1 Introduction

The Sele coastal plain, located in the southern Campania Region (southern Italy) between the towns of Salerno and Agropoli (Fig. 1), is the result of the sedimentary aggradation in a Plio-Quaternary tectonic depression located along the western Tyrrhenian margin of the southern Apennine Chain (Cinque et al. 2009; Pappone et al. 2009).

Fig. 1
figure 1

Schematic map of the Sele Plain showing (1) the last Interglacial and (2) Holocene coastal ridges, (3) Paestum Travertines and (4) main normal faults (barbs indicate downthrown side). The dashed rectangles delineate the sectors of the plain (Dx and Sx Sele) where the boreholes with chrono-altimetrical markers, used for the vertical land component evaluation, are located

Full size image

The plain that extends over an area of about 400 km2 is bounded seawards by a narrow sandy coastal strip and landwards by the Lattari and Picentini Mts. (on the N) and by the Cilento Mts. (on the SE). Further seawards, the coastal plain is characterized by an elongated beach dune ridge formed during the last Interglacial period (MIS 5, between 130 and 70 ky BP; Brancaccio et al. 1988), known as Gromola paleoridge (Cinque et al. 2009 and Fig. 1).

Close to the present coastline, a composite sandy ridge occurs representing the evolution of a Holocene barrier-lagoon system (Cinque et al. 2009; Pappone et al. 2009; Amato et al. 2011). It disappears inland under a flat, muddy, back dune depression. Moving from inland to the sea, the dune system shows at least three orders of paleoridges: Laura ridge dated from 5.3 to 3.6 ky BP, Sterpina I and Sterpina II ridges, dated, respectively, older than 2.6 ky BP and about 2 ky BP (Brancaccio et al. 1986 and 1988; Barra et al. 1998; and 1999). These sandy ridges constitute a discontinuous dune system with a mean height of about 3 m a.s.l interrupted by rivers and man-made drainage channels. The back-ridge depressions, only recently drained, are spread over a large area of the plain, with a mean height of about 0.50 m/1.5 m asl. The naturalness of the area is quite well preserved, but some sectors are influenced by human activity, critical since the 1950s (Alberico et al. 2011).

These main features point out the fragility of the study area to risks due to sea-level changes.

For the next century, the global sea level should rise at a maximum value of 59 cm according to IPCC (2007) and of about 140 cm, according to Rahmstorf (2007) in relation to the global model used. These values, the vertical land movements and the effects of past deglaciation, were quantified and summed to define the Relative Sea-level Rise (RSLR) for 2050 and 2100 AD at Sele Plain. Three different scenarios were identified and used to depict the areas that could be interested by coastal erosion and marine inundation.

A preliminary approach, mainly based on the topographic elevation analysis of study area, was used to define both the areas potentially subject to inundation (lowland zones close to the shoreline) and to erosion (shoreline retreat).

2 Sea-level change assessment

The relative sea-level change at Sele plain has been considered as the sum of the glacio-hydro-isostasy, the changes in ocean volume (eustatic) and the vertical land movements.

2.1 Glacio-hydro-isostasy

We used as glacio-isostatic component, the sea level predicted for the past 100 years for 40 sites (site 9 for Sele plain, Lambeck et al. 2011) located along the Italian coasts, since it could be considered as a tool to assess the relative sea level in the next future. At Sele plain, the sea level is rising, as a consequence of past deglaciations, at a rate of 0.44 mm/years (see Table 1a, b).

Table 1 Vertical component of the Dx Sele (a) and Sx Sele (b) in the predicted sea-level rise
Full size table

2.2 Change in ocean volume

To consider the effects of the future change in oceans volume, caused by global warming and ice melting, we used the highest sea-level rise rates predictions. In particular, in our model, we have adopted a sea-level rise prediction for 2100 AD of 5.9 mm/year proposed in A1F1 IPCC (IPCC 2007) and of 14 mm/year proposed by Rahmstorf (2007).

2.3 Vertical land movements

The vertical land component was evaluated with a chrono-stratigraphic study, integrated by a detailed geomorphological analysis, based on cartographic (1:5,000 in scale) and photo-aerial interpretation. About 200 cores very close to the present day shoreline (dashed polygons in Fig. 1) were analyzed in the study area. The cores interpretation was supported by paleoecological/paleo-environmental (mollusks, ostracods, benthic foraminifera and pollens) and tephrostratigraphical analyses as well as by 14C datings (Barra et al. 1998, 1999; Cinque et al. 2009; Amato et al. 2011), and according to Amato et al. (2011), different sedimentary units (dunal sands, beach sands, upper shoreface sands, lagoonal silts and clays, fluvial-marshy sands, silts and clays) were identified.

This approach allowed to recognize several horizons and biological events that were used as paleo-sea-level markers, well constrained by tephra layers (Amato et al. 2011) and by 14C AMS datings (Barra et al. 1998 and 1999; Cinque et al. 2009; Amato et al. 2011).

The paleo-sea-level markers were first corrected taking into account the drilling and ecological-induced uncertainties (according to Lambeck et al. 2004) and then compared with the Holocene-predicted sea-level curve relative to the Sele plain (site n. 9 in Lambeck et al. 2011).

Differential uplift/subsidence trends into the plain were identified. The data set have been grouped into two sectors: the Dx Sele river (from the Sele mouth to the Tusciano mouth) and the Sx Sele river (from the Sele mouth to the Paestum archeological area, Figs. 1, 2b; Table 1b). For each of these two areas, the long-term vertical movements were evaluated.

Table 2 Sea-level rise scenarios in the years 2050 and 2100 for Dx Sele (a) and Sx Sele (b)
Full size table
Fig. 2
figure 2

Comparison between the observed relative sea-level data in the Dx Sele (a) and in the Sx Sele (b) sector, with their vertical uncertainties, and the Versilia (northern Tyrrhenian coast) predicted sea-level curve (redrawn after Lambeck et al. 2011). In blue, the positive values in meters (showing uplift), and in orange, the negative values (showing subsidence)

Full size image

For the Dx Sele, our data suggest two trends: an uplift with an average rate of ca. 0,8 mm/year, from ca. 10 ky BP to ca. 6.0 ky BP, and a subsidence with an average rate of ca. 0,4 mm/year, from ca. 5.0 ky BP to ca.1.0 ky BP.

For the Sx Sele, our data showed an uplift trend with an average of 3.5 mm/year; at ca. 11–10 ky BP, of 0.65 mm/year at ca. 8.0–7.0 ky BP, and an average of 0.2 mm/year at ca. 4.0–2.5 ky BP.

Therefore, the data allowed us to identify an uplift of the whole Sele plain during the first part of the Holocene (up to ca. 6.0 ky BP) and a subsequent different behavior in the two sectors. The different vertical land movements into the plain seem to continue until to the present days, as evidenced by Vilardo et al. (2009), with geodetic data derived by PS-InSar interpretation (1992–2001 time period).

In order to evaluate the future sea-level rise, two different values of vertical land movements were considered: −0.82 mm/year for the Dx Sele and +0.2 mm/year for the Sx Sele. These values represent the highest ones recorded in the plain in the last 5 ky, and they must be regarded with caution, considering that the main values for −0.4 mm/year and +0.22 mm/year for Dx and Sx sectors, respectively, of the Sele plain.

2.4 Sea-level rise scenarios

The zones of Sele coastal plain that could be hit by marine inundation and coastal erosion in the years 2050 (time window of 50 years) and 2100 AD (time window of 100 years) were identified by using three possible scenarios of RSLR labeled with letter from A to C (Table 2a, b). These future events were defined as the concourse of the three components (glacio-hydro-isostasy, changes in ocean volume and vertical land movements), described in the previous paragraph.

The scenario “A” defined for the years 2050 is the outcome of a sea-level rise of 5.9 mm/year (scenario A1F1, IPCC 2007). The scenario “B” is identical to the scenario A, except that the time window considered is of 100 years. The scenario “C” is characterized by a sea-level rise of 14 mm/year (Rahmstorf 2007) and a time window of 100 years.

3 Marine inundation assessment

The potentially flooded zones were evaluated with a static model based on the comparison of Land Digital Elevation Model, (LDEM) and sea level in 2050 AD and 2100 AD (Mazria and Kershner 2007; Poulter and Halpin 2007; Rowley et al. 2007). The zones with the sea level exceeding the underlying ground surface elevation were identified as inundation-prone. We considered also the effect of the LDEM mean square error (± 0.5 m, see paragraph 4.1) on the inundation areas assessment; in fact, each flood area was delineated three times, considering three conditions: morphology of the study area, corresponding to: (a) the LDEM; (b) the LDEM minus 0.5 m and (c) the LDEM plus 0.5 m.

3.1 Land digital elevation model

The LDEM was defined by using the topographic data (elevation points and contour lines) of the Technical Map of the Campania Region, at scale 1:5,000, published in 2001. The study area encompasses the topographic height from 0 to 25 m asl. It represents, in fact, the zone exposed in the next future to potential erosion and/or inundation, caused by sea-level changes (Gornitz et al. 1982; IPCC 2007; Lambeck et al. 2011).

The procedure used to obtain the LDEM is based on the exploration of the elevation data to search for blunder, systematic and random errors. The blunder errors, that are the points with very high or very low values respect to the average elevation range of the study area (USGS 1997), were easily recognized and removed by selecting the points with values outside the elevation range (0–25 m) typical of the onshore coastal area. The systematic errors instead were not easily detectable (Brown and Bara 1994; Carter et al. 1994), because they mainly correspond to irrigation networks and are probably due to an erroneous registration of elevation of these artefacts in the contour line layers. To remove them, all points located within a distance of 5 m from both sides of irrigation network and irrigation lines have been selected and eliminated. The elimination of the random errors was complicated because their elevation only slightly differs from the adjacent points (USGS 1997). These errors were identified by using their distribution on the map of the elevation points, plotted in a distribution diagram available in the Geostatistical Analyst Module of ArcGis software. Selecting each bar of the histogram, it has been possible to control if the data on a map are isolated points or groups of points. In the first case, the points were eliminated, in the second were preserved, but only if they represented local variation in morphology. Eliminated the spurious elevation points, by using the Kriging interpolation method, the LDEM of the Sele plain coastal area was generated (Fig. 3).

Fig. 3
figure 3

Digital Elevation Model of onshore Sele coastal plain from the Picentino to the Solofrone river mouths

Full size image

This raster model has a 10-m cell size and a mean square error of ±0.26 m, defined by using a cross-validation method (Liu et al. 2009; Wise 2011). External-validation was also carried out comparing the elevation values of 23 Regional and IGM vertices, falling inside the study area, with the interpolated elevation values extracted by the LDEM for the same points. This analysis highlighted a mean square error of ±0.54 m.

3.2 Marine inundation map

The lowlands of Sele coastal plain that will be hit by marine inundation in the years 2050 AD and 2100 AD were identified by using three scenarios of RSLR, labeled with letter from A to C (Table 2a, b).

In this preliminary study, the susceptibility of Sele coastal plain to inundation events was oversimplified and mainly based on the topographic height analysis. The LDEM shown in Fig. 4 highlights a dune system characterized by a mean topographic height of about 3 m asl, where only a few dune crests reach 5 m in height, and a large back dune areas with a topographic height ranging from 0.5 to 1.5 m asl.

Fig. 4
figure 4

3D visualization of the Sele coastal plain

Full size image

These last zones have the highest probability to be inundated because of their topographic height and the discontinuities in the dune system.

The extension of areas susceptible to inundation is related to the sea-level scenarios identified in this paper with letters “A”, “B” and “C”, and to the effect of the LDEM mean square error.

The extension of inundated areas (Fig. 5; Table 3) increases from scenario A (ca. 2 km2) to C (ca. 18 km2) because the scenario “A” corresponds to the shortest time windows (50 years) considered, while the scenarios “B” and “C” are related to the longest time interval (100 years) and to the different values of RSLR due to global warming. The last scenario foresees the highest potential sea-level rise in Sele plain.

Fig. 5
figure 5

Inundation susceptibility in the a, b and c scenarios, according to the LDEM. Black areas are inundable coastal sectors

Full size image
Table 3 Extension (in km2) of lowland areas prone to inundation
Full size table

Finally, the same maps were drawn taking into account the effect of the mean square error of ±0.5 m in the LDEM. In particular, the maps of the Figs. 6, 7 were drawn considering the error –0.5 m (hereafter LDEM−) and of +0.5 m (hereafter LDEM+), respectively.

Fig. 6
figure 6

Inundation susceptibility for a, b and c scenarios, considering a topographic model affected by a mean standard error of +50 cm (LDEM+). Black areas are inundable coastal sectors

Full size image
Fig. 7
figure 7

Inundation susceptibility for a, b and c scenarios, considering a topographic model affected by a mean standard error of −50 cm (LDEM−). Black areas are inundable coastal sectors

Full size image

For the mean square error −0.5 m, the extension of inundated zones is the widest of the study area for all three scenarios (Table 3), while with a mean square error of +0.5 m, the scenarios A1 and B1 do not show inundation zones except for the C1 scenario.

4 Coastal erosion

In order to predict the effects of the sea-level rise on the Sele Plain shoreline, profiles of eight beaches from the Tusciano river mouth to the Agropoli promontory (Fig. 8) were realized by using a DGPS survey (Di Paola 2011). Along these profiles, the Davidson-Arnott (2005) flooding methods were used to estimate the probable coastal retreats (Table 4). This methodology differs from the classical Bruun rule (Bruun 1962). The hypotheses of Davidson-Arnott (2005) are as follows:

  • The beach and foredune are eroded as a result of the profile modification because of sea-level rise, while the junction between the beach and the dune migrates landwards and upwards to keep pace with the rising sea level.

  • A net onshore migration of sediment in the nearshore profile keeps pace with rising sea level. The outer part of the nearshore is eroded, and the point of closure moves landwards and upwards to keep pace with sea-level rise and with the landward migration of the shoreline.

  • All the sediments eroded from the dune are transferred landwards, and as a consequence, the foredune migrates inland. Because the volume transferred landwards is equal to the volume eroded, the dune will maintain its overall volume.

Fig. 8
figure 8

Location of topographic profiles along the Sele coastline

Full size image
Table 4 Amount of coastal retreat in the Sx and Dx Sele, applying the Flooding methodology of Davidson-Arnott (2005), which can be related to the scenarios showed in Fig. 6
Full size table

This model allows the displacement of the shoreline to be independent from the actual shape of the beach profile and thus, whether the bars are present or not (Allison and Schwartz 1981), the displacement depends only from the tan β as follow:

$$ R = S \cdot \frac{1}{\tan \beta } $$

where R is the shoreline retreat, S is the RSLR and β the slope of the profile between the dune ridge and the closure depth. The RSLR of scenarios “A,” “B” and “C” was used to derive the mean retreat for the left and right side of Sele river (Table 4).

The predicted beach erosion shows a mean landward retreat of the shoreline of about 20 m for the 2050 AD (scenario “A”) and of 40 m for the year 2100 AD (scenario “B”). However, in the scenario “C,” adopting Rahmstorf (2007) prediction, the landward retreat could reach up to 93 m.

The method was based on a static approach; however, these predictions can be even worse in the case of incoming natural (i.e. extreme storms) or man-induced (i.e. deficit of coastal sediment supply) events.

5 Conclusions

The response of Sele coastal zone to the ongoing changes in ocean volume due to climatic changes combined with the vertical land movement and the glacio-hydro-isostasy has been here evaluated.

The comparison of the Holocene paleo-sea levels with those of the tectonically stable areas (sensu Lambeck et al. 2011) allowed to identify, during the Holocene, a gently uplifting in the SE sector of the coastal plain, with an average of 0.22 mm/year, and a subsidence for the sector between the Sele and Tusciano river mouths, with an average of −0.4 mm/year. In this work, we have used the highest rates of vertical movements corresponding to –0.82 mm/year for the Dx Sele and +0.2 mm/year for the Sx Sele (Table 1a, b).

These values, combined with the sea-level rise predictions (IPCC 2007; Rahmstorf 2007), due to global warming and the local effects of the glacio-hydro-isostasy (Lambeck et al. 2011), have allowed to identify three sea-level rise scenarios.

The use of a simple static inundation model allowed to identify the lowland areas that could be inundated. The extension of these areas ranges from a minimum of 0.42 km2, for the 2050 AD, to a maximum of 7.6 km2 for the 2100 AD. In particular, these areas are mainly concentrated in the NW sector of Sele Plain, where the back-ridge zones are characterized by topographic elevation lower than 2 m above asl and should be considered as the major flood prone areas in the plain. Furthermore, because the extension of areas prone to inundation were demonstrated sensitive to the error associated to the DEM of Sele coastal zone, in the future, a more detailed study, based on a DEM with higher resolution, is necessary.

Moreover, it is important to consider that major historical earthquakes can cause very rapid coseismic vertical displacements (i.e. described in Mastronuzzi and Sansò 2002), and their effect can produce vertical movement over a time windows of different years constituting the primary cause of coastal flooding (Ballu et al. 2011).

The coastal retreat, due to sea-level rise predicted for the years 2050 and 2100 AD, was also calculated, according to the flooding method (Davidson-Arnott 2005). In this case, the mean retreat values vary from a minimum of ca 19 m to a maximum of ca 93 m. In the latter case, the landward retreat of the coastline could be very severe for the future dune ridge preservation, since the mean width of the present backshore does not exceed 80 m.

Finally, this study shows a high sensitivity of the Sele plain to the future RSLR, particularly in the NW sector, where the back-ridge area, characterized by a less elevated and flat morphology, could undergo major flood risks respect to the SE sector. The above geologic hazard needs to be carefully taken into account in the future management of this territory.