Evolution of Coastal Environments under Inundation Scenarios Using an Oceanographic Model and Remote Sensing Data

Abstract

A new methodology to map Italian coastal areas at risk of flooding is presented. This approach relies on detailed projections of the future sea level from a high-resolution, three-dimensional model of the Mediterranean Sea circulation, on the best available digital terrain model of the Italian coasts, and on the most advanced satellite-derived data of ground motion, provided by the European Ground Motion Service of Copernicus. To obtain a reliable understanding of coastal evolution, future sea level projections and estimates of the future vertical ground motion based on the currently available data were combined and spread over the digital terrain model, using a GIS-based approach specifically developed for this work. The coastal plains of Piombino-Follonica and Marina di Campo (Tuscany Region), Alghero-Fertilia (Sardinia), and Rome and Latina-Sabaudia (Lazio Region) were selected as test cases for the new approach. These coastal stretches are important for the ecosystems and the economic activities they host and are relatively stable areas from a geological point of view. Flood maps were constructed for these areas, for the reference periods 2010–2040, 2040–2070, and 2040–2099. Where possible, the new maps were compared with previous results, highlighting differences that are mainly due to the more refined and resolved sea-level projection and to the detailed Copernicus ground motion data. Coastal flooding was simulated by using the “bathtub” approach without considering the morphodynamic processes induced by waves and currents during the inundation process. The inundation zone was represented by the water level raised on a coastal DTM, selecting all vulnerable areas that were below the predicted new water level. Consequent risk was related to the exposed asset.

1. Introduction

Throughout the Quaternary period, coastal areas have undergone substantial transformations driven by many factors, including climate variability, ground motion, and other natural and anthropogenic factors and the consequent relative sea level rise (RSLR) [1,2,3,4]. Morphological coastal changes are studied worldwide due to their social and economic impacts [5,6,7,8,9,10,11,12].

The current understanding about past, present, and future SLR has been recently summarized by the Intergovernmental Panel on Climate Change (IPCC; [13]). The main findings are that (1) SLR has been faster in the past century compared to previous periods; (2) SLR has accelerated in the past two decades; (3) sea level (SL) will continue to rise over the 21st century, and in the worst emission scenario (business as usual), the global SL will increase by about 1 m by 2100; and (4) the regional mean relative SLR will also continue to rise throughout the 21st century, and will be accompanied by an increase in the number of extreme events, yielding increases in the frequency and severity of coastal flooding and erosion in low-lying areas. It should be noted that these global estimates do not include the effects of isostasy and tectonics, and should consequently be corrected, where possible, to include these factors [4,14,15]. The IPCC provides, usually once every three years, detailed reports which include, among other things, eustatic data concerning the melting of glaciers; the most recent report, issued in 2022, includes predictions up to the year 2100. These are global data that do not consider other geological movements due to volcanism, tectonics, or isostasy [16,17,18,19].

In addition to SLR, changes in coastal topography can be influenced by geological processes, such as tectonics, hydro-isostatic movement, volcanic activity, sediment flux, subsidence, and other related phenomena [17,20,21,22], which act on a variety of time scales. This study does not attempt to disentangle the individual contributions of these geological factors. Instead, it leverages satellite-based Interferometric Synthetic Aperture Radar (hereinafter InSAR) observations to estimate the average trend of the vertical ground motion within the areas investigated. This approach is facilitated by data provided by the Copernicus European Ground Motion Service (EGMS) [23,24], representing by far the most advanced and updated product for the characterization of the coastal environment. Through this methodology, we aim to provide a comprehensive assessment of how these geological dynamics collectively modulate flooding hazard, thereby enriching our understanding of coastal vulnerability in the face of climatic and geophysical changes.

A detailed knowledge of the local SLR and ground motion is essential for accurately evaluating flood hazard in low-lying coastal areas [2,4,13,25,26]. Furthermore, climate change exerts significant influence on wind regimes and storm surges, potentially exacerbating flooding [27,28]. Recent examples of coastal flooding in several countries are documented [29,30,31,32]. RSLR can also produce serious damage to the built environment [27,32,33,34,35,36], and this calls for the adoption of coastal protection and adaptation measures [37], based on a scientific assessment of the risks [38,39,40,41,42].

Despite the availability of global web-platforms such as the NOAA Sea-Level Rise Viewer, the Aqueduct Global Flood Analyser, and the Climate Central Coastal Risk Screening Tool [43,44], which have played a pivotal role in raising public awareness about the immediate and future impacts of climate change and SLR, they still provide only large-scale information, leading to a potential overestimation of future flood risks [45,46,47,48]. At the European level, scientifically validated data are provided by the Land, Marine and Climate Copernicus services, and by platforms such as Envri-FAIR and SAVECOAST [20,49,50,51,52]. The CoCliCo European Project [53] is currently developing a new, high-resolution, homogeneous mapping of the flooding risk all along the European coasts. Sophisticated hazard and risk assessment tools are also used in France and the Netherlands for national coastal risk planning and prevention [48,54,55].

At the regional scale, a variety of local platforms have been developed globally to address coastal vulnerabilities. However, this scenario presents a fragmented landscape of solutions and methodologies. Notable examples include the Coastal Storm Modelling System (CoSMoS) in the San Francisco Bay area [27,56,57]. These initiatives reflect targeted efforts to create adaptable and region-specific tools, yet they also highlight the challenge of achieving a cohesive approach to coastal risk management on a global scale.

Reliable estimates of the flooding hazard do require reliable estimates of the local SLR [58,59]. The climate models used for the latest IPCC report may be adequate at global scale, but SL results from these models often need further refinement in marginal seas. This applies to the Mediterranean Sea, which is connected to the Atlantic Ocean through the Gibraltar Strait, a narrow passage hosting complex, small-scale dynamics induced by the local bathymetry and tides [60,61,62,63,64]. Therefore, studying SL variability in the Mediterranean Sea requires the use of high-resolution regional models that downscale external forcings, initial and boundary conditions of global climate models [65,66,67,68,69,70]. Adloff et al. [71] demonstrated that incorporating improved SL information at the Atlantic lateral boundary significantly enhances the reliability of results. Along these lines, a high-resolution model of the Mediterranean Sea circulation, with an appropriate grid refinement in the Gibraltar Strait, capable of addressing the complex, hydraulically driven local dynamics, was implemented [72]. A future climate scenario simulation (2006–2100) was realized with that model, named MED16, whose results are discussed in Sannino et al. (2022) [72]. SL projections from this simulation are used in the present investigation.

Our methodology for addressing flood risk unfolds in three distinct phases. In the first phase, we utilize the EGMS and SLR data to identify coastal areas that are most vulnerable to RSLR in the forthcoming decades. To this end, we developed a Geographic Information System (GIS)-based tool which integrates the vertical ground motion data, the simulated SLR data, and topography data from a high-resolution digital terrain model (DTM) to predict the vulnerability of coastal areas subject to inundation. The subsequent phase delves into a detailed assessment of assets highly exposed to flood events. The final phase involves in situ field surveys, aimed at improving the quality of the morphological and geological reconstruction initially based only on satellite data. Such surveys may also allow to discern among all the geological components in action, such as tectonics, subsidence, lithospheric sediment loading, and compaction, glacial adjustment and changes in groundwater resulting from water exploitation [73].

In the present study, we detail the methodologies employed in conducting the first and second phase, specifically focusing on the coastal area of Piombino-Follonica and Marina di Campo, both within the Tuscany Region, Alghero-Fertilia in Sardinia, and Rome along with Latina-Sabaudia in the Lazio Region. These selected areas are notable for their low-lying features in coastal stretches characterized by high stability and a low rate of ground motion variation. We do not consider the effects of storm surge, tides, or meteorological processes that may affect the short-term evolution of coastal landscapes through morphodynamic processes and subsequent topo-bathymetric changes [74,75]. Onshore evaporation may influence the flooding process and hydrogeology of coastal areas [76], but its contribution is not considered because our analysis does not include the effect of the short-term variation in SL (i.e., tides and storm surge). To assess the impact of the geological factors on flood hazard, we used satellite InSar observations conducted over the past six years. This approach enabled us to estimate the average trend of vertical ground movement within the targeted areas.

Three of the regions under study (Fertilia, Marina di Campo, and Piana Pontina) have been recently analysed in other studies [2,25]. However, our findings reveal significant differences in the delineation of areas exposed to inundation within these sites when compared to their analyses. These disparities largely arise from updated local sea level rise (SLR) estimates, now refined through the utilization of a high-resolution model specifically developed for the Mediterranean Sea circulation, and the use of the EGMS Copernicus dataset.

The new methodological approach, implemented in the present research, aims to map coastal areas potentially subject to flooding, providing information on the evolutionary trends of the region, as support in planning adaptation strategies [46,47,77,78], and considering the different assets exposed to RSLR [79].

2. Study Area

With more than 7900 km of coastline, Italy encompasses several coastal areas potentially vulnerable to inundation from RSLR in the coming decades [16,18]. The present research focuses on five of these areas (Figure 1), characterized by significant environmental and economic value.

The first two sites, Marina di Campo and Piombino-Follonica, are located in the Tuscany Region (central Italy). The coastal areas of Tuscany host about 13% of the population of the region, along with significant touristic activities that leverage the rich environmental and cultural local resources [21,80]. The Gulf of Follonica includes almost 35 km of fine sandy coast that has experienced considerable erosion rates in the past decades, leading to shoreline retreat in many segments [80,81], while coastal processes have been deeply affected by anthropogenic sediment management [21,82,83,84]. Tuscany also includes a group of stunning small islands that constitute the Tuscan Archipelago. These islands are mainly characterized by high and rugged coastlines, including some sandy beaches, such as the one facing the Gulf of Marina di Campo on Elba Island, which is about 2 km long. As for the oceanographic context, it is noted that during autumn and winter, the main feature of marine surface circulation along the Italian western coast is the northern current, which flows cyclonically from the southern Tyrrhenian to the Latium coasts [85]. However, at about 42°N, after bordering the Bonifacio cyclone, the current moves away from the coast, heads towards the Corsica Strait (west of Elba), and then crosses it to merge with the wide cyclonic cell of the Liguro-Provencal basin. Thus, the coasts of Tuscany are not directly exposed to this current, but are subject to coastal circulations induced by it, and by the presence of numerous islands (see Bendoni et al., 2022 [86]). The current stops in spring, and in summer an anticyclonic circulation typically develops in the Corsica Strait area, which basically separates the Northern Tyrrhenian from the Ligurian Sea [87,88], and may extend into the coastal region of Tuscany.

The third site that we consider is the Alghero-Fertilia littoral, located in the bay of Alghero, on the northwestern coast of Sardinia. It is characterised by a 4 km sandy shore forming an arc with an NNW-SSE orientation. The bay is bounded by the harbour of Alghero to the south and by the smaller Fertilia harbour, at the inlet of Calich Lagoon, to the north [89]. The lagoon of Calich is fed by three small rivers, it has been partially reclaimed for agricultural use, and its connection to the sea has been armoured by the harbour of Fertilia. Detailed studies of emerged and submerged beach and seafloor habitat mapping have been conducted by the University of Sassari, highlighting complex feedback mechanisms controlling the evolution of coastal geomorphology and the relevance of anthropogenic impacts [89,90]. The extension of the lagoon was further reduced in the late 1980s. The northwestern part of Sardinia is exposed to persistent systems of waves of considerable energy, which could be suitable for the deployment of wave energy converters [91]. However, the strong indentation of the coast to the southeast of Capo Caccia provides a very good shelter for the Gulf of Alghero-Fertilia, preventing wave propagation from the dominant south-eastward direction into the Gulf, where very low wave energy flux is typically observed.

The fourth and fifth study areas are the coastal stretches of Rome (Ostia—Fiumicino) and Latina-Sabaudia, part of the Pianura Pontina (Pontina coastal plain). The Pontina coastal plain has a length of about 50 km and a width of about 20 km from the foot of the Lepine ridges to the Tyrrhenian Sea. Near the coast, it hosts the Circeo National Park, a protected area characterized by beach–dune systems that are the remnant of a barrier island that separated the coastal plain from the open sea during the Holocene sea level rise [92]. The protected coastal area of the Circeo National Park is 24 km long, arched in shape, 80–250 m wide, and partially vegetated [93]. Its morphological evolution is influenced by sediment exchanges between the seaward face of the foredune and the upper part of the backshore under the effect of incoming winds and wave motion varying between 150° and 285° [94]. The local river network has a natural component, with watercourses running from the Apennine ridge to the Tyrrhenian Sea to the N and S of the strip. There is also an artificial network of canals dug during the land reclamation of the 1930s to drain swamp water into the sea [2].

The coastline of Rome is characterised by the mouth of the Tiber River, from which the study area extends southwards for about 18 km. This stretch of coastline is characterised for about 10 km by the tourist port of Ostia, followed by a beach resort with a hinterland of the built-up area of Ostia and, beyond the Canale dei Pescatori, the urbanised areas between the Pineta di Castel Fusano and the coast. Southward, the Castel Porziano Natural Reserve is characterised by a beach–dune system [95]. The coastline has an erosive tendency, varying over time, but generally more accentuated near the mouth of the Tiber River, which has led to a strong reduction in the emerged beach and the implementation of numerous protection and nourishment measures [95]. Finally, we note that in the study areas, tidal effects are limited, as in most of the coastal regions of the Mediterranean Sea (exceptions are the north Adriatic and the Gulf of Gabes). Tidal ranges are typically 40–45 cm and are mostly associated with semidiurnal components. In

Table 1. Ranges of the astronomical tide in 2023 in four stations close to the study areas (data from ISPRA). Piombino-Follonica and Marina di Campo refer to Livorno (Tuscany), Alghero-Fertilia refers to Porto Torres (western Sardinia), Rome refers to Anzio (central Lazio), and Latina Sabaudia refers to Gaeta (southern Lazio).

Station	Piombino-Follonica	Marina di Campo	Alghero-Fertilia	Rome	Latina Sabaudia
Minimum (m)	−0.244	−0.244	−0.216	−0.268	−0.277
Maximum (m)	0.229	0.229	0.193	0.240	0.246

, we give tidal ranges over the year 2023 for each study area, referring to the four mareographic and wave buoys closest to them provided by the “Istituto Superiore per la Protezione e la Ricerca Ambientale” (ISPRA; https://indicatoriambientali.isprambiente.it/it/acque-marino-costiere-e-transizione/altezza-della-marea-astronomica-lungo-le-coste-italiane; last access 25 June 2024).

3. Data and Methodology

A GIS tool for the implementation of flood maps was realized, which can in principle be used in any coastal area of the Mediterranean Sea covered by DTMs. The input datasets used to create raster images of flood maps of the five study areas are listed in

Table 2. List of input datasets considered in the implementation of inundation scenarios. (Last access to URL 25 June 2024).

Dataset	Year	Source	Link
MED16 Ocean model	2010–2099	ENEA	Not Applicable
Topography of Italian coastal areas	2008–2012	National Geoportal	http://www.pcn.minambiente.it/mattm/
Digital Terrain Model of Sardinia	2008	Sardinia Region Geoportal	https://www.sardegnageoportale.it/areetematiche/modellidigitalidielevazione/
Bathymetry of coastal areas	2012	Bathymetric LiDAR up to 40 m of depth	http://www.pcn.minambiente.it/mattm/
European Ground Motion Service	2016–2021	COPERNICUS Land Monitoring Service	https://land.copernicus.eu/pan-european/european-ground-motion-service
CORINE Coastal Zones 2012 and 2018	2010–2014 2017–2019 Changes 2012–2018	COPERNICUS Land Monitoring Service	https://land.copernicus.eu/local/coastal-zones

; they are SLR from the Med 16 model, DTMs of the National and Regional Geoportals, EGMS, and Land cover/Land Use databases.

3.1. MED16 Ocean Model

MED16 is a high-resolution implementation of the ocean global circulation model by the Massachusetts Institute of Technology (MITgcm model; [96]), suitable for long-time climatic integrations. The model, recently implemented at ENEA, covers the Mediterranean–Black Sea systemwith a horizontal resolution of 1/16° (about 7 km) and 100 vertical levels. The horizontal resolution is further increased at the Gibraltar and Turkish Straits, to adequately resolve the complex local dynamics. Another distinctive feature of the present implementation is the inclusion of the main tidal forcing, both local and propagating from the Atlantic. The model has been validated considering 34 tide gauges. Three simulations of the Mediterranean Sea climate were performed: a hindcast simulation (1980–2010), a historical simulation (1981–2005), and a future climate simulation under the RCP8.5 scenario. The three runs were forced using regional high-resolution (about 12.5 km) downscaling performed with the SMHI-RCA4 atmospheric regional model [97], constrained by the ERA-Interim reanalysis [98] and by present climate and future climate runs of the CMIP5 global model HadGEM2-ES [99], respectively. Further details about the numerical implementation, together with a first analysis of the model results, can be found in Sannino et al. (2022) [72].

The purpose of the hindcast simulation was to show that the model can reproduce the climate variability of the past decades. Hindcast fields were found in very good agreement with the observations, and the model was able to correctly reproduce even near-shore sea level variations. Under the RCP8.5 future scenario, the temperature is projected to generally increase, while the surface salinity decreases in the portion of the Mediterranean affected by the penetration of the Atlantic stream and increases elsewhere. The warming of sea waters results in the partial inhibition of deep-water formation. The scenario simulation allows for a detailed characterization of the regional patterns of future sea level due to ocean dynamics and indicates a relative sinking of the Mediterranean with respect to the Atlantic, more pronounced than the current one.

3.2. Topography and Bathymetry

Except for the Sardinia case study, the information on topography used in this work comes from a very high-resolution DTM, available through the National Geoportal, whose realization has been funded by the Italian Ministry of Environment and Energy Security within the framework of an extraordinary plan for environmental remote sensing. The DTM is based on the results of a Lidar survey at 2 m of resolution performed in 2010, which covered the 1st and 2nd order river courses (hierarchical order given in the IGM river catalogue) of the continental Italian territory and coastal areas [100]. The model employs the WGS84—EPSG:4326 geographic projection and has been widely used in the past decade [33,100,101]. For the Sardinia test case, a DTM at 5 m of resolution was instead used, derived from a Lidar survey on the coastal zone and inland urban centres owned by the Regional Administration of Sardinia. In the present study, 33,102,778 topographic cells were used to analyse the vulnerability of coastal areas to RSLR. To better represent the underwater environment, a bathymetric dataset was also used as cartographic background. For most of the Italian Coast, bathymetric data were provided by the National Ministry for Environment and Energy Security (Table 1; [102]).

The accuracy of the digital terrain model is a relevant issue as the planimetric accuracy of the LIDAR dataset is about ±15 cm on the vertical component. Nevertheless, in situ surveys indicate that the real accuracy can be higher. An example is the work carried out in the Circeo National Park [103], where the calibration of the LiDAR dataset was performed using DGPS measurements. These studies suggest that accuracy is higher on the emerged beach and lower on the vegetated foredunes and on the submerged beach [93,94].

3.3. Ground Motion

In November 2022, the Copernicus Land Monitoring Service released a new Pan-European product, EGMS, which is based on the elaboration of Synthetic Aperture Radar (SAR) data acquired by Sentinel-1 constellation satellites [24]. EGMS allows users to investigate ground motion (both vertical and horizontal) with unprecedented accuracy, over spatial scales ranging from that of a nation or a city to that of a single infrastructure [104]. Both maps and time series are provided; the latter allow for the computation of mean annual vertical velocities in the regions of interest. The current dataset, which will be updated at yearly frequency, covers the period January 2016–December 2021, with a time step of one datum every 12 days until October 2016 and one datum every 6 days from October 2016 to December 2020, thus collecting almost 250 records over 6 years (time interval 2016–2021).

EGMS offers three different map products which are described in Crosetto et al. (2020) [23]. In the present work, the Ortho L3 product was used: it is an update of the calibrated product where data are collected in a square area of 100 × 100 m in ETRS89-LAEA coordinates for the whole European territory. It combines different satellite view angles to provide calibrated motions, in two separate layers, purely vertical and purely horizontal (E-W), resulting in the direction of any observed displacement [23,24].

The ground motion component of our GIS tool uses EGMS data (n. 8607 cells). Within the DTM area, the available data of ground motion were acquired directly from the Copernicus system raster with a spatial resolution of 100 × 100 m. The EGMS raster points do not cover the entire study area. So, the GRASS r.fillnulls interpolation algorithm [105,106,107] was used to produce information where needed. The algorithm used retains the original data and fills in the gaps using spline interpolation [108,109]. The result of the interpolation procedure is shown in the first two panels of Figure 2 for the study area of Rome. The upper panels provide a comparison of the grid covered by the Copernicus Land Monitoring Service (left) and our interpolation to have displacement values for each 100 × 100 m grid within the study areas. The lower panel shows the ground motion displacement (mm) measured by the Sentinel satellite (blue) and the interpolated values (brown) along the A-B transect.

3.4. Land Cover

The CORINE (COoRdination of INformation on the Environment) Land Cover inventory, based on satellite imagery classification and photointerpretation, is one of the basic pan-European datasets developed by the Copernicus Land Monitoring Service; it was initiated in 1985 and constantly updated from 1990 to 2018 [110,111]. The CORINE Coastal Zones Land Cover (CZLC) product used in this study (see the official guidelines “CZ Nomenclature Guideline, Issue: 1.2” for details, [112]) covers a 10 km inland buffer zone along the European coasts (a total area of about 715.000 Km²) with high resolution (70 × 70 m), providing detailed satellite-derived information about the different types of coastal land cover and usage. The product is available as a vector file in Lambert Azimuthal Equal Area, ETRS89-extended/LAEA Europe—EPSG:3035 geographic projection (m).

The dataset has changed several times over the decades; here, we use the version resulting from the last major update, in 2018, that now differentiates 71 thematic LC/LU classes. Five levels of category are possible. In particular, we refer to level one (of five) where eight categories of cover can be identified as follows: (1) Urban; (2) Cropland; (3) Woodland and Forest; (4) Grassland; (5) Heathland and Scrubs; (6) Open Space with Little or No Vegetation; (7) Wetland; and (8) Water (Büttner et al., 2021 and references therein; Figure 3).

3.5. Processing Chain

To estimate the future coastal evolution, we combined the SL variations predicted by MED16 with projections of the vertical ground motion based on the available EGMS data. The model SL was spatially averaged in the test regions and translated to the coastal area using a simple “bathtub” approach [113,114]. On the other hand, the available 6 years of EGMS data were used to compute present time average velocities of vertical displacement in the five areas of interest, which were then assumed to remain constant during the timespan of the scenario. This is a limitation of the analysis because future geological events could modify the present rates of displacement. However, since the dataset will be constantly updated, uncertainty will eventually reduce in time.

The MED16 SL data and the EGMS displacements were both spread on the high-resolution DTM to obtain the overall movement yearly from 2010 to 2099. The area of interest (AoI) was extended for 10 Km inland from the coastline, and CORINE was then used to study the impacts of the SLR on the exposed infrastructure and natural environments in the test areas. A general workflow of the methodology is represented in Figure 4.

The GIS procedure we implemented makes use of the open-source software QGIS 3.22 and later. The Atlas function present in the “Layout” section of QGIS was used to create the custom maps with dynamic data. To create the diagrams, the QGIS plugin “Data Plotly” was used, which allows dynamic charts to be inserted both within the GIS project and in the “Layout” section with customized plugins. All the input data were managed in a GIS project and tailored to the UTM WGS84 coordinate system, ETRS89-extended Lambert Azimuthal Equal Area (LAEA) Europe. Using the graphic modeller available in QGIS, a specific and flexible tool was implemented that collects the input data, allows data processing, and produces inundation maps for the areas of interest. The computational steps involved are reported in Figure 5.

The GIS procedure was used to compute the maximum extension of inundated areas, by using the highest SL during the periods 2010–2040, 2040–2070, and 2040–2099 for the inundation scenarios.

When raster images of first-level flood maps are available (basemap is provided by OpenStreetMap of the OSM Foundation and bathymetry of ISPRA), Corine Land Cover service provides a LC/LU dataset for areas along the marine coastline of the EEA39 countries. So, the area of inundation considers the classes of CZ. In the present study, we focused on the impact of inundation scenarios by calculating the typology of terrain under sea level by the end of this century (time interval 2070–2099).

4. Results

4.1. SLR in the Study Areas

Figure 6 shows yearly time series of the projected SLR for the five study areas in the proximity of the coastline; the three periods previously indicated are marked by vertical lines. The time series resemble each other, but there are some differences between them, both in terms of SL values and time variability. In the last decades of the simulation, the highest SL values are mostly found in the Alghero-Fertilia site, on the western side of Sardinia. In the following paragraphs, we will present inundation maps for the three periods, showing values for the year in which the maximum of SL was attained. Note that these years do not necessarily coincide with the last year of each time-interval (see horizontal lines in the figure).

4.2. Topography

In the present study, we analysed more than 193 km² of coastal areas, creating a separate (2 × 2 m) DTM for each study area. The DTMs show that the study areas are characterized by a variety of coastal natural environments, such as emerged beaches, beach–dune systems, river mouths, and, in some cases, by the presence of critical infrastructures (energy plants, harbours, and airports). It should be noted that in the DTM of the National Geoportal, the coastal strip surveyed is rather narrow, whereas the lower resolution DTM (5 × 5 m) used for Alghero-Fertilia extends more on shore.

4.3. Ground Motion

Ground motion presents a very high spatial and temporal variability (Table 2 and Figure 7). The EGMS data do not entirely cover the study areas; the maximum coverage is 59% (Follonica), but it is smaller in other regions (e.g., Pontina Plain, where it is about 25%). Therefore, we used accurate interpolations to fill the gaps.

The middle panel in Figure 7 shows, for each of the study areas, the average values of ground motion in the time interval 2016–2021. These values are negative (subsidence) and decreasing in an almost linear way, indicating that it may be reasonable to project the observed trends in the next decades. The upper and lower panels of the figure show the minimum (positive) and maximum (negative) values of the ground motion over the study areas. The latter are much higher than the average values (up to about 15 cm in the Rome site; see also

Table 3. Minimum, medium, and maximum displacement (mm) released by the EGMS platform within the studied areas in the time interval 2016–2021. Note that upward displacements have a positive sign (+), while negative displacements have a negative sign (−).

	Minimum Values (mm)
Study Area	2016	2017	2018	2019	2020	2021
Piombino-Follonica	18.80	17.20	21.00	27.00	31.60	36.50
Marina di Campo	17.30	14.50	13.00	11.60	14.60	15.50
Alghero-Fertilia	17.40	2240	2190	23.50	24.90	28.30
Rome	20.40	23.00	15.10	14.60	17.90	14.50
Latina-Sabaudia	16.00	14.10	13.00	13.00	23.20	33.20
	Medium Values (mm)
Study Area	2016	2017	2018	2019	2020	2021
Piombino-Follonica	−0.80	−2.00	−3.20	−4.50	−6.20	−8.40
Marina di Campo	−0.80	−2.50	−4.20	−6.20	−8.50	−10.90
Alghero-Fertilia	−0.80	−2.30	−3.50	−4.70	−6.00	−7.30
Rome	−1.00	−3.00	−4.90	−6.90	−9.40	−12.10
Latina-Sabaudia	−1.10	−3.20	−5.50	−7.30	−8.70	−9.60
	Maximum Values (mm)
Study Area	2016	2017	2018	2019	2020	2021
Piombino-Follonica	−22.70	−33.10	−52.30	−67.40	−82.30	−95.70
Marina di Campo	−20.80	−24.90	−34.90	−37.10	−41.60	−51.50
Alghero-Fertilia	−19.70	−27.20	−43.80	−60.60	−72.50	−89.10
Rome	−15.40	−60.70	−78.60	−102.30	−125.70	−148.00
Latina-Sabaudia	−20.00	−31.00	−38.60	−44.20	−54.90	−61.80

).

4.4. Inundation Maps

In this section, we present inundation maps for the three reference periods, for each study area. The flooded areas are represented by the water level rise on the coastal DTM of 2010, without considering the morphological effects during the inundation processes [51]. Raster images of cumulative flood maps are shown in Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12. Surface Under Sea Level (SUSL) refers to areas that will be at a lower elevation compared to the mean sea level of 2010, under future scenario conditions, and they are indicated by the blue colour. In general, all maps show an increase in the flooded areas through time, but the consequences in terms of vulnerability and risks vary from region to region, as described below. Surface Under Sea Level (SUSL) refers to the baseline digital terrain model (DTM) of 2010.

The area of Piombino-Follonica (Figure 8) has a surface of 24.4 km²; it contains 6,095,979 DTM cells and 2417 EGMS cells (59% coverage). The study area is dedicated to agriculture, harbour activities, and energy production. The effects of cultivation and tillage (ploughing) are evident on the shore. Progressive flooding scenarios, due to RSLR, can be observed in a series of subparallel narrow strips of cultivated land.

The area of Marina di Campo (Figure 9) has a surface of 11.1 km²; it contains 2,49,679 DTM’s cells and 1087 EGMS’s cells (51% coverage). This coastal plain is characterized by a network of natural and artificial waterways and irrigation channels and hosts an important airport infrastructure. Flood maps show that during the 21st century, the most vulnerable portions of land are adjacent to the airport and to the banks of water courses.

The area of Alghero-Fertilia (Figure 10) has a surface of 68.5 km²; it contains 2,739,980 DTM’s cells and 6804 EGMS’s cells (37% coverage). The Calich, which is both a fishpond and a lagoon, is among the most important coastal wetlands of Sardinia [89]. It overlooks the coast of Alghero and is connected to the sea by a channel where the remains of a Roman bridge can be found. Flood maps show that portions of its banks will be affected by flooding, with consequent ecological changes for flora and fauna. By the end of the century, the armoured mouth and Fertilia Harbour could be submerged.

The area of Rome (Figure 11) has a surface of 51.6 km²; it contains 12,901,334 DTM’s cells and 5153 EGMS’s cells (44% coverage). The greatest coastal risk focuses on the Tiber River mouth and the backshore of the Municipalities of Fiumicino (to the north of the mouth), and Ostia-Torvaianica (to the south of the mouth), with impacts on both urbanized and natural areas.

The area of Latina-Sabaudia (Figure 12) has a surface of 37.9 km²; it contains 9,485,264 DTM’s cells and 3789 EGMS’s cells (25% coverage). Here, there is an increasing flooded region, which results in the enlargement of existing coastal lakes. Starting from 2040, most of the wetland will evolve into ponds and shoal environments connecting coastal lakes and reducing the extension of emerged land. Such a trend will probably impact urban areas and coastal infrastructures (roads) in the time interval 2070–2100 in the northern part of the study area.

The Relative SLR and the surface of submerged areas for each of the five study areas at different time horizons are reported in

Table 4. Inundation data in reference years. Relative Sea Level Rise (RSLR) is a representative value that consider SLR and an average value of GM within the studied area. Surface Under Sea Level (SUSL) refers to the baseline digital terrain model (DTM) of 2010.

Study Area	Piombino-Follonica			Marina di Campo			Alghero-Fertilia
Year	2040	2070	2099	2040	2070	2099	2040	2070	2099
RSLR (cm)	21.44	44.86	71.22	23.5	48.73	76.84	22.39	46.91	74.42
SULS (km2)	1.36	2.72	4.57	0.04	0.06	0.09	0.99	1.13	1.44
Study Area	Rome			Latina-Sabaudia
Year	2040	2070	2099	2040	2070	2099
RSLR (cm)	23.62	48.87	76.98	25.05	51.62	81.26
SULS (km2)	1.99	3.31	5.84	9.43	12.05	14.84

.

4.5. Assessment of Exposed Assets

Land flooding impacts both natural and anthropic environments, in ways that may be difficult to predict. In the case of ecosystems, complex feedback mechanisms and highly articulated interrelations may occur [115,116]. In the case of critical infrastructures (CIs), coastal defences (e.g., jetties and groins), and confined disposal facilities (CDFs) of harbours, which generally show a gradual tendency to sink into the seabed, the time evolution of the ground motion may be nonlinear, and therefore difficult to predict using linear extrapolations of the present trends.

Nevertheless, planning and adaptation strategies can be guided by the knowledge of the main present characteristics of the vulnerable areas [77]. Using the Corine Land Cover, we determined the assets exposed to inundation within the five coastal areas (see

Table 5. Breakdowns in percentage values of the most exposed assets due to the RSLR during the 21st century.

	Piombino-Follonica				Marina di Campo				Alghero-Fertilia
Terrain Under Sea Level	2010	2040	2070	2099	2010	2040	2070	2099	2010	2040	2070	2099
Cropland	26.60	35.20	39.60	44.60	1.54	1.03	0.96	1.52	2.83	0.68	1.92	6.36
Woodland and Forest	12.30	18.30	17.30	18.30	1.54	4.38	4.19	6.69	0.53	0.45	1.52	5.37
Grassland	0.00	0.01	0.05	0.10	-	-	-	-	2.49	0.59	0.98	2.39
Heathland and Scrubs	0.01	0.03	0.30	0.46	0.14	0.21	0.88	17.90	-	0.12	0.52	1.49
Open Space 1	2.98	2.64	2.74	3.11	43.40	48.80	45.70	31.20	11.60	1.40	2.57	4.07
Wetland	17.60	18.00	23.00	19.50	-	-	-	-	-	-	-	-
Water	35.30	21.20	12.40	8.15	45.90	39.60	36.90	22.00	68.80	95.40	88.50	73.30
Urban	5.19	4.57	4.55	5.83	7.55	6.03	11.40	20.60	13.70	1.34	3.96	6.98
	Rome				Latina-Sabaudia
Terrain Under Sea level	2010	2040	2070	2099	2010	2040	2070	2099
Cropland	31.70	42.70	43.20	38.60	29.30	27.80	25.80	24.50
Woodland and Forest	12.10	8.71	6.85	7.75	2.04	3.62	5.21	6.50
Grassland	0.13	0.24	0.96	3.07	15.30	22.30	28.40	31.20
Heathland and Scrubs	0.01	0.12	0.36	0.39	-	0.00	0.00	0.00
Open Space 1	5.35	6.25	6.84	6.14	0.41	0.64	1.14	1.58
Wetland	2.10	3.72	5.61	5.14	3.66	3.16	2.74	2.42
Water	35.80	25.20	17.90	11.80	48.00	39.80	32.30	26.60
Urban	12.80	13.10	18.20	27.10	1.28	2.73	4.44	7.16

¹ Open space with little or no vegetation.

and Figure 13). Cropland, Open Space with Little or No Vegetation (where present, like in the study area of Alghero-Fertilia-) are the most impacted ecosystems, followed by Wetland (in Follonica and Pontina’s plains) and Water.

5. Discussion and Conclusions

In the present paper, we presented the implemented methodology that is flexible, exportable, and reliable, with the aspiration of making a significant contribution to the evolution of coastal risk assessment and adaptation tools through what we hope will be an accessible climate service in the future.

The projections provided by IPCC often lack the regional specificity required to accurately characterize SLR in the Mediterranean Sea. To address this gap, our research used projections from the MED16 model. This model boasts a high-resolution coverage of the Mediterranean basin, including a very high representation of the complex tidal dynamics at the Strait of Gibraltar which determine significant differences between the Mediterranean SLR and the global one [72]. The high resolution of the numerical model also allows researchers to capture spatial variations of SLR along the coasts of the basin.

To consider the geological factors that contribute to the flooding hazard, we used the synthetic data on vertical ground motion from EGMS, which provides the state-of-the-art product for the remote characterization of the coastal environment.

The high-resolution projections of SLR and vertical ground motion were then inserted in a GIS procedure that also includes very detailed information on the local topography, to produce flooding maps in the sites in consideration. Finally, an in-depth assessment of the categories of exposed assets was performed using the Corine LC data. We found, perhaps not surprisingly, that the main assets exposed to the risk of flooding are wetlands, backshore and coastal plains, and coastal infrastructures.

Two of the test areas considered in the present study (Alghero-Fertilia and Marina di Campo) were already investigated by some of the authors in the past, using the same DTM, with different results concerning the flooding risk. Those previous works used SLR data from Rahmstorf (2007) [117] and from the IPCC 2013 [118] projections, including the contribution of tectonics and GIA. In Fertilia, the results of potential submersion areas were, respectively, 2.29 and 1.89 km², against the 1.44 km² of the present study [25]. In Marina di Campo, the results of potential submersion areas were, respectively, 0.39 and 0.14 km², against the 0.09 km² of the present study (

Table 6. Comparison of terrain under sea level (km²) in 2100 predicted by using previous scenarios (Rahmstorf, 2007 [117] and IPCC, 2013 [118]; Antonioli et al., 2020 * [25]) and Med 16 Model of Sannino et al. (2022 [72]; present study **).

	Marina di Campo		Fertilia
	SLR (cm)	SUSL (km2)	SLR (cm)	SUSL (km2)
Rahmstorf 2007 *	143.2	0.39	145.2	2.29
IPCC 2013 *	102.2	0.14	102.2	1.89
MED16 2022 **	58.3	0.09	62.6	1.44

; Figure 14).

The flooded surfaces predicted using the MED16 results are somewhat smaller than those estimated before (the bigger differences are with respect to the predictions based on Rahmstorf, 2007 [117]). Figure 14 shows that the changes in the SLR contribution are amplified by the local topography, determining large differences between the two sites. So, for these two tectonically stable low coastal areas, we can say that the main factor influencing the different results is the lower SLR projection simulated by MED16 compared to the IPCC 2013 and Rahmstorf 2007 scenarios.

For all study areas, we are already processing multisource data in order to verify the congruency between long-term geological data and short-term remote sensing data. Some limitations of the present approach should be pointed out. As already noted, the nominal vertical accuracy of the Italian DTM we used appears not negligible (±15 cm), even though the real one can be smaller. This aspect will soon be improved, since in 2026 a new, higher resolution LIDAR mapping of the entire Italian coast will be available thanks to the Marine Ecosystem Restoration (MER) Project which will allow for the reduction of the topographic uncertainties [119].

The main limitation concerning the ground motion component is associated with the short duration of the present EGMS dataset, but the yearly updates that are foreseen will alleviate this restriction; having longer time series will be of help in understanding whether it is reasonable to extrapolate the present rates into the future. Recently, the dataset was extended to 2021, so the estimates of vertical ground displacement are now based on 6 years (2016 to 2021). This is a small step forward compared to the analyses we recently presented [120] for different study areas by using the same methodological approach. However, it should be noted that a European product validated by Copernicus is already a great step forward, and this product will be further improved when data from the most recent satellite constellations will be available [121,122].

We tested our new methodology on coastal areas that were recently indicated as nearly stable ones [123], where we may expect that EGMS trends are more reliable. A recent investigation [124] has shown that in the Pontina coastal plain there is very close agreement between the EGMS dataset and geological records from the MIS 5.5 (Tyrrhenian), and this result is in agreement with other studies [123,124,125,126]. For other areas, we are already processing multisource data in order to verify the congruency between long-term geological data and short-term remote sensing data. Clearly, there are also unstable coastal regions, such as Campi Flegrei and Calabria, where the analysis should be complemented with detailed in situ measurements.

It should be noted that in the central Mediterranean area, and particularly in Italy, vertical movements due to volcanism, tectonics, and isostasy can be calculated using geological and instrumental methods [16,17,18,19]. The maximum highstand of MIS 5.5, which occurred 118 thousand years ago, is the most used geological method, and many scientific papers and databases about Italy are available in the literature [3,17,123,127]. The rates coming out from those geological markers range from +1.7 to −1.2 mm/year [128]. The geological markers are very useful because they average, for the past 118 thousand years, the result from different movements, including co-seismic ones from earthquakes, with the recurring event time of 50–100 years. The vertical movements, due to glacio-hydro isostasy (GIA) are well known and modelled for the central Mediterranean area, showing values from −0.1 to 0.6 mm/year [17,22]; in Northern Europe (Norway, Sweden, and Finland), instead, higher values are recorded with an uplift rate larger than 1 cm/year. Moreover, some other vertical movements could happen, caused by subsidence, uplift, and proximity to volcanoes or eruptive sites. As an example, the area belonging to the town of Venice, Italy, has experienced a relative sea level rise of 25.3 cm in the past hundred years in contrast to the rate of 13.1 in other stable areas of the Mediterranean Sea [16,18,30].

Another source of uncertainty is related to the accuracy of the SLR modelling estimates. Here, we considered results from a single high-resolution ocean model, but, in principle, it would be useful to have more simulations (ensemble runs), eventually with coupled regional models, which would allow for a self-consistent description of the interactions between the atmosphere and the ocean (e.g., Anav et al., 2024 [129]). Recent developments of a new high-resolution regional climate model for the Mediterranean region (MESMAR; [130]) are improving the already available coupled climate tools over the Mediterranean region. Moreover, one has to carefully consider the improvements in the models and in the parameterizations that will be available in the near future, as well as the progress in the discussions concerning some crucial points which are still debated, such as the possibility of a future change of regime in global ice melting (Sweet et al., 2017 [131] and references therein).

The new product described in this work could be of help in capitalizing the efforts that the scientific and European communities are making to improve and share platforms of environmental data and information. The improved platforms will be part of an increasingly reliable methodology that we hope will become a climate service capable of assessing the future impacts of climate change and plan appropriate prevention and adaptation strategies [49,50,76,132]. The implemented approach is ready to be tested in other regions bordering the Mediterranean Sea. We are currently implementing a specific tool for the assessment of damages due to flooding for critical infrastructures (CIs; [128,133]) and agriculture [134].

In general, we can conclude that coastal geomorphology is strongly influenced both by SLR and ground motion due to geological factors such as tectonics, volcanic activities, sediment input, and subsidence. The variability of some of these factors, such as subsidence, can also be influenced by human activities, and can occur on different time scales. In fact, there are areas in which subsidence is happening more rapidly than SLR, and its impact is not fully appreciated on a global scale. The most advanced satellite-derived product for characterizing the coastal environment can provide a first level of guidance to concentrate more detailed investigations on hot spots and, hopefully, reveal how the higher resolution of data and the different geological components may influence the results.