Sea level rise projections up to 2150 in the northern Mediterranean coasts

Vertical land movements (VLM) play a crucial role in affecting the sea level rise along the coasts. They need to be estimated and included in the analysis for more accurate Sea Level (SL) projections. Here we focus on the Mediterranean basin characterized by spatially variable rates of VLM that affect the future SL along the coasts. To estimate the VLM rates we used geodetic data from continuous global navigation satellite system stations with time series longer than 4.5 years in the 1996–2023 interval, belonging to Euro-Mediterranean networks and located within 5 km from the coast. Revised SL projections up to the year 2150 are provided at 265 points on a geographical grid and at the locations of 51 tide gauges of the Permanent Service for Mean Sea Level, by including the estimated VLM in the SL projections released by the Intergovernmental Panel on Climate Change (IPCC) in the AR6 Report. Results show that the IPCC projections underestimate future SL along the coasts of the Mediterranean Sea since the effects of tectonics and other local factors were not properly considered. Here we show that revised SL projections at 2100, when compared to the IPCC, show a maximum and minimum differences of 1094 ± 103 mm and −773 ± 106 mm, respectively, with an average value that exceeds by about 80 mm that of the IPCC in the reference Shared Socio-economic Pathways and different global warming levels. Finally, the projections indicate that about 19.000 km2 of the considered Mediterranean coasts will be more exposed to risk of inundation for the next decades, leading to enhanced impacts on the environment, human activities and infrastructures, thus suggesting the need for concrete actions to support vulnerable populations to adapt to the expected SL rise and coastal hazards by the end of this century.


Introduction
Tide gauge measurements and satellite radar altimetry data show that the global mean sea level (GMSL) over the past two centuries is rising at faster rates with respect to the last millennia due to global warming (Vermeer and Rahmstorf 2009, Lambeck et al 2010, Church and White 2011, Kemp et al 2011, DeConto et al 2021, Palmer et al 2021).Changes in GMSL are due to the joint effect of thermosteric, barystatic, glacio-hydro-isostatic and land-hydrology components (Lambeck and Purcell 2005).The rate of rise of the GMSL was 1.7 [1.3-1.9]mm yr −1 , over the period 1901-2015, 3.7 [3.2-4.2]mm yr −1 over the period 2006-2018 and it will likely reach 5.2-12.1 mm yr −1 in the period 2080-2100, for the lowest and highest CO2 emission scenarios (Fox-Kemper et al 2021).Since at a given geographical location both the SL and the land surface are not static over time, the observed SL change along the coasts results from the combined action of solid Earth and oceanographic variations driven by geophysical and climate processes.At regional scale the SL can show large changes with respect to the GMSL due to the contributions of dynamic effects, glacial isostatic adjustment (GIA) and natural and anthropogenic Vertical land movements (VLM) that depends on the geographical location (Lambeck et al 2010, Kopp et al 2014, Vecchio et al 2019, www.ipcc.ch).In particular, coastal subsidence and uplift due to tectonics and other natural or anthropogenic factors, act at rates up to several mm yr −1 .These effects can change the amplitude and impact of the SL rise by locally increasing or reducing its effects on the coastal areas (Lambeck et al 2011, Bucx et al 2015, Higgins 2016, Vousdoukas et al 2018, Oppenheimer et al 2019, Scardino et al 2020, 2022, Anzidei et al 2021, Jevrejeva et al 2023).Since the SL rise is a major factor of hazard in coastal regions where cultural, socio-economic and environmental losses are already occurring (Syvitski et al 2009, Strauss et al 2014, Kulp and Strauss 2019, Horton et al 2020, Tay et al 2022), understanding and including the VLM in the SL projections for the next decades become crucial for a more detailed estimation of the flooding hazard and socio-economic consequences in coastal areas.
The SL of the Mediterranean is rising since 1880 (Fenoglio-Marc et al 2012, 2013, Wöppelmann and Marcos 2012, Tsimplis et al 2013, Anzidei et al 2014, Meli et al 2023) affecting the low elevated coastal plains, river deltas and beaches with consequent coastal erosion and retreat and salinization of the water table, thus representing a significant factor of hazard for coastal populations and infrastructures (Antonioli et al 2017, Anzidei et al 2017, Nienhuis et al 2023).In addition, the Mediterranean region is the seat of historical cities (e.g.Venice, Istanbul, Alexandria among others), heritage sites and highvalue environmental and industrial areas, all exposed to severe coastal hazard and flooding (Reiman et al 2018, Anzidei et al 2020, Strauss et al 2021, Scardino et al 2022).Detailed SL projections for the next decades in the Mediterranean are therefore needed for building up realistic marine flooding scenarios (Rizzo et al 2022 and references therein) and making people, stakeholders, and decision makers aware of the ongoing phenomenon for a conscious management of the coastal zone (Loizidou et al 2023).
The most updated projections released by the Intergovernmental Panel On Climate Change (IPCC) in the 6th Assessment Report (AR6, Fox-Kemper et al 2021) refer to global and regional sea levels and consider the contributions of several geophysical sources.In particular, the regional projections include the contribution of a long-term VLM extracted from the longest time series of sea level as obtained by selected tide gauges of the Permanent Service for mean Sea level (PSMSL, https://psmsl.org/).In this study we show and discuss the SL projections in the Mediterranean basin by revisiting the IPCC AR6 projections up to 2150 including the linear trend of the current rates of VLM as estimated from the analysis of ground displacement time series of continuous global navigation satellite system (GNSS) networks operating in the Euro-Mediterranean region.The sea level projections in a gridded format at a resolution of 0.5 • in latitude and longitude for the north Mediterranean basin and at the location of selected PSMSL stations, are provided to map the future variability of the SL along the coasts.

The Mediterranean basin
The Mediterranean region is the result of the evolution of the convergence between the African and Eurasian plates across an east-west boundary that has been active since the Late Cretaceous (Dewey et al 1973, Le Pichon et al 1988, Jolivet and Faccenna 2000, Faccenna et al 2014).These plates are still colliding at a velocity of a few millimetres per year as indicated by geodetic data (Devoti et al 2017, Serpelloni et al 2022).The seismicity of the region is characterized by well-defined seismic belts underlining the boundaries of minor plates and a complex pattern of crustal deformations (Jackson and McKenzie 1988, Rebai et al 1992, Vannucci and Gasperini 2004, Serpelloni et al 2007, 2022) that include active and quiescent volcanoes (volcano.si.edu).The present-time kinematics of the Mediterranean basin is characterized by several active tectonic structures in the region The present-time kinematics of the Mediterranean basin is characterized by several active tectonic structures in the region (Faccenna et al 2014).Among these, the Hellenic Arc system in the eastern Mediterranean, the subduction of the Calabrian Arc and the volcanic arc of the Aeolian Islands in the central Mediterranean are the most relevant.In this complex tectonic framework, geological and geodynamic processes are causing continuous and episodic earthquake-related crustal deformations that also affect the coastal areas (Ferranti et al 2010, Anzidei et al 2014).
The Mediterranean region has also experienced large sea level changes driven by the growth and decay of the large ice sheets culminating in the Last Glacial Maximum about 20.000 years ago.This results in the associated isostatic effect as the Earth ′ s gravity, shape and rotation respond to the changing surface loads of ice and water (Peltier 2001, Lambeck and Purcell 2005, Spada 2017).Direct information about current trends of VLM in the Mediterranean regions can be inferred from instrumental data through seismic strain analysis of focal mechanisms for about the last 140 Direct information about current trends of VLM in the Mediterranean regions can be inferred from instrumental data through seismic strain analysis of focal mechanisms for about the last 140 years and crustal deformation analysis based on continuous GNSS stations (Anzidei et al 2014).

VLM from GNSS data
The Mediterranean region is characterized by spatially variable horizontal and vertical crustal deformations, with large deviations in volcanic areas.To evaluate the impact of VLM on the SL, the vertical component of the crustal velocity field along the coasts of the Mediterranean region has been estimated by the analysis of the geodetic signal recorded at GNSS networks, managed by different institutions, in the Euro-Mediterranean and African areas (Esposito et al 2015, Devoti et al 2017, Marino et al 2022, Serpelloni et al 2022).Since we focus on the effects of VLM on the SL changes along the coasts, only GNSS stations located within 5 km from the coastline have been selected.Among these, to get a stable estimate of the vertical geodetic velocities, only time series longer than 4.5 years in the time interval 1996-2023 have been considered.This choice agrees with results from synthetic data suggesting that time series longer than 4.5 years are needed to get velocity accuracy of the order of mm yr −1 (Masson et al 2019).Analysis of GNSS data in the Euro-Mediterranean region shows that a median length of 6-7 years, with standard error ± 2.5 years, provides velocity estimates consistent with the ones obtained from longer time series (>10 years) (Serpelloni et al 2022).When the lower bound (4.5 years) is considered, velocity estimates are still consistent with results from synthetic data.The choice of 4.5 years cutoff value has thus been driven by the studies mentioned above and by the need to have a sufficiently large number of stations for the following analysis.In the same way the choice of a threshold distance of 5 km represents a good compromise between having a sufficiently large number of stations close to and a reasonable distance from the coast.This selection resulted in 638 GNSS stations (figure 1).Unfortunately, the geographical distribution of the selected stations is not homogeneous since their number and density are higher in the North Mediterranean while only a few of them are operational in North Africa (at least concerning public available data), leading to only a partial picture of the current crustal motion across the Mediterranean region.Nevertheless, the spatial variability of the measured velocities highlights the different tectonic and volcanic environments of the Mediterranean region characterized by variable rates of VLM that can affect the SL change along the coasts.
GNSS velocities and uncertainties at the coastal stations have been obtained following the approach described in Serpelloni et al (2022), that includes: (i) the GNSS phase data reduction by GAMIT software (Herring et al 2010), (ii) by the GLOBK software and (iii) the time-series analysis.The reference frame has been obtained by minimizing coordinates and velocities of the IGS global core stations, while estimating a seven-parameter transformation with respect to the IGS realization of the ITRF2014 reference frame (Altamimi et al 2011).A maximum likelihood estimation method (Bos et al 2013) has been used to evaluate the stochastic noise content in the time series.Measured vertical velocities are in the range −17 (in the northern Adriatic) +36 mm yr −1 (in the Phlegrean fields).The uncertainties on the velocity estimates have been calculated using a white and a flicker noise model (Serpelloni et al 2022), which are suitable for the GNSS data in the Mediterranean region (Santamaría-Gómez et al 2011, Nguyen et al 2016).In figure 1 of the supplementary material a map of the vertical velocity uncertainties at each GNSS station is shown.Larger uncertainties are associated to noisier or shorter data.However, since the uncertainty decreases the more the length exceeds 2.5 years (Serpelloni et al 2022), only a limited number of stations show uncertainties higher than 1 mm yr −1 .The mean uncertainty is 0.65 mm yr −1 .
In the Mediterranean area, the VLM pattern reflects the ongoing tectonic and non-tectonic processes.For example, the Venice lagoon (Italy) is undergoing land subsidence due to the combined effects of GIA, sedimentary, tectonic, and anthropogenic processes that locally accelerate the SLR.Conversely, along the sedimentary coast of Levant (Israel) and in the volcanic area of Phlegrean fields (Italy), the current uplift is slowing down the SLR.

Sea level projections
To evaluate the effect of the VLM on future sea levels, we considered the regional sea level projections provided in the 6th Assessment Report of the IPCC (Kopp et al 2023, Fox-Kemper et al 2021, Garner at al 2021), spanning the interval 2020-2150 (https://zenodo.org/record/6382554).In this work we used the medium confidence data that include only processes for which there is a general agreement in the scientific community (see e.g.Fox-Kemper et al 2021 for a detailed definition).AR6 projections are provided on both a 1 • × 1 • worldwide grid and at the location of several tide gauges of the PSMSL network.AR6 projections are obtained by combining the contributions of several geophysical sources: the Antarctic Ice Sheet, the Greenland Ice Sheet (GIS), glaciers, land water storage, ocean dynamics (including thermal expansion) and VLM.For each gridpoint and PSMSL station, the quantile functions (107 quantiles from 0 to 1 in 0.01 increments) are available for the distribution of the total sea level change levels and rates and for each single contribution.The methods for deriving the sea level changes as a function of the quantiles and geographical location are not straightforward and can be different for each individual contribution.For example, to derive the contribution of the GIS, different SL projections are produced, by computing the probability distributions for two different ice sheet models and combined in a 'probability box' which represents the upper and lower bounds of the distribution.These methods are described in detail in the IPCC AR6 report (Fox-Kemper et al 2021).
The data are provided for all the scenarios considered in the AR6, namely the Shared Socioeconomic Pathways (SSP1-1.9,SSP1-2.6,SSP2-4.5, SSP3-7.0,SSP5-8.5) and the five global warming levels (Tlim: 1.5 • , 2 • , 3 • , 4 • and 5 • C, higher than the current mean global temperature).Unlike the former IPCC projections, which included only the component of VLM associated with the GIA, the AR6 projections include a constant long-term background rate of change.The latter, including both the GIA and other long-term VLM contributions, is extrapolated from historical tide gauge data to global coverage using a spatiotemporal statistical approach (Kopp et al 2014) and assumed to be the same for each scenario and constant over the considered time span.However, this approach can lead to an underestimation of the VLM in areas not yet monitored by tide gauge stations (Keogh and Törnqvist 2019) that can be affected by significant rates of land subsidence or uplift (e.g.due to fluid extraction, soil compaction, tectonics, or volcanism) (Minderhoud et al 2017).
To obtain more accurate SL projections in the Mediterranean area the AR6 projections have been revisited by including the trend of the VLM derived from the ground velocities from GNSS data.We assume that the VLM is linear, namely it will progress up to 2150 with the same rate as today.This assumption is reasonable, over a time scale of centuries, for the signal due to the regional secular tectonics (GIA and plate tectonics) (see e.g.Kato 1983, Bird et al 2009, Santamaría-Gómez and Mémin 2015, Palin et al 2020).Moreover, since the recurrence times of strong magnitude earthquakes (M > 5.5) in the Mediterranean area, producing vertical deformations on the coast capable of deviating the secular trend of the tectonic signal, is estimated to be hundreds or thousands of years (see e.g.DISS Working Group, 2021, and reference therein), we assumed that the VLM rate is not significantly affected over the time range considered for the revised SL projections (130 years).Unfortunately, it is not possible to define a recurrence time for volcanic events.In volcanic areas (as for example the Phlegraean Fields), the approximation of linear VLM trend depends on the dynamics of the volcanic structure that can produce local VLMs with time varying direction and rate.Despite these limitations, the assumption we made in this study is the most reasonable one, namely the current VLM directions and rates will remain constant over the time span of the SL predictions.We remark that in the IPCC AR6 report a similar linear approximation for the VLM is made worldwide thus suffering for the same uncertainties in volcanic areas.
The revised SL projections are calculated by adding the measured VLM to the median values of the AR6 IPCC projections data which do not include the VLM contribution (file ar6-regional_novlmconfidence.zip).The uncertainties on the revised projections are evaluated at each grid box as the sum of the squared uncertainties from the AR6 IPCC projections without VLM (as the distribution spread estimated from the 32nd and 68th percentiles) and the squared error from the GNSS velocity measurements.The choice of 4.5 years as minimum threshold for the GNSS time series does not affect the sensitivity of the VLM estimates with respect to the tide gauge estimates used by IPCC.The comparison of GNSS and IPCC based rates, as a function of the GNSS series length (see supplementary material), shows no reduction in sensitivity for time series shorter than 6-7 years.At the same time no difference in sensitivity between tectonically active and volcanic areas is found.

Results and discussion
By analogy to the original IPCC AR6 data, the revised sea level projections in the range 2020-2150, are computed at different geographical locations: 1. Sea levels are evaluated at each PSMSL tide gauge location by considering the VLM rate from the closest GNSS stations (figure 2 panel (b)).2. After rescaling the IPCC data to a 0.5 • × 0.5 • grid, the VLM contribution in each cell is obtained as the median of the VLM rates from the GNSS stations falling into the cell (figure 2 panel (a)).
If the cell does not encompass any GNSS station, the corresponding sea level projection is not calculated.
The revised SL at 2100 (relative to 2020) in the SSP5-8.5 scenario for each grid point and at the PSMSL stations are shown in panels (c) and (d) of figure 2. Examples of the sea level projections at four PSMSL tide gauges and at the four grid points encompassing them are shown in figures 3 and 4 for three SSP (namely 1-2.6, 3-7.0 and 8-8.5) and three global warming Tlim scenarios (namely 1.5 • , 3 • , 5 • ), respectively.The sea levels in 2050, 2100 and 2150 (relative to 2020) for all the considered IPCC scenarios are shown in tables 3 and 4 of the supplementary material.A comparison between the VLM contribution as provided by IPCC and from GNSS data, at the same PSMSL stations and grid points of figures 3 and 4, is shown in figure 5.The four chosen locations represent an example of sites where high rate of land subsidence (Venice and Po Delta/Marseille), and uplift (Hadera/Galilee coast and Phlegraean fields) are observed.For gridded data the comparison between revised and original IPCC AR6 SL projections (the latter are shown in figure 3 of the supplementary material) show significant differences (∆SL) at 2100 varying between 1094 ±103 mm and −773 ± 106 mm with an average value of 49 ± 6 mm (see figure 2  The observed differences between SL calculated on the grid and at PSMSL stations mainly arise from the different approach used to estimate the VLM trend, that is the median over a cell and the measurement at a given GNSS station.This difference is particularly remarkable in the north Adriatic coast of Venezia/Po delta (figures 3 and 4 panels (a)-(b)) and in the central Tyrrhenian coast of the Napoli/Phlegrean fields (figures 3 and 4 panels (g)-(h)) where the measured VLM rates are −1.08 ± 0.60/-13.6 ± 0.2 and 0.34 ± 0.79/9.5 ± 0.2, respectively.The geological setting of both these areas is different and quite complex.The first is characterized by a general subsidence due to tectonics, soil compaction and ground fluid extraction, with a patchy VLM pattern with rates varying from −18.15 ± 0.19 to −2.78 ± 0.06.Spatially varying VLM is also found in the grid cell encompassing the Phlegrean fields area, characterized by active volcanic activity, where rates range between −0.43 ± 0.26 and 36.6 ± 1.0.The high variability of the VLM rate together with the high density of GNSS stations in this area cause the noticeable differences between the VLM calculated as the median value over a cell and the measurement at a single GNSS station thus resulting in significant differences in the corresponding SL projections.In the Phlegrean fields the area affected by volcanic deformation is very limited with an extension less than about 4 km 2 .This implies that calculating the VLM rate as the median values of all the GNSS measurements falling into the cell, whose size is about 55 × 42 km 2 , also stations that are not located in the volcanic area contribute to the median VLM.Measurements at these stations, where the approximation of linear trend is more suitable, can compensate for the contribution of the stations in the volcanic area.Thus, in areas affected by volcanic activity the SL projections on the grid can provide more reliable results with respect to the ones obtained by picking up a single GNSS station.
The above results indicate that in the areas of the Mediterranean considered in this study, the VLM contribution from IPCC AR6 on average underestimates the effective VLM resulting in lower or higher future sea levels in zones undergoing active subsidence and uplift.In some areas, such as the Po delta and the Phlegraean fields where VLM rates are particularly high, the differences are extremely large.
Figure 6 shows the maps of the SL at 2100 and 2150 (relative to 2020) for the considered Mediterranean coasts in the most optimistic (SSP1-1.9 and Tlim1.5 • ) and pessimistic (SSP5-8.5 and Tlim 5 • ) scenarios.Areas undergoing high rates of land subsidence result in a SL up to 2.9 m at 2150 (relative to 2020) in the worst scenarios.These regions correspond to low lying coastal zones such as river deltas, lagoons, reclamation areas (e.g. the Po delta in Italy, the Thessaloniki plain in Greece and the Rhone delta in France, among others) and volcanic zones (e.g. the Aeolian archipelago and Pantelleria Island in Southern Italy).Conversely, the areas of the Galilee coast (Israel) and the Phlegrean fields (Italy), show a slower sea level rise due to land uplift in response to tectonics and volcanism that characterize these areas.Due to the high uplift rate, in the most optimistic scenarios, both the Galilee coast and Phlegrean fields area show a SL variation at 2150 (relative to 2020) of about 0.5 and −0.7 m, respectively.
The SL projections are more reliable in coastal areas characterized by continuous deformations mainly induced by steady tectonic processes, such as Po, Ebro, and Rhône deltas, Chalastra plain and the coast of Israel.On the other hand, for regions characterized by intense volcanic activity, such as the Aegean Sea around Santorini, Ikaria, and Rhodes islands (Greece), the linear approximation for the VLM can be less precise due to the time variability of the deformation.In this respect, the Aeolian islands represent an exception since they are characterized by a generalized and linear subsidence due to the peculiar activity of the volcanic arc.Each panel shows:-the revised sea level projections, including the VLM estimate performed in this paper, in three SSP scenarios (full lines) and corresponding uncertainties (shadowed areas of the same tone of the lines);-the original IPCC projections for the most optimistic and pessimistic scenario among those shown for the revised projections (dashed lines).No uncertainties are shown for the original projections.
Along the coasts of Turkey, part of Slovenia and Croatia and in North Africa, the lack of available geodetic data (except for Alexandria-Egypt and some stations in Morocco) prevents the calculation of revised SL projections in these areas.Previous studies, based on geological and geo-archaeological observations along the coasts of North Africa (Anzidei et al 2011a, Bouaziz et al 2003, Benjamin et al 2017 and references therein), indicate that this area is characterized by an overall vertical tectonic stability in the last 125 ka BP and especially during the late Holocene, implying that significant deviations from the SL projections estimated by the IPCC by 2100-2150 are not expected.Conversely, similar studies along the southwest coasts of Turkey, have shown the relevant role of tectonic land subsidence in the flooding of coastal zones (Lambeck 1995, Anzidei et al 2011b, Kızıldag et al 2023).

Exposed zones
By combining the SL projections with the available topographic elevation, it is possible to identify the coastal zones of the Mediterranean most exposed to flooding in the next decades.These correspond to Each panel shows:-the revised sea level projections, including the VLM estimate performed in this paper, in three Tlim scenarios (full lines) and corresponding uncertainties (shadowed areas of the same tone of the lines);-the original IPCC projections for the most optimistic and pessimistic scenario among those shown for the revised projections (dashed lines).No uncertainties are shown for the original projections.low elevated areas where the exposure to flooding increases the lower is their elevation above the mean sea level.By making use of the geoprocessing facilities provided by GIS and the availability of datasets of territorial data from EUDEM (a combination of SRTM and ASTER GDEM for the European part and SRTM only for the North-African part from Morocco to Egypt and the Levant for Cyprus, Israel, and Lebanon, https://asterweb.jpl.nasa.gov/gdem.asp), the lowest elevation coastal zones of the Mediterranean region have been identified.The territorial analysis considered about 44.000 km of Mediterranean coastlines, including islands.The analysis of EUDEM data involved the selection of three coastal altimetric classes above the sea level for the three intervals [−3, 0], [0, 1] and [1, 2] m.Each selected area was then identified on high-resolution satellite images of Digital Globe (Digital Globe Quickbird, which is roughly 65 cm pan-sharpened), to check the morphological correspondence with the altimetric analysis.The analysis identified 163 main coastal plains distributed in 15 countries facing the Mediterranean coasts (figure 7) that correspond to an area of about 38.529 km 2 and a coastal length of about 7613 km (table 5 of the supplementary material).For these areas the flooding is often favored by lack of natural barriers (e.g.dune belts) and gentle sloping coastal morphology (<10%).Several of these areas are affected by land subsidence, coastal erosion and anthropogenic pressure that accelerate the flooding process, exposing them to increasingly coastal hazard in the next decades, including heritage sites (Reimann et al 2018).The most exposed are those of Egypt (12.879 km 2 ), Italy (10.060 km 2 ) and France (3.681 km 2 ).In these countries, the river deltas of Nile (Egypt, about 12.879 km 2 ), Po (Italy, about 4.000 km 2 ) and Rhone (France, about 2.000 km 2 ), which are undergoing high rates of land subsidence due to natural (soil compaction) and anthropogenic (ground fluid exploitation) processes (Syvitski et al 2009, Bohannon 2010, Cenni et al 2021, Pont et al 2002), are at high risk of flooding in 2050-2100-2150, as shown by the revised SL projections show in figure 6 It is worth noting that most of the population living along the coasts of the Mediterranean is not aware of SL rise, land subsidence and related coastal hazard that impact on the environment, coastal infrastructures, and human activities (Loizidou et al 2023).In addition, the current resident population in the Mediterranean countries is about 450 million with a possible rise to 700 million by 2100.The higher population density is found in large coastal cities as well as thousands of small villages placed even below 2 m above the sea level which are highly exposed to SL rise (Papathanassiou and Gabrielides 1999).

Conclusions
In this study we showed that VLM can significantly increase the future sea levels along the coasts of the Mediterranean basin.The results of the analysis, which includes geodetic rates of VLM in the SL projections, indicate that the new estimates of SL show significant deviations, up to the 62%, from the IPCC AR6 projections.
This implies that in the following years, about 19.000 km of low-lying coastal plains of the Mediterranean will be increasingly exposed to marine flooding and coastal hazard.The effects of the VLM are particularly relevant in river deltas and lagoons and in the unstable volcanic areas, such as the Aeolian islands and Aegean Sea.A large portion of the coast of Italy, Greece, Spain, and France is subsiding thus accelerating the SL rise.The coast of Israel is slightly uplifting and the VLM will locally counterbalance the SL rise.A similar behavior is observed in the volcanic area of the Phlegrean fields (Italy), which is currently uplifting at a high rate of about 9.5 mm yr −1 , posing critical conditions for the local coastal infrastructures and human activities.The lack of available public geodetic data for the coasts of Turkey, part of Slovenia, Croatia and North Africa prevented the calculation of revised SL projections.For these areas further analysis is needed to assess the current trend of VLM since the SL rise will in any case have consequences on the low elevated coasts.
The revised SL projections for the Mediterranean discussed in this letter can support adaptation options for coastal planning for multiple case scenarios of SL rise.Moreover, the information provided in this study can be relevant to mitigate the risk of damaging flooding, which will be increasingly common in case of storm surges and the eventual tsunamis.Finally, the revised SL projections suggest the need to improve science-based dissemination about the impacts of the SL rise toward the coastal populations of the Mediterranean to raise their awareness on this global threat.The data that support the findings of this study are openly available at the following URL/DOI: www.savemedcoasts2.eu/index.php/en/results.

Figure 1 .
Figure 1.Rate of vertical land motion at the GNSS stations considered in this study.
panel (e)).In the examples of figures 3 and 4, ∆SL at 2100/2150 varies between −0.77/−1.2m for the Phlegraean fields to more than 1.1/1.8m for the Po Delta.Values of ∆SL for the PSMSL stations are shown in figure 2 panel (f).∆SL ranges between 438 ± 103 mm and −180 ± 100 mm at 2100 with an average value of 49 ± 6 mm.In the examples of figures 3 and 4, the VLM, although with slightly different values, shows the same sign of the trend for the stations of Venezia Punta della Salute and Napoli Mandracchio while the measured VLM rate is opposite to the IPCC one at Hadera and Marseille (figure 5, panels (c) and (e)).

Figure 2 .
Figure 2. (a) GNSS VLM rate at the grid points.VLM rates on the geographical grid are shown in figure 2 of the supplementary material.(b) GNSS VLM rate at PSMSL stations ordered by increasing longitude.(c) Revised SL at 2100 (relative to 2020) in the SSP5-8.5 scenario at the grid points and (d) at PSMSL stations.(e) Difference between revised and IPCC AR6 SL at 2100 at the grid points and (f) at PSMSL stations.(The correspondence between grid point index and latitude-longitude, PSMSL station index and the station name are shown in tables 1 and 2 of the supplementary material).

Figure 3 .
Figure3.Sea level projections relative to 2020 at four PSMSL tide gauge stations indicated in the title (left column) and at the cells of the grid encompassing the PSMSL stations (right column).The coordinates of the center of each cell are indicated in the title.Each panel shows:-the revised sea level projections, including the VLM estimate performed in this paper, in three SSP scenarios (full lines) and corresponding uncertainties (shadowed areas of the same tone of the lines);-the original IPCC projections for the most optimistic and pessimistic scenario among those shown for the revised projections (dashed lines).No uncertainties are shown for the original projections.

Figure 4 .
Figure 4. Sea level projections relative to 2020 at four PSMSL tide gauge stations indicated in the title (left column) and at the cells of the grid encompassing the PSMSL stations (right column).The coordinates of the center of each cell are indicated in the title.Each panel shows:-the revised sea level projections, including the VLM estimate performed in this paper, in three Tlim scenarios (full lines) and corresponding uncertainties (shadowed areas of the same tone of the lines);-the original IPCC projections for the most optimistic and pessimistic scenario among those shown for the revised projections (dashed lines).No uncertainties are shown for the original projections.

Figure 5 .
Figure 5.Comparison between VLM contributions from IPCC (blue line) and GNSS (orange line) data at the four PSMSL stations (left column) and grid points (right column) of figures 3 and 4. Error bars represent the uncertainty of the measurements for GNSS data, and the distribution spread from the 32nd and 68th percentiles for the IPCC VLM contribution.

Figure 6 .
Figure 6.Sea level projections at 2100 (left column) and 2150 (right column) relative to 2020, for SSP and Tlim scenarios at the PSMSL tide gauges (green dots) and on a geographical grid of 0.5 • × 0.5 • in the Mediterranean Sea (color scale).The maps are obtained by combining the IPCC AR6 projections at regional scales with the VLM derived from GNSS data.Only grid points that include GNSS stations are represented.The VLM in each cell is obtained by calculating the median of all the GNSS measurements contained in the given cell.Panel (a)/(b): RCP1-1.92100/2150; panel (c)/(d): Tlim 1.5 • C 2100/2150; panel (e)/(f): RCP5-8.5 2100/2150; panel (c)/(d): Tlim 5 • C 2100/2150.

Figure 7 .
Figure 7.The main coastal plains of the Mediterranean basin are underlined in red.The geospatial analysis highlights that the total area of the coastal plains of the Mediterranean region potentially prone to be flooded is about 38 529 km 2 .The most affected countries are Egypt (12.879 km 2 ), Italy (10.060 km 2 ) and France (3.681 km 2 ).
-The sea level projections associated with the Intergovernmental Panel on Climate Change Sixth Assessment Report are available at the following link: https://zenodo.org/record/6382554. -The sea level projections developed in this work are available at the website www.savemedcoasts2.eu/index.php/en/results.The VLM rates on the geographical grid are available at the same link.-The VLM rates at the selected PSMSL stations are available in table 2 of the supplementary material.