The First Relative Sea Level Rise and Storm Surges Scenarios up to 2150 CE for the Coasts of Monterosso and Vernazza, Cinque Terre National Park (Liguria, Italy)

Highlights

What are the main findings?

• The study provides the first assessment of the exposure of the villages of Monterosso and Vernazza to sea level rise and storm surges up to 2150 CE.
• The SLR could reach up to 1.17 m in 2150 CE under high greenhouse gas emissions (IPCC SSP5-8.5 scenario).

What are the implications of the main findings?

• The changes projected in the coming years will cause a significant retreat of the coastline and beaches, with implications for harbor functionality and the local economy.
• Adaptation and mitigation actions are required to protect the coastal stretches where valuable tourism, maritime, and transport infrastructures are located.

Abstract

The UNESCO World Heritage coastal zone of Cinque Terre (Liguria, northern Italy) is increasingly threatened by ongoing sea-level rise. To assess expected sea levels up to 2150 CE under the IPCC Sixth Assessment Report (AR6), we carried out a first evaluation of potential coastal flooding for Monterosso and Vernazza under different Shared Socioeconomic Pathways (SSPs), integrating high-resolution topography and bathymetry, geodetic data, historical tide-gauge evidence, and storm-surge modeling. The historical sea-level analysis indicates a non-stationary rising trend for the Ligurian sector. Relative sea-level rise (RSLR) projections were computed for SSP1-2.6, SSP3-7.0 and SSP5-8.5, including local land subsidence, and were used to map the potential inundated areas for 2030, 2050, 2100, and 2150 CE. In 2150, projected RSLR is expected to range from 0.60 m to 1.17 m, corresponding to flooded surfaces of 9931 m² and 22,079 m² in the SSP1-2.6 and SSP5-8.5 scenarios, respectively. Because site-specific long-term run-up observations are not available for formal calibration at the two study sites, the storm-surge simulations are interpreted as scenario-based hazard envelopes. Even within this framework, storm surges with return times of 1 and 100 years in the SSP5-8.5 scenario cause maximum wave run-up in the range of 5.12 m and 13.36 m. The results show that narrow pocket beaches and low-elevation harbor areas are the most critical receptors and that adaptation measures should focus on quay elevation, drainage/backwater management, and the protection of transport and tourism infrastructures.

1. Introduction

The Cinque Terre National Park (CTNP), part of the UNESCO “World Heritage List” [1], together with Portovenere and its islands, is located along the north-western sector of the Tyrrhenian coast of Italy. The CTNP is delimited from the Vara Valley to the east, the Gulf of La Spezia to the southeast, and the Levanto area to the north, while its western side faces the Ligurian Sea in the Northern Tyrrhenian Sea (Figure 1). The area has been inhabited since historical times and is characterized by five small villages built on steep rocky cliffs or along narrow floodplains. From north to south, the villages of Monterosso, Vernazza, Corniglia, Manarola, and Riomaggiore are distributed along 15 km of coastline, between Punta Mesco and Punta Montenero (Figure 1). The CTNP has a resident population of approximately 3500 total inhabitants, while several million tourists visit the area each year. In addition to the CTNP, the Cinque Terre Marine Protected Area (CTMPA) was established in 1997 and includes three regulated coastal zones (A, B, and C), where activities are limited in accordance with Italian national park laws. The offshore boundary of the CTMPA approximately follows the 50 m isobath in the southern sector and the 80 m isobath in the northern sector (Figure 1).

From a geological point of view, the Cinque Terre area belongs to the Northern Apennines and faces the Tyrrhenian Sea. The geological setting consists of five overlapping lithological units of oceanic crust and sedimentary formations. These include: (i) ophiolites and turbidite sequence (Gottero unit outcropping at Punta Mesco area), as well as (ii) pelitic rocks, claystone with limestones, and silty sandstones (Ottone, Canetolo and Marra units outcropping in the western side of Monterosso and in the central sector of the CTNP). Most of the coastline is composed of turbidites (Macigno formation) and, to a lesser extent, limestones belonging to the Tuscan Nappe. These tectonic units are arranged in large, overturned southwest-verging folds, which are extensively segmented by tectonic structures [2].

The area shows a rugged morphology, with a maximum elevation of 815 m a.s.l. on Mount Malpertuso (Figure 1). The main geomorphological features of the emerging sector of CTNP are small catchments characterized by high-energy, narrow, short streams along deep valleys (Figure 1). Most of the villages are located at the mouths of the narrow fluvial systems, which transport debris to the sea, particularly after heavy rainfall events.

Fluvial deposits, together with linear and areal erosional phenomena, shape the surrounding slopes. However, the original morphology has been largely modified by human interventions aimed at creating agricultural terraces. Pervasive human activity, documented since the Middle Ages, has reshaped the original topography through the construction of 6000 km of low dry-stone walls. This extensive terracing represents a prominent example of landscape modification that influences surface runoff, erosion, and sediment deposition [3,4,5].

Many studies have highlighted the role of terraced landscapes in enhancing slope stability, whereas the lack of maintenance of dry stone walls plays an important role in making terraced slopes susceptible to instability [6]. Nowadays, about 40% of the area of the CTNP is occupied by terraced slopes, up to 500 m a.s.l.

Warm, dry summers and temperate winters are typical for this sector of Liguria. The average annual rainfall is approximately 1000 mm; however, intense and concentrated precipitation often triggers flash floods that transport large quantities of debris to the coast, including large rocky blocks. On 25 October 2011, an extreme flash flood hit the villages of Vernazza and Monterosso with a cumulative rainfall of 539 mm. This event triggered several landslides, causing extensive damage to villages, roads and local infrastructure, resulting in 13 fatalities. The resulting debris flows reached the coast, forming small deltas along the shoreline that were subsequently eroded by the wave action. These deposits represent the primary source of sediment supplying the beaches of the Cinque Terre coast [7].

Although Mediterranean-scale and Italian case studies have already quantified relative sea-level rise and coastal submersion for low-lying plains, islands, archeological sites, and selected urban coasts [8,9,10,11,12,13,14,15,16], comparable analyses for steep rocky pocket beach systems such as Cinque Terre are still lacking. This gap matters because cliffed coasts with narrow beach prisms, constrained backshore space, and critical infrastructures concentrated at stream mouths are expected to respond differently to sea-level rise than low-gradient coastal plains. Accordingly, considering the local geomorphology, observed vertical land movements, the historical sea-level trend, and the IPCC AR6 regional projections, we carried out a first evaluation of potential coastal flooding under different Shared Socioeconomic Pathways (SSPs). The scientific novelty of the study lies in extending relative sea-level rise and storm-surge scenarios up to 2150 CE for Monterosso and Vernazza. The results are intended to support long-term disaster-risk reduction and coastal management by providing a site-specific framework for integrating climate scenarios into local planning and adaptation.

2. Subaerial and Submarine Morphology

The entire coastal stretch of the CTNP is characterized by mountainous areas with steep gradients, affected by instability phenomena that affect the entire coastal slope. These slopes are cut by deep incisions that, during intense rainfall events, generate flash floods with high destructive potential [17] for the five villages. Rocky cliffs account for more than 80% of the coastline, while the remaining sectors consist of shores and artificial structures [5]. Cliffs are vertical to sub-vertical, reaching heights of up to 100 m and often lacking shore platforms at their base [4]. East of Riomaggiore, the coastal slope becomes gentler and is characterized by pebbly to gravelly deposits along the shoreline. Small bays are typically bounded by headlands, and the largest settlements have developed near these limited beach areas. The coastline of Monterosso al Mare and Vernazza trends approximately east–west. It includes five pocket beaches, each extending a few hundred meters and interrupted by rocky outcrops and anthropogenic structures that have modified the natural coastal morphology. Cliff dynamics remain active due to persistent wave action, particularly from the south-western sector. In this area, the intense erosion combined with human activity has removed most evidence of past landslides. Despite this, a large landslide known as “Guvano” remains partially active between Vernazza and Corniglia. It extends for approximately 350 m, from the hamlet of S. Bernardino to the shoreline [3].

The general bathymetry of the study area shows a smooth and regular deepening of a narrow continental shelf from 5 to 7 km wide. Beyond the shelf edge, the seabed becomes more complex due to the canyon heads that incise the seafloor and reach depths of up to 650 m. Small harbors, protected by artificial breakwaters, are located in Riomaggiore, Manarola, Vernazza, and Monterosso.

The submarine sector of Monterosso al Mare is characterized by a gentle and slightly articulated morphology. On the western side, between Punta Mesco and Gigante, an infralittoral prograding wedge occurs at depths between 10 m and 15 m, approximating the present base level of storm waves [18] (Figure 1).

In shallow water, the summit of the infralittoral prograding wedge shows a flat morphology (0.4° of slope), characterized by rocky outcrops and blocks originating from cliff erosion and by an extensive Posidonia Oceanica meadow that extends to about 22 m water depth. Beyond its edge, an abrupt scarp with a slope of approximately 15° is linked with the progradation of the prism, at about 30 m water depth. In contrast, the submerged area between the Gigante at Monterosso and Punta Corone is mostly flat. Here, the seafloor consists of mobile sediments with limited Posidonia oceanica patches located up to 27 m water depth. The seabed gently slopes from about 1.7° to 3° from west to east. At approximately 6 m water depth, artificial reef breakwaters for coastal defense are present, locally altering the natural seafloor morphology.

The submerged area offshore of Vernazza is characterized by a flat morphology, with a slope of about 1°, near the edge of the continental shelf. In contrast, the coastal zone is indented and consists of steep, blocky deposits derived from cliff erosion and subaerial landslides. These deposits also form the foundation of the artificial jetty, which exhibits steep slopes ranging from 10° to about 30°. The pocket beach south of Vernazza, supplied by a channel during heavy rainfall events, has undergone significant erosion and has nearly disappeared. Overall, the small beaches of the Cinque Terre receive a limited sediment supply. Only extreme fluvial events, such as flash floods, provide the sediment necessary for their persistence. Notably, the flash floods of 25 October 2011 led to the formation of small, submerged deltas that were rapidly removed by coastal erosion, indicating a strong interaction with the wave action coming from the southwest (Scirocco and Libeccio wind directions) and coastal dynamics.

In the submarine area of the CTNP, from the coastline to a depth of about 15 m, high-resolution multibeam data reveal blocky facies produced by subaerial landslides, distributed along the entire investigated sector (over 50 occurrences). These deposits extend seaward from the coastline for distances ranging between 50 m and 150 m, with average slopes of about 10 degrees. Beyond this zone, the seabed becomes smoother, with gentle slopes of approximately 1–2° toward the continental shelf break.

3. Materials and Methods

To provide a detailed analysis of the expected sea level rise (SLR) and flooding scenario up to 2150 AD for the Cinque Terre area, we planned a multidisciplinary workflow structured into five main phases.

(1) UAV surveys were conducted to collect high-resolution imagery of the present-day land surface, enabling the generation of ultra-high-resolution Digital Elevation Models (DEMs) and orthomosaics for the three selected coastal sectors. We then calibrated and integrated these datasets. These datasets were subsequently calibrated and integrated with topographic LiDAR data (see Section 3.1 and Section 3.2).

(2) High-resolution multibeam bathymetry data were collected during three surveys to produce very high-resolution digital marine models (see Section 3.3).

(3) A geodetic analysis was performed to estimate current rates of vertical land movement (VLM) using data from the nearby GNSS stations (LASP, La Spezia; LRCI, Lerici) combined with ground velocity measurements from InSAR data provided by the Copernicus European Ground Motion Service (available at https://land.copernicus.eu/en/products/european-ground-motion-service, accessed on 8 January 2026) (see Section 3.4 and Supplementary Materials Figures S1 and S2).

(4) Geodetic and topographic datasets were integrated with the regional IPCC-AR6 projections (SSP1-2.6, SSP3-7.0 and SSP5-8.5 climatic scenarios) to estimate the upper bounds of the expected sea levels for 2050, 2100, and 2150 CE (see Section 3.4 and Section 3.5).

(5) Significant wave heights associated with return periods (RT) of 1 and 100 years were modeled using WAVEWATCH (WW3) version 6.07 software (see Supplementary Materials Files S1 and S2). Within this analysis, high-resolution bathymetry and digital terrain models were considered (see Section 3.6 and Section 3.7).

Using the outputs of these phases, we created detailed maps of the current and projected coastline positions and estimated the multitemporal evolution of the inland extent of the marine flooding and shoreline changes along the investigated coastal sectors.

In addition, we analyzed storm surge scenarios (SS) in both present-day and future sea-level rise projections.

3.1. UAV Surveys and Digital Elevation Model Modeling

Figure 2 shows the data processing workflow to produce ultra-high-resolution DEMs and ortho-images, suitable for the coastal hazard assessment and SLR scenarios.

To generate high-resolution DEMs of the investigated subaerial coastal areas, we planned and conducted a low-elevation aerial photogrammetry survey using a UAV [19]. We acquired aerial images using a DJI INSPIRE 2 drone equipped with a Zenmuse camera (DJI, Shenzhen, China) in early October 2017. 865 validated, high-resolution RGB images (24 megapixels each) were acquired from a flight altitude of about 70 m, which was the maximum allowed in Italy at the time of surveys. The survey covered an area of approximately 0.6 km² with an average ground sampling distance between 2.05 cm and 2.48 cm. Forward and lateral overlaps were 70% and 60%, respectively. We carried out a dedicated topographic survey to measure a set of Ground Control Points (GCPs) to support the aerial photogrammetric processing. We measured a total of 42 GCPs, including both temporary artificial markers specifically deployed for the survey (e.g., ground targets) and stable, clearly identifiable natural or man-made features already present in the study area (e.g., rock outcrops or structural corners), which were used as control points. We performed the measurement using the GNSS/RTK technique with a rover receiver and CORS reference stations using a LEICA 1230 Global Navigation Satellite System (GNSS) receiver equipped with a Leica AX1202 antenna (Leica Geosystems AG, Heerbrugg, Switzerland). The nearest reference station named LASP, located near the city of La Spezia, was chosen from the SmartNet network (http://it.smartnet-eu.com). The baseline length ranged between 10 and 13 km. Measurement accuracy was about 2–3 cm in the horizontal components and 3–5 cm (or better) in the vertical component for each GCP, with a mean RMS error better than 1 cm horizontally and 2 cm vertically.

3.2. UAV Data Processing

We generated orthoimages in the WGS84 projection using Pix4D^® (version 4.4.12) and MetaShape^® (version 1.7.5). Both software packages are widely used to process terrestrial, submarine and aerial imagery acquired by high-resolution cameras onboard UAVs or aircraft [20]. Images are converted into high-resolution point clouds, surface models and ortho-mosaic images, textured 3D and simplified CAD models, usable for a wide range of GIS and CAD applications (Figure 2).

To reference elevations above the sea level, we converted altitude values into orthometric altitude, taking into account the Italian Geoid undulation. Specifically, we corrected GCPs elevations collected in the WGS84 reference system for the instrumental height offset between the antenna (mounted on the top of the pole) phase center and the ground surface. We then converted the data to orthometric elevations using the ConveRgo software, released by the Italian Istituto Geografico Militare (www.igmi.org), which includes the latest Geoid model of Italy ITALGEO2005. Finally, the vertical Datum was referred to the mean sea level at the tide gauge station of Genova in 1942 AD [21].

During the processing, we estimated aerial triangulation and the exterior orientation (position and rotation angles) of each image (Figure 2). We then generated the DEM through two steps: point cloud generation and the production of the DEM Grid. Using the estimated exterior orientation, we produced individual orthoimages, which we subsequently mosaicked into a single orthophoto to create high-resolution DEMs.

The combination of high-resolution, low-altitude UAV imagery and a dense, well-distributed GCP network enabled high-accuracy georeferencing of the point cloud and the generation of high-resolution digital terrain models.

The average ground sampling distance was estimated with accuracy better than 2.48 cm/pixel in the Monterosso and Vernazza areas, confirming the high-resolution of the orthophotos. We used these data to extract topography, which we combined with multibeam bathymetry to generate a continuous DEM covering the whole coastal area investigated in this study.

3.3. High-Resolution Multibeam Bathymetry (MBES)

We conducted high-resolution multibeam bathymetry surveys in September 2016 and October 2017 along the coast of CTNP, covering water depths between 0.1 m and 50 m. The surveys were performed using a 7 m long vessel equipped with a pole-mounted Teledyne RESON SeaBat 7125 SV2 (TELEDYNE RESON, Slangerup, Denmark) (400 kHz) with a footprint size of 1° × 0.5. Boat positioning and inertial data (IMU) were collected using an Applanix PosMV Wave Master V5 (Trimble Applanix, Richmond Hill, ON, Canada), while sound speed water parameters were recorded using two Valeport sound speed probes, both continuously operating along the water column to the maximum depth investigated [22]. Real-time vessel positioning was provided by a temporary GNSS master base station located on land in the municipality of Monterosso and transmitted via UHF radio signals. We recorded raw GNSS data at both the master base station (Trimble SPS855 receiver equipped with a Zephir 2 antenna, Trimble Inc., Westminster, CO, USA) and at the onboard rover (Applanix POSMV WM) for Post-Processing Kinematics (PPK) corrections processed using PosPAC MMS software. We processed a master-station, GNSS raw data against the permanent GNSS RING network (https://webring.gm.ingv.it/, accessed on 9 January 2026) framed in RDN (Rete Dinamica Nazionale 2008, https://www.igmi.org/en/direzione-geodetica/progetto-rdn-rete-dinamica-nazionale, accessed on 9 January 2026), with respect to the closest permanent station at Lerici (LRCI station).

After processing GNSS and inertial data (SBET data), we replaced real-time navigation data with post-processed solutions using Caris Hips and Sips 11.4 hydrographic professional software (Teledyne CARIS, Fredericton, NB, Canada), achieving centimetric accuracy in planimetry and altitude. To obtain the official national bathymetric orthometric elevation, we converted the collected GNSS data from ellipsoid elevation to orthometric elevation using the ITALGEO2005 model, which is the official Italian geoid model. During the last survey, we used a Merlin laser scanner mounted on the boat’s cabin, interfaced with the Applanix system for motion compensation and GNSS positioning. The Applanix data were processed with PosPac MMS and the LiDAR data were calibrated and processed with the PDS2000^® acquisition software (Teledyne RESON, Rotterdam, The Netherlands). The objective of the LiDAR survey, integrated with the MBES survey, was to cover the entire coastal area of the Cinque Terre Park and generate a land–sea Digital Terrain and Marine Model (DTMM) useful for future comparisons following gravitational instability events along the entire coastal area.

3.4. Geodetic Data

Geodetic analysis for the estimation of the current VLM included tidal correction for the zero-elevation reference and the high water level (HWL), along with GNSS and InSAR data from the Copernicus European ground Motion Service—EGMS—(https://land.copernicus.eu/en/products/european-ground-motion-service, accessed on 26 February 2026). To estimate the current vertical crustal velocity, we used the nearest GNSS stations, LASP (La Spezia) and LRCI (Lerici), both belonging to the RING network managed by INGV (https://webring.gm.ingv.it/, accessed on 26 February 2026). We obtained the crustal velocity at these stations following the approach described in [23]. The vertical velocity was estimated by applying a classic trajectory model [24] to displacement time series computed in the IGb20 reference frame, which is the latest GNSS realization of the ITRF2020 global reference frame [25]. Both stations show weak (sub-mm/yr) subsidence with rates of −0.57 ± 0.27 mm/yr at LASP and −0.8 ± 1 mm/yr at LRCI. The LASP provided the most robust estimate, based on a time-series longer than 16 years (Figure 3), whereas the shorter LRCI record, ~3.8 years long, did not allow robust vertical ground velocity estimation [26] [14]. EGMS data indicated average VLM between −1.18 mm/yr and −1.4 mm/yr along the Cinque Terre coast, particularly in the Monterosso and Vernazza areas, with RMSE values of 1.07 and 1.4 mm, respectively. The subsiding trend estimated by InSAR exceeds GNSS measurements by less than 1 mm/yr (Figure 3 and Supplementary Materials Figure S1a,b). Anyway, this discrepancy falls within the accuracy of the technique, as reported in [14], and may also reflect the shorter time recordings of the available InSAR observations (2019–2023) compared with the GNSS record at LASP.

In addition to GNSS and InSAR data, we analyzed monthly mean sea-level records from the tide gauge stations of Genova and La Spezia. The first, which is representative of the historical sea level records for this area, is managed by the Italian Hydrographic Institute of the Navy (Istituto Idrografico della Marina Militare), while the second is managed by the Italian Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA) (www.mareografico.it, accessed on 26 February 2026). The aim was to verify whether the recent regional behavior is consistent with a rising or stationary sea-level signal. We focused on the historical Genova series, available since 1884, and used La Spezia as a local consistency check. The analysis shows a clear positive trend at Genova of 1.33 ± 0.08 mm/yr over 1884–2025, whereas the La Spezia record does not show a meaningful trend over 2010–2025 because the series is short and discontinuous (see Figure S2 in the Supplementary Materials). Therefore, the local historical evidence supports a non-stationary rising sea-level background for the Ligurian sector, while confirming that the La Spezia record is presently insufficient for robust trend estimation. Because of the short duration of the recordings at the tide gauge of La Spezia and the interruption between 2015 and 2019, we considered the trend estimated at Genova as the most representative long-term sea-level indicator for the investigated area.

3.5. Relative Sea-Level Rise Projections and Flooding Scenarios for 2030, 2050, 2100 and 2150 CE

Sea-level projections used in this study are derived from AR6 datasets on a 1° × 1° worldwide grid. These projections combine multiple geophysical contributions (e.g., AIS, GIS, glaciers, land water storage, thermal expansion and GIA), each represented through probabilistic distributions (107 quantiles per grid point). Uncertainty in the total projected sea-level change (at the grid cell including the Cinque Terre area) is quantified using a quadrature approach, assuming independence among error sources. Specifically, the total uncertainty is calculated as (i) uncertainty from AR6 projections (excluding the GIA contribution) [27,28], estimated from the spread between the 32nd and 68th percentiles (i.e., an approximation of ±1 standard deviation), (ii) observational uncertainty from GNSS and InSAR-derived velocity measurements. The AR6 uncertainty reflects the combined spread of contributing processes, while observational uncertainties capture measurement errors in vertical land motion estimates.

Therefore, to estimate future local sea level projections for the Cinque Terre coast, we incorporated the contribution of VLM into regional AR6 sea-level projections for the interval 2030–2150. We retrieved the data available from the NASA Sea Level Portal (https://sealevel.nasa.gov/), previously available through the JPL Physical Oceanography Distributed Active Archive Center (PO-DAAC, https://podaac.jpl.nasa.gov/).

Data were provided for the Shared Socioeconomic Pathways (SSP1-2.6, SSP3-7.0 and SSP5-8.5) and the global warming levels (Tlim = 2 °C, 4 °C and 5 °C) scenarios considered in IPCC AR6 (www.ipcc.ch). In this study, we followed the approach described in [29] and refined the projections by incorporating a mean current VLM trend of 0.99 mm/yr, as estimated from the LASP GNSS station and InSAR data (see Section 3.4, Figure 4, and Supplementary Materials Figure S1a,b).

We assumed a linear continuation of the VLM rate up to 2150 in the absence of unpredictable tectonic activity. Since the latter effects can be long-term episodic events, a linear trend for the next few decades was considered the best approximation for incorporating the VLM signal into the relative sea-level projections presented in this study.

3.6. Storm Surge Analysis

Storm-surge and wave-driven coastal flooding scenarios were derived by combining an expeditive statistical characterization of offshore storm conditions with process-based nearshore hydrodynamic simulations. Offshore boundary conditions were extracted from the Mediterranean hindcast/forecast system developed at the University of Genoa [30], based on WRF wind forcing and WAVEWATCH III wave modeling for 1979–2016. For the Cinque Terre coastal area, grid point 000330 was used as a reference. The forcing variables transferred to the nearshore simulations were significant wave height (Hs), peak period (Tp), and mean wave direction, coupled with the relative sea-level increments associated with the RSLR scenarios. High-resolution bathymetry, merged land–sea topography, and present-day coastal geometry were used as morphologic inputs. The long-term wave climate is dominated by storms approaching from the south-west sector (Figure 5A), while the joint distribution of Hs and Tp shows the expected increase in Tp with Hs, with the most frequent sea states at low Hs and Tp around 4–8 s and energetic events extending towards Tp of about 10–12 s (Figure 5B).

Extreme conditions were quantified through a Peak Over Threshold approach, selecting storm peaks above the 0.98 percentile and enforcing a minimum inter-arrival time of 12 h between independent events. Return levels of Hs were estimated by fitting a Generalized Pareto Distribution (GPD), producing the omni-directional return-period curve and associated confidence bounds (Figure 6). The derived extreme wave conditions, combined with the RSLR increments, were propagated to the nearshore and surf zone using Delft3D with online coupled FLOW-SWAN modeling [31], which accounts for two-way wave-current interaction and wave setup. Simulations were performed in depth-averaged (2DH) mode, with SWAN run in quasi-nonstationary hot-start configuration. A full site-specific calibration/validation of the nearshore model against measured run-up or water-level time series at Monterosso and Vernazza was not possible because long, event-resolving observational series are not available for the two sites. For this reason, the modeling framework should be interpreted as a scenario-based hazard assessment rather than as a deterministic reconstruction of single historical events. Nevertheless, model forcing is grounded in a regional hindcast previously validated for the Mediterranean [30], while the consistency of offshore wave conditions is independently supported by the good agreement between the OS-IS measurements and the ISPRA buoy data in the adjacent sector (Section 3.7; Figure 7). The main sources of uncertainty are therefore associated with (i) the statistical extrapolation of extremes, (ii) the use of an omni-directional analysis, (iii) the 2DH simplification, and (iv) the assumption of present-day morphology in long-term simulations; where relevant to local morphodynamics, sediment-transport and bed-update routines were activated through reduced-complexity settings to provide first-order feedback on bathymetric sensitivity.

3.7. OS-IS: A Dedicated Sea-Wave Monitoring System

The storm-surge and wave-modeling analysis described in the previous section for coastal hazard assessments strongly depends on the availability of observational data, primarily from offshore buoys. In addition to these data, to have further observations to strengthen our analyses, we experimented with Ocean Seismic—Integrated Solution (OS-IS) System [32], an innovative monitoring approach. This is based on the measurement of micro-seismic signals generated by sea waves and recorded by a high-sensitivity three-axis accelerometer installed onshore, typically in the basement of existing buildings, without any device deployed at sea. The operating principle relies on the well-established physical mechanism by which the interaction of ocean waves, particularly counter-propagating wave components, induces pressure oscillations at the seabed that generate micro-seismic signals, detectable several kilometers inland [33]. By processing the power spectral density of the recorded ground acceleration and applying calibrated physical or data-driven models, sea-state parameters, such as significant wave height and wave period, can be retrieved.

Within the framework of an international project (RAISE project Spoke 3—Sustainable environmental caring and protection technologies, https://www.raiseliguria.it/en/), four OS-IS stations have been deployed along the coast of Liguria at Livorno, Genoa, Andora (Savona) and Bonassola (La Spezia). The latter is located close to the Cinque Terre and, given its location, it is particularly relevant for the present study. A comparison between the significant wave height (Hs) measured by the OS-IS station and the corresponding observations from the ISPRA wave buoy of La Spezia (45° 55′ 45″ Lat N; 09° 49′ 40″ Lon E) is presented in Figure 7. This buoy is managed by Istituto Superiore per la Protezione e la Ricerca Ambientale (https://www.isprambiente.gov.it/it), and is located offshore the investigated area of Cinque Terre. The analysis shows the good agreement between the OS-IS data of wave height measurements with buoy observations, thus supporting and strengthening the storm surge analyses reported in Section 3.6.

4. Results

The estimated RSLR projections from 2030 to 2150 CE for the chosen socioeconomic pathways and global warming levels along the Cinque Terre coast (Figure 4) correspond to the flooding scenarios derived from the high-resolution marine and terrestrial topography and coastal cross sections shown in Figure 8, Figure 9, Figure 10 and Figure 11, with quantitative values reported in

Table 1. Multitemporal (2024–2150) extension of expected flooding areas according to different RSLRs and SSP scenarios for Monterosso (a) and Vernazza (b). Right columns: Minimum and maximum expected extension (m²) of flooding areas for storm surge scenarios with respect to minimum and maximum expected water levels (m) in 2150 for the SSP5-8.5 scenario.

(a)
Monterosso
Year	Ordinary Condition with Relative Sea Level Rise (RSLR)						Storm Surge 2024 and 2150 (SSP5-8.5)
	SSP1-2.6		SSP3-7.0		SSP5-8.5		Affected Area (m2)		MSL (m)
	Underwater Area (m2)	RSLR (m)	Underwater Area (m2)	RSLR (m)	Underwater Area (m2)	RSLR (m)	Min	Max	Min	Max
2024	0	0	0	0	0	0	69,389	80,911	5.42	7.67
2030	844	0.05 ± 0.06	844	0.05 ± 0.06	896	0.05 ± 0.06	71,183	81,646	5.60	7.86
2050	1454	0.14 ± 0.11	1677	0.17 ± 0.11	1755	0.18 ± 0.11	71,183	81,646	5.60	7.86
2100	4105	0.39 ± 0.28	7028	0.59 ± 0.28	8629	0.68 ± 0.28	73,885	83,590	6.10	8.35
2150	7201	0.60 ± 0.52	15,175	1.04 ± 0.52	17,227	1.17 ± 0.52	76,175	86,667	6.59	8.84
(b)
Vernazza
Year	Ordinary Condition with Relative Sea Level Rise (RSLR)						Storm Surge 2024 and 2150 (SSP5-8.5)
	SSP1-2.6		SSP3-7.0		SSP85		Affected Area (m2)		MSL (m)
	Underwater Area (m2)	RSLR (m)	Underwater Area (m2)	RSLR (m)	Underwater Area (m2)	RSLR (m)	Min	Max	Min	Max
2024	0	0	0	0	0	0	15,986	17,081	9.47	12.20
2030	357	0.05 ± 0.06	357	0.05 ± 0.06	390	0.05 ± 0.06	16,042	17,217	9.65	12.38
2050	741	0.14 ± 0.11	865	0.17 ± 0.11	907	0.18 ± 0.11	16,042	17,217	9.65	12.38
2100	1811	0.39 ± 0.28	2687	0.59 ± 0.28	3095	0.68 ± 0.28	16,221	17,525	10.15	12.88
2150	2730	0.60 ± 0.52	4510	1.04 ± 0.52	4852	1.17 ± 0.52	16,379	18,161	10.64	13.36

.

The historical sea-level analysis discussed in Section 3.4 indicates that the Ligurian sector is characterized by a long-term rising signal rather than by stationary behavior, which supports the physical plausibility of the projected RSLR scenarios.

The SLR levels reported in Table 1 for the three SSP scenarios include the mean land subsidence of −0.57 ± 0.27 mm/yr, inferred from the LASP GNSS station. We selected this station because it provided a longer and more continuous time series than the LRCI station and showed, within the uncertainties of the applied techniques, comparable trends with InSAR data from the Copernicus EGMS. Our results indicate that, under the SSP5-8.5 scenario, the maximum expected RSLR relative to 2020 reaches 0.05 ± 0.06 m, 0.18 ± 0.11 m, 0.68 ± 0.28 m, and 1.17 ± 0.52 m for 2030, 2050, 2100, and 2150, respectively (Table 1). These values should be interpreted as scenario-based hazard envelopes because they combine AR6 regional sea-level uncertainty, local VLM uncertainty, and a static representation of present-day topography and bathymetry. Given the gentle slope of the beaches at Monterosso and Vernazza, this rise is expected to trigger a continuous retreat of the coastline position. In particular, at the upper limit of the projected RSLR for 2150 in the SSP5-8.5 scenario, significant land flooding is expected (Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12 as well as Table 1).

Concerning storm surges under RSLR conditions, our analysis shows that water levels can reach maximum wave run-up values of 8.3 m and 11.7 m for the investigated sections under current mean sea-level conditions, while the highest compound levels are obtained when extreme events are superimposed on the 2150 SSP5-8.5 scenario (Figure 13, Figure 14, Figure 15 and Figure 16 and Table 1; see also Supplementary Materials Files S1 and S2). These values should be interpreted as upper-bound scenario estimates of combined wave setup and run-up rather than as deterministic predictions for a single future storm.

Figure 11. Cross-sections of the expected RSLR up to 2150 across the coast of Vernazza. For the location of the sections, see Figure 14. Distance along the cross sections is reported in meters (m).

Figure 11. Cross-sections of the expected RSLR up to 2150 across the coast of Vernazza. For the location of the sections, see Figure 14. Distance along the cross sections is reported in meters (m).

Figure 12. Extension of the expected flooded surfaces: (a,b) for Monterosso and (c,d) Vernazza. Results are based on the RSLR projections for the SSP1-2.6 (blue), SSP3-7.0 (green) and SSP5-8.5 (red) climatic scenarios for 2030, 2050, 2100 and 2150. For the low emission scenario SSP5-2.6, the estimated flooded surface at Monterosso is 7201 m², while at Vernazza, it is 2730 m². For the high emission scenario SSP5-8.5, the flooded surface at Monterosso is 17,227 m² and 4852 m² at Vernazza. In (c,d), are reported the comparison of the flooded surfaces (in m²) between the ordinary (in black) and the Storm Surge (in grey) conditions for the SSP5-8.5 shared socioeconomic pathway.

Figure 12. Extension of the expected flooded surfaces: (a,b) for Monterosso and (c,d) Vernazza. Results are based on the RSLR projections for the SSP1-2.6 (blue), SSP3-7.0 (green) and SSP5-8.5 (red) climatic scenarios for 2030, 2050, 2100 and 2150. For the low emission scenario SSP5-2.6, the estimated flooded surface at Monterosso is 7201 m², while at Vernazza, it is 2730 m². For the high emission scenario SSP5-8.5, the flooded surface at Monterosso is 17,227 m² and 4852 m² at Vernazza. In (c,d), are reported the comparison of the flooded surfaces (in m²) between the ordinary (in black) and the Storm Surge (in grey) conditions for the SSP5-8.5 shared socioeconomic pathway.

Figure 13. (A) Location of the land–sea altimetric sections along the bay of Monterosso. In (B) a detail of (A) across Monterosso promontory. Present-day potential scenario for a storm surge with return times of 100 years with respect to the current mean sea level. M1–M9 are the cross sections presented in Figure 9 for the storm surge assessment. Numbers are the maximum runup values. See color palette for reference and Supplementary Materials (Figure S1a,b and Files S1 and S2). (Zone UTM32 N; the length of the cross sections is reported in meters).

Figure 13. (A) Location of the land–sea altimetric sections along the bay of Monterosso. In (B) a detail of (A) across Monterosso promontory. Present-day potential scenario for a storm surge with return times of 100 years with respect to the current mean sea level. M1–M9 are the cross sections presented in Figure 9 for the storm surge assessment. Numbers are the maximum runup values. See color palette for reference and Supplementary Materials (Figure S1a,b and Files S1 and S2). (Zone UTM32 N; the length of the cross sections is reported in meters).

Figure 14. (A) Storm surge scenario at Monterosso for the year 2150 in RSLR condition for an extreme event with return times of 100 years, considering the SSP5-8.5 scenario. M1–M9 are the cross sections for the storm surge assessment presented in Figure 9. In (B) a detail of (A) across Monterosso promontory. Numbers are the maximum runup values. See color palette for reference and Supplementary Materials Files S1 and S2. (Zone UTM32 N; the length of the cross sections is reported in meters).

Figure 14. (A) Storm surge scenario at Monterosso for the year 2150 in RSLR condition for an extreme event with return times of 100 years, considering the SSP5-8.5 scenario. M1–M9 are the cross sections for the storm surge assessment presented in Figure 9. In (B) a detail of (A) across Monterosso promontory. Numbers are the maximum runup values. See color palette for reference and Supplementary Materials Files S1 and S2. (Zone UTM32 N; the length of the cross sections is reported in meters).

Figure 15. (A) Present-day potential scenario for a storm surge with return times of 100 years with respect to the current mean sea level for Vernazza. V1–V3 are the cross sections for the storm surge assessment, presented in Figure 11. Numbers are the maximum runup values. In (B) a detail of (A) across the harbor. See color palette for reference and Supplementary Materials (Figure S1a,b and Files S1 and S2). (Zone UTM32 N; the length of the cross sections is reported in meters).

Figure 15. (A) Present-day potential scenario for a storm surge with return times of 100 years with respect to the current mean sea level for Vernazza. V1–V3 are the cross sections for the storm surge assessment, presented in Figure 11. Numbers are the maximum runup values. In (B) a detail of (A) across the harbor. See color palette for reference and Supplementary Materials (Figure S1a,b and Files S1 and S2). (Zone UTM32 N; the length of the cross sections is reported in meters).

Figure 16. (A) Storm surge scenario for Vernazza for the year 2150 in RSLR condition for an extreme event with return times of 100 years, considering the SSP5-8.5 scenario. V1–V3 are the cross sections for the storm surge assessment presented in Figure 11. Numbers are the maximum runup values. In (B) a detail of (A) across the harbor. See color palette for reference and Supplementary Materials Files S1 and S2. (Zone UTM32 N; the length of the cross sections is reported in meters).

Figure 16. (A) Storm surge scenario for Vernazza for the year 2150 in RSLR condition for an extreme event with return times of 100 years, considering the SSP5-8.5 scenario. V1–V3 are the cross sections for the storm surge assessment presented in Figure 11. Numbers are the maximum runup values. In (B) a detail of (A) across the harbor. See color palette for reference and Supplementary Materials Files S1 and S2. (Zone UTM32 N; the length of the cross sections is reported in meters).

5. Discussion

The rocky coast of Cinque Terre is a sediment-limited, infrastructure-constrained system in which even moderate RSLR can trigger disproportionate functional impacts. This behavior differs from that of low-lying Mediterranean coastal plains, where hazard is mainly expressed as large horizontal inundation surfaces under relatively small vertical increments [13,15]. In Cinque Terre, by contrast, the critical transition is the loss of narrow pocket beaches and low-elevation harbor and service areas, which removes the already limited accommodation space between sea and steep hinterland. In this respect, the setting is more comparable to other Mediterranean pocket beach and island environments, such as Lipari [27], than to broad coastal plains, although the concentration of rail, port and tourism functions at the mouths of narrow catchments makes Cinque Terre particularly exposed.

The progressive depletion of the beach prisms expected by 2150 at Vernazza and Monterosso is important not only because it enlarges the flooded footprint, but because it alters nearshore dissipation processes. These beaches are fed episodically by coarse sediment delivered during flash-flood events, while for long periods they remain supply-limited. Under RSLR, the active profile is forced landward and upward, beach width decreases, and wave breaking occurs closer to quay walls, piers, retaining structures, and railway-related assets. This promotes a transition from partial dissipation across the beach face to more reflective conditions, increasing swash excursion, overtopping propensity, and local toe scour. Therefore, the relevant process is not only static submersion, but the coupling between sedimentary dynamics, wave transformation, morphology and local geology.

The interaction between RSLR and storm surges is also nonlinear. A higher mean water level allows larger waves to reach the shoreline before breaking, increases wave setup, and amplifies run-up on steep reflective beaches and armored waterfronts. In confined settings such as Monterosso and Vernazza, even decimetric increases in mean water level can therefore translate into disproportionately larger hydraulic loads on low-elevation infrastructures. This interpretation is consistent with previous Mediterranean studies showing that VLM and local coastal geometry can substantially modulate the effects of regional sea-level rise [10,11,12,14]. In the Cinque Terre case, the absolute VLM rate is modest, but when accumulated to 2150, it becomes non-negligible and acts as an additional accelerator of coastal exposure.

In addition to long-term trends, climate-driven SLR manifests at high frequency too, in the form of extreme weather events, such as severe storm surges. For example, the storm surge of 30 October 2018 destroyed the ports of Rapallo and Portofino [34] not far from Cinque Terre; similarly, in January 2026, a storm surge due to Mediterranean Hurricane Henry devastated the coast of southern Italy with impressive force, destroying roads, houses, railway lines, and port infrastructure (https://www.climameter.org/20260120-mediterranean-cyclone-harry, accessed on 9 March 2026). In recent years, such events have become increasingly more frequent and energetic, causing significant economic and social damage [35].

The maps and the cross-sections reported in Figure 9, Figure 11, Figure 12, Figure 13, Figure 14, Figure 15 and Figure 16 show the first detailed flooding scenarios associated with RSLR along the investigated coastal areas during the 2030–2150 period under three different expected socioeconomic considered pathways. VLM accelerates the coastal submergence, with direct consequences for coastal morphology and infrastructure. In this regard, the socioeconomic resilience of the Cinque Terre strongly depends on the railways of the Rete Ferroviaria Italiana [36], which might be increasingly compromised by sea level rise and escalating wave energy. Extreme storm surge scenarios indicate a maximum wave run-up of 13.48 m, a threshold that threatens tunnel apertures and overhead contact lines. This risk is exacerbated by sea surface rising temperatures (+0.045 °C/year), as recorded by the Smart Bay Santa Teresa observatory [37]. In fact, warmer waters contribute to storm surge intensification, accelerated saline corrosion, and increased structural damage due to debris impact within the tunnels, potentially causing systemic service disruptions. In addition to railway lines, infrastructural vulnerability is more severe at terminal stations located at the outlets of narrow catchments. Rising sea levels can induce “backwater effects” that impede the efficient discharge of flash floods. This hydraulic synergism, as reported in [38], risks turning railway underpasses into preferential routes for coastal flooding, posing risks to logistics and passenger safety. In this context, the transition toward predictive maintenance and adaptive engineering represents a key requirement to prevent the systemic collapse of the regional infrastructure network [39].

Because coastal hazards remain poorly understood by local populations [40,41], high-resolution datasets and visualization tools, such as the predictive maps presented in this study, can improve awareness among citizens and stakeholders [42]. The flooding hazards presented in this study can support local stakeholders and decision-makers in developing appropriate mitigation and adaptation strategies to address future SLR impacts. Our study details previous regional [38] and local scenarios [9,10,11,12,13,14,15,16] of RSLR along Mediterranean coasts.

5.1. Monterosso

The area of Monterosso consists of an open bay to the west- southwest, with a beach zone bordered by high slopes from 30° up to 65°. The coastal area consists of three pocket beaches confined between the docks and the breakwater (Figure 8).

The westernmost one has a subaerial beach extension of about 160 m and a depth of about 50 m and is bordered by a small pier and an artificial structure used as a parking lot. The second, larger central one has a width of about 400 m and a depth of about 30 m. The third, located in the eastern sector, is about 170 m wide and about 28 m deep. The average slope for all three beaches is about 5°. The small piers and harbor area placed at the east and north of the bay are located at elevations of between about 0.7 and 3 m and host roads, pavements, and other facilities for mooring, commercial shipping and nautical tourism.

In Figure 8, the expected flooded area for the worst scenario, SSP5-8.5, for 2030 (yellow), 2050 (orange), 2100 (red), and 2150 (dark red), is shown. Maps and cross sections show how docks and beaches will be severely flooded by 2100, while by 2150, about the whole harbor area will likely be submerged. In particular, the cross-sections in Figure 9 highlight that in the absence of any reinforcement works and the raising of the docks by 2150, the expected level of submersion could cause loss of use of the port infrastructures and the full beach retreat during storm surges. The external part of the breakwater that protects the parking area will be partially submerged and will then be increasingly exposed to storm surges, becoming unable to protect the port basin. Figure 13 and Figure 14 show the storm surge (SS) scenarios for return times of 1 and 100 years, respectively. A similar impact is expected by 2150 for an SSP3-7.0 scenario (Figure 8). In the case of the most optimistic scenario, SSP1-2.6, the dock is still mostly safe while a large part of the beach will retreat (Figure 8 and Figure 9).

5.2. Vernazza

The small coastal tract of the Vernazza village has variable elevations and steeper slopes than in Monterosso, with coastal elevations ranging from the sea-level up to several meters above sea level. Here, the coastline is indented and characterized by elevated cliffs consisting of rocky outcrops with slopes of up to 87°. In this area, there are two pocket beaches located in small semi-enclosed inlets, one in the village of Vernazza with a breakwater and a pier for tourist boats. While the other is located at the foot of the rocky cliff that hosts part of the Vernazza village at the top. The topography of these two small beaches is characterized by steep slopes and consists of large pebbles and boulders, with average gradients between 3° and 9°. The coastal area includes housing structures at elevations from about 1 m a.s.l., with temporary accommodation facilities for bathing and touristic activities located close to sea-level. The southern beach is now protected by an artificial reef, but it has undergone significant changes over time due to sudden deposits produced by flash floods, resulting in the formation of small deltas alternating with severe erosion caused by sea storms.

Figure 10 shows the expected flooded area for the SSP5-8.5 scenario between 2030 and 2150. The two pocket beaches will already be partially flooded by 2100, while almost all of them will be exposed to submersion by 2150 (Figure 10, Figure 11, Figure 15 and Figure 16 and Table 1). The cross-sections in Figure 11 show that by 2150, the expected RSLR will likely cause the submersion of both pocket beaches and the loss of use of the touristic infrastructures.

Given the height of most of the buildings above sea-level, they will not be exposed to submersion, even in the worst-case scenario of SSP5-8.5 (Figure 10, Figure 11 and Figure 12). On the other hand, the pocket beaches will be severely affected by the RSLR, even in the more optimistic scenarios SSP1-2.6 and SSP3-7.0 (Figure 10, Figure 11 and Figure 12).

5.3. Sources of Uncertainty and Limitations

Four uncertainty domains should be considered when interpreting the results. First, the RSLR projections inherit the uncertainty ranges of the IPCC AR6 regional scenarios and of the adopted local VLM trend. Second, the historical sea-level analysis relies mainly on the long Genova record, whereas the La Spezia series is too short and discontinuous to constrain local secular behavior with the same robustness. Third, the storm-surge component depends on statistical extrapolation of offshore extremes and on the transfer of those conditions to the nearshore through a depth-averaged Delft3D setup, without site-specific long-term run-up measurements for formal calibration at Monterosso and Vernazza. Fourth, the inundation maps are based on present-day topography and bathymetry and do not fully resolve long-term morphodynamic feedbacks, such as beach lowering, episodic sediment inputs from flash floods, cliff-derived block supply, or structural modification of waterfronts.

These limitations do not invalidate the comparative value of the scenarios, but they require that the results be read as plausible hazard envelopes rather than as exact forecasts of future shoreline position or overtopping depth for a specific event. In practical terms, the maps are likely robust for identifying the most exposed sectors and for comparing the relative increase in hazard through time and across SSPs, whereas local engineering design would require event-scale calibration, dedicated nearshore monitoring, and repeated updates of the land–sea morphology.

6. Conclusions

Due to the combined effect of regional sea-level rise, local subsidence, steep coastal morphology, and limited sediment accommodation space, the Cinque Terre coast is expected to experience a marked increase in marine hazard well before 2150 CE. The long historical tide-gauge record from Genova indicates that the regional background signal is already rising, and the projections presented here show that even a modest additional vertical increment can produce substantial retreat of pocket beaches and recurrent flooding of low-elevation harbor and service areas.

Monterosso and Vernazza are vulnerable in different but complementary ways. At Monterosso, the main criticalities concern the harbor, beaches, parking area, and low quay elevations; at Vernazza, the greatest impacts are concentrated on the two pocket beaches and waterfront tourism infrastructures, while most of the historical village remains at safer elevations except during the most severe compound events. In both cases, storm surges superimposed on RSLR substantially increase exposure, confirming that static inundation alone underestimates future hazard.

The main planning implication is that adaptation should be spatially targeted and process-aware. Priority measures include protection or elevation of low quays and service areas, improved drainage and backflow control at underpasses and outlets, monitoring and warning thresholds for storm-driven operational shutdowns, and periodic updating of coastal morphology and sea-level scenarios.

Future work should include site-specific calibration with dedicated nearshore observations and explicit morphodynamic simulations, especially for those sectors where railway infrastructures run close to the shoreline and where compound marine–fluvial interactions may amplify damage.

In Monterosso, piers, breakwaters, beaches, and the parking area—located on a low-elevation promontory—will be increasingly exposed to RSLR, particularly during storm surges. This exposure will likely limit their functionality during extreme events, potentially already before 2100 CE. Although not specifically addressed in this study, other coastal sectors crossed by the railway lines connecting the Cinque Terre villages may also be significantly affected by RSLR and deserve further investigation.

Finally, we recognize the need to evaluate current coastal erosion processes to better understand and predict the future evolution of the coastline as a result of sea level rise. However, this study is beyond the scope of this work and requires further specific investigations in the submerged and emerged sectors of this area, which have not yet been conducted in detail.

The approach presented in this study can be applied to coastal areas worldwide that are exposed to SLR due to the combination of global warming and vertical land movements. In this regard, land planners and decision-makers should integrate SLR projections and flooding scenarios, such as those reported in this study, into coastal management strategies to ensure the preparedness of local populations and the protection of coastal infrastructures against future changes.