quality → additional monitoring or treatment required (D-O1) → environmental impacts (CE) and increased operational costs (CA, CF & CR)	X		X		X
DIM3	PRCPTOT	Damage to exposed parts of the dam due to low water levels	Prolonged low reservoir levels → D-P1 exposure of upstream dam surfaces → D-A1 erosion or deterioration of exposed materials due to waves, temperature and UV radiation → increased inspection and maintenance requirements (CA) and higher repair costs (CF & CR)	X				X
DIM4	CDD	Desiccation and shrinkage of clay core causing seepage and piping	Prolonged drought conditions → D-A1 desiccation and shrinkage of clay core and embankment materials → cracking and increased seepage paths → risk of piping and internal erosion → potential structural instability (CA) and downstream impacts (CH, CE, CS, CF & CR)	X	X	X	X	X
DIM5	PRCPTOT, CDD	Instability or slumping of the upstream dam face	Repeated wetting and drying cycles associated with reservoir level fluctuations → D-A1 instability or slumping of upstream dam face → reduced structural reliability → increased maintenance and repair needs (CA) and higher operational costs (CF & CR)	X				X
DIM6	CDD	Increased irrigation demand during drought conditions	Drought and prolonged dry periods → D-O1 increased irrigation water demand → increased withdrawals and reduced reliability of water supply (CS) → economic losses in agriculture (CF & CR)				X	X

and

Table 4. Impacts of extreme precipitation and floods, relevant impacts chains and affected risk areas.

Symbol	Climate Indicator	Impact	Impact Chain	CA	CH	CE	CS	CF & CR
FIM1	Rx1day	Overflow and flooding risk	Extreme precipitation events → F-I increased inflows to the reservoir → F-P1 rapid increase in reservoir water levels → F-P2 overflow and increased flooding risk → downstream impacts on population and environment (CH, CE, CS) and economic losses (CF & CR)	X	X	X	X	X
FIM2	Rx1day	Overtopping of the dam	Extreme inflow and rapid reservoir filling → F-P1 rapid rise in reservoir water level → F-A1 overtopping of the embankment dam → erosion and possible dam breach → severe downstream impacts (CH, CE, CS, CF & CR)	X	X	X	X	X
FIM3	Rx1day	Seepage and piping due to rapid water level rise	Rapid reservoir level rise during floods → F-P1 rapid water level fluctuations → F-A1 increased pore pressure and seepage within embankment → piping and internal erosion risk → potential dam failure (CA) with downstream impacts (CH, CE, CS, CF & CR)	X	X	X	X	X
FIM4	R20mm, Rx1day	Damage or malfunction of spillway structures	High inflow and discharge velocities → F-P1 increased flow through spillway system → F-A2 structural stress or deterioration of spillway components → reduced discharge capacity → increased maintenance needs (CA, CF & CR)	X				X
FIM5	R20mm, Rx1day	Sediment and debris transport	Heavy rainfall and runoff → F-I increased sediment loads and debris transport → F-P1 sediment accumulation and obstruction of hydraulic structures → damage or malfunction of components (CA) and environmental impacts (CE)	X		X		X
FIM6	R20mm	Degraded water quality due to sediments and turbidity	Intense rainfall and runoff → F-I increased turbidity and sediment inflow → F-P1 deterioration of water quality → need for additional monitoring or treatment (F-O1) → environmental impacts (CE) and operational costs (CF & CR)			X		X
FIM7	R20mm, Rx1day	Damage to auxiliary structures and equipment	Flood flows and debris → F-A3 damage to pipelines, valves, intake structures or monitoring equipment → reduced operational reliability (CA) → repair and maintenance costs (CF & CR)	X				X
FIM8	R20mm	Damage to access roads and site accessibility	Heavy rainfall and local flooding → F-S3 erosion or damage to access roads → reduced accessibility for inspection and maintenance (CS) → increased restoration costs (CF & CR)				X	X

, twenty representative climate–impact chains are presented using the typology of impacts defined in Stamou et al. [60]. These chains describe the main mechanisms through which (i) temperature-related impacts (TIM1–TIM6), (ii) drought-related impacts (DIM1–DIM6), and (iii) flood-related impacts (FIM1–FIM8) may affect the five risk areas of D&R systems, along with the relevant climate indicators used in the risk assessment. These indicators include maximum temperature (TXm in °C), hot days (HD; annual count of days with daily maximum temperature > 30 °C), tropical nights (TR; annual count of days with daily minimum temperature > 20 °C), annual count of days with daily maximum temperature > 35 °C (TX35), annual total precipitation on wet days (PRCPTOT in mm), consecutive dry days (CDD; maximum number of consecutive days with daily precipitation less than 1 mm in a year), annual count of days when precipitation is ≥ 20 mm (R20mm), and annual maximum one-day precipitation (Rx1day in mm) [60]. The information in Table 2, Table 3 and Table 4 provides the basis for the subsequent climate risk assessment, in which the likelihood of hazards and the magnitude of their consequences are evaluated using a semi-quantitative risk matrix approach.

2.2. Consequences Analysis

The assessment of climate risks for D&R systems requires a multi-dimensional evaluation of potential consequences, reflecting the wide range of impacts associated with infrastructure malfunction or failure. As shown in Table 1, Table 2, Table 3 and Table 4, these impacts extend beyond structural damage to include effects on human safety, the environment, service provision, and economic performance, in line with established dam safety frameworks [11,13,14] and the climate-proofing guidance of the European Commission [58].

In this study, consequences are assessed using the five-level scale of the climate-proofing framework of the EC [58], ranging from insignificant (score = 1) to catastrophic (score = 5), which is extended to include representative quantitative indicators for each risk area, for which thresholds are defined to enhance transparency and applicability to the Almopeos D&R system. These indicators include repair costs relative to asset value (CA), population at risk (CH), spatial extent and recovery time of environmental impacts (CE), irrigation deficit (CS), and economic losses (CF & CR). The resulting classification framework is presented in

Table 5. Magnitude of consequence across the five risk areas (based on EC [58]).

Score	Magnitude	Asset Damage (CA)	Safety and Health (CH)	Environmental Impacts (CE)	Service Disruption (CS)	Financial Impacts (CF & CR)
1	Insignificant	<1% damage (negligible)	No population at risk	Negligible impact, localized, immediate recovery	<5% irrigation deficit (no impact)	<2% economic loss
2	Minor	1–5% damage (minor repair)	<10 people, minor injuries	<1 km affected, recovery < 1 month	5–15% deficit (minor restrictions)	2–10% economic loss
3	Moderate	5–15% damage (moderate repair)	10–100 people at risk, serious injuries possible	1–5 km affected, recovery < 1 year	15–30% deficit (moderate impact)	10–25% economic loss
4	Major	15–40% damage (major repair)	100–1000 people at risk (high risk)	5–20 km affected, recovery > 1 year	30–60% deficit (severe shortage)	25–50% economic loss
5	Catastrophic	>40% damage or structural failure	>1000 people at risk or fatalities	>20 km affected, long-term or irreversible impact	>60% deficit (system failure)	>50% economic loss

, where qualitative levels are directly linked to quantitative thresholds.

This combined approach ensures consistency with EC [58] while enabling a more explicit and reproducible assessment based on system-specific characteristics. Moreover, while EC [58] expresses financial impacts in terms of turnover of the infrastructure operator, the present study adopts an alternative approach based on agricultural production losses. This adaptation reflects the primary function of the Almopeos D&R system, which is irrigation water supply, and its role in supporting regional economic activity, and therefore provides a more appropriate measure of financial impact.

2.3. Likelihood Analysis

The likelihood analysis evaluates the probability of occurrence of the climate hazards considered in the assessment of D&R systems within the selected planning horizon. The classification presented in

Table 6. Scale of the likelihood analysis for the climate hazards (based on EC [58]).

Score	Term	Qualitative Estimation	Quantitative Estimation
1	Rare	Hazard is highly unlikely to occur	5%
2	Unlikely	Hazard is unlikely to occur	20%
3	Moderate	Hazard is as likely to occur as not	50%
4	Likely	Hazard is likely to occur	80%
5	Almost certain	Hazard is very likely to occur	95%

is based on the five-level qualitative scale proposed by the EC [58], to which scores ranging from 1 (rare) to 5 (almost certain) are assigned. The assessment is based on the distributions of key climate indicators, which are derived from available climate information, including historical data, climate projections, and expert judgement, and expresses the degree of confidence that a given hazard may occur.

Likelihood is evaluated for the main groups of hazards identified in the vulnerability analysis: temperature increase and heat waves, decreased precipitation and drought conditions, and extreme precipitation and flood events [60].

2.4. Risk Analysis

The overall climate risk for each impact chain is determined by combining the magnitude of consequences and the likelihood of occurrence using a semi-quantitative risk matrix approach. This approach provides a transparent and practical framework for comparing risks across different hazard categories and types of consequences. The risk score (R) is calculated as:

$$\mathrm{R}=\mathrm{C}\times \mathrm{L}$$

(1)

where C is the consequence score and L is the likelihood score, both ranging from 1 to 5. The resulting risk scores range from 1 to 25.

Risk scores are classified into four qualitative levels: Low, Moderate, High, and Very High that correspond to risk scores 1–4, 5–9, 10–15 and 16–25, respectively, as shown in

Table 7. Classification of risk scores.

Risk Score	Risk Level
1–4	Low
5–9	Moderate
10–15	High
16–25	Very High

. This classification supports the prioritization of climate risks and the identification of impact chains requiring further analysis or mitigation. However, impact chains with similar overall risk scores may differ substantially in their risk attributes. In particular, low-likelihood high-consequence (LLHC) events, such as dam overtopping or internal erosion leading to possible failure, are not considered equivalent in engineering significance to more frequent operational risks with moderate consequences. Therefore, the risk matrix is used in this study primarily as a screening and prioritization tool, while the interpretation of results also considers the likelihood–consequence profile of each impact chain and the specific safety requirements associated with dam failure modes.

Although semi-quantitative risk matrix approaches may involve a degree of expert judgement in the definition of likelihood and consequence levels, they are widely used in water resources and dam safety studies due to their transparency, simplicity, and practical applicability [13,14,62].

2.5. Assessment of Adaptation Measures

Following the risk assessment, adaptation measures are identified and classified to reduce the most significant risks and enhance the climate resilience of a D&R system. The identified measures are then assessed and prioritized, and the preferred measures are integrated into the D&R design and/or operation to improve climate resilience [59].

The identification and classification of measures are guided by the Key Type Measures (KTMs) framework of the European Climate Adaptation Platform (Climate-ADAPT) [40]. The KTM framework groups adaptation options into categories including: (i) grey infrastructure measures, such as structural upgrades of hydraulic components; (ii) nature-based solutions (NbSs), including catchment restoration and erosion control; (iii) management and operational measures, such as improved reservoir operation, monitoring, and maintenance; (iv) policy and institutional measures, including regulatory and planning instruments; and (v) information and capacity-building measures, such as data collection, training, and stakeholder engagement.

In D&R systems, these measures aim to reduce either the likelihood or the consequences of climate-related risks, while enhancing the adaptive capacity of the system. Their application to the Almopeos D&R system is presented in Section 3.

3. Application of the Methodology

3.1. The Case Study of Almopeos D&R System

The study area is located along the Almopeos River in northern Greece, where the river flows through a narrow valley before entering the Giannitsa plain. The planned dam is situated approximately 4 km north of the settlement of Kali, at the outlet of the gorge. The reservoir collects runoff from the upstream Almopia basin, which includes both mountainous and lowland areas. A plan view of the system is shown in Figure 2.

The Almopeos reservoir has a total storage capacity of approximately 35.5 million m³. The dam is an earthfill embankment structure with a height of about 61.0 m and a crest length of 245.0 m. The embankment consists of an impervious clayey silt core surrounded by coarse-grained shells derived from nearby borrow areas, while the upstream face is protected by a rockfill layer against wave action and water level fluctuations. The foundation consists of alluvial deposits treated by grouting to reduce seepage.

The primary function of the reservoir is irrigation water supply for the surrounding agricultural areas. Inflow is provided by the Almopeos River, whose ecological status upstream is classified as good, although its chemical status is considered less than good according to the River Basin Management Plan of Western Macedonia.

The spillway system comprises an inflow channel, a spillway equipped with fusegates, a collector channel, a drop channel, a stilling basin, and an escape channel returning flow to the downstream river. Most structures are constructed of reinforced concrete, while the escape channel is partly excavated in natural soil and protected with stone lining. The fusegates increase the effective discharge capacity during extreme flood events. Additional hydraulic structures include a vertical concrete well for water abstraction, pipelines for irrigation release, environmental flow, and sediment flushing, as well as a Howell–Bunger valve for flow regulation. Sediment flushing is performed periodically during early winter rainfall events. The system also includes an administration building with automation and control systems, as well as monitoring equipment such as piezometers, extensometers, inclinometers, and accelerometers. Access is provided by two rural roads on both sides of the river. Although no permanent staff is present on site, regular inspection and maintenance visits are planned. Environmental flow released downstream ranges between 0.04 and 1.1 m³/s, depending on hydrological conditions. For clarity, the annotations shown in Figure 2 correspond to the following main components of the system: (1) dam, (2) spillway, (3) collector channel, (4) drop channel, (5) stilling basin, (6) escape channel, (7) water abstraction well, (8) outlet water channel, (9) flushing pipe, (10) flushing sluice gate, (11) administration building, (12) irrigation pipe, (13) access road to the administration building, (14) access road to the spillway, and (15) Almopeos River axis.

Based on these characteristics, the methodology described in Section 2 is applied in the following subsections.

3.2. Selection of Relevant Impact Chains for the Almopeos D&R System

Table 2, Table 3 and Table 4 present representative climate–impact chains describing the main mechanisms through which temperature increase, drought conditions, and extreme precipitation events may affect D&R systems. However, their relevance depends on the characteristics and operation of the specific system considered. Therefore, prior to the risk assessment, the impact chains were screened for their applicability to the Almopeos D&R system. This selection was based on the main characteristics of the project described in Section 3.1 and the following criteria:

• Relevance to key climate–infrastructure interactions commonly reported in the literature, such as changes in inflows, evaporation losses, structural stresses, and extreme hydraulic loading conditions [41,62];
• Coverage of the main components of D&R systems, including inputs, functions, assets, outflows, and supporting infrastructure, following the system typologization of Stamou et al. [61];
• Representation of impacts across the five risk areas considered in the assessment.

Based on these criteria, eleven impact chains were identified as the most relevant for the Almopeos system, including three temperature-related chains (TIM1–TIM3), three drought-related chains (DIM1, DIM2, and DIM4), and five chains associated with extreme precipitation and floods (FIM1–FIM5), which are summarized in

Table 8. Selected impact chains and justification.

Group of Hazards	Selected Impact Chains	Main Justification
Temperature increases and heat waves	TIM1, TIM2 & TIM3	Irrigation use, water quality sensitivity, evaporation losses, increased demand
Decreased precipitation and drought	DIM1, DIM2 & DIM4	Reduced inflows, water quality deterioration at low levels, clay-core desiccation
Extreme precipitation and floods	FIM1, FIM2, FIM3, FIM4 & FIM5	Earthfill dam safety, spillway performance, sediment/debris transport, flood loading

.

These selected chains represent the dominant climate-impact mechanisms affecting water availability, water quality and dam safety, including embankment desiccation, seepage risk, overtopping, piping and sediment transport. They form the basis for the subsequent consequence and risk assessment.

3.3. Consequence Analysis of Almopeos D&R System

The magnitude of consequences for the selected eleven impact chains (Table 2, Table 3 and Table 4 and 8) was evaluated using the five-level scale defined in Table 5 for the five risk areas introduced in Section 2.1 and the corresponding system-specific indicators, namely repair costs relative to asset value (CA), population at risk (CH), spatial extent and recovery time of environmental impacts (CE), irrigation deficit (CS), and economic losses (CF & CR).

3.3.1. Asset Damage (CA)

For temperature-related impact chains (Table 2), increased water temperature (TIM1) leads to algal growth and obstruction of infrastructure components, resulting in minor maintenance costs (1–5% of asset value, CA = 2), while TIM2 and TIM3 do not directly affect structural integrity. For drought-related chains (Table 3), reduced reservoir levels (DIM1 and DIM2) expose structural elements, leading to minor deterioration (CA = 2). In contrast, prolonged drought (DIM4) may cause desiccation and cracking of the clay core, requiring major repair works (15–40%, CA = 4). For extreme precipitation chains (Table 4), flooding and hydraulic loading (FIM1 and FIM4) result in significant structural stress (CA = 4), while internal erosion (FIM3) produces similar damage levels. Overtopping (FIM2) may lead to structural failure (CA = 5), whereas sediment transport (FIM5) results in minor damage (CA = 2).

3.3.2. Safety and Health (CH)

Temperature-related (TIM1–TIM3) and most drought-related chains (DIM1 and DIM2) do not pose direct safety risks (CH = 1), while DIM4 presents low risk (CH = 2). For extreme precipitation chains (Table 4), flooding (FIM1) is expected to have limited direct impact on human safety and is therefore classified as moderate (CH = 3). For dam failure scenarios, overtopping (FIM2) and piping/internal erosion (FIM3), the assessment is based on dam-break analysis results provided in the design studies of the project, which include both one-dimensional and two-dimensional hydraulic modelling of flood wave propagation downstream of the dam. These analyses indicate rapid flood wave propagation, high flood depths (up to approximately 20–28 m near the dam), and significant inundation extent, affecting settlements and agricultural areas downstream. Based on these results, the affected population is estimated to range between approximately 150 and 800 people, considering the spatial distribution of settlements within the inundation area and typical rural population densities. The risk to human safety is further increased due to the short arrival times of the flood wave, estimated at approximately 1–5 min near the dam and less than two hours in downstream areas. Consequently, both overtopping (FIM2) and piping (FIM3) are classified as high to very high risk (CH = 4–5), while spillway malfunction (FIM4) is considered moderate (CH = 3).

3.3.3. Environmental Impacts (CE)

Temperature-related chains (Table 2) affect water quality over distances of approximately 1–5 km (TIM1, CE = 3), while TIM2 and TIM3 result in more localized impacts (CE = 2). Drought-related chains (Table 3) affect river reaches of 1–5 km (DIM1–DIM2, CE = 2–3), while DIM4 has no direct environmental impact (CE = 1). Under drought and warming conditions, reduced inflows and lower reservoir levels may also affect downstream ecological conditions by increasing water temperature, reducing dilution capacity, and worsening water-quality conditions in the receiving river reach. Given that the environmental flow released downstream ranges between 0.04 and 1.1 m³/s, prolonged periods of low inflow combined with increased irrigation demand may place additional stress on ecological flow maintenance, particularly during the dry season. Extreme precipitation chains (Table 4) produce the most significant environmental impacts. Flooding (FIM1) is expected to affect river reaches of 5–20 km (CE = 4), while dam failure scenarios, including overtopping (FIM2) and internal erosion (FIM3), are associated with high-magnitude impacts along several kilometers downstream, as indicated by dam-break analysis and flood propagation modelling from the project design studies. These impacts include rapid changes in flow regime, erosion and sediment transport, and disturbance of aquatic habitats. Sediment transport (FIM5) is classified as moderate (CE = 3), reflecting increased turbidity and geomorphological changes within the affected river corridor. Overall, the environmental consequences are classified as moderate to high, depending on the impact chain and spatial extent.

3.3.4. Service Disruption (CS)

Service disruption was assessed based on irrigation deficit, considering the relationship between water demand and reservoir storage capacity. For an annual irrigation demand of approximately 60–70 million m³ and a storage capacity of 35.5 million m³, as defined in the project design studies, the system is highly sensitive to changes in inflow and demand. Temperature-related chains (Table 2) increase irrigation demand (TIM3), resulting in deficits of 30–40% (CS = 4), while increased evaporation losses (TIM2) lead to deficits of 10–20% (CS = 3). Drought-related chains (Table 3) have the strongest effect: reduced inflows (DIM1) result in deficits of 35–50% (CS = 4), while DIM2 and DIM4 have limited impact (CS = 2). These deficit ranges imply substantial pressure on irrigation reliability for the approximately 12,000 ha served by the system, especially during peak demand periods, with likely effects on irrigation scheduling, cropping choices, and the capacity of farmers to maintain stable production under prolonged dry and hot conditions. Extreme precipitation chains (Table 4) may cause temporary or permanent disruption of the system. Flooding (FIM1) leads to deficits of 20–40% (CS = 4), mainly due to temporary operational disruption and infrastructure damage. In contrast, dam failure scenarios, including piping/internal erosion (FIM3) and overtopping (FIM2), as analyzed in the dam-break modelling studies, result in partial or complete system failure. Piping (FIM3) is associated with severe disruption (CS = 4), while overtopping (FIM2) leads to complete loss of system functionality and irrigation supply (>60%, CS = 5).

3.3.5. Financial and Reputational Impacts (CF & CR)

Financial impacts were evaluated based on agricultural production losses associated with irrigation deficits. According to the agricultural study of the project, the irrigated area is dominated by high-value crops, particularly tree crops (58.1%) and vegetables (10.8%), while maize and other crops account for smaller shares. Based on this crop distribution and typical production values, the total annual agricultural production is estimated to range between approximately 20 and 35 million Euros. Accordingly, irrigation deficits result in proportional economic losses depending on their severity and duration. From a social perspective, these losses are directly relevant to farmer livelihoods, since reduced irrigation reliability may affect crop yields, farm income stability, and the economic viability of irrigated agriculture in the command area, particularly for water-dependent high-value crops. Temperature and drought-related chains (Table 2 and Table 3) lead to minor to major losses (CF = 2–4), consistent with the deficit ranges identified in Section 3.3.4. In contrast, extreme precipitation chains (Table 4) produce the highest financial impacts. Flooding (FIM1) leads to moderate losses (CF = 3), while piping/internal erosion (FIM3) results in major losses (CF = 4). Overtopping (FIM2), associated with complete system failure and loss of irrigation supply, leads to very high financial impacts (CF = 5).

Reputational impacts (CR) are considered moderate to high, particularly in failure scenarios, due to potential effects on public trust, institutional credibility, and stakeholder confidence.

The resulting consequence scores for all impact chains are summarized in

Table 9. Consequences scores for the selected impact chains and risk areas.

Group of Hazards	Impact Chains	CA	CH	CE	CS	CF & CR	Overall
Temperature increases and heat waves	TIM1—Water quality degradation	2	1	3	2	2	3
	TIM2—Evaporation losses	2	1	2	3	2	3
	TIM3—Increased irrigation demand	1	1	2	4	4	4
Drought conditions	DIM1—Reduced reservoir storage	2	1	3	4	4	4
	DIM2—Water quality deterioration	2	1	3	2	2	3
	DIM4—Clay core desiccation/seepage	4	2	1	2	2	4
Extreme precipitation and floods	FIM1—Overflow and flooding	4	4	4	4	3	4
	FIM2—Overtopping	5	5	4	5	5	5
	FIM3—Piping/internal erosion	4	4	4	4	4	4
	FIM4—Spillway malfunction	4	3	2	2	2	4
	FIM5—Sediment and debris transport	2	1	3	2	2	3

, where the maximum value across the five risk areas defines the overall consequence score.

3.4. Likelihood Analysis of Almopeos D&R System

The likelihood analysis is based on the distributions of representative key indicators that correspond to the main hazards groups identified in the vulnerability analysis. The following indicators were selected in the present analysis:

• TX35 (temperature increase and heat waves): number of days per year with maximum temperature exceeding 35 °C;
• CDD (drought conditions): maximum number of consecutive dry days per year;
• Rx1day (extreme precipitation): maximum daily precipitation per year.

3.4.1. Climate Change Scenarios

Two greenhouse gas emission scenarios were considered: SSP2-4.5 and SSP5-8.5, representing intermediate and high-emission pathways, respectively. For each scenario, two future periods were analyzed (2041–2060 and 2081–2100), resulting in four scenario–period combinations. The analysis was based on daily high-resolution (1 km × 1 km) statistically downscaled climate projections from four CMIP6 global climate models (GCMs): UKESM1-0-LL (r1i1p1f2), developed by the Met Office Hadley Centre (MOHC); MIROC-ES2L (r1i1p1f2), developed by the Centre for Climate System Research (University of Tokyo), JAMSTEC, and NIES; CanESM5 (r1i1p1f1), developed by the Canadian Centre for Climate Modelling and Analysis; and INM-CM4-8 (r1i1p1f1), developed by the Russian Academy of Sciences. This set of models was selected from the available CLIMADAT-hub simulations [63] because it provides consistent daily temperature and precipitation outputs for the selected scenarios and periods, while also representing different modelling centers and, therefore, part of the inter-model variability of CMIP6 projections. The corresponding simulations were statistically downscaled within the framework of the CLIMADAT-hub project, following the methodology described in Varotsos et al. [64], using the CLIMADAT-GRid high-resolution (1 km × 1 km) gridded observational dataset for Greece as the reference dataset [54]. Although the use of four models does not span the full range of CMIP6 uncertainty, it provides a representative subset of regionally processed projections suitable for the semi-quantitative comparative analysis carried out in this study. A formal multi-model weighting scheme was not applied, because the objective was not to produce probabilistic climate projections, but to examine the response of the risk framework to a plausible ensemble of downscaled climate futures.

It should be noted that the present study does not perform a new station-by-station historical validation or ranking of the individual parent GCMs for the Almopeos area. Instead, it uses statistically downscaled climate projections produced within the CLIMADAT-hub framework, whose applicability relies on the underlying bias-adjustment and downscaling procedure and on the use of the CLIMADAT-GRid observational dataset as reference. Therefore, the climate inputs used here should be understood as regionally processed scenario data intended for climate-impact assessment rather than as raw GCM outputs directly applied at the local scale. Accordingly, the resulting likelihood estimates should be interpreted as scenario-based semi-quantitative indicators rather than as exhaustive representations of the full uncertainty range of future climate conditions.

3.4.2. Empirical Distributions of Climate Indicators

For each climate indicator, annual values from all climate models and simulation years within each scenario–period combination were combined into a single ensemble dataset representing the range of possible future conditions. This dataset captures both interannual variability and inter-model differences. The variability of each indicator was quantified using empirical exceedance probability functions, defined as:

$$\mathrm{P}\left(\mathrm{X}>\mathrm{x}\right)={\displaystyle \frac{\mathrm{N}(\mathrm{X}>\mathrm{x})}{{\mathrm{N}}_{\mathrm{tot}}}}$$

(2)

where N(X > x) is the number of years in which the indicator exceeds the threshold x, and N_tot is the total number of years in the sample.

This empirical approach avoids assumptions regarding theoretical probability distributions and enables a direct, data-driven estimation of likelihood based on the frequency of occurrence of the selected climate indicators in the ensemble of model simulations. The resulting exceedance probability distributions for all indicators and scenario–period combinations are presented in

Table 10. Exceedance probabilities (%) and corresponding likelihood levels (scale 1–5) for key climate indicators (TX35, CDD and Rx1day) across scenarios SSP2-4.5 and SSP5-8.5 and future periods.

Threshold	SSP2-4.5		SSP5-8.5
TX35	2041–2060	2081–2100	2041–2060	2081–2100
10	5 (91.3%)	5 (96.3%)	5 (96.3%)	5 (100.0%)
15	4 (83.8%)	5 (92.5%)	5 (92.5%)	5 (100.0%)
20	4 (72.5%)	5 (90.0%)	5 (86.3%)	5 (100.0%)
25	4 (68.8%)	4 (82.5%)	4 (80.0%)	5 (100.0%)
30	3 (48.8%)	4 (80.0%)	4 (76.3%)	5 (100.0%)
35	3 (36.3%)	4 (66.3%)	4 (62.5%)	5 (100.0%)
40	3 (23.8%)	3 (56.3%)	3 (47.5%)	5 (98.8%)
CDD	2041–2060	2081–2100	2041–2060	2081–2100
30	3 (56.3%)	3 (56.3%)	4 (63.8%)	4 (78.8%)
50	2 (11.3%)	2 (15.0%)	2 (15.0%)	3 (26.3%)
60	2 (6.3%)	2 (7.5%)	2 (7.5%)	2 (11.3%)
70	1 (3.8%)	1 (3.8%)	1 (2.5%)	2 (8.8%)
90	1 (0.0%)	1 (1.3%)	1 (0.0%)	1 (2.5%)
110	1 (0.0%)	1 (0.0%)	1 (0.0%)	1 (1.3%)
130	1 (0.0%)	1 (0.0%)	1 (0.0%)	1 (1.3%)
RX1d	2041–2060	2081–2100	2041–2060	2081–2100
30	3 (28.8%)	3 (23.8%)	3 (33.8%)	3 (32.5%)
50	1 (0.0%)	1 (3.8%)	1 (1.3%)	1 (2.5%)
70	1 (0.0%)	1 (0.0%)	1 (0.0%)	1 (0.0%)
100	1 (0.0%)	1 (0.0%)	1 (0.0%)	1 (0.0%)
130	1 (0.0%)	1 (0.0%)	1 (0.0%)	1 (0.0%)

.

3.4.3. Definition of System-Based Thresholds for Climate Indicators

Critical thresholds were defined to link climate indicators with system performance based on physical system characteristics, irrigation demand, and hydrological response.

For temperature-related hazards, the threshold for TX35 was derived from irrigation demand analysis using project data. Average daily demand during typical summer conditions (June and August) is approximately 0.33–0.37 million m³/day, corresponding to a representative value of about 0.35 million m³/day, while peak demand during July reaches approximately 0.43 million m³/day. This increase of about 20–25% is consistent with enhanced evapotranspiration under high-temperature conditions [65]. In addition, sustained high-temperature periods are expected to affect the system not only through increased irrigation demand, but also through enhanced reservoir evaporation and reduced summer inflows, which act synergistically to increase pressure on available storage. Considering the concentration of irrigation demand within the May–September period, the limited inflows during summer, and the cumulative effect of high-temperature days, a threshold of 20 days per year is adopted. This value represents sustained periods of elevated demand leading to system stress and is consistent with observed durations of heatwaves in Mediterranean regions [39,66,67]. Thus, the TX35 threshold should be interpreted as a practical system-stress indicator integrating the combined effects of heat on water demand, evaporation losses, and seasonal inflow reduction, rather than as an indicator of irrigation demand alone.

For drought-related impacts, the threshold for CDD was derived from a simplified reservoir water balance using project data. The peak irrigation demand is approximately 0.43 million m³/day, while the total reservoir storage is approximately 35.5 million m³, resulting in a theoretical supply duration of about 70–80 days under zero inflow conditions. Accounting for environmental flow requirements (0.04–1.1 m³/s), operational constraints and residual inflows, this duration reduces to approximately 60–70 days. Since shorter dry periods (20–30 days) are common in Mediterranean climates and do not represent critical conditions, a threshold of 60 consecutive dry days is adopted, corresponding to the onset of significant system stress. This is consistent with drought classifications reported in the literature [39,52,68]. The selected threshold is therefore intended to represent a practical indicator of critical within-year stress conditions for reservoir operation. However, it is acknowledged that the present analysis does not explicitly account for the cumulative effects of consecutive dry years, which may further aggravate storage deficits and system vulnerability in Mediterranean environments. Such interannual persistence effects are important and should be examined in future extensions of the methodology.

For flood-related hazards, the threshold of Rx1day was defined based on simplified rainfall–runoff relationships and empirical evidence on rainfall extremes. Daily rainfall that exceeds approximately 50 mm/day is generally associated with significant runoff generation in Mediterranean catchments and represents the onset of conditions requiring operational attention. However, it should be acknowledged that, although statistical downscaling improves the spatial detail of the projections and makes them more suitable for local-scale applications, short-duration precipitation extremes may still be underestimated because the driving GCMs remain limited by coarse native resolution and by their imperfect representation of convective processes. Consequently, the selected threshold should be interpreted as an operational indicator of hydrological response rather than a strict design-level criterion, and the derived exceedance probabilities should be interpreted with caution.

3.4.4. Likelihood Probability and Scores of Hazards

The likelihood of occurrence of the main climate hazards was evaluated based on the exceedance probabilities of the selected climate indicators and their corresponding system-based thresholds. The resulting likelihood scores for the selected impact chains are summarized in

Table 11. Likelihood probability and scores for the selected impact chains and risk areas.

Group of Hazards	Indicator	Impact Chain	SSP2-4.5		SSP5-8.5
			2041–2060	2081–2100	2041–2060	2081–2100
Temperature increases and heat waves	TX35	TIM1—Water quality degradation	4 (72.5%)	5 (90.0%)	5 (86.3%)	5 (100.0%)
		TIM2—Evaporation losses
		TIM3—Increased irrigation demand
Decreased precipitation and drought	CDD	DIM1—Reduced reservoir storage	2 (6.3%)	2 (7.5%)	2 (7.5%)	2 (11.3%)
		DIM2—Water quality deterioration
		DIM4—Clay core desiccation/seepage
Extreme precipitation and floods	Rx1day	FIM1—Overflow and flooding	1 (0.0%)	1 (3.8%)	1 (1.3%)	1 (2.5%)
		FIM2—Overtopping
		FIM3—Piping/internal erosion
		FIM4—Spillway malfunction
		FIM5—Sediment and debris transport

.

Temperature-related hazards (TX35) exhibit consistently high exceedance probabilities across all scenarios and future periods. For the selected threshold of 20 days per year, probabilities range from approximately 72.5% to 100% (Table 10), corresponding to likelihood levels 4 to 5 (Table 11). This indicates that high-temperature conditions leading to increased irrigation demand are expected to become a persistent feature of the system. These findings are consistent with projections of more frequent and prolonged heatwaves in the Mediterranean region [39,66,67] confirming that temperature increase represents the dominant climate driver affecting system performance.

Drought-related hazards, evaluated using a threshold of 60 consecutive dry days (CDD), show relatively low exceedance probabilities, generally ranging between approximately 6% and 18% (Table 10), corresponding to likelihood level 2 (Table 11). This suggests that while moderate dry periods are common in Mediterranean climates, prolonged drought conditions leading to critical system stress occur less frequently. These findings are consistent with projections of increasing drought variability and persistence in Southern Europe [39,68], while highlighting the importance of distinguishing between frequent dry spells and severe drought events.

In contrast, extreme precipitation events (Rx1day) exhibit low exceedance probabilities for thresholds relevant to system operation, generally below 5% (Table 10), corresponding to likelihood level 1 (Table 11). However, these results should be interpreted with caution. Climate models are known to have limitations in representing short-duration precipitation extremes, which are often associated with convective processes and may be underestimated [53,69]. In addition, flood generation depends on multiple factors, including antecedent soil moisture, rainfall duration and spatial variability, which are not fully captured by single-day precipitation indicators. Consequently, although extreme precipitation is classified as a low-likelihood hazard based on the selected indicator, its potential contribution to dam safety risk remains important due to the high consequences associated with failure scenarios. This highlights the need for a precautionary interpretation of flood-related risks in dam safety assessments.

Overall, the likelihood analysis indicates that the climate hazard regime of the Almopeos D&R system is dominated by temperature increase and, to a lesser extent, drought conditions, while extreme precipitation represents a lower-probability but potentially high-impact hazard. The differences in likelihood across scenarios and future periods are illustrated in Figure 3. The figure shows a clear increase in the likelihood of temperature-related hazards from SSP2-4.5 to SSP5-8.5 and from the mid-century to the late-century period, confirming that TX35 becomes an increasingly dominant climate driver under stronger warming conditions. In contrast, drought-related hazards represented by CDD exhibit a more moderate increase across scenarios and periods, while extreme precipitation hazards represented by Rx1day remain associated with comparatively low likelihood levels for the selected threshold. This visualization supports the interpretation of Table 10 and Table 11 by making the relative differences among indicators, scenarios, and periods more explicit.

The robustness of the likelihood classification was also examined through a ±10% sensitivity analysis of the adopted system-based thresholds, namely TX35 = 20 days/year, CDD = 60 consecutive dry days, and Rx1day = 50 mm/day, as shown in

Table 12. Sensitivity of exceedance probabilities and likelihood classes to ±10% variation in the adopted system-based thresholds.

Indicator	Threshold	SSP2-4.5		SSP5-8.5
		2041–2060	2081–2100	2041–2060	2081–2100
TX35	18 d	4 (76.2%)	5 (91.2%)	5 (91.2%)	5 (100.0%)
	20 d	4 (72.5%)	5 (90.0%)	5 (86.2%)	5 (100.0%)
	22 d	4 (71.2%)	5 (86.2%)	5 (83.8%)	5 (100.0%)
CDD	54 d	2 (8.8%)	2 (10.0%)	2 (12.5%)	2 (15.0%)
	60 d	2 (6.3%)	2 (7.5%)	2 (7.5%)	2 (11.3%)
	66 d	1 (3.8%)	1 (3.8%)	1 (5.0%)	2 (10.0%)
Rx1day	45 mm	1 (3.8%)	1 (3.8%)	1 (3.8%)	2 (6.3%)
	50 mm	1 (0.0%)	1 (3.8%)	1 (1.3%)	1 (2.5%)
	55 mm	1 (0.0%)	1 (2.5%)	1 (0.0%)	1 (2.5%)

Note: The adopted thresholds are 20 days/year for TX35, 60 consecutive dry days for CDD, and 50 mm/day for Rx1day. The sensitivity analysis was performed using a ±10% variation around each adopted threshold. Values are presented as likelihood score (exceedance probability).

. For TX35, the corresponding thresholds of 18 and 22 days/year produced only moderate changes in exceedance probabilities, while the likelihood class remained unchanged in all scenario–period combinations, indicating a robust classification. For CDD, the corresponding thresholds of 54 and 66 consecutive dry days showed somewhat greater sensitivity: the likelihood class remained unchanged for the lower threshold but decreased by one class for the higher threshold in several scenario–period combinations. For Rx1day, the tested thresholds of 45 and 55 mm/day resulted in low exceedance probabilities in all cases, and the likelihood class changed only once, for the lower threshold under SSP5-8.5 in 2081–2100. Overall, the sensitivity analysis indicates that, although the exact likelihood scores may vary with threshold selection, the main interpretation of the results remains unchanged, with temperature increase identified as the dominant hazard driver, drought as a secondary but important source of system stress, and extreme precipitation as a comparatively low-likelihood hazard under the selected indicator and thresholds.

The resulting likelihood estimates should also be interpreted in the context of the uncertainties associated with the underlying climate projections, including model structure, emission scenario, and the performance of the downscaling procedure. This is particularly important for precipitation-related indicators, whose local-scale representation remains more uncertain than that of temperature-related indicators. Accordingly, the likelihood classes derived in this study are intended as scenario-based semi-quantitative estimates supporting comparative risk assessment, rather than deterministic predictions of future local conditions.

3.5. Risk Assessment of Almopeos D&R System

The overall risk associated with the groups of climate hazards and impact chains affecting the Almopeos D&R system was evaluated by multiplying the consequences scores (Table 9) with the corresponding likelihood scores (Table 11). The resulting risk scores and levels for each impact chain are presented in

Table 13. Risk scores for the three groups of hazards and the related impact chains.

Group of Hazards	Impact Chains	Consequences Score	Likelihood Score	Risk Score	Risk Level
Temperature increases and heat waves	TIM1—Water quality degradation	3	5	15	High
	TIM2—Evaporation losses	3	5	15	High
	TIM3—Increased irrigation demand	4	5	20	Very High
Decreased precipitation and drought	DIM1—Reduced reservoir storage	4	2	8	Moderate
	DIM2—Water quality deterioration	3	2	6	Moderate
	DIM4—Clay core desiccation/seepage	4	2	8	Moderate
Extreme precipitation and floods	FIM1—Overflow and flooding	4	1	4	Low
	FIM2—Overtopping	5	1	5	Moderate
	FIM3—Piping/internal erosion	4	1	4	Low
	FIM4—Spillway malfunction	4	1	4	Low
	FIM5—Sediment and debris transport	3	1	3	Low

Note: For some impact chains, especially low-likelihood high-consequence (LLHC) events such as FIM2 (overtopping), a moderate overall risk class in the matrix does not imply low engineering importance. In the case of FIM2, the moderate risk classification results from the combination of catastrophic consequences (C = 5) and rare likelihood (L = 1), according to the adopted semi-quantitative risk matrix. Nevertheless, such events require particular precautionary control, monitoring, and emergency planning because of their potentially severe dam safety implications.

.

Temperature-related hazards (TX35), which exhibit very high likelihood levels (level 5; Table 11), result in high to very high risk across all associated impact chains (TIM1–TIM3). In particular, increased irrigation demand (TIM3) reaches the highest risk level (Very High), reflecting both the high probability of occurrence and the significant impact on system operation. Water quality degradation (TIM1) and evaporation losses (TIM2) are also classified as high risk. These results confirm that temperature increase constitutes the dominant risk driver for the system.

Drought conditions, which were evaluated using a threshold for CDD of 60 consecutive dry days, are associated with lower likelihood levels (score = 2; Table 11), but relatively high consequences. As a result, drought-related impact chains (DIM1, DIM2, DIM4) are classified as moderate risk (Table 13). This indicates that although severe drought events are less frequent, their impacts on reservoir storage, water quality, and structural behavior can be significant when they occur. The results highlight the importance of considering both likelihood and consequence in risk evaluation, particularly for water resource systems sensitive to prolonged dry conditions.

Extreme precipitation events (Rx1day) exhibit low likelihood levels (score = 1; Table 11), resulting in low to moderate risk levels across the corresponding impact chains (FIM1–FIM5). Most flood-related risks are classified as low, except for overtopping (FIM2), which reaches a moderate risk level due to its high consequence despite the low probability of occurrence. These results should be interpreted with caution, as climate models are known to underestimate short-duration precipitation extremes, and flood-generating processes depend on additional factors such as antecedent conditions, rainfall duration, and catchment response [53,69]. In particular, overtopping (FIM2) and piping/internal erosion (FIM3) represent typical low-likelihood high-consequence (LLHC) events. Although their likelihood scores are low in the present climate-indicator-based analysis, they correspond to dam safety failure modes with potentially catastrophic consequences and therefore require special consideration in engineering decision-making. For this reason, these impact chains are not interpreted solely on the basis of their overall risk class, but also as safety-critical scenarios requiring precautionary control through spillway adequacy, seepage surveillance, inspection, and emergency preparedness.

Overall, the risk assessment indicates that the Almopeos D&R system is primarily exposed to temperature-driven increases in irrigation demand, with drought conditions representing a secondary but important risk factor. Extreme precipitation and dam safety failure scenarios, however, constitute safety-critical LLHC risks, since, although their probability is low under the scenarios considered, their consequences are severe enough to justify precautionary structural, monitoring, and emergency-response measures. These findings provide a basis for prioritizing adaptation and risk management measures, with emphasis on demand management, system flexibility, and resilience to prolonged dry conditions. The resulting risk levels should nevertheless be interpreted as semi-quantitative estimates that depend on the adopted system-based thresholds; although moderate variations in threshold values may affect individual likelihood classes, they do not alter the overall identification of the dominant climate risk drivers for the Almopeos D&R system.

3.6. Assessment of Adaptation Measures for Almopeos D&R System

Based on the climate risk assessment presented in Section 3.5, a set of adaptation measures was identified to reduce the most significant climate risks affecting the Almopeos D&R system.

3.6.1. Identification of Adaptation Measures

The identification of these measures follows the KTMs framework of Climate-ADAPT [40], as introduced in Section 2.5.

The dominant risks are primarily associated with increased irrigation demand and evaporation losses under high-temperature conditions (TIM2 and TIM3), as well as reduced water availability during prolonged dry periods (DIM1). Secondary risks include dam safety under drought conditions (DIM4) and low-probability but high-consequence flood events (FIM1–FIM5).

The identified adaptation measures are grouped according to the KTM categories as follows:

• Management and operational measures (KTM-M), which include adaptive reservoir operation rules, improved irrigation scheduling, and measures to increase irrigation efficiency in the command area.
• Grey infrastructure measures (KTM-G), which entail the maintenance and upgrading of spillway components (including fusegates), and reinforcement of drainage and seepage control systems.
• Information and capacity-building measures (KTM-I) that deal with the enhanced monitoring of seepage, pore water pressures, and water quality, as well as with the development of flood forecasting and early warning systems.
• Nature-based solutions (KTM-N), such as catchment management interventions aimed at reducing erosion and sediment inflow into the reservoir.
• Policy and institutional measures (KTM-P). These measures, although not explicitly developed in this study, include regulatory and planning measures that support efficient water use and risk-informed dam operation and are implicitly relevant. Representative instruments include volumetric water pricing, drought-contingency allocation rules, and water-use quotas during shortage periods, which have been identified as useful mechanisms for reducing drought impacts in Mediterranean basins [70].

This classification ensures consistency with the Climate-ADAPT framework while linking adaptation options to the dominant risk drivers identified in the system.

3.6.2. Appraisal of Adaptation Measures

The identified measures were analyzed in terms of their role in reducing either the likelihood or the consequences of the main climate-related risks.

Reservoir operation and irrigation management (KTM-M). Given that the reservoir supplies irrigation water to approximately 12,000 ha of agricultural land, priority is given to measures that improve water management efficiency under increased demand and reduced inflows. Adaptive reservoir operation rules and improved irrigation scheduling directly reduce the consequences of increased irrigation demand and evaporation losses (TIM2 and TIM3), which represent the highest risk levels identified in Table 13. Such approaches are widely recognized as effective strategies for climate adaptation in water resources systems [40,41].

Dam safety under drought conditions (KTM-G and KTM-I). Prolonged dry periods may lead to desiccation and cracking of the clay core, increasing seepage and piping risk (DIM4). Adaptation measures include enhanced monitoring of seepage and pore water pressures (KTM-I), systematic inspection of the embankment, and maintenance of drainage systems (KTM-G). In this context, the monitoring system should focus on representative parameters such as seepage discharge, pore-water pressure, phreatic surface position, and deformation trends derived from piezometers, extensometers, inclinometers, and related instrumentation. Such parameters are widely used in dam-safety monitoring and can support early detection of abnormal seepage behavior [41,71]. These measures reduce the likelihood of structural degradation and are consistent with established dam safety practices [11,41].

Dam safety under extreme precipitation events (KTM-G and KTM-I). Although extreme precipitation events exhibit low likelihood (Table 11), they may lead to severe consequences, particularly overtopping (FIM2). Measures include inspection and maintenance of spillway components (KTM-G), verification of discharge capacity under future conditions, and the development of early warning systems and emergency action plans (KTM-I). These measures reduce consequences and improve preparedness [10,14].

Sediment and water quality management (KTM-N and KTM-I). Impacts related to water quality degradation and sediment transport (TIM1, DIM2 and FIM5) are addressed through enhanced monitoring (KTM-I), optimization of sediment management practices, and catchment-scale interventions (KTM-N). These measures support long-term system functionality and environmental performance [41].

Overall, the measures represent a combination of operational, structural, monitoring and catchment-based interventions targeting both dominant and secondary risks.

3.6.3. Prioritization of Adaptation Measures

Following the appraisal stage, the identified measures were evaluated and prioritized using a qualitative multi-criteria approach that considers their effectiveness in reducing climate risks, technical feasibility, and implementation cost. Such approaches are widely applied in climate adaptation planning for water infrastructure systems, as they support decision-making under uncertainty and limited data availability [19,40]. The results of this evaluation are summarized in

Table 14. Evaluation and prioritization of adaptation measures for the Almopeos.

Adaptation Measure	KTM (Category)	Impact Chains	Risk Level	Effectiveness	Feasibility	Cost	Priority
Adaptive reservoir operation rules	KTM-M (Management)	TIM2, TIM3 & DIM1	High-Very High	High	High	Low	High
Improved irrigation efficiency	KTM-M (Management)	TIM2, TIM3 & DIM1	High-Very High	High	Moderate	Moderate	High
Enhanced seepage monitoring and Instrumentation	KTM-I (Monitoring)	DIM4	Moderate	High	High	Low	High
Maintenance and upgrading of spillway (e.g., fusegates)	KTM-G (Structural)	FIM1–FIM4	Low-Moderate	High	High	Moderate	High
Flood forecasting and early warning system	KTM-I (Monitoring)	FIM1–FIM3	Low–Moderate	Moderate–High	Moderate	Moderate	Medium
Sediment management and catchment interventions	KTM-N (NbS)	DIM2 & FIM5	Low–Moderate	Moderate	Moderate	Moderate	Medium
Water quality monitoring	KTM-I (Monitoring)	TIM1, DIM2 & FIM5	Low–High	Moderate	High	Low	Medium

.

Effectiveness refers to the ability of each measure to reduce the likelihood or the magnitude of the identified climate risks, particularly those associated with increased irrigation demand and evaporation losses under high-temperature conditions (TIM2 and TIM3), reduced reservoir storage (DIM1), and dam safety under prolonged drought conditions (DIM4) and extreme events (FIM1–FIM3). Technical feasibility reflects the degree to which each measure can be implemented within the existing design, operational practices and institutional framework of the Almopeos D&R system. Cost represents the approximate level of financial resources required for implementation and is expressed qualitatively as low, moderate or high.

As shown in Table 14, the prioritization indicates that operational and monitoring interventions, corresponding mainly to the management and operational (KTM-M) and information and capacity-building (KTM-I) categories, represent the most effective and feasible options for addressing the dominant climate risks. In particular, adaptive reservoir operation rules and improvements in irrigation efficiency are identified as high-priority measures, as they directly address the highest-risk impact chains related to increased water demand and evaporation losses (TIM2 and TIM3), as well as reduced water availability (DIM1). Similarly, enhanced monitoring of seepage and embankment conditions and systematic inspection practices are considered essential for managing structural risks associated with prolonged drought conditions (DIM4), offering high effectiveness at relatively low implementation cost. However, the prioritization of adaptation measures is not based solely on the numerical risk class. It also reflects the different nature of the identified risks. In particular, dam safety failure modes such as overtopping and piping/internal erosion are treated as safety-critical low-likelihood high-consequence (LLHC) risks. Accordingly, measures such as spillway maintenance and upgrading, seepage monitoring, systematic inspection, and flood warning and emergency planning are considered essential even when the corresponding risk score is lower than that of operational risks. In addition, policy instruments such as volumetric pricing, allocation rules, and water-use quotas can support the effectiveness of operational measures by moderating demand during dry periods and improving the practical implementation of drought-response strategies [70].

Measures related to structural safety, such as maintenance and inspection of spillway components, including fusegates, also rank among the high-priority options (Table 13). Although these measures primarily address flood-related hazards with low likelihood, they are critical for reducing the consequences of extreme events, particularly overtopping and piping (FIM1, FIM2 and FIM3), and are therefore necessary within a precautionary dam safety framework [10,14].

In contrast, measures such as flood forecasting and early warning systems, sediment management, and water quality monitoring are assigned medium priority (Table 14). These measures contribute to enhancing system resilience and operational performance but are primarily associated with impact chains characterized by moderate or lower risk levels. In particular, nature-based solutions (KTM-N), such as catchment management interventions to reduce sediment inflow, play a supportive role in improving long-term system sustainability, while information-based measures (KTM-I) related to water quality monitoring contribute to maintaining environmental performance.

Overall, the prioritization highlights that adaptation planning for the Almopeos D&R system should focus primarily on measures that improve water management efficiency and system flexibility under increasing temperature and demand pressures, while also ensuring adequate monitoring and structural preparedness for less frequent but potentially high-consequence events. This combination of measures reflects a balanced adaptation strategy consistent with current best practices in climate risk management for water infrastructure systems.

4. Discussion

The climate risk assessment performed for the Almopeos D&R system highlights the importance of considering both operational and structural impacts of climate change on D&R systems. The results indicate that the dominant risks are associated primarily with increased irrigation demand and evaporation losses under high-temperature conditions, as well as reduced water availability during prolonged dry periods, while extreme precipitation events represent low-probability but high-consequence hazards.

In the case of the Almopeos D&R system, the highest risk levels are associated with increased irrigation demand (TIM3), reduced reservoir storage (DIM1), and potential desiccation of the clay core during prolonged drought periods (DIM4). These results reflect the strong dependence of the system on water availability and are consistent with projections for Mediterranean regions, where increasing temperatures and hydrological variability are expected to intensify water scarcity and affect reservoir operation [39,40]. Under such conditions, the balance between water supply and demand becomes the critical factor controlling system performance.

The results are broadly consistent with recent studies from the Mediterranean region, which indicate that increasing temperature, higher irrigation demand, reduced water availability, and more persistent drought conditions are among the dominant climate pressures on water-storage systems and irrigated agriculture [40,42,43,72]. In this respect, the high and very high-risk levels identified in this work for temperature-related impacts, and the moderate but important risk levels associated with drought-related impacts, are consistent with the broader regional picture of increasing water stress under climate change. At the same time, the comparatively low likelihood but high consequence of extreme precipitation-related dam-safety hazards is also consistent with Mediterranean conditions, where flood-generating events are less frequent than drought and heat-related stresses but may still produce severe impacts when they occur. Compared with many regional studies that focus mainly on water availability, drought indices, or individual operational impacts, the proposed framework offers the additional advantage of linking climate indicators to system-specific thresholds, combining likelihood and consequence in a transparent semi-quantitative risk matrix, and directly connecting the resulting risk profile with adaptation prioritization.

At the same time, extreme precipitation events remain a key concern for dam safety. Although their likelihood is low according to the climate projections used in this work, their potential consequences are severe, particularly in the case of overtopping and piping. This finding is consistent with previous studies identifying overtopping as one of the leading causes of dam failures worldwide [2,41]. It also highlights the need to account for low-probability, high-consequence events in dam safety assessments, even when their contribution to overall risk is limited.

The results underline the importance of combining structural and non-structural adaptation measures. Operational measures, such as adaptive reservoir management and improved irrigation efficiency, are particularly effective in addressing the dominant risks related to increased demand and water scarcity. At the same time, structural and monitoring measures, including spillway maintenance, seepage monitoring and enhanced inspection procedures, are essential for managing safety risks associated with extreme hydrological conditions. This combination of measures reflects current best practice in climate adaptation for water infrastructure systems [11,41]. This distinction is important because a matrix-based classification alone may place LLHC dam safety events in the same qualitative class as more frequent operational risks, despite their fundamentally different engineering implications. In practice, operational risks are prioritized according to their expected effect on system performance and water supply reliability, whereas LLHC failure scenarios require precautionary control because they are associated with possible loss of structural integrity and severe downstream impacts. Therefore, the present assessment combines matrix-based risk ranking with engineering judgement on the nature of each risk and its control priority.

From a methodological perspective, the semi-quantitative risk matrix approach proved to be a practical tool for integrating climate projections, infrastructure characteristics and expert judgement within a consistent assessment framework. The use of climate indicators and empirically derived exceedance probabilities allowed a transparent estimation of likelihood levels without requiring full probabilistic modelling. This approach is particularly suitable for screening-level or intermediate assessments where data availability is limited.

Nevertheless, several limitations should be acknowledged. First, the likelihood assessment is based on statistically downscaled CMIP6 climate projections and is therefore subject to uncertainties associated with the driving climate models, emission scenarios, and the downscaling and bias-correction procedures. Although the statistical downscaling increases the spatial resolution of the projections, the underlying climate model data may still underestimate short-duration precipitation extremes because such events are strongly influenced by processes that are not always well represented in coarse-resolution global climate models. This limitation is directly relevant to the low exceedance probabilities obtained for the Rx1day indicator and implies that flood-related risks may be underestimated. Second, the consequence evaluation relies partly on engineering judgement, particularly in the definition of thresholds and impact severity. Third, the analysis focuses on a representative subset of impact chains rather than an exhaustive set of all possible failure mechanisms. Although this approach ensures practical applicability, it may omit less probable or indirect pathways. Fourth, a further limitation concerns the definition of the system-based thresholds used for likelihood analysis. Although these thresholds were selected on the basis of project data, physical system characteristics, and literature evidence, they remain simplified representations of more complex hydro-climatic processes. In particular, the TX35 threshold does not resolve all compound effects of prolonged heat on irrigation demand, evaporation, and inflow reduction in a process-based manner, while the CDD threshold does not explicitly incorporate the persistence of consecutive dry years. A more detailed treatment of such compound and multi-year effects would be a useful extension of the present framework and could further improve the robustness of risk estimates. Fifth, the selected ensemble of four downscaled CMIP6 models does not cover the full uncertainty range of possible future climate conditions and was not combined with a formal multi-model weighting procedure. Although the selected models provide a useful subset of regionally processed projections for comparative risk assessment, a larger ensemble, together with more systematic threshold sensitivity analysis, would provide a more comprehensive characterization of uncertainty. Sixth, the applicability of the selected climate-model ensemble at the site scale remains a limitation. Although the study uses high-resolution statistically downscaled CMIP6 projections developed within an established regional framework, it does not include a separate historical performance ranking of the individual parent GCMs specifically for the Almopeos basin. Such a site-specific comparative evaluation of model skill for temperature and precipitation would be a useful extension of the present work and could further support the refinement of likelihood estimates, particularly for precipitation extremes.

Despite these limitations, the methodology provides a structured and transparent framework for climate risk assessment of D&R systems. Its application to the Almopeos case demonstrates how climate change considerations can be systematically incorporated into dam safety and water resources management. The approach is transferable to other systems, particularly in Mediterranean regions, where increasing temperatures, drought conditions and hydrological variability are expected to significantly affect water infrastructure performance.

5. Conclusions

This study developed and applied a structured framework for assessing climate-related risks in D&R systems, integrating climate projections, system-based thresholds, and a semi-quantitative risk matrix approach.

The application to the Almopeos D&R system demonstrates that temperature-driven increases in irrigation demand constitute the dominant climate risk, leading to high to very high impacts on system operation. Drought conditions represent a secondary risk, becoming critical under prolonged dry periods that affect reservoir storage. In contrast, extreme precipitation events are characterized by low likelihood, although their potential consequences remain important for dam safety.

The results indicate that adaptation strategies should prioritize operational measures that improve water use efficiency and system flexibility, supported by monitoring and structural interventions to ensure safety under both drought and extreme conditions.

Methodologically, the study highlights the value of combining empirical climate information with system-specific thresholds to support transparent and practical risk assessments under uncertainty. The proposed approach is transferable to other water infrastructure systems, particularly in regions exposed to increasing temperature and hydrological variability.

Overall, the findings support the integration of climate risk considerations into dam operation and safety planning, with emphasis on managing demand-driven stresses while maintaining preparedness for low-probability, high-consequence events.