*Article* **Natural and Anthropogenic Factors Shaping the Shoreline of Klaipeda, Lithuania ˙**

**Vitalijus Kondrat \*, Ilona Šakurova, Egle Baltranait ˙ e˙ and Loreta Kelpšaite-Rimkien ˙ e˙**

Marine Research Institute, Klaipeda University, Universiteto Ave. 17, LT-92294 Klaip ˙ eda, Lithuania; ˙ ilona.sakurova@ku.lt (I.Š.); egle.baltranaite@ku.lt (E.B.); loreta.kelpsaite-rimkiene@ku.lt (L.K.-R.) **\*** Correspondence: vitalijus.kondrat@ku.lt

**Abstract:** Port of Klaipeda is situated in a complex hydrological system, between the Curonian ˙ Lagoon and the Baltic Sea, at the Klaipeda strait in the South-Eastern part of the Baltic Sea. It has ˙ almost 300 m of jetties separating the Curonian Spit and the mainland coast, interrupting the main path of sediment transport through the South-Eastern coast of the Baltic Sea. Due to the Port of Klaipeda reconstruction in 2002 and the beach nourishment project, which was started in 2014, the ˙ shoreline position change tendency was observed. Shoreline position measurements of various periods can be used to derive quantitative estimates of coastal process directions and intensities. These data can be used to further our understanding of the scale and timing of shoreline changes in a geological and socio-economic context. This study analyzes long- and short-term shoreline position changes before and after the Port of Klaipeda reconstruction in 2002. Positions of historical ˙ shorelines from various sources were used, and the rates (EPR, NSM, and SCE) of shoreline changes have been assessed using the Digital Shoreline Analysis System (DSAS). An extension of ArcGIS K-means clustering was applied for shoreline classification into different coastal dynamic stretches. Coastal development has changed in the long-term (1984–2019) perspective: the eroded coast length increased from 1.5 to 4.2 km in the last decades. Coastal accumulation processes have been restored by the Port of Klaipeda executing the coastal zone nourishment project in 2014. ˙

**Keywords:** Baltic Sea; Port of Klaipeda; shoreline changes; DSAS; clusterization; regime shift detection ˙ ; dredging; sand nourishment

### **1. Introduction**

Erosion is a significant problem affecting sandy beaches that will worsen with climate change and anthropogenic pressure. Sandy shorelines are highly dynamic due to altering wave conditions, sea levels and winds, geological factors, and human activity [1]. Therefore, identifying the most vulnerable areas to erosion is crucial for nearshore communities since it could significantly affect their socio-economic state through destruction of infrastructure, loss of land and property on the coast, and valuable beach areas used for recreation.

Shore regeneration is a slow process lasting for more than one year, while erosion usually occurs in a matter of a few days, making it difficult to detect visually. As short-term measurements do not reflect actual multi-annual dynamic trends, studies involving several shoreline decay and regeneration cycles are necessary to determine long-lasting changes in the shoreline dynamics. Typically, coastal research to assess and predict long-term shoreline dynamics and the erosion rates is based on the data covering up to 10 years (short-term), 10–60 years (medium-term), and more than 60 years (long-term) of shoreline position [2–4].

Shoreline dynamics depend on different causes, mainly on the sediments in the sea-land system [5–7]. Furthermore, the different coastal stretches have particular favorable hydrometeorological conditions for the accumulation or erosion processes. The rapid urbanization of the coastal zone has a significant impact on shoreline development [8–10]. Sustainable coastal development requires knowledge of the coastal processes

**Citation:** Kondrat, V.; Šakurova, I.; Baltranaite, E.; Kelpšait ˙ e-Rimkien ˙ e, L. ˙ Natural and Anthropogenic Factors Shaping the Shoreline of Klaipeda, ˙ Lithuania. *J. Mar. Sci. Eng.* **2021**, *9*, 1456. https://doi.org/10.3390/ jmse9121456

Academic Editor: Carlos Daniel Borges Coelho

Received: 18 November 2021 Accepted: 17 December 2021 Published: 20 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

combined with incorruptible urbanization and properly chosen shoreline erosion mitigation methods [10,11]. Often, an insufficient understanding of the coastal processes causes costly incidents.

A number of studies [8,12,13] show the impact of anthropogenic factors in particular port activities on shoreline positions. Erosion and accumulation are naturally occurring processes that often coincide in a dynamic equilibrium [14]. However, increasing anthropogenic pressure at the coast has disrupted the natural development of the coast, accelerating erosion processes in some places and causing accumulation in others [14]. Analysis of shoreline changes is a well-developed field that has progressed complex data processing and analytical protocols [15]. However, quantifying coastal development trends is only one aspect of the problem; it is necessary to understand the drivers of change and address local impacts in a broader regional context that is important from a decadal to a centennial timescale [15]. Understanding the causes of atypical coastal development is important to make sustainable coastal zone management plans. Such knowledge is crucial not only for the coastal dynamics experts, but also for the port managers, as it can serve as the basis for future decisions on how to reduce port damage to the coasts.

This paper analyses the shoreline dynamic in the context of climate change and increased anthropogenic pressure, focusing on identifying long- and short-term shoreline movement tendencies and identifying the direct impact zone of the Port of Klaipeda. ˙ As well as answering the question of whether and how shoreline evolution is affected by the artificial sand nourishment carried out in accordance with the Port of Klaipeda ˙ management plan.

#### **2. Study Site**

The Lithuanian coast of the Baltic Sea represents a generic type of almost straight, relatively high-energy, actively developing coasts that (i) contain a large amount of fine mobile sediment; (ii) are open to predominating wind and wave directions; and (iii) are exposed to waves from many directions [16]. The study area extends 10 km from Klaipeda ˙ seaport jetties to the north and 10 km to the south. This particular area was chosen based on the following aspects: (i) the broad demand spectrum of recreational uses [17]; (ii) the high risk of coastal erosion [18,19]; (iii) the possibility of direct and indirect anthropogenic impacts [20,21].

The South-Eastern coast of the Baltic Sea is formed by the presence of the Port of Klaipeda [ ˙ 21,22]. Historically, the Port of Klaipeda has been known from the 13th century ˙ when vessels of Lubeck and Bremen merchants used to moor in the small port neighboring the Klaipeda castle [ ˙ 23]. Port expansion to the Klaipeda strait started in 1745, and the ˙ chronicle of 1797 mentions that Port of Klaipeda consists of the Dane river port and a big ˙ water basin in the strait of the Curonian Lagoon. In the 19th century, wooden jetties were constructed [22]. 1924–1939 was a period when Klaipeda seaport was at its flourishing ˙ peak—new stony jetties and quays were assembled [24,25]. Since the occurrence of the first jetties, ongoing coastal engineering problems were encountered relating to wave exposure, siltation within the port, extensive dredging requirements, and seiching within the confines of the present harbor [22,26].

After the construction of the first port jetties, at the end of the 19th century, the shoreline moved seawards significantly on both sides of the jetties [20]. This insight raises doubts about the predominant sedimentary direction from south to north [6]. The dumping of the dredged sand can partly explain this accumulation tendency in the northern part of the jetties from the Klaipeda strait [ ˙ 22]. Up until the beginning of the 20th century, sand dredged from the port had been dumped at shallow depths north of the jetties, initiating coast accumulation [22].

After the prolongation and construction of new concrete jetties at the beginning of the 20th century (works finished till 1934) [21] alongside changed dredging policies [13], observations were made that sand dredged next to the port jetties returns into the inlet and continues dredging works to ensure the depth of the entrance channel.

Due to depth restrictions in the Danish Straits, vessels with a maximum draft of 16.5 m, and in some cases, vessels with a draft of up to 17 m can enter the Baltic Sea. Another limitation for ships entering the Baltic Sea is the bridge height to about 65 m entering the strait of the Great Belt, which connects the Danish islands of Zeeland and Funen. These restrictions prevent vessels with a draft greater than 16.5 m from entering the Baltic Sea from those of Class Panamax (Baltmax). The long-term competitiveness and sustainability of the Klaipeda seaport can be ensured only by increasing the technical capability of the ˙ port to receive and service ships of the maximum capacity [27].

Therefore, in 1999, the final design for reconstruction of the Port of Klaipeda jetties ˙ was established. The seaport jetties system was reconstructed by narrowing the entrance channel and changing the position of the northern pier. In 2002 the northern pier was extended by 205 m (up to 733 m) and the southern pier by 278 m (up to 1374 m) (Figure 1) [28]. At the same time, the entrance channel was dredged to a depth of 14.5 m. According to the recent port development plan, the entrance channel will be dredged up to 17 m by 2023.

**Figure 1.** The Klaipeda seaport jetties before and after the reconstruction of 2002, ( ˙ **a**) 1997, and (**b**) 2005.

The Port of Klaipeda, located at the Klaip ˙ eda strait (Figure ˙ 2) (South-Eastern coast of the Baltic Sea), divides the Lithuanian coast into two geologically and geomorphologically different parts: southern—the coast of the Curonian spit, northern—mainland coast (Figure 3) [29]. Port jetties interrupt the main sediment transport path and significantly influence the Lithuanian coast's northern (38.49 km long) part [6,20]. Only Quaternary sediments are found on the Lithuanian coast of the Baltic Sea [6,30]. From the geological point of view, the mainland coast and the Curonian Spit coast are not homogenous (Figure 4). The geological structure of the mainland coast was mainly determined by the sediments formed during the last few glaciations. The sediments of the Curonian Spit coast were formed in the Baltic Sea basin—starting with the Baltic Ice Lake and ending with the modern Baltic Sea [6,30].

**Figure 2.** Location of the study site in the south-eastern Baltic Sea, A: the Curonian Spit coast, B: the mainland coast.

The sandy sediments form the part of the Curonian Spit coast: this Lithuanian coastal sector is characterized by accumulation relief [6]. The mainland coast of Lithuania is geologically heterogeneous: the northern part of the mainland coast is mainly formed of fine-grained sand (0.25–0.1 mm), while the southern and central parts of the mainland coast are formed by the medium-grained (0.5–0.25 mm) and coarse-grained (1–2.5 mm) sand [6,14]. A detailed description of the Lithuanian coast geomorphological and geological structure is provided by Bitinas et al. (2005)

According to the granulometric analysis of sediment samples from 2019 along the study area (Figure 4), on the Curonian Spit coast (A, a), very well and moderately sorted (σ = 1.21–1.47 mm) fine sand (Md = 0.20–0.37 mm) prevails, while on the mainland coast (B, b, c), the sorting of the sediments differs in a cross-shore profile. In profile *b,* moderately well-sorted (σ = 1.44 mm) medium sand (Md = 0.32 mm) prevails in a shoreline area, well-sorted (σ = 1.19 mm) slightly very fine gravelly medium sand (Md = 0.21 mm) prevails in a beach area, and moderately well-sorted (σ = 1.47 mm) sand prevails (Md = 0.36 mm) in a foredune area. In profile *c*, poorly sorted (σ = 3.84 mm) very fine gravelly fine sand (Md = 0.24 mm) prevails in a shoreline area, poorly sorted (σ = 10.69 mm) medium gravelly fine sand (Md = 0.23 mm) prevails in a beach area, and poorly sorted (σ = 18.21 mm) sandy very fine gravel prevails (Md = 1.12 mm) in a foredune area.

**Figure 3.** Study site shoreline features: (**a**) the Curonian Spit coast Smiltyne I beach ( ˙ © I. Šakurova); (**b**) the Curonian Spit coast Smiltyne I beach ( ˙ © I. Šakurova); (**c**) the mainland coast Giruliai beach (© L. Kelpšaite-Rimkien ˙ e); ( ˙ **d**) the mainland coast Melnrage I beach (© V. Kondrat). ˙

> During the dredging of the Klaipeda strait, the glaciogenic moraine deposits and ˙ alluvial sediments are mainly excavated—sand (0.002 mm 10–30%–2 mm 50%) and silt (0.002 mm 10–30%–2 mm 30–50%). All lithological sediment types are dumped in dumping area I (Figure 4) at a depth of 45–50 m. The II dumping area (Figure 4) is intended only for the dumping of sandy sediments—fine (0.25–0.1 mm > 50%) and aleuritic (<0.063 mm 10–30%) sand at a depth of 28–35 m. Since 2001, clean sand that meets sanitary–hygienic requirements excavated from the port entrance channel has been dumped in the dumping area III (Melnrage-Giruliai) at a 4–6 m depth. This area is intended to replenish the sediment ˙ balance and restore beach sand reserves [24].

**Figure 4.** Map of Quarternary sediment type of coastal area and dumping zones of dredging material. Lg III B glaciolacustrine sediments of the Baltic Ice Lake (fine sand); lg III bl—glaciolacustrine sediments (various sand); lgt III bl—marginal glaciolacustrine deposits (fine sand); m IV L—Litorina Sea sediments (fine sand); v IV—aeolian deposits (fine sand); m IV a—nearshore sediments (extra fine sand (0.05—0.1 mm)); m IV b—nearshore sediments (fine sand (0.1—0.25 mm)); m IV c—nearshore sediments (gravel with sand); gII-III—glacial deposits of Middle and Upper Pleistocene (unseparated), glacial loam, boulders and gravel (washed till). I—distant dumping area; II—near dumping area; III—nearshore dumping area (adapted from Bitinas et al., 2004). Grain-size composition of surface sediments at Smiltyne I ˙ (**a**), Melnrage I ( ˙ **b**), and Karkle ( ˙ **c**). Orange color line—western part of the dunes (foredune); blue line—the middle of the beach; red line—a dynamic shoreline in July of 2019.

#### **3. Materials and Methods**

#### *3.1. Analysis of Cartographical Data*

In this paper, we evaluate a period of 35 years of shoreline position variation tendencies for 1984–2019. All shoreline position changes were determined using the available high accuracy (1:10,000) cartographic data for the years: 1984, 1990, 1995, and 2005 (Table 1) obtained from Lithuania's National Land Service under the Ministry of Agriculture and GPS survey data for 2015 and 2019. The shoreline position was established at the middle of the swash zone by dual-band GPS receiver "Leica 900".

Shoreline position changes were analyzed with the ArcGIS extension DSAS v. 5.0 (Digital Shoreline Analysis System) package, developed by the United States Geological Survey (USGS) [31,32]. The DSAS is executed in five steps: (1) shorelines digitizing and uniforming to WGS-84 coordinate systems (UTM Zone 34); (2) computation of the uncertainties; (3) baseline creation and transects generation; (4) computation of distances between baseline and shorelines at each transect; and (5) computation of shoreline change statistics.

Three statistical parameters—net shoreline movement (NSM), end-point rate (EPR), and the shoreline change envelope (SCE)—were estimated and analyzed along with each transect every 25 m along the shoreline (796 transects in total). NSM values report the net change of the shoreline in the study period between the oldest and most recent shoreline.

EPR rate (m/yr) indicates change rates between the earliest and most recent shoreline positions. SCE capacity provides the envelope of shoreline variability, and it is the only measure of the total shoreline change among all the available shoreline positions [33].

**Table 1.** Shoreline positioning and detection errors. Ed—digitization error, Ep—pixel error, Es—sea-level fluctuation error, Ec—shoreline line detection or resolution errors, Etc—T-sheets plotting errors, Er—rectification error, Ut—shoreline capture error.


### *3.2. Data Reliability and Limits of Uncertainty*

The shoreline position is highly variable in short time scales due to heavy storms, waves, and wind setup when extreme natural variations induce significant temporary shoreline retreat. Mapping the historical shorelines introduces additional uncertainties [34]. Although most researchers have similar techniques for estimating shoreline value changes, the methodology used to estimate changes varies considerably, significantly altering the accuracy and reliability of the data collected or determined. The dynamics of the shore itself may also cause certain differences and inaccuracies in shoreline surveys. Therefore, the values of the same shoreline determined by two independent scientists in the same field of science can vary considerably in their size and accuracy [35].

The most significant differences in the data occur during the digitization and processing of cartographic material. The differences in the values of shoreline changes may also occur due to the different statistical research methods used to determine the degree of shoreline change (shoreline change rate). The primary data and the analysis methods are the main factors defining the shoreline variations and accuracies. Therefore, prior to choosing a statistical research method, it is imperative to estimate the errors in determining the shoreline position in the cartographic material [36].

In this study, we determined three shoreline positioning and detection errors (Table 1) based on [14,36,37]:

The error in the position of the shoreline when determining in the T-Sheets:

$$\text{Ut} = \pm \left( \text{Ed}^2 + \text{Ep}^2 + \text{Etc}^2 + \text{Es}^2 + \text{Ec}^2 \right)^{1/2} \tag{1}$$

The positioning error of the shoreline in orthophotos equals:

$$\text{Ut} = \pm \left( \text{Er}^2 + \text{Ed}^2 + \text{Ep}^2 + \text{Es}^2 + \text{Ec}^2 \right)^{1/2} \tag{2}$$

GPS data error:

$$\mathbf{U}\mathbf{t} = \pm \left(\mathbf{E}\mathbf{s}^2 + \mathbf{E}\mathbf{c}^2\right)^{1/2} \tag{3}$$

Here: Ut—shoreline capture error, Er—rectification error, Ed—digitization error, Ep pixel error, Ets—photo plan creation error, Ec—shoreline line detection or resolution errors, Eg—georeferencing error; Es—sea-level fluctuation error; Etc—T-sheets plotting errors.

The shoreline uncertainty limit for different periods is equal to the sum of the shoreline fixation errors for different periods:

$$\sum \mathbf{Ut} = (\mathbf{Ut}\mathbf{n}\_1 + \mathbf{Ut}\mathbf{n}\_2 + \mathbf{U}\mathbf{t}\mathbf{n})^{1/2} \tag{4}$$

Here n1, n2,—shoreline detection errors for different periods.

The shoreline uncertainty threshold (minimum time criterion) in the statistical methods for deter-mining shore change (EPR) equals:

$$\sum \text{Ut/n} \tag{5}$$

Here n—research period.

#### *3.3. Clusterization*

K-Mean cluster analysis for the net shoreline movement (NSM) values was applied to identify shoreline zones with similar evolution tendencies [38]. The K-means algorithm is a simple and popular clustering approach used in various applications [39]. It is a point-based clustering approach that starts with cluster centers located initially in arbitrary locations and goes through each stage of the cluster center to reduce the cluster error [39–41].

$$\mathbf{E} = \sum \left\| \mathbf{X\_i} - \mathbf{m\_i} \right\|^2 \tag{6}$$

where E is the sum of squared errors for all objects in the data, X<sup>i</sup> is the point in a cluster, and m<sup>i</sup> is the mean of cluster k<sup>i</sup> . The objective of K-means is to minimize the sum of squared errors over all k clusters. The algorithm first places k points in the space represented by the objects clustered as initial group centroids. The second step is to assign each object to the nearest cluster center. Then, the mean of each cluster is calculated to obtain a new centroid. These steps are repeated until the centroids do not change. The within-cluster sum of squares measures the variability of the observations within each cluster. In general, a cluster with a small sum of squares is more compact than a cluster with a large sum of squares [38,39]. Clusters with higher values exhibit more significant variability of the observations within the cluster [38,39]. The number of clusters is chosen based on the elbow method [38], whose main idea is to define groups such that the total intra-cluster variation (or the total sum of squares within clusters (WSS)) is minimized. In this case, the elbow of the curve is formed for the five clusters (Figure 5).

**Figure 5.** The number of clusters is chosen based on the within-cluster sum of squares parameter.

#### *3.4. Analysis of Meteorological Data*

The meteorological data (annual mean wind speed and direction) of the 1960–2019 time period were analyzed to detect the wind direction's regime shift. The meteorological data were acquired from the Marine Environment Assessment Division of the Environmental Protection Agency (EPA) and derived from the Klaipeda coastal meteorological station ˙ (Figure 2) under the Lithuanian Ministry of Environment's environmental monitoring program. The program has been prepared in line with the legislation of the European Union.

A STAR (Sequential T-test Analysis of Regime Shifts) algorithm was applied to determine regime shifts in the analyzed time series (https://www.beringclimate.noaa.gov/ (accessed on 10 October 2021)). The algorithm was built upon a sequential *t*-test that can signal the possibility of a real-time regime shift [42]. The algorithm can process the data regardless of whether it is presented in anomalies and/or absolute values or not. It can automatically calculate regime shifts in large sets of variables [42].

For this study, the following set of input parameters were used: cut-off length (I) was set to 10 years; Hubert's weight parameter (HWP) was set to 1. HWP determined the weight of outliers in the calculation of average values of the regime shift. The confidence level was set to 0.1.

### **4. Results**

#### *4.1. Long-Term Shoreline Changes*

NSM for the entire study period **1984–2019** showed (Figure 6) that 60.43% of the shoreline was accumulative, 20.98% erosive, and 18.59% was stable or within the range of uncertainty ±9.08 m (Table 2). Generally, the studied coast can be described as accumulative with the average 14.46 ± 1.92 m shoreline movement offshore tendency; the average shoreline movement velocity was 0.42 ± 0.03 m/year.

**Figure 6.** Net Shoreline Movement (NSM) rates 1984–2019 short-term vs long-term tendencies on the Curonian Spit coast (**A**) and the mainland coast (**B**). Annual shoreline change rates are shown on the transects graph. Purplish color tones on the transects indicate a trend of coastal erosion, while green tones indicate a trend of accretion, and grey color indicates shoreline variation values in its positioning and detection uncertainty range. Numbers and lines on the A and B coasts indicate transects distribution along the study site.


**Table 2.** Shoreline uncertainty range.

\* T-Sheets; \*\* Orthophotos; \*\*\* GPS.

Comparing trends of shoreline changes in **1984–2019**, we found that the accumulation processes on the shores of the Curonian Spit accounted for 96.12% (396 out of 412) of transects. The shoreline moved towards the sea at an average speed of 1.01 ± 0.02 m/year (Figure 7), with the highest rates of the EPR 2.15 m/year. The NSM value was 35.97 ± 0.69 m, stable shoreline changes were found in 3.64% of transects and erosions in 0.24% of transects. The highest intensity of erosion processes at the Curonian Spit was recorded in 1984–1995. The negative shoreline shift towards the mainland was found in 6.07% (25 out of 412) of transects, where the average NSM value was −19.38 ± 2.50 m. Stable shoreline changes were found in 18.69% (77 of 412) of transects, and accumulation was detected in 75.24% (310 of 412) of transects with an accumulation rate of 2.17 ± 0.05 m/year, NSM value was 23.86 ± 0.52 m.

**Figure 7.** Graph showing the distribution of EPR (**a**) and wind rose (**b**) for 1984–2019.

In **1984–2019**, accumulation processes occurred in 22.14% (85 out of 384) of transects on the mainland coast. The shoreline shifted towards the sea within 20.30 ± 1.04 m, with an average speed of 0.57 ± 0.03 m/year (Figure 7). Erosion during this period accounted for 43.23% (166 out of 384) of transects, and the shoreline shifted towards the mainland at an average velocity of −0.70 ± 0.02 m/year; the NSM value was −24.84 ± 0.74 m. Stable shoreline was found in 34.64% (133 of 384) of transects. Significant coastal erosion extends at the northern pier of the Port of Klaipeda ˙ −56.9 m in transect 413 (Figure 6). Accumulation processes in the accesses of Port of Klaipeda piers changed to intensive ˙ erosion, which in 2019 covered 700 m (28 transects) of the coast; the total NSM in them was −28.28 m, the EPR value was −0.76 ± 0.04 m/year.

#### *4.2. Short-Term Shoreline Changes*

Comparison of the shoreline changes in 1984–1990 and 1984–2019 showed that the area of eroded coast increased 2.7 times, from 1.50 km (60 transects) to 4.15 km (166 transects). The effect of accumulation processes in 1984–2019 was recorded in 85 transects instead of 145 transects in 1984–1990. The accumulation rate decreased from 4.33 ± 0.11 m/year to 0.57 ± 0.03 m/year. The area of stable shores decreased from 3.325 km (133 transects) to 4.475 km (179 transects).

During the **1984–1990** period (Figure 8), the overall shoreline change was positive the coast moved seawards on average 23.95 ± 0.76 m. During this period, the predominant wind direction was W, WSW, and the average wind speed variated from 0 to 16 m/s.

**Figure 8.** Graph showing the distribution of EPR (**a**) and wind rose (**b**) for 1984–1990.

Accumulation was detected in all transects of the Curonian Spit coast, where the shoreline moved seawards by 12.99–65.08 m with an average velocity of 6.56 ± 0.08 m/year. On the mainland coast, the shoreline position changes were observed within the range of determination ± 12.42 m and can be considered as quasi-stable.

Coastal erosion was observed in a 1.5 km (60 transects) area to the north in the 6.2 km from the northern seaport jetty. The shoreline moved landward at an average velocity of −3.97 ± 0.13 m/year. The most significant negative change occurred in the 672nd transect and reached −41.58 m. Accumulation occurred in 37.8% of transects on the mainland coast, and here the shoreline moved seawards, with an average velocity of 4.33 ± 0.11 m/year.

In the **1990–1995** period (Figure 9), the coast has been intensively eroded, with the predominant 0–16 m/s W, WSW, SW wind direction. The shoreline moved landwards in 620 (77.9%) from 796 transects with an average of −22.85 ± 0.46 m. Significant changes in shoreline movement were observed in the immediate proximity of the seaport jetties. In the Curonian Spit coast, the maximum value of NSM was −100.85 m and was detected in the 412rd transect, next to the southern Klaipeda seaport jetty (Figure ˙ 8). The most significant shoreline movement landwards was observed in a 250 m (402–412 transects) coastal area to the south from the southern seaport jetty. Here the shoreline moved toward land on average 77.88 ± 1.11 m with an average velocity of (EPR) −15.58 ± 0.22 m/year.

**Figure 9.** Graph showing the distribution of EPR (**a**) and wind rose (**b**) for 1990–1995.

65.9% of transects on the mainland coast can be described as erosive. The average change velocity reached −4.09 ± 0.10 m/year, and the shoreline moved landwards about −20.45 ± 0.51 m. The quasi-stable coast was observed in 131 transects (34.1%), and an average EPR value was 0.44 ± 0.08 m/year. The most significant shoreline movement >30 m was detected in the 419–443 transect. The maximum value was observed in the 424th transect and reached 49.61 m (EPR −9.92 m/year).

The following ten years, **1995–2005**, with the predominant SW, SSW, and WSW (0–16 m/s velocity) winds (Figure 10), had accumulative tendencies at the Curonian spit coast. The coast started recovery after the previous erosive period. Furthermore, hurricane Anatoly, which occurred in December 1999 [20], was not visible in the coastal evolution processes. It is evident that the quasi-stable part became erosive during the last five years at the mainland coast, and all other parts stayed accumulative.

**Figure 10.** Graph showing the distribution of EPR (**a**) and wind rose (**b**) for 1995–2005.

The total change of the shoreline in the studied area in 1995–2005 was positive and amounted to 6.72 ± 0.39 m with an EPR value of 0.67 ± 0.04 m/yr. The Curonian Spit coast was characterized as accumulative. Here accumulation processes were observed in 320 transects from 412, and the accumulation rate was 1.70 ± 0.044 m/yr. Erosion was observed in 27 transects (650 m). From 304 to 320 the transect EPR value was −1.00 ± 0.03 m/yr. From 277 to 282, the EPR value reached −0.86 ± 0.10 m/yr. The significant accumulation rate of 4.15 m/yr. (NSM 41.52) was noted in the immediate proximity of the jetties.

In the next five years, **2005–2010** (Figure 11), wind accumulation processes prevailed, with the WSW, SW, S, SE (0–12 m/s). In 61.1% of transects, the shoreline moved seawards with an averaged velocity of 2.12 ± 0.05 m/yr., and NSM value reached 10.62 ± 0.25 m.

**Figure 11.** Graph showing the distribution of EPR (**a**) and wind rose (**b**) for 2005–2010.

Accumulation processes were more frequent on the Curonian Spit coast, which was observed in 67.7% of transects. The average velocity of shoreline movement seawards was +2.42 ± 0.07 m/yr. During 2005–2010 the shoreline erosion on the Curonian Spit coast occurred only in 10.40% of transects that amounted to 1075 m out of 10.3 km. The significant erosive coastal stretch was found in the southern part of the Curonian Spit between 10 and 34 transects. In the 625 m section, the shoreline moved landwards, on average −12.82 ± 0.29 m (EPR −2.56 ± 0.26 m/yr). The maximum value of NSM was noted in the 26th transect and reached −26.69 m.

On the mainland coast, accumulation was detected in 53.9% (270 out of 384) of transects, and the shoreline moved towards the sea by an average of 8.62 ± 0.28 m. The average EPR value was 1.73 ± 0.06 m/yr. Stable shoreline changes or changes in the shoreline determination uncertainty range within ±0.69 m/yr were detected at 119 or 31% of transects. Coastal erosion was recorded in 15.1% of transects (58 transects), in which the shoreline moved landwards at an average speed of −2.01 ± 0.19 m/yr. The most significant adverse changes in the shoreline position were found between 413 and 446 transects. In this 850 m-long coast stretch, the shoreline shifted to the mainland on average by −9.64 ± 0.28 m (EPR was −1.93 ± 0.06 m/yr).

During the **2010–2015** period (Figure 12), with the predominant WSW, SW, S, SE (0–12 m/s) winds, accumulation processes were noticed in 94.9% of transects (391 out of 412 transects) on the coast of the Curonian Spit, in which the shoreline moved seawards at an average speed of 3.40 ± 0.09 m/yr. In 50% of transects (206 out of 412 transects), the shoreline shifted from land to sea by an average of 27.82 ± 0.04 m (NSM). The maximum value of NSM reached 49.67 m in the 318th transect.

**Figure 12.** Graph showing the distribution of EPR (**a**) and wind rose (**b**) for 2010–2015.

On the mainland coast, erosive processes were observed during 2010–2015. Negative tendencies of shoreline displacement landwards were recorded in 47.4% of transects (182 out of 384), in which the shoreline generally shifted at an average speed of −0.51 ± 0.07 m/yr. The significant shoreline movement towards land was recorded in the 1175 m shoreline section north of the northern seaport jetty (between tr. 413 and 459). The average EPR value was −1.49 ± 0.01 m/yr, and the average NSM value was −8.63 ± 0.07 m; the maximum value of EPR was −2.57 m/yr in 421 transects, and the maximum NSM value was −14.84 m. The section of the shore from 746 to 796 transects stands out. This shore of 1275 m in 2010–2015 moved towards the sea in total −5.10 ± 0.07 m, and the erosion rate reached −0.88 ± 0.01 m/yr. The central part of the mainland coast was mainly formed by accumulation processes, which accounted for 41.9% of all transects (182 out of 284). The average accumulation rate in these transects was 1.14 ± 0.06 m/yr, the value of NSM was 6.80 ± 0.34 m. Stable shoreline fluctuations of about ± 0.19 m/yr were recorded in the 41st transect.

During the last analyzed period **2015–2019** (Figure 13), the predominant wind direction was WSW, SW, SWS, S, SSE (0–12 m/s velocity) and all of the coast was erosive. Over these 4 years, the shoreline moved seawards in 80.9% of transects (644 out of 796) with the average EPR value −4.24 ± 0.12 m/yr., and NSM — −15.91 ± 0.46 m.

**Figure 13.** Graph showing the distribution of EPR (**a**) and wind rose (**b**) for 2015–2019.

On the Curonian Spit coast, erosion processes were detected at 97.3% of transects (401 out of 412) and were 3 times more intense than on the mainland. Here the EPR value reached −5.72 ± 0.15 m/yr., and the NSM respectively was −1.80 ± 0.07 m/yr.

The mainland coast moved seawards in 63.3% of transects (243 out of 384). In the southern part of the mainland coast, 105 transects (27.3%) were accumulative with an average velocity of 1.30 ± 0.08 m/yr; here, the NSM value was 4.86 ± 0.29 m.

In 2015, the Klaipeda seaport authorities started a nearshore nourishment project in ˙ front of the mainland coast (Figure 2). As a result, the additional sediments in the longshore sediment transport system led to milder coastal erosion on the mainland coast.

#### *4.3. Clusterization*

K-Means cluster analysis was used to group the transects to identify stretches of shoreline with similar development tendencies. Net Shoreline Movement (NSM) values over the study period were grouped into five clusters (Figure 14). The NSM and SCE values and results of the cluster analysis distinguish different processes in different stretches of the Curonian Spit and the mainland coast and reflection of the influence of Klaipeda seaport ˙ piers on the morpho-lytodynamic processes of the coast.

**Figure 14.** Graph showing the distribution of shoreline change envelope (SCE) (gray line) and net shoreline movement (NSM) (black line) along the study area for 1984–2019, and five clusters: cluster No. 1 (CL1), cluster No. 2 (CL2), cluster No. 3 (CL3), cluster No. 4 (CL4), cluster No. 5 (CL5).

The SCE corresponds closely with the NSM, implying that progressive and continuous change is more common than cyclical or reverse behavior in the spatial pattern of shoreline variability along the Curonian Spit. This stretch of coast connects Clusters No. 2 and No. 5, where the shoreline shifted towards the sea at an average of 38.93 ± 1.53 m and 27.66 ± 2.17 m, respectively (Table 1). Both clusters indicate accumulation processes on the coast. In cluster No. 2, the accumulation rate was 1.10 ± 0.04 m/yr., the SCE range was 65.14 m. In cluster No. 5, the shoreline moved towards the sea at an average velocity of 0.78 ± 0.06 m/yr. The SCE ranged between 38.01 m and 102.62 m (64.61 m). Moreover, on the coast of the Curonian Spit, Cluster No. 4 enters the southern port pier impact zone, which includes 27 transects (675 m long shoreline), where the shoreline may have different trends onshore dynamics at different times depending on hydrometeorological conditions. During the study period, the total change of the shoreline position in this cluster was positive and reached 20.74 ± 5.52 m, and the accumulation speed was 0.58 ± 0.16 m/yr. NSM values in this cluster ranged from −11.66 to 37.07 m.

The SCE closely corresponds with NSM along the mainland coast, except for the 445 and 547 transect section. The section of Cluster No. 1 is alternating, mainly due to anthropogenic activity, such as beach nourishment.

The majority (67.2%) of the mainland coast transects belong to cluster No. 1 (No. 2—3.1%, No. 3—29.7%). Four coast sections can be distinguished in this area, where the shoreline has different movement tendencies in the transects in the 675 m long section of the coast (from 415 to 442 tr.) North of the northern port jetty, erosion processes took place during the study period. The average erosion rate (EPR) was −0.64 ± 0.04 m/year, and the NSM value was −24.59 ± 1.31 m. The NSM range covered values from −4.19 m to −43.49 m, with a mean SCE of 56.74 ± 0.96 m. From 445 to 547 transects, the shoreline position changed at an average speed of 0.47 ± 0.01 m/year. The total NSM in transects was 16.67 ± 0.36 m. from −0.33 m to 47.25 m. SCE from 11.8 m to 47.25 m. In 2014–2018, by order of the Klaipeda ˙ seaport Authority, 237.78×10<sup>3</sup> <sup>m</sup><sup>3</sup> of sand was dumped on the coast near the beaches of Melnrage-Giruliai (Figure ˙ 2).

Another group of transects from 519 to 619 in Cluster No. 1 showed slightly negative shoreline position changes, in which the shoreline moved towards the mainland during the study period by −0.05 ± 0.01 m/yr., the mean NSM value was −1.93 ± 0.30 m. SCE ranges from 15.78 m to 26.37 m, NSM from −16.07 to 10.73 m. In the northern part of cluster No. 1, from 736 to 796 transects, changes in the shoreline influenced by erosive processes were recorded. Here the shoreline changed at an average velocity of −0.20 ± 0.02 m/yr. NSM was −7.15 ± 0.72 m (from −29.23 to 25.7 m), SCE covered an overall change of 23.83 ± 0.32 m and ranged from 12.82 m to 37.52 m.

Cluster No. 3 covers the central part of the mainland coast and indicates transects in which negative trends in shoreline dynamics have occurred during the study period. The shoreline of the 117 transects of this cluster moved towards the mainland at an average velocity of −0.64 ± 0.05 m/yr. The overall change in NSM was −22.70 ± 1.74 m.

This indicates the accretion processes in the Curonian Spit coast. The clusterization approach also suggests the accretion processes on the Curonian Spit coast with positive values of SCE and NSM (Table 3).



#### *4.4. Meteorological Data Analysis*

Changes in the wind direction are determined as the primary driver for sediment transport and drive coastal erosion [1,16,43,44]. The long-term wind direction and velocity at the studied area were analysed to indicate such changes.

The time series of yearly mean wind direction at Klaipeda is presented in Figure ˙ 15, and demonstrates changes in the regime of wind direction in the 1960–2019 period and suggests that at least two regime shifts have occurred during this period. The regime shift timings are found using a cut-off length of 10 years and Hubert's weight parameter of 1 [42]. This method detected that from 1960 till 1992, the wind direction on average was 216◦ (SW), then an average direction shifted to 188◦ (S), and the recent shift that occurred in 2011 was to 177◦ (S). The applied Rodionov regime shift method indicates that the average wind direction is shifting to the southern direction.

**Figure 15.** A shift in the annual average wind direction in Klaipeda in 1960–2019. ˙

The first observed regime shift in the mean values of wind direction occurred in 1992 (Figure 15). At this point, we observed that the wind direction shifted to the west–south direction. This change in the regime coincides with the changes in the shoreline that occurred when erosion was observed both on the Curonian Spit and on the mainland coast. Another detected regime shift occurred in 2011 with the same shift to the southern direction. During this period on the mainland coast, erosion processes were observed, and accumulation prevailed on the Curonian Spit coast.

The frequency distribution (Figure 16) of the predominant wind direction at Klaipeda ˙ in the 1960–2019 period determines that the predominant wind, up to1995, was 270◦ (W). The applied Rodionov shift detection method (Figure 15) confirms that in 1995 the predominant wind direction shifted to 209◦ (SSW).

**Figure 16.** Frequency of occurrence wind directions at Klaipeda in 1960–2019. ˙

#### **5. Discussion**

The Port of Klaipeda jetties location interrupts the natural longshore sediment trans- ˙ port path from the south to north at this point of the South-East Baltic Sea [6,23,45,46]. This should create favorable conditions for the two different processes: accumulation on the Curonian Spit south of the jetties and erosive—north of the jetties. Although the long-term analysis of shoreline changes in the whole study area indicates a total positive shoreline shift towards the sea, on the average velocity of 0.43 ± 0.03 m/yr, over the 35 years, the shoreline had different trends in both geomorphological and temporal changes. From the long-term perspective, the 10 km long Curonian Spit coast to the south of the southern Klaipeda seaport jetties is attributed to the accumulating coastal stretch. The mainland ˙ coast encompassing the northern part of the study site is affected by erosive processes.

The jettie's seaport systems on a straight sandy shore block the natural littoral drift [47,48], which determines the development of shoreline configurations. Typically, an up and down littoral drift is formed when hard breaking structures interrupt the predominant sediment transport direction. Due to the prevailing W and SW winds off the coast of Lithuania, sand transport is directed from south to north [49–52]. As a result, up-drift accretion occurs on the Curonian Spit coast on the south side of the jetties. Down-drift erosion occurs after losing its replenishment to maintain stability on the mainland coast (on the north side of the jetties).

The morphological changes of sandy beaches occur rapidly on a spatio-temporal scale as a response to natural (wind direction and speed, wave climate, sea-level fluctuations, etc.) processes [53]. Signs of climate change in the Baltic Sea can be more than just seawater level rise [54–56], increase in storminess [1], but also changes in the predominant wind and wave climate [43]. The climate change indicator in the wind regime is characterized as increasing in the wind velocity or intense wind events and changes in the predominant wind direction. This indicates changes in the cyclone patches over the Baltic Sea [57]. Changes in the wind direction and wave climate can alter longshore sediment transport magnitude and the dominant direction [58,59].

During this study, changes were observed in the predominant wind direction since 1992 (Figure 14), when the first regime shift occurred. The second shift in the wind direction regime was observed in 2012 (Figure 14). Significant changes in the predominant coastal evolution processes were observed after the wind direction shifts. Observed shifts of wind direction regime correspond with short-term changes of shoreline dynamics.

Shifts of wind direction regimes influenced intensified coastal erosion on both the Curonian Spit and the mainland coasts. In particular, the change in the wind direction regime influenced the short-term development of the Curonian Spit coast. In the periods of 1990–1995 and 2015–2019, the degree of erosion on this coast reached the respective levels of 4.57 ± 0.09 and 4.24 ± 0.12 m/year. The shoreline movement tendency of the 19th century was observed when the shoreline shifted towards the sea on both the Curonian Spit and the mainland coast [21]. This tendency reoccurred in the period of 2015–2019, on the usually accumulative Curonian Spit coast, which became erosive, while the average rate of erosion processes on the mainland coast decreased. In order to identify shoreline movement changes related to shifts in hydrometeorological conditions, a detailed investigation of wave climate (height, direction, period), sea-level fluctuations, and stormy events is required. Wave climate is driven by the wind climate [1,60] combined with the wind-driven coastal currents, and these are the major drivers for erosion and sedimentation, especially along the sandy sections of sandy beaches, dunes and soft moraine cliffs [2,61]. Future coastal process predictions are complicated as potential changes in the long-term mean and extreme wind speeds have a high uncertainty rate [1,62].

Moreover, significant changes in shoreline dynamics were observed in periods after the 2002 Klaipeda seaport reconstruction. Intensive erosion was observed on the mainland ˙ coast in the nearest proximity to the seaport jetties. Erosion after the reconstruction is acknowledged in other authors' [13,63] research. However, nowadays, as well as in the past, the main factor for the coastal erosion processes was attributed to dredging works in the Klaipeda seaport and especially in port jetty area [ ˙ 22,64].

Dredging works are carried out to maintain proper water levels in fairways, waterways, and ports. Work related to the extraction of bottom sediments includes various areas of activity related to their extraction, transportation, storage, cleaning, and practical use. Dredging works disturb the natural integrity of bottom sediments (benthos) and directly and indirectly impact all marine environment elements [65,66]. Sediments excavated from the Baltic Sea coast are stored in designated areas at sea or on land. Such sites are usually located near port areas for economic motives [65]. Current environmental trends encourage the recycling or practical use of excavated sediments. One of the essential practical advantages is the beach nourishment with extracted sand if it meets the established physical and chemical properties. Artificial sand nourishment can be used as a coastal erosion mitigating measure by adding sediments directly to the coast or supplementing the natural longshore sediment transport budget.

In 2014–2018, by order of the Klaipeda seaport Authority, 237.78 ˙ <sup>×</sup> <sup>10</sup><sup>3</sup> <sup>m</sup><sup>3</sup> of fine sand was dumped on the nearshore beaches of Melnrage-Giruliai at 4–6 m depth ( ˙ Figure 4). The extracted sediments from the Klaipeda strait were used to restore the mainland sediment ˙ budget and replenish the coast. Beach sand nourishment is a widely known method to widen and restore the subaerial beach and decrease coastal erosion [67–69]. The nourishment material redistribution is driven by local hydrodynamic conditions (waves and currents). The predominant longshore current is directed from south to north along the Lithuanian coast [49,51]. Therefore, to mitigate the disrupted natural sediment transport by Klaipeda seaport jetties, the sediment dumping areas are located north of Klaip ˙ eda seaport ˙ jetties (Figure 4). The grain size distribution of the sand is dominated by grains with a size of 0.1–0.25 mm, representing 70–96% of grains with an Md between 0.14 mm and 0.22 mm, which corresponds precisely to the composition of the beach sand. Such sand composition detected on the mainland coast indicates that the nourishment material is transported in a predominant longshore direction and significantly influences cross-shore profile evolution.

Understanding the short- and long-term variability of the shoreline changes could help design shore nourishment in such a way that anthropogenic activity would be carried out in coherence with natural processes rather than in conflict [70,71]. Usually, shoreline change rates are best suited for the quasi-linear trend analysis. However, values of the shoreline variation are often non-linear and have different trend reversals. It is possible to single out the behaviors of certain groups that have the same or similar tendencies of change when using a joint shoreline change rates trend and cluster-based segmentation analysis.

According to K-means clustering of long-term changes in five different short-term periods in 796 transects, 369 transects covering clusters No. 2 and No. 5 are essentially distributed at the Curonian Spit and indicate accumulation processes. The positive dynamic characteristics of this coastal stretch are essentially in line with the multi-year shoreline changes in this coast type. Moreover, they reflect the main geomorphological and sedimentary conditions of the Curonian Spit.

The Klaipeda seaport impact zone was reflected in clusters No. 1 and No. 5. Still, ˙ cluster No. 1 identifies significant anthropogenic activities or impacts on the mainland coastal stretch due to shore replenishment. At the same time, on the mainland coast further from the direct port jetties impact area [20,28], Cluster No. 3 shows the presence of other factors with a more significant impact on the shoreline evolution. The trend in the SCE indicator also distinguishes the accumulative stretch of shore from 445 to 550 transects, which proves the impact of damping of the dredged sand from the Klaipeda strait. ˙

#### **6. Conclusions**

Forecasting and continuous estimation of the intensity of the sandy South-Eastern Baltic Sea coast dynamics are essential to customizing coastal development management methods and techniques that affect the nature and economics of the coastal environment. The analysis of long- and short-term shoreline changes should provide the required knowledge for reducing the extent of the anthropogenic intervention factors into the natural coastal system with long-lasting consequences.

This study aims to qualitatively and quantitatively identify the sandy South-Eastern Baltic Sea coast shoreline evolution tendencies. The reconstruction of Klaipeda jetties ˙ disrupted the settled equilibrium stage, interrupted the longshore sediment transport, and activated erosion processes. As a result, in the long-term (1984–2019) perspective, the northern part of the coast became abrasive, eroded coast length increased three times, from 1.5 to 4.2 km.

Assessment of short-term shoreline changes combined with K-means cluster analysis has helped identify the direct impact zone of the Port of Klaipeda. In this study, short- ˙ term shoreline changes correspond with shifts in wind direction and reflect the effect of the dredging works in the Klaipeda strait. The research helped identify the part of ˙ the mainland coast (transects from 445 to 550) that acquires other dynamic properties of the shore—accumulation. Although according to the hydrometeorological and lithogeomorphological characteristics and the impact of the port, erosion processes should prevail. It occurs due to coastal zone nourishment works. Therefore, this site needs continuous research because it is sensitive to anthropogenic and meteorological conditions. It also requires regular monitoring of the coast nourishment, as the development of coastal infrastructure, coastal use for recreational purposes, and planning of coastal protection measures depend on it.

**Author Contributions:** Conceptualization, V.K.; methodology, V.K., L.K.-R. and I.Š.; software, V.K.; validation, V.K., I.Š. and E.B.; formal analysis, V.K.; investigation, I.Š.; data curation, V.K., E.B. and I.Š.; writing—original draft preparation, V.K.; writing—review and editing, I.Š. and E.B.; visualization, V.K. and I.Š.; supervision, L.K.-R. All authors have read and agreed to the published version of the manuscript.

**Funding:** The APC was funded by Lithuanian science foundation project "Development of doctoral studies" Nr. 09.3.3-ESFA-V-711-01-0001. Kelpšaite-Rimkien ˙ e was also supported by the Baltic ˙ Research Programme (EEA Financial Mechanisms 2014–2021) project "Solutions to current and future problems on natural and constructed shorelines, eastern Baltic Sea" (EMP480).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We would like to thank the Klaipeda State Seaport Authority for supporting ˙ this research and providing data.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **The Making of a Gravel Beach (Cavo, Elba Island, Italy)**

**Irene Cinelli 1,2, Giorgio Anfuso <sup>3</sup> , Enrico Bartoletti 2,4, Lorenzo Rossi 2,5 and Enzo Pranzini 1,3,\***


**Abstract:** This paper presents the history and evolution of the different projects carried out from 1999 to 2008 at Cavo beach in the Elba Island, Italy. The village of Cavo almost completely lost its beach in the 1970s due to the reduction of sedimentary input, and the backing coastal road was defended by a revetment and two detached breakwaters. Such severe erosion processes continued in the following years and impeded any possibility of beach tourist development. In 1999, a project based on the removal of existing breakwaters and beach nourishment works based on the use of gravel as borrow sediment and the construction of two short groins to maintain nourished sediment, raised environmental concern and did not find the approval of the stakeholders. They were worried about the characteristics of the sediments, i.e., waste materials from iron mining rich in red silt and clay. Such sediment fractions made the sea red during the nourishment and deposited on the *Posidonia oceanica* meadow in front of the beach, with a potential environmental impact. Furthermore, they cemented the gravel fraction forming a beach rock. Between 2006 and 2008, these materials were covered with better quality gravel, extending and raising the beach profile, which required the elevation and lengthening of the two existing groins. Beach evolution monitoring following the second project, based on morphological and sedimentological data acquired before, during and after the works, demonstrated the great stability of the newly created beach. The wider beach has allowed the construction of a promenade and the positioning, in summer, of small structures useful for seaside tourism, increasing the appeal of this village. Data presented in this paper shows an interesting study case, since few examples exist in international literature regarding gravel nourishment projects monitoring and evolution.

**Keywords:** beach nourishment; coastal erosion; gravel beaches; sediment budget; shore protection structures

### **1. Introduction**

In small islands, pocket beaches quite often represent one of the most important tourist assets [1], and this is even more true for those in the Mediterranean Sea [2,3], where a strong transition from traditional activities (agriculture and fishing) to tertiary activities (almost exclusively tourism) occurred in the 20th century [4,5]. Furthermore, in small islands, pocket beaches with a limited sediment stock are extremely vulnerable to sedimentary input reduction [6], which is frequently produced by the abandonment of cultivated lands [7]. In those sites where attractive landscape values support beach tourism, shore protection projects based on the emplacement of hard structures should be limited and artificial nourishment preferred, at best associated with small containment structures. This is the most sustainable option to preserve the natural scenic beach value and all beach-related activities [8–10].

**Citation:** Cinelli, I.; Anfuso, G.; Bartoletti, E.; Rossi, L.; Pranzini, E. The Making of a Gravel Beach (Cavo, Elba Island, Italy). *J. Mar. Sci. Eng.* **2021**, *9*, 1148. https://doi.org/ 10.3390/jmse9101148

Academic Editor: Carlos Daniel Borges Coelho

Received: 15 September 2021 Accepted: 11 October 2021 Published: 19 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Beach nourishment projects on small islands are quite complex because their limited surface and complex orography often prevent finding suitable natural borrow materials in land deposits, and the importing of sediments from the mainland is extremely expensive [11]. In addition, environmental constraints do not allow shelf sediment dredging and, in any case, mob-demob cost for deep operating dredges is not justified by the small volumes of sediments usually required. Riverbed quarrying, which however is forbidden in Italy, reduces sediment input to the coast and cannot be a solution to contrast coastal erosion. Beach nourishment projects on small islands are quite complex because their limited surface and complex orography often prevent finding suitable natural borrow materials in land deposits, and the importing of sediments from the mainland is extremely expensive [11]. In addition, environmental constraints do not allow shelf sediment dredging and, in any case, mob-demob cost for deep operating dredges is not justified by the small volumes of sediments usually required. Riverbed quarrying, which however is forbidden in Italy, reduces sediment input to the coast and cannot be a solution to contrast coastal erosion.

This is the most sustainable option to preserve the natural scenic beach value and all

Therefore, one option is rock crushing to produce gravel to create coarse sediment beaches, whose pros are: Therefore, one option is rock crushing to produce gravel to create coarse sediment beaches, whose pros are:

Higher stability than sand beaches [12]; Higher dry beach expansion at a given fill volume [13]; Clearer water, since there are no fine sediments that can be suspended [14]; No wind erosion [15]; No sticking on the beachgoers' skin [16]. Higher stability than sand beaches [12]; Higher dry beach expansion at a given fill volume [13]; Clearer water, since there are no fine sediments that can be suspended [14]; No wind erosion [15]; No sticking on the beachgoers' skin [16].

However, some cons must be considered, e.g.,: However, some cons must be considered, e.g.,:

*J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 2 of 3

beach-related activities [8–10].

Less ease in walking and lying on the beach and entering the sea [17]; Steeper swash zone, which is an obstacle for elderly, children and disabled people [17]; Reduced play possibilities for children and limitations to beach games and sports [18,19]. Less ease in walking and lying on the beach and entering the sea [17]; Steeper swash zone, which is an obstacle for elderly, children and disabled people [17];Reduced play possibilities for children and limitations to beach games and sports [18,19].

In Italy, gravel and grains have been used both to build a new beach where it was completely lost, e.g., at Cala Gonone [20], Marina di Pisa [21], or to expand an eroding sand beach, e.g., at Massa [22] and Terracina [23]. The case considered in this paper refers to a pocket beach in front of a small village named Cavo, in the eastern coast of the Elba Island, in the Thyrrhenian Sea, Italy (Figure 1). In Italy, gravel and grains have been used both to build a new beach where it was completely lost, e.g., at Cala Gonone [20], Marina di Pisa [21], or to expand an eroding sand beach, e.g., at Massa [22] and Terracina [23]. The case considered in this paper refers to a pocket beach in front of a small village named Cavo, in the eastern coast of the Elba Island, in the Thyrrhenian Sea, Italy (Figure 1).

**Figure 1.** Location map of central-northern part of Italy (**a**) and Elba Island (**b**) (Google Earth). **Figure 1.** Location map of central-northern part of Italy (**a**) and Elba Island (**b**) (Google Earth).

At such beach, in the 1970s, sand was almost completely lost, and a revetment was positioned adjacently to the coastal road and two detached breakwaters were constructed, thus preventing access to the sea and creating very dangerous nearshore conditions (Figure 2). After an attempt to restore the original mixed sediment sand and gravel beach carried out in 1999 (Figure 3), which created a hard surface disliked by tourists, and environmental concern for the water turbidity on the *Posidonia oceanica* meadow, a pure gravel beach was constructed in 2006–2008 (Figure 3) with the satisfaction of beachgoers and local stakeholders. At such beach, in the 1970s, sand was almost completely lost, and a revetment was positioned adjacently to the coastal road and two detached breakwaters were constructed, thus preventing access to the sea and creating very dangerous nearshore conditions (Figure 2). After an attempt to restore the original mixed sediment sand and gravel beach carried out in 1999 (Figure 3), which created a hard surface disliked by tourists, and environmental concern for the water turbidity on the *Posidonia oceanica* meadow, a pure gravel beach was constructed in 2006–2008 (Figure 3) with the satisfaction of beachgoers and local stakeholders.

*J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 3 of 17

**Figure 2.** The central part of the beach at Cavo in June 1995. **Figure 2.** The central part of the beach at Cavo in June 1995. **Figure 2.** The central part of the beach at Cavo in June 1995.

years after the end of the second nourishment). **Figure 3.** Shoreline (zero isobath) evolution at Cavo from 1997 (before the first nourishment) to 2018 (ten years after the end of the second nourishment). **Figure 3.** Shoreline (zero isobath) evolution at Cavo from 1997 (before the first nourishment) to 2018 (ten years after the end of the second nourishment). In red sectors number.

**Figure 3.** Shoreline (zero isobath) evolution at Cavo from 1997 (before the first nourishment) to 2018 (ten

The artificial fill was not accompanied by a specific monitoring, as frequently happens for coastal defence projects in Italy. However, for this case, a three-year monitoring was commissioned to the University of Florence (Italy) by the works contracting entity (the Provincia di Livorno, i.e., the provincial administration), and results of such studies appeared only in grey literature and essentially concerned administrative and project management issues. Ten years later, a survey of Cavo beach was commissioned by the Regione Toscana (i.e., the regional administration) within the framework of a wide regional project focused on the monitoring of all regional beaches. Despite the nonhomogeneous temporal spacing of acquired data regarding Cavo beach evolution, this paper describes the different coastal projects actuations and assesses the effectiveness of the nourishment project carried out in 2006–2008, giving only some indications on a previous project performed in 1999 for which data are extremely poor. Results, which demonstrate the fill stability and the approval of the type of material by beachgoers, are of interest to coastal planners and can be useful for designing new gravel beaches or to expand eroding ones in similar environments.

#### **2. The Beach at Cavo (Elba Island, Italy)**

The beach of Cavo (Figure 1) is located inside a bay facing NNE (fetch width from 32◦ N to 107◦ N) sheltered by the Central Tuscany coast. Fetch length at the extremes of the angle is 6.3 and 17.0 nautical miles (nmi), respectively (13.4 nmi along the bisector of the opening angle 69.5◦ N). No physical or virtual wave gouges are present in this sea sector, and data on wave climate can be obtained from the "Wind and Wave Atlas of the Mediterranean" [24], where the nearest point is 22 nmi south of Cavo at 25 nmi from the coast, thus has a longer fetch than the real one observed at Cavo. At such point, waves approach from the 30–120◦ N sector and significant wave height (Hs) > 2.0 m approach from 30◦ N and represent 0.2% of records. Tidal range is 36 cm at the Livorno Gouge, on the continental coast [25].

### *2.1. The Loss of the Beach*

Cavo is a little village located on the eastern side of Elba Island (Figure 1) where, since Roman times, the main traditional activity has been iron mining, flanked by some agriculture consisting mainly of vine cultivation, the main occupation in the rest of the Island.

After WWII, all the island recorded a transformation of the economic activity, from agriculture to tourism, but such a shift took place a bit later on the eastern side, since mining activity lasted, although with reduced production, until 1981 [26]. All the beaches at Elba Island are eroding because sediment input was reduced when the crops were abandoned, and the forest grew [27–29]. On the eastern side, additional sediment input to beaches was linked to mining activity because waste materials from excavations were abandoned on the slopes of mountains and thus easily transported by run-off processes to the coast [30]. Therefore, the beach of Cavo and others on the eastern coast, formed thanks to land erosion of cultivated areas and erosion of quarry waste deposits, but when both activities were interrupted, coastal retreat also interested this part of the island. In addition, a small pier with impermeable root was created on the southern side of the bay as a docking structure for ferries connecting the island to the continent and a small marina added at its northern side. These structures interrupted the limited longshore sediment transport and contributed to the urban beach disappearance, a loss only apparently compensated by the expansion of the beach placed updrift, the latter being in a marginal area of lesser tourist value.

At the same time, this village also started to look for new opportunities in 3S (Sun, Sea and Sand) tourism, with few hotels and some second houses, but the beach was almost inexistent and could not support this activity.

#### *2.2. The Making of the Gravel Beach*

At the end of the 1990s, the beach in front of the village of Cavo was constituted by only two narrow strips of sand, one close to the northern headland and one leaned on the marina downcoast dock. In addition, the coastal road was defended by two detached breakwaters and by a revetment (before 1954) making difficult and extremely dangerous the use the beach for bathing activity (Figure 2). *J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 5 of 17

> The wide beach, which is now in front of the village, is the result of a complex and controversial story that presents technical and legal aspects, both of which were important for the achievement of the design solution that led to the current configuration of the coastline. This story can be synthetized in two projects: the first carried out in 1999 and the second between 2006 and 2008; the latter being the main topic of this paper. At the end of the 1990s, the beach in front of the village of Cavo was constituted by only two narrow strips of sand, one close to the northern headland and one leaned on the marina downcoast dock. In addition, the coastal road was defended by two detached breakwaters and by a revetment (before 1954) making difficult and extremely dangerous

#### 2.2.1. The First Project (1999) the use the beach for bathing activity (Figure 2). The wide beach, which is now in front of the village, is the result of a complex and

In 1999 a project was carried out to create a 10-meter-large beach in front of the street wall. The previous detached breakwaters and the revetment were removed, and two short groins constructed to divide in three parts the 497 m long coastal sector delimited by the headland to the north and the marina to the south (Figure 3). controversial story that presents technical and legal aspects, both of which were important for the achievement of the design solution that led to the current configuration of the coastline. This story can be synthetized in two projects: the first carried out in 1999 and the second between 2006 and 2008; the latter being the main topic of this paper.

Due to the lack of suitable natural aggregate deposits on the Island, wastes of the old iron mines were used as borrow material. They consisted of unsorted sand and gravel, with a high percentage of fines (silt and clay = 13%) formed by yellow-red iron oxides [29]. The presence of heavy minerals in excess respect to environmental regulations was later assessed as well. 2.2.1. The First Project (1999) In 1999 a project was carried out to create a 10-meter-large beach in front of the street wall. The previous detached breakwaters and the revetment were removed, and two short groins constructed to divide in three parts the 497 m long coastal sector delimited by the headland to the north and the marina to the south (Figure 3). Due to the lack of suitable natural aggregate deposits on the Island, wastes of the old

The nourishment was carried out just before the tourist season, but results were not those expected: the sea water acquired a red colour that persisted during all the summer and several tourists cancelled their hotel reservations. The *Posidonia oceanica* meadow, present in the nearshore, was covered by a thin layer of clay and the risk of some permanent ravages was raised. Local stakeholders claimed environmental and economic damages, and the case arrived at the court. Monitoring of the Posidonia proved that no long−lasting injury was done, and economic loss not motivated. As far as heavy minerals are concerned, further analyses showed that all the beaches present along the eastern side of the Island have similar concentrations, mostly deriving from more than two thousand years of mining activity. iron mines were used as borrow material. They consisted of unsorted sand and gravel, with a high percentage of fines (silt and clay= 13%) formed by yellow-red iron oxides [29]. The presence of heavy minerals in excess respect to environmental regulations was later assessed as well. The nourishment was carried out just before the tourist season, but results were not those expected: the sea water acquired a red colour that persisted during all the summer and several tourists cancelled their hotel reservations. The *Posidonia oceanica* meadow, present in the nearshore, was covered by a thin layer of clay and the risk of some permanent ravages was raised. Local stakeholders claimed environmental and economic damages, and the case arrived at the court. Monitoring of the Posidonia proved that no long−lasting injury was done, and economic loss not motivated. As far as heavy minerals are concerned, further analyses showed that all the beaches present along the eastern side of the

In the months after the nourishment, the borrow material compacted and became impermeable because something similar to a beach rock was formed (Figure 4) so that, during the following winter storms, run up water was not able to infiltrate into beach sediments and reached the coastal road [31]. Permeability measurements were performed by the University of Florence in three points of the beach (north, centre, south), giving a permeability coefficient (Ks) between 5.0 <sup>×</sup> <sup>10</sup>−<sup>6</sup> and 1.3 <sup>×</sup> <sup>10</sup>−<sup>7</sup> m/s (typical of silty to silty-clay sediments). Island have similar concentrations, mostly deriving from more than two thousand years of mining activity. In the months after the nourishment, the borrow material compacted and became impermeable because something similar to a beach rock was formed (Figure 4) so that, during the following winter storms, run up water was not able to infiltrate into beach sediments and reached the coastal road [31]. Permeability measurements were performed by the University of Florence in three points of the beach (north, centre, south), giving a permeability coefficient (Ks) between 5.0 × 10−6 and 1.3 × 10−7 m/s (typical of silty to siltyclay sediments).

**Figure 4.** The beach at Cavo in January 2002, after the nourishment performed with iron rich materials with puddles showing the low permeability of the sediments (**a**); outcropping of the consolidated and impermeable fill material layer (**b**).

#### 2.2.2. The Second Project (2006–2008) was considered but the risk of favouring further fines and heavy minerals offshore dispersion discouraged such a solution. This led to the design a new gravel beach, large and

2.2.2. The Second Project (2006–2008)

In the following years, the eventuality of removing all the material still present ashore was considered but the risk of favouring further fines and heavy minerals offshore dispersion discouraged such a solution. This led to the design a new gravel beach, large and high enough to allow run-up water percolation even during extreme storms. This solution forced a modification of the groin configuration by elevating the crest and extending it offshore to host a higher and wider berm (Figure 3). Nourishment comprised approx. 30,000 m<sup>3</sup> (*ca.* 80 m3/m for the project sector) of gravel 4.0–4.5 phi (16–24 mm) in mean size (Figure 5). To not increase water turbidity, local authorities asked to maintain the quantity of fines (<0.063 mm) lower than 2%, as occurred for many projects carried out in Italy in sensible sites (e.g., at Cala Gonone, Sardinia [20]). high enough to allow run-up water percolation even during extreme storms. This solution forced a modification of the groin configuration by elevating the crest and extending it offshore to host a higher and wider berm (Figure 3). Nourishment comprised approx. 30,000 m3 (*ca.* 80 m3/m for the project sector) of gravel 4.0–4.5 phi (16–24 mm) in mean size (Figure 5). To not increase water turbidity, local authorities asked to maintain the quantity of fines (<0.063 mm) lower than 2%, as occurred for many projects carried out in Italy in sensible sites (e.g., at Cala Gonone, Sardinia [20]). As in the 1999 project, aggregates were transported by truck and downloaded directly on the beach and later distributed with a bulldozer, pushing them even into the water.

In the following years, the eventuality of removing all the material still present ashore

*J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 6 of 17

**Figure 4.** The beach at Cavo in January 2002, after the nourishment performed with iron rich materials with puddles showing

the low permeability of the sediments (**a**); outcropping of the consolidated and impermeable fill material layer (**b**).

**Figure 5.** Fill materials used at Cavo: on the left side the iron rich mining waste, on the right side the new carbonate sandstone of the second nourishment. One euro coin for reference at the centre of the photo. Work initiated in January 2006 with the groins modification and ended in May 2008, **Figure 5.** Fill materials used at Cavo: on the left side the iron rich mining waste, on the right side the new carbonate sandstone of the second nourishment. One euro coin for reference at the centre of the photo.

with interruptions during the summer tourism seasons. Circa 25,000 m3 of gravel were deposited. In addition, a small (unknown) volume of very fine sand dredged at the har-As in the 1999 project, aggregates were transported by truck and downloaded directly on the beach and later distributed with a bulldozer, pushing them even into the water.

bour entrance was discharged in the sectors near the marina, but being a volume moved inside the area, it did not change the overall sedimentary budget. Dry beach expansion, on the 497−meter−long coast, was 3777 m2. Mean shoreline progradation was 7.48 m, with important differences from the southern and central sectors (10 to 14 m) and the northern one (less than 3 m) where the original beach was not nour-Work initiated in January 2006 with the groins modification and ended in May 2008, with interruptions during the summer tourism seasons. Circa 25,000 m<sup>3</sup> of gravel were deposited. In addition, a small (unknown) volume of very fine sand dredged at the harbour entrance was discharged in the sectors near the marina, but being a volume moved inside the area, it did not change the overall sedimentary budget.

ished with iron-rich materials and did not necessitate additional protection. In spring 2008, a volume of 3,000 m3 was placed to complete the project [31]. Part of the gravel was deposited in front of the swash zone, as feeding groins [16] that theoretically allow to have a more natural beach profile since wave action should move grains onshore, after a first phase in which grains are in situ cleaned and rounded. Dry beach expansion, on the 497-meter-long coast, was 3777 m<sup>2</sup> . Mean shoreline progradation was 7.48 m, with important differences from the southern and central sectors (10 to 14 m) and the northern one (less than 3 m) where the original beach was not nourished with iron-rich materials and did not necessitate additional protection. In spring 2008, a volume of 3000 m<sup>3</sup> was placed to complete the project [31].

An extensive beach scraping was performed in March 2009 to redistribute sediments accumulated in December 2008 in front of the promenade wall during a severe storm, but similar works are frequently carried out at the beginning of the summer season to flatten Part of the gravel was deposited in front of the swash zone, as feeding groins [16] that theoretically allow to have a more natural beach profile since wave action should move grains onshore, after a first phase in which grains are in situ cleaned and rounded.

An extensive beach scraping was performed in March 2009 to redistribute sediments accumulated in December 2008 in front of the promenade wall during a severe storm, but similar works are frequently carried out at the beginning of the summer season to flatten the storm berm crest that constitutes an obstacle to dive for children and elderly or disabled people. This activity does not influence the fill stability assessment, since beach volume is not modified, and the profile soon adapts to the autumn–winter storms.

The overall beach expansion and stabilization allowed to provide the coastal road with a large sidewalk with a balustrade and trees, which transformed it into a promenade. In this way, part of the fill material is delimitated by the promenade wall and is no more part of the beach sediment stock. This partially explains why the volume of sediments added resulting from the comparison of the two first surveys (pre- and post-works) is lower than what was actually deposited on the beach. At the root of the two groins, where the beach was expected to be wider and stable, the promenade was expanded into semi-circular exedras (Figure 6). All this transformed the coastal landscape and triggered a revival of the tourist activity of this location. ume is not modified, and the profile soon adapts to the autumn–winter storms. The overall beach expansion and stabilization allowed to provide the coastal road with a large sidewalk with a balustrade and trees, which transformed it into a promenade. In this way, part of the fill material is delimitated by the promenade wall and is no more part of the beach sediment stock. This partially explains why the volume of sediments added resulting from the comparison of the two first surveys (pre- and post-works) is lower than what was actually deposited on the beach. At the root of the two groins, where the beach was expected to be wider and stable, the promenade was expanded into semicircular exedras (Figure 6). All this transformed the coastal landscape and triggered a revival of the tourist activity of this location.

*J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 7 of 17

the storm berm crest that constitutes an obstacle to dive for children and elderly or disabled people. This activity does not influence the fill stability assessment, since beach vol-

**Figure 6.** One of the two exedras constructed along the promenade (Google Earth). **Figure 6.** One of the two exedras constructed along the promenade (Google Earth).

#### **3. Materials and Methods 3. Materials and Methods**

As explained in the Introduction, different topographic, bathymetric and sedimentological surveys (not homogeneously spaced in time, Table 1) are available to assess the evolution of nourishment projects realised in 1999 and in 2006–2008 at Cavo. They were performed within (i) a scientific research agreement between the Province of Livorno and the Earth Science Department of the University of Florence and (ii) the surveys commissioned in 2018 by the Regione Toscana to monitor several beaches at Elba Island, including the one at Cavo. As explained in the Introduction, different topographic, bathymetric and sedimentological surveys (not homogeneously spaced in time, Table 1) are available to assess the evolution of nourishment projects realised in 1999 and in 2006–2008 at Cavo. They were performed within (i) a scientific research agreement between the Province of Livorno and the Earth Science Department of the University of Florence and (ii) the surveys commissioned in 2018 by the Regione Toscana to monitor several beaches at Elba Island, including the one at Cavo.

Shoreline position was acquired by means of GPS surveys (LEICA system RX 900,

Leica Geosystems, Heerbrugg, Swiss), in June 1997 and RTK-GPS (GPS NRTK 1250, Leica **Table 1.** Available data concerning Cavo beach area.


Shoreline position was acquired by means of GPS surveys (LEICA system RX 900, Leica Geosystems, Heerbrugg, Swiss), in June 1997 and RTK-GPS (GPS NRTK 1250, Leica

Geosystems, Heerbrugg, Swiss) in the following surveys: June 1997, March 2002, July 2006, June 2007, April 2008, May 2009 and June 2018 (Figure 3, Table 1). Bathymetric surveys, consisting of 34 single beam (Hydrotrac, Teledyne Odom, Slangerup, Denmark) profiles each, were performed in March 2002, June 2007 and May 2009, whereas a multibeam survey (Seabat 7k, Teledyne Reason, Slangerup, Denmark) was performed in 2018. Together with the 2002, 2007 and 2009 surveys, sediment samples (N = 91, 42 and 35) were collected with a Van Veen grab along 6 profiles from +2 m to −5 m, and grain size analysis via dry sieving was performed to obtain Folk and Ward (1957) textural parameters [32].

Shoreline evolution was analysed dividing the coast in 9 sectors, each approximately 55 m long, 3 in each cell in which the beach is divided by the two groins. Mean shoreline displacement was computed for each sector (Table 2) using the Surface Based Analysis (SBA), since the traditional Profile Based Analysis (PBA) was not considered reliable due to the nonlinear shape of the shoreline given by the groins [33]. Surface measurements were performed with QGIS Rel. 3.0.

Beach morphology evolution was studied by comparing pairs of surveys with Surfer Rel. 14 and producing vertical changes maps for the time intervals 2002–2007, 2007–2009 and 2009–2018. Although surveys were performed with standard calibrations (check bar, tide and draft, pitch and roll) and linked to geodetic points, further corrections were done on sea-true points located on some rocky shoals at the border of the bay. Despite all the above, the accuracy of vertical changes in bathymetric maps was approximately 20 cm [34] and, therefore, such maps were only used for a semi-quantitative assessment. Sedimentological maps were drawn to represent Mean size (Mz) and Sorting (σ<sup>I</sup> ) parameters, and a Mean size *vs* Depth graph was plotted.

No measured wave data are available for this bay, and reference can be done only with two buoys operated by the Tuscany Region, one near Gorgona Island, 75 km NNE of Cavo, the other near Giannutri Island, 90 km SE of Cavo, but considering that this coast is exposed to the East, towards the nearby continent (approx. 10 km), wave energy is significantly lower. However, after the last survey, on 31 December 2018a storm with significant wave height (Hs) of 5.40 m at Gorgona and 6.50 m at Giannutri was registered, the highest since the buoys were installed (2008).

#### **4. Results**

#### *4.1. General Considerations on the First Nourishment Evolution*

As previously stated, no specific monitoring was performed on this project, and beach transformation could be evaluated only by the comparison between two surveys done by the University of Florence in 1997 (two years before the fill), within a regional study on Elba Island beaches erosion, and one in 2002 as a basis of a second project. A further survey, performed in June 2006 (before the second nourishment), allows to assess fill evolution in the following four years.

The first nourishment induced a notable dry beach expansion, with the March 2002 beach 9.8 m wider on average respect to the June 1997 one (Figures 3 and 7; Table 2).

Obviously, fines were lost only from the upper part of the deposits, which strongly modified optical properties of the nearshore water that acquired a deep red colour induced by the fact that the grainsize fraction was mostly composed by clay (and not silt). Actually, a very limited fill volume was dispersed.

From March 2002 to July 2006 the beach was as a whole stable, with only some sediment shift from the side sectors to the central ones (no. 4 to no. 7; Figure 3; Table 1), favoured by the low groins height.

**Figure 7.** Location of sectors (**a**) and sectors and total mean beach width evolution (**b**) from 1997 to 2018. First nourishment 1999; second nourishment 2006–2008. **Figure 7.** Location of sectors (**a**) and sectors and total mean beach width evolution (**b**) from 1997 to 2018. First nourishment 1999; second nourishment 2006–2008.


**Table 2.** Mean shoreline displacement (m) in the different periods from June 1997 to June 2018 in the 9 considered beach sectors and in the entire beach. **Table 2.** Mean shoreline displacement (m) in the different periods from June 1997 to June 2018 in the 9 considered beach sectors and in the entire beach.

#### modified optical properties of the nearshore water that acquired a deep red colour induced by the fact that the grainsize fraction was mostly composed by clay (and not silt). *4.2. Second Nourishment Evolution Assessment* Shoreline Displacement

Actually, a very limited fill volume was dispersed. From March 2002 to July 2006 the beach was as a whole stable, with only some sediment shift from the side sectors to the central ones (no. 4 to no. 7; Figure 3; Table 1), favoured by the low groins height. The survey performed in June 2007 shows the beach as it was immediately after the end of the main works (i.e., without the 3000 m<sup>3</sup> of the 2008 additional filling) and its comparison with the July 2006 allows to quantify the enlargement artificially obtained.

*4.2. Second Nourishment Evolution Assessment*  Shoreline Displacement This beach nourishment was not homogeneous along the coast but concentrated where the beach was narrower. The different dry beach expansion is visible in Figures 3 and 7 and in Table 2: external sectors 1, 2 and 9 received a limited or null volume of sediments, whereas the central ones obtained more (Table 2). Such distribution of nourished sediments was essentially aimed to cover the previously deposited iron-rich material (1999 nourishment), more than to expand the beach.

Mean shoreline progradation was 7.48 m, ranging from 2.24 (sector 1) to 13.39 m (sector 7). Adding these values to what was obtained with the 1999 nourishment, the beach was 18.42 m wider on average than it was in 1997. However, part of this increment will be later used to enlarge the coastal road adding a wide footstep and two exedras.

From June 2007 to April 2008, i.e., in the first winter after the nourishment, all the sectors of the beach recorded moderate erosion (Figure 3): overall beach surface reduction was approximately 1189 m<sup>2</sup> , with a mean beach retreat of 2.26 m.

Concerning beach morphology, the beach profile constructed during the nourishment works was flat and wide; in the following months, after the impacts of winter waves, storm berm crests were observed. This involved the migration of material from the swash zone to a more internal position, with a reduction of the beach surface, but without changes in gravel volume, as proved by elevation change maps (Figure 8). *J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 11 of 17

**Figure 8.** Beach vertical changes (m) at Cavo considering the 2002−2007 (**a**) and 2007–2018 (**b**) surveys. **Figure 8.** Beach vertical changes (m) at Cavo considering the 2002−2007 (**a**) and 2007–2018 (**b**) surveys.

*4.2. Nearshore Morphology Evolution*  To assess the morphological evolution of the beach, and the sediment budget of the emerged and the submerged parts, 3D topo-bathymetric models produced in each survey In the following year (April 2008–May 2009) advancement is recorded in all the sectors with an average dry beach expansion of 3.42 m, resulting in a beach wider than the one obtained immediately after the works.

were compared limiting the offshore part to that in which full overlapping was possible, i.e., down to ca. 5 m water depth (Figure 8). From May 2002 to June 2007 the direct effect of the nourishment is visible with beach When comparing the evolution of the different sectors (Figure 3), a very homogeneous behaviour is observed and characterised by synchronous accretion and erosion in the period 2006–2009, showing that no significant beach rotation occurred in each cell.

surface rising both on the dry beach and on the submerged profile, especially in the central cell, where most of the volume was deposited. At the centre of this cell a lobe is evident,

where the main feeding groin was positioned (Figure 9).

It is only in the nine following years (May 2009–June 2018) that the three cells show a slightly different behaviour. In the northern one (sectors 1–3), sector no. 1 accretes more than the others, which results in a slight clockwise rotation, although within a general beach progradation. On the contrary, the central cell (sectors 4–6) behaves in an opposite way, with the first sectors eroding and the latter accreting, with an anticlockwise rotation and limited beach erosion. In the southern cell (sectors 7–9) a widespread but limited beach accretion is measured, especially at the central segment (no. 8, Figure 3).

All those mentioned changes represent small variations (between +4 m and −2 m) for a long period, with negative values in the central cell only, which is the most exposed to the incoming waves. All these data show that waves strong enough to move these coarse sediments approach the coast almost orthogonally.

Unfortunately, no high temporal resolution data are available for these nine years, but several inspections and photographs show that the beach was almost stable, so much that the municipality built public toilets at the base of the promenade and gave some surfaces in concession to carry out commercial activities (e.g., beach bars).

A stable and attractive beach was the goal of the project, and collected data, although incomplete, show that this was achieved. Its expansion, necessary to cover the nourishment of 1999, was a welcome side effect of the project.

#### *4.3. Nearshore Morphology Evolution*

To assess the morphological evolution of the beach, and the sediment budget of the emerged and the submerged parts, 3D topo-bathymetric models produced in each survey were compared limiting the offshore part to that in which full overlapping was possible, i.e., down to ca. 5 m water depth (Figure 8).

From May 2002 to June 2007 the direct effect of the nourishment is visible with beach surface rising both on the dry beach and on the submerged profile, especially in the central cell, where most of the volume was deposited. At the centre of this cell a lobe is evident, where the main feeding groin was positioned (Figure 9). *J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 12 of 17

**Figure 9.** Northward (**a**) and southward (**b**) views of the remains of the filling groins on 18 June 2008. **Figure 9.** Northward (**a**) and southward (**b**) views of the remains of the filling groins on 18 June 2008.

In the following eleven years (2007–2018), limited morphological changes occurred, except the disappearance of what remained of the filling groin that experienced a concentrated lowering of ca. 1.5 m (evident the blue area in the central cell in Figure 7). However, it must be considered that small morphological changes and volumetric variations at the beginning of the tourist season are associated with artificial beach smoothing carried out to eliminate storm berms crest (Figure 10), as a result the emerged profile is expanded and amounts of sediments are moved to the nearshore. Therefore, beach scraping makes the In the following eleven years (2007–2018), limited morphological changes occurred, except the disappearance of what remained of the filling groin that experienced a concentrated lowering of ca. 1.5 m (evident the blue area in the central cell in Figure 7). However, it must be considered that small morphological changes and volumetric variations at the beginning of the tourist season are associated with artificial beach smoothing carried out to eliminate storm berms crest (Figure 10), as a result the emerged profile is expanded and amounts of sediments are moved to the nearshore. Therefore, beach scraping makes the

interpretation of the post-nourishment beach evolution more complex, but it is a proce-

dure frequently carried out also in sand beaches with relevant tourist use.

**Figure 10.** Storm berm crest on the Northern Cell, October 2018 (Google Earth).

trenches excavated on the berm.

Beach volume increase on the dry beach and decrease on the wet one in the 2007– 2018 period can be explained with the onshore sand moving, very likely induced by the higher permeability and porosity of the gravel fill, as observed in other gravel nourishments [21]. Similar to what was observed by previous authors, sand patches are seldom present in the swash zone and sand saturates the pores of the gravel as evident in some

*J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 12 of 17

**(a) (b)** 

**Figure 9.** Northward (**a**) and southward (**b**) views of the remains of the filling groins on 18 June 2008.

interpretation of the post-nourishment beach evolution more complex, but it is a procedure frequently carried out also in sand beaches with relevant tourist use. amounts of sediments are moved to the nearshore. Therefore, beach scraping makes the interpretation of the post-nourishment beach evolution more complex, but it is a procedure frequently carried out also in sand beaches with relevant tourist use.

In the following eleven years (2007–2018), limited morphological changes occurred, except the disappearance of what remained of the filling groin that experienced a concentrated lowering of ca. 1.5 m (evident the blue area in the central cell in Figure 7). However, it must be considered that small morphological changes and volumetric variations at the beginning of the tourist season are associated with artificial beach smoothing carried out to eliminate storm berms crest (Figure 10), as a result the emerged profile is expanded and

**Figure 10.** Storm berm crest on the Northern Cell, October 2018 (Google Earth). **Figure 10.** Storm berm crest on the Northern Cell, October 2018 (Google Earth).

Beach volume increase on the dry beach and decrease on the wet one in the 2007– 2018 period can be explained with the onshore sand moving, very likely induced by the higher permeability and porosity of the gravel fill, as observed in other gravel nourishments [21]. Similar to what was observed by previous authors, sand patches are seldom present in the swash zone and sand saturates the pores of the gravel as evident in some trenches excavated on the berm. Beach volume increase on the dry beach and decrease on the wet one in the 2007–2018 period can be explained with the onshore sand moving, very likely induced by the higher permeability and porosity of the gravel fill, as observed in other gravel nourishments [21]. Similar to what was observed by previous authors, sand patches are seldom present in the swash zone and sand saturates the pores of the gravel as evident in some trenches excavated on the berm.

Tentative sediment budget estimation was performed and presented in Table 3. Data concerning the dry beach volumes are reliable since topographic measurements have the accuracy of few centimetres (i.e., gravel size) but, unfortunately, it is not the case for the nearshore. As previously said, accuracy of bathymetric data is approximately ±10 cm, which corresponds to changes in height of 20 cm between surveys. Since the study area surface is approximately 40,000 m<sup>2</sup> , a volume change of ca. 8000 m<sup>3</sup> is within the accuracy of the methodology used.


**Table 3.** Beach volumetric changes during the studied period (m<sup>3</sup> ).

However, according to our data, the artificial input of ca. 30,000 m<sup>3</sup> of gravel did not produce an equivalent increase in the beach sediment budget. One reason could be that part of the gravel forms now, together with stones, asphalt and different aggregates, the promenade embankment, which is out of the area considered in this study. Analysis of dry beach evolution confirms the stability of nourished sediments, which is not demonstrated when the sedimentary balance is calculated up to a depth of 5 m.

#### *4.4. Sedimentological Evolution*

A first sedimentological study was performed in June 2007 (Figure 11a), i.e., at the end of the second nourishment work, when sediments have not yet been sorted by wave action

and still have their original size (between −4 and −3 phi; 16–8 mm). A very thin strip of coarse to medium sand (0.0–2.0 phi) runs on the seaside of the gravel, but its bi-modal distribution shows that it is a mixture of the fill gravel and the native sand previously present on the nearshore and in the two lateral segment of the bay. Pure sand is present only offshore of the investigated depth. *J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 14 of 17

**Figure 11.** Sediment mean size (Mz) in June 2007 (**a**) and May 2009 (**b**). **Figure 11.** Sediment mean size (Mz) in June 2007 (**a**) and May 2009 (**b**).

A second sedimentological survey was performed in May 2009, before beach raking, to assess whether offshore gravel displacement occurred or not. Results (Figure 11b) show that all the gravel remained in the dry beach or in the nearshore close to the shoreline, with no sediments coarser than 2.0 phi (0.250 mm) offshore of the 2 m isobath, confirming the cross-shore sorting of sediments observed in other natural [34–37] and artificially nourished beaches [21,37]. The coarsest grains of the nourishment, between −5 and −4 phi (16 and 32 mm) were found in the central cell, all along the beach step and, between −1.5 m and −2.0 m, in the point where the filling groin was located. There, coarse and very well sorted lag deposits are present. In addition to this, evident is a lobe of 2–3 phi (0.250–0.125 mm) sediments in the central area, possibly due to a limited migration of the fine grains present in the fill material.

Both 2007 and 2009 grain size data show that below the 2 m isobath, where wave energy is lower, no gravel is present (Figure 12).

**Figure 12.** Mean size (phi) vs. Depth (m) for the sediments at Cavo in June 2007 and May 2009.

**Figure 11.** Sediment mean size (Mz) in June 2007 (**a**) and May 2009 (**b**).

**(a) (b)** 

**Figure 12.** Mean size (phi) vs. Depth (m) for the sediments at Cavo in June 2007 and May 2009. **Figure 12.** Mean size (phi) vs. Depth (m) for the sediments at Cavo in June 2007 and May 2009.

#### **5. Discussion**

The discontinuity of the data, with a long temporal gap from 2009 to 2018, the known but not quantified reshaping of the beach carried out to better satisfy tourism purposes, and the absence of in situ wave climate data, make the study of the evolution of this gravel nourishment very difficult.

Nevertheless, it is important to share this experience within the scientific and technical community, since this kind of nourishment is more and more frequently performed, being cost effective in terms of shore protection and sustainability.

It is sure that the fill proved to be very stable since, after twelve years, almost all the deposited volume is still on the beach, and the beach width is approximately the same as it was after the nourishment.

The behaviour of gravel to gather on the dry part of the profile, as occurred in other gravel nourishments, is confirmed together with the fact that in the case of highly permeable beach sediments, sand can approach the coast and fill the intergranular space [20,21].

Interviews done to beach goers [38] show that this kind of sediments are strongly appreciated, and traditional frequenters of this beach recognize that the coastal environment has improved, for water transparency, beach width (Figure 13) (that gave the possibility to expand the promenade and provide beach services, Figure 14), and for the fact that grains do not stick to the skin. As grain sharpness is concerned, only 30% of the interviewed claimed it as a negative element.

terviewed claimed it as a negative element.

terviewed claimed it as a negative element.

The discontinuity of the data, with a long temporal gap from 2009 to 2018, the known but not quantified reshaping of the beach carried out to better satisfy tourism purposes, and the absence of in situ wave climate data, make the study of the evolution of this gravel

The discontinuity of the data, with a long temporal gap from 2009 to 2018, the known but not quantified reshaping of the beach carried out to better satisfy tourism purposes, and the absence of in situ wave climate data, make the study of the evolution of this gravel

Nevertheless, it is important to share this experience within the scientific and technical community, since this kind of nourishment is more and more frequently performed,

Nevertheless, it is important to share this experience within the scientific and technical community, since this kind of nourishment is more and more frequently performed,

It is sure that the fill proved to be very stable since, after twelve years, almost all the deposited volume is still on the beach, and the beach width is approximately the same as

It is sure that the fill proved to be very stable since, after twelve years, almost all the deposited volume is still on the beach, and the beach width is approximately the same as

The behaviour of gravel to gather on the dry part of the profile, as occurred in other gravel nourishments, is confirmed together with the fact that in the case of highly permeable beach sediments, sand can approach the coast and fill the intergranular space [20,21]. Interviews done to beach goers [38] show that this kind of sediments are strongly appreciated, and traditional frequenters of this beach recognize that the coastal environment has improved, for water transparency, beach width (Figure 13) (that gave the possibility to expand the promenade and provide beach services, Figure 14), and for the fact that grains do not stick to the skin. As grain sharpness is concerned, only 30% of the in-

The behaviour of gravel to gather on the dry part of the profile, as occurred in other gravel nourishments, is confirmed together with the fact that in the case of highly permeable beach sediments, sand can approach the coast and fill the intergranular space [20,21]. Interviews done to beach goers [38] show that this kind of sediments are strongly appreciated, and traditional frequenters of this beach recognize that the coastal environment has improved, for water transparency, beach width (Figure 13) (that gave the possibility to expand the promenade and provide beach services, Figure 14), and for the fact that grains do not stick to the skin. As grain sharpness is concerned, only 30% of the in-

being cost effective in terms of shore protection and sustainability.

being cost effective in terms of shore protection and sustainability.

*J. Mar. Sci. Eng.* **2021**, *9*, x FOR PEER REVIEW 15 of 17

**5. Discussion** 

**5. Discussion** 

nourishment very difficult.

nourishment very difficult.

it was after the nourishment.

it was after the nourishment.

**Figure 13.** The beach at Cavo before (**a**) and after (**b**) the works (Courtesy www.elbaworld.com). **Figure 13.** The beach at Cavo before (**a**) and after (**b**) the works (Courtesy Elbaworld). **Figure 13.** The beach at Cavo before (**a**) and after (**b**) the works (Courtesy www.elbaworld.com).

**Figure 14.** A small bar (a) and toilets (b) on the new gravel beach (April, 2017). **Figure 14.** A small bar (a) and toilets (b) on the new gravel beach (April, 2017). **Figure 14.** A small bar (**a**) and toilets (**b**) on the new gravel beach (April, 2017).

#### **6. Conclusions**

The beach at Cavo was almost completely lost in the 1970s and a revetment and two detached breakwaters were built to protect the coastal road. However, it made this coastal segment very dangerous for diving and unsuitable for the tourist industry. An attempt to give the site a beach again was carried out with iron ore waste, which was rejected by the stakeholders and posed several environmental issues.

The gravel beach, created to cover this ugly and potentially dangerous material, was effective at this end and proved to be very stable and appreciated by the beachgoers.

The wide beach, the transparent coastal water, the new promenade (Figure 13), and beach services (Figure 14), were all made possible thanks to the choice to build a gravel beach, demonstrating that coarse sediment nourishment can be a viable solution to beach erosion, especially in areas where sand is not available, or its use should be accompanied with harder shore protection structure.

#### **7. Update**

The 31 October 2018, storm severely hit the entire Tuscany coast, so much that the region implemented an emergency plan to allow municipalities to dredge sediment in the nearshore to feed the beach in order to allow tourist activity in the following summers. Approximately 11 million euros were spent at that end for 17 small projects.

The beach of Cavo was very little affected by the storm, with a shoreline retreat of approximately 2 m (data from Regione Toscana), and the coastal road was not reached by the waves; therefore the gravel beach proved to be not only suitable for tourism, but also an effective shore protection structure.

**Author Contributions:** Conceptualization, E.P. and E.B.; methodology, E.P., L.R. and I.C.; software, I.C.; validation, G.A., E.B. and L.R.; formal analysis, I.C.; writing—original draft preparation, I.C.; writing—review and editing, G.A., L.R., E.P.; supervision, E.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable. The study did not involve humans or animals.

**Data Availability Statement:** Data supporting reported results can be asked to the first author.

**Acknowledgments:** This research is a contribution to the Ibero-American Beach Management and Certification Network—PROPLAYAS and to the Andalusia PAI Research Group 'RNM-3280 .

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

