Medium-term dynamics of a Middle Adriatic barred beach

In the recent years, attention has been paid to beach protection by means of soft and hard defenses. Along the Italian coasts of the Adriatic Sea, sandy beaches are the most common landscapes and around 70 % of the Marche-Region coasts (central Adriatic), is protected by defense structures. The longest free-from-obstacle nearshore area in the region includes the beach of Senigallia, characterized by a multiple barred beach, frequently monitored during the last decades. The bathymetries surveyed in 2006, 2010, 2011, 2012 and 2013 show long-term stability, confirmed by a good adaptation of an analyzed 5 stretch of the beach to the Dean-type equilibrium profile, though a strong short-/medium-term variability of the wave climate has been observed during the monitored periods. This suggests a slight influence of wave forcing on the long-term profiles, which seems to only depend on the sediment size. Further, the medium-term dynamics of the submerged bars and their geometric features have been related to the wave climate collected, during the analyzed temporal windows, by a wave buoy located 40km off Senigallia. An overall interpretation of the complete dynamics, i.e. hydrodynamics (buoy data), sediment 10 characteristics (equilibrium-profile A parameter) and morphodynamics (bathymetric surveys), suggests that the wave climate is fundamental for the morphodynamic changes of the beach in the medium term: waves coming from NNE/ESE, characterized by a larger/smaller steepness and by a larger/smaller relative wave height, induce seaward/shoreward bar migration, as well as bar smoothing/steepening. Moving southeastward, the bar dimension increases, while the equilibrium profile shape suggests the adaptation to a decreasing sediment size in the submerged beach. This is probably due to the presence of both the harbor 15 jetty and river mouth North of the investigated area.


Introduction
Our communities are experiencing a series of problems and difficulties related to the inundation risk in the coastal areas, the protection of nearshore regions, the use of beaches for tourist and recreational activities. In the last decades, an increasing attention has been paid to short-and long-term predictions associated with the climate change effects, which are strictly related to the above-mentioned aspects (e.g., see Houghton et al., 2010;Ranasinghe et al., 2013). In fact, such predictions are associated 20 with both the mean sea-level rise and the more frequent sea storms, also occurring during the summertime. The understanding of the main physical processes driven by such changes is fundamental for (i) the modeling of the nearshore dynamics, also in terms of rapid morphological changes of the beach (e.g., Postacchini et al., 2016b), (ii) the correct prediction of coastal flooding (e.g., Villatoro et al., 2014), (iii) the proper design of protection solutions (e.g., Lorenzoni et al., 2016) and (iv) the correct analysis of future scenarios in the coastal area (e.g., see Benetazzo et al., 2012;Lionello et al., 2012). 25 1 Several studies (e.g., Benavente et al., 2006;Walton and Dean, 2007) showed that a proper representation of the local bathymetry is fundamental both to correctly predict the seabed changes induced by wave/current forcing and to design efficient solutions for the coastal protection. Hence, typical bedforms of unprotected sandy beaches should be taken into due account.
In particular, submerged subtidal bars usually form on bottom slopes within 0.005-0.03 and their height ranges between some centimeters to meters (Leont'ev, 2011). In semi-protected and open coasts, two-dimensional longshore bars are quite common 5 and have been extensively studied, though the complex mechanisms of generation and migration are not yet completely understood. Generation of submerged bars can be ascribed to three different mechanisms, i.e. wave breaking, infragravity waves and self arrangement (Wijnberg and Kroon, 2002), while the bar migration depends on several coastal processes and has been investigated both in the field (e.g., Ruessink et al., 1998), numerically (e.g., Dubarbier et al., 2015) and through laboratory experiments (e.g., Alsina et al., 2016). It has been observed that swash-zone slope, grain size and wave characteristics play an 10 important role. The influence of the former on the bar dynamics has only been observed during laboratory experiences, after an ad hoc manual reshaping of the swash zone (Baldock et al., 2007;Alsina et al., 2012). On the other hand, field observations confirmed that the grain size could be important in the bar migration rates, due to the larger sediment transport induced by finer sands (Goulart and Calliari, 2013), while the wave characteristics are fundamental for the bar migration direction. In particular, the wave breaking over the bars leads to the generation of a deep return flux, known as undertow, which promotes a seaward 15 motion. As an example, Gallagher et al. (1998) observed, near Duck (North Carolina), an intensified wave breaking occurring over the bar during storms, this inducing a large undertow inshore of the bar that pushed it seaward. Conversely, a shoreward bar migration was also observed under small waves, during less energetic states (see also Goulart and Calliari, 2013).
While numerical simulations well reproduced the offshore migration during severe conditions, some difficulties arose when reproducing the onshore bar motion during mild wave conditions (Gallagher et al., 1998;Plant et al., 2004), this suggesting 20 that not all the processes involved in the bar migration were clearly understood and correctly simulated, e.g., lower-frequency waves. Further, Ruessink et al. (1998), who analyzed the cross-shore sediment transport and morphological changes occurring in the nearshore area of Terschelling (Netherlands), stated that the role of the infragravity waves have not been completely understood. In particular, it was fairly clear that during energetic conditions, the suspended load dominated over the bedload and the morphodynamics were controlled by undertow and, probably, infragravity waves: the latter, more important during 25 breaking than during calm conditions, mobilize large amounts of sediments, which are then advected offshore by the undertow.
The importance of infragravity waves is confirmed by other authors, and a detailed study about their influence on the bar dynamics was undertaken by Aagaard et al. (1994) using field data collected at Stanhope Lane Beach (Canada). They stated that the sediment transport induced by infragravity waves may be either shoreward or seaward, and suspended sediments are mainly transported towards antinodes in the water surface elevation. However, the contribution of infragravity waves on both sediment 30 transport and sandbar motion can be neglected on time scales of years, i.e. when dealing with medium-term morphodynamics (Ruessink and Terwindt, 2000).
With the purpose to characterize the sandbar migration, an important parameter has been recently introduced. This is the local relative wave height, i.e. the ratio between local wave height H and water depth over the bar crest h cr . Values smaller than ∼ 0.3 promote landward migration, while values larger than 0.6 promote seaward migration (Houser and Greenwood, 2005). 35 In particular, along the Dutch coast (Ruessink et al., 1998;Ruessink and Terwindt, 2000), a relative wave height H s /h cr = 0.33 represented the onset of breaking, with H s being the local significant height. Hence, H s /h cr > 0.33 referred to breaking intensification and undertow increase, leading to seaward bar migration. While H s /h cr < 0.33 indicated dominance of short waves and wave skewness, leading to shoreward bar migration. The analysis of the velocity moments and sediment transport confirmed the correlation between medium-term wave conditions and short-term sediment transport measurements (Ruessink 5 and Terwindt, 2000).
From a physical point of view, the increase of both H s /h cr and breaking intensification produces an increase of the breaking wave celerity (e.g., see , this leading to an intensification of the shoreward volume flux, hence to a wave setup (e.g., see Soldini et al., 2009) and to the following increase of the undertow velocity (e.g., see Kuriyama and Nakatsukasa, 2000).

10
Only few literature studies have been carried out to investigate the seasonal and annual scale of the beach dynamics (e.g., Ruggiero et al., 2009). Some field observations confirmed a cyclic behavior of multiple bars (Ruessink and Terwindt, 2000;Goulart and Calliari, 2013), mainly characterized by three stages, i.e. initial generation, seaward migration and final degradation. Conversely, other authors observed a continuous landward motion, until bar-shore welding, even during storm events (Aagaard et al., 2004). While the offshore migration is promoted by the undertow dominance in the net transport balance, as 15 already stated, the onshore migration is probably enhanced by storm surge: this increases both skewness and phase coupling and reduces the undertow contribution.
The present study describes the seabed evolution of a natural unprotected beach stretch of Senigallia (Marche Region, Italy), a touristic town of the Italian Middle Adriatic. The available bathymetries, covering the last decade, and the wave climate, enable us to analyze the medium-term morphological evolution of the beach, including the geometry and migration of the 20 submerged bars, as a function of the wave forcing. To the authors' knowledge, this is the first study on the medium-term beach evolution and bar migration occurring in a sandy beach of the Adriatic Sea, a semi-enclosed basin characterized by small tidal excursions (∼ 40cm) and reduced wave heights, if compared to, e.g. the Dutch coastal areas.
The manuscript is divided as follows. Sect. 2 and Sect. 3 illustrate, respectively, the investigated site and the available data.
Results are presented in Sect. 4 and discussed in Sect. 5. Some conclusions close the paper. 25 2 Description of the site The analyzed coast is part of the longest unprotected beach of the Marche Region, which extends from the estuary of the Misa River, whose final reach is highly engineered and adjacent to the Senigallia harbor, to ∼ 3.5 km North of the Esino River estuary, hence for a total length of ∼ 12 km (Fig. 1). As observed during recent field experiments, the Misa River estuary is dynamic throughout the year, especially during sea storms driven by NNE winds, which mobilize a large amount of sediment 30 and generate significant erosion/deposition patterns nearby the rigid structures (Brocchini et al., 2015(Brocchini et al., , 2017. The investigated site is characterized by a swash-zone slope in the range 1 : 30-1 : 40, an array of submerged bars in a water depth h = 0-3 m, The analysis of the beach morphology, using the concept of the equilibrium beach profile (Dean, 1991), describes the longterm beach equilibrium of a natural beach, i.e. the balance between erosive and accretive forcing, through where h is the water depth and x the distance to shoreline. A is a dimensional shape parameter, directly related to the median grain diameter d 50 (Hanson and Kraus, 1989). Notice that Eq. 1 also leads to the estimate of the so-called "fitting depth", i.e. the water depth at which the measured profile collapse over the equilibrium profile. Though recent models account for further parameters, like seasonal changes (Inman et al., 1993) or the generation of submerged bars (Holman et al., 2014), their application is fairly difficult and it has been demonstrated that Eq. 1 properly represents the long-term natural profile, to be 5 used for coastal engineering purposes (e.g., Walton and Dean, 2007;Soldini et al., 2013). and, following Hanson and Kraus (1989), d 50 ∼ 0.15 mm, while the smallest occur ∼ 3.9 km South of the harbor (profile 66), where A ∼ 0.060 and d 50 ∼ 0.13 mm. Such values are in agreement with the fine sand characterizing the submerged beach (Lorenzoni et al., 1998a). It has been observed that, throughout the coast surveyed in 2006, the natural beach well adapts to the Dean-type equilibrium profile. This is confirmed by the following campaigns (2010-2013), when a good adaptation still 15 exists, the values of A remain almost constant in time and decrease moving southward. Further, the fitting depth increases from the harbor to the "Rotonda", i.e. the pile-mounted permeable structure within profiles 10 and 11, and decreases South of the "Rotonda". This suggests a sediment motion occurring at larger depths in correspondence of the structure, that partially (and locally) influences the beach evolution and bar migration.
Although the present study aims at investigating the nearshore area, where both cross-shore and alongshore sediment trans-20 port contributions determine the short-to long-term equilibrium of the shallow beach, a regional framework may also be taken into account. In general, the sediment transport throughout the Adriatic Sea is influenced by a number of factors. Specifically, the western Adriatic coast is characterized by large depositions nearby the rivers (e.g., at the Misa River estuary, as described by Brocchini et al., 2017) and especially close to the Po Delta. Further depositions occur north of the Gargano Peninsula, due to the Western Adriatic Coastal Current (WACC, e.g., see Harris et al., 1998;Sherwood et al, 2004), which is responsible of 25 the suspended sediment transport. In the same regional framework and for depths greater than 10m, while Bora-induced waves provide large sediment fluxes, Scirocco-induced waves lead to sediment flux reduction, though sediment suspension increases due to significantly energetic conditions.

Experimental data
The natural beach of Senigallia was characterized by a number of bathymetric surveys since the 80s. More recently, due to 30 a specific requirement of the Marche Region, a detailed survey of the nearshore region of Senigallia was undertaken in June 2006, both North and South of the harbor, such areas being respectively characterized by a protected and an unprotected beach.

6
The surveys cover the nearshore region up to a depth of 6 m and a total length of 4.3 km, most of which (∼ 3.9 km) South of the harbor (Fig. 1a).
Between 2010 and 2013, after the modification of the harbor entrance, annual bathymetric surveys up to a depth of 6 m were carried out by the municipality of Senigallia on a 2.5 km-long area covering part of the protected and part of the unprotected beaches.

5
The available bathymetric surveys enabled us to extract 18 cross-shore profiles which characterize the unprotected beach for about 1 km. The bathymetries have been used for the analysis of the wave-climate-induced morphological changes, i.e. bed variations between two consecutive surveys, in terms of bar migration and geometry. It is worth noting that bathymetries could have been surveyed just after an intense storm, which promotes significant morphological changes. However, the medium-term climate is the sum of a number of energetic and calm states occurring between two consecutive surveys. Hence, a bathymetry 10 does not depend on any specific event, nor on a specific season/month (e.g., see the significantly different beach profiles surveyed in February 2010 and February 2011, illustrated in Fig. 2), but on the sum of the contributions of all such events to the overall morphological change observed in the chosen time range. Further, the separation between the morphological effects induced by long-term and short-term events is difficult, especially in semi-enclosed basins like the Adriatic Sea, which is characterized by an extremely variable climate, with significantly large deviations of the wave characteristics from the mean 15 values, even during a storm. Hence, the aim of the present work is that of analyzing the morphological changes and discussing the cumulative effect of all events occurring between consecutive surveys. Such an analysis is also useful to demonstrate that the beach evolution can be predicted when a limited number of surveys is available, a typical condition for coastal municipalities.
From the analysis of both surveys and satellite data, the submerged bars remain for a stretch of ∼ 12 km. Further, moving southeastward, the sediment size changes, with a transition from sand to gravel occurring ∼ 6 km South of the harbor (Loren-20 zoni et al., 1998b). Hence, the initially two-dimensional longshore bars of the investigated area get closer to the shoreline, thus switching to three-dimensional (see Fig. 1b, where the location of the bars is highlighted by both foam and suspended sediment induced by the waves breaking over them). However, the ∼ 1 km-long area South of the harbor can be taken as representative of the sandy beaches characterizing the Middle Adriatic Sea and will be analyzed in the next sections.

25
The following sections illustrate the results obtained from the analysis of the seabed variation using the available bathymetric Both migration and geometry of the submerged bars are also discussed.

Wave climate
Except for the period 2006-2010, during which the waverider did not work, the wave climate referring to the considered time  8 coming from either ESE, i.e. forced by Levante-Scirocco winds, or NNE, i.e. forced by Bora winds, which thus correspond to the predominant waves of such a coastal area. Waves from NW are also frequent, but less energetic. While the wave frequency (blue lines) is fairly well distributed and homogeneous, the wave energy (red lines) is characterized by sharper peaks in correspondence of the dominant directions and by a reduced distribution elsewhere.
It is well known that the wave climate for the extra-tropical regions at intermediate latitudes, like that of the Adriatic Sea, is 5 characterized by the presence, at the soil level, of closed dynamical systems, as cyclones and anticyclones. Usually, soil weather systems are connected to a movement with an upper-level wavy structure, that slowly migrates eastward. So, the presence of migrating temporal troughs and ridges alternates during the year. Troughs are linked to low atmospheric pressure areas, with colder air and a sequence, usually, of cyclones. Ridges are linked to high pressure areas, with warmer air and anticyclonic, more stable, weather. Specifically, the Bora is a cold and dry wind usually linked to a well-developed anticyclone on the central 10 or northern Europe and a relative low pressure on the Mediterranean Sea. It is more frequent and very intense during the winter.
Conversely, the Scirocco is a southern warm wind, which is dry in Africa, then becomes wet passing on the Mediterranean Sea, and finally generates big sea storms with important surges and persistent swell. Scirocco intensities are less than the Bora, but generate longer and more enduring waves.
In the studied site, the weather is not characterized by two distinct (seasonal) behaviors, rather by a pronounced temporal 15 variability of the wave climate during the year: the two peaks illustrated in Fig. 3a-d do not refer to the prevalent conditions occurring, respectively, in summer and winter, but mainly refer to the most severe winter storms (Fig. 3e), the summertime being characterized by milder wave conditions, due to less strong winds and slowly changing wind directions during storms (see also Brocchini et al., 2015). Further, the fairly well distributed frequency, with respect to the more peaked energy flux, indicates that the annual variability of storms is not bound to the seasonal variability of wave climate. This can be observed in  Brocchini et al. (2017), who observed two consecutive storms in January 2014, the first due to Bora winds and the second, after three days, due to winds coming from WNW and N. 25 With reference to both frequency and energy flux, a statistic analysis of the main sectors has been undertaken for each selected time period, as in the following steps: the wave climate during the whole time range is analyzed to obtain the energy distribution illustrated in Fig. 3b-d, the most energetic direction is chosen and associated to a specific sector, i.e. (105-135) • for ESE or  • for NNE, the waves falling in the chosen sector are analyzed to get the most energetic wave-height ranges, 30 the most frequent wave-period ranges associated to such heights are chosen.
In detail, since Fig. 3b and Fig. 3d show that the ESE forcing dominates in 2010-2011 and 2012-2013, only the (105-135) • sector has been analyzed. Conversely, the NNE forcing dominates in 2011-2012, hence this has been associated to  • . In the former case, the largest energetic contributions (more than 60% of the total) are ascribed to significant wave heights in the range H m0 = (1-3) m (2010-2011) and H m0 = (1. 5-3.5) m (2012-2013). The most frequent waves falling in such ranges are characterized by mean periods T m = (4-5.5) s (2010-2011) and T m = (4.5-6) s (2012-2013). Peak periods are, respectively, T p = (6-7.5) s and T p = (7-8.5) s. In 2011-2012, the largest energetic contribution (> 60%) belongs to a narrower waveheight range, i.e. H m0 = (1-2.5) m, which corresponds to most frequent waves falling in wider ranges T m = (3.5-5.5) s and With the purpose of characterizing each time interval with specific wave features, the most energetic direction (ESE or NNE) associated with the most probable wave-height class gives the most probable wave-period class. As an example, Tab. 1 shows that in 2012-2013 the largest energy-flux distributions characterize the ranges H m0 = 1.5-2m (16.56%) and H m0 = 3-3.5m (16.02%). However, we believe that the former is more representative, as more probable height-period classes exist (see Tab. 2).

10
In particular, 10.51% of all waves are characterized by H m0 = 1.5-2m and T m = 5.0-5.5s, while waves with H m0 = 3-3.5m are not so frequent.  The described procedure leads to the following mean values, which represent the most probable combinations (H m0 , T m ) and (H m0 , T p ), related to the most energetic waves (ESE or NNE). As expected, due to the available fetch length (see also Fig. 1a), a larger wave steepness (H m0 /L p0 , where L p0 is the deep-water peak wavelength) occurs during the NNE-dominated periods.
It is worth noting that in 2010-2011, though ESE is the most energetic direction (∼ 26%), the NNE contribution (∼ 21%) 20 is also important (Fig. 3b). The analysis of the NNE direction suggests that the most energetic waves are characterized by a reduced height range, i.e. H m0 = (1.5-2.5) m, associated to periods T m = 3.5-5.5s and T m = 5.5-7s. If we look at the mean values of the most energetic and frequent classes, we get Such a result demonstrates that the NNE sector provides waves steeper if compared to the ESE sector, whether or not it 25 represents the most energetic sector.

Bathymetric surveys
The available bathymetries have been overlapped using ArcGIS software and the difference in the bed depth has been estimated between each pair of consecutive surveys. Hence, Fig. 4 illustrates the difference between the bed depth measured in 2010 and   The shoreline is fairly stable and, in the medium-term, oscillates in the cross-shore direction less than 20 m (Fig. 5a), with the largest motions occurring in 2006-2010 (advance) and 2011-2012 (retreat). To properly reconstruct the bar migration, the crest locations are overlapped to the color maps of Fig. 4.
Further, each of the 18 cross-shore profiles have been characterized by means of (also refer to Fig. 2): (i) the shoreline position from a fixed point (s sh ), (ii) the distance of each bar crest from both fixed point (s cr ) and shoreline (x cr = s cr −s sh ), and (iii) the bar geometry, i.e. crest (h cr ) and trough (h tr ) depths. The location of both bar crest s cr and shoreline s sh are illustrated in Fig. 5a. Since it is evident (Fig. 4)  at profiles 17-18. Further, the outer bars (•) do not seem to be strongly influenced by the "Rotonda", as small local changes 30 occur in the crest location between profiles 10 and 12 (Fig. 5a), while the depth ratio slightly increases moving South (Fig. 5b).
It is worth noting that Fig. 5, i.e. both bar alignment and geometry, invokes the existence of two regions where bars behave differently, one North (between profiles 4 and 9) and one South (profiles 14-18) of the "Rotonda". This also means that the complex hydrodynamics and beach morphology induced by the structures lead to discontinuities, like those observed for the outer bar in 2010 and 2012, and different beach responses in the two regions (e.g., see Shand et al, 2001). Transition regions also exist, one due to the jetty (profiles 1-4), the other due to the "Rotonda" (profiles 10-13).
While the depth variation of Fig. 4 is representative of the volume changes occurred at each point of the domain, the crossshore profiles at different alongshore locations more clearly illustrate the volume changes occurred between two consecutive surveys. In particular, since the ESE forcing slightly dominates in 2010-2011 and Fig. 3b suggests a bimodal behavior of the 5 wave climate, three profiles collected in 2010 (blue line) and 2011 (green line) are analyzed. They represent the region located between the jetty and the "Rotonda" (Fig. 6a, profile 6), that around the "Rotonda" (Fig. 6b, profile 10) and that far from the "Rotonda" (Fig. 6c, profile 18). In addition, the cumulative volume change is illustrated (red dashed line), with the aim to explain how the sediment is transported through the cross-shore profile, but notice that the alongshore sediment losses are not accounted for in such an approach. The volume change (V C) at the j-th cross-shore location (j = 1...n, with n the number of 10 points along the x axis), referring to two profiles surveyed at times k f and k i , may be expressed as where the volume variation at the j-th location between k i and k f is with the volume (per unit length) at the j-th location V j being calculated as the product between the profile discretization 15 in the cross-shore direction (∆x) and the profile elevation with respect to a horizontal reference system (z b,j ): V j = z b,j ∆x.
In the example of Fig. 6, time indexes are k f = 2011 and k i = 2010. Notice that at j = 1, i.e. x = 0, the volume change is Along the undisturbed profile (Fig. 6c), the volume change is positive between x = 0 and x ∼ 120m, i.e. up to the 2010 inner bar, this suggesting an increase of the upper beach and nearshore area due to that bar and an overall sediment balance in the 20 range x = 0 − 120m, as V C| x∼120m = 0. Further, the volume change is also positive between x ∼ 150 and x ∼ 210m, i.e. up to the 2010 intermediate bar, and between x ∼ 210 and x = 800m. Such an analysis suggests that three distinct regions exist.
The first region is characterized by both the migration of the inner bar and the increase of the upper beach and nearshore area.
The migration of the intermediate bar occurs in the second region. The third region deals with a significant beach reshaping involving the outer bar. Similar results have been found for the profile surveyed between the rigid structures (Fig. 6a), where 25 both inner and intermediate bars contribute to the nearshore/upper beach change. Finally, the "Rotonda" significantly affects the sediment balance throughout the analyzed profile (Fig. 6b), as the volume change never goes to zero. In particular, the upper beach change mainly depends on the inner bar, as the volume accretion (i.e. V C increase) occurring at x = 0 − 90m partially derives (∼ 39%) from the volume erosion (i.e. V C decrease) occurring at x = 90 − 115m. Hence, the structure significantly affects the observed sediment transport, which is characterized by both cross-shore and alongshore contributions, and promote 30 an increase of the closure depth, in agreement with the fitting depth increase nearby the structure (see Sect. 2).
The inspection of the cross-shore profiles and volume changes referring to the other time intervals confirms the existence of the three above-mentioned regions and, far from the "Rotonda", the inner and intermediate bars mainly contribute to the volume change in the upper beach and nearshore area. Further, the permeable structure significantly influences the volume change in 2011-2012 and less in 2012-2013 (e.g., see the regularity of both intermediate bar crest alignment and relative seabed variation in Fig. 4d, which seem not to be affected by the "Rotonda"). Finally, while in 2010-2011 the balance throughout the profile, i.e. V C| x=800m , is small far from the structures (Fig. 6a, c) and large close to the "Rotonda" (Fig. 6b), this suggesting an important alongshore sediment transport localized nearby the structure, in 2011-2012 the farthest profiles (e.g., profile 18) 5 are almost in equilibrium (V C| x=800m ∼ 0), while the alongshore contribution is important between jetty and "Rotonda", this promoting an overall beach erosion (V C| x=800m 0). The V C observed in 2012-2013 suggests a slight alongshore contribution throughout the domain.

Bar characterization
The previous data have been used to introduce a detailed analysis of the nearshore morphodynamics, especially the bar geometry and migration. Dimensionless parameters are introduced to analyze the bar geometry (e.g., see Grunnet and Ruessink, 5 2005). In Fig. 7a, the dimensionless bar height H bar /h cr is plotted against the dimensionless bar width W bar /s cr , where the bar dimensions are defined as: W bar = 2(s cr − s tr ). The analysis of the longshore distribution of the bar geometry can be undertaken accounting for the bar cross-shore area which is made dimensionless using both depth and distance to shore of the bar crest. the reduced number of sections at which bars occur. In addition, it can be noticed that the shape parameter A of Eq. 1 (solid thin lines and triangles), always decreases from left to right, suggesting a sediment-size reduction moving southward.
Hence, though Figure 7a illustrates a natural data scattering due to the beach variation both in time and space (e.g., see Grunnet and Ruessink, 2005), a best-fit polynomial curve well represents the geometrical characterization of outer and middle  represented. This probably means that the inner bars are significantly influenced by the rigid structures and are not characterized by homogeneous alongshore distributions, while middle and outer bars are locally influenced by the structures, but only during specific time periods (e.g., see the influence of the jetty on the first two points of the 2011 intermediate bar, Figure 7b).

Bar dynamics
As suggested by several studies, the generation of subtidal bars may depend on three different mechanisms, i.e. i) breakpointrelated, ii) infragravity-waves-related, and iii) self-organisational mechanisms (e.g., see Wijnberg and Kroon, 2002;Leont'ev, 2011). From the results presented in Sect. 4.1 and 4.3, the bar dynamics in this area might be influenced by either the first or the second mechanism, while the self-organisation seems negligible. In fact, in agreement with Wijnberg and Kroon (2002), 10 such a mechanism cannot explain the bar re-generation between 2012 and 2013 (Fig. 4d), after a general beach smoothing and the partial bar destruction occurred in 2011-2012 ( Fig. 2 and Fig. 4c).
The destructive nature of the NNE storms significantly affects the bar geometry (beach smoothing), as well as the migration (seaward rather than shoreward), this being strongly influenced by the different wave features (waves coming from NNE were higher and steeper than those coming from ESE), which force the breaking to occur at different locations. Hence, the difference 15 in terms of characteristics of the incoming sea-storm waves directly reflects on the beach morphology, this underlining that the medium-term bar dynamics in the Adriatic sandy beaches are mainly governed by wind waves and breakpoint mechanisms.
Furthermore, steep NNE waves are associated with not excessive storm surges, while less steep ESE waves are associated with larger surges, due to the larger fetch which characterizes this wave direction in the Adriatic Sea. As an example, two consecutive intense storms occurred in December 2010, the former coming from ESE, the latter from NNE, were characterized 20 by maximum surges of, respectively, 80 and 43 cm, measured within the protected basin of the Ancona harbor (data from Rete Mareografica Nazionale, ISPRA, http://www.mareografico.it). This leads to larger water depths over the crest (h cr ) and smaller relative wave heights (H/h cr ) during ESE than during NNE waves. In fact, wave propagation from the offshore to the outer bar depth enables one to estimate the local wave inclination (α l ), and also the local wave height H m0,l . This may be done using either simple analytical models, which accounts for wave refraction and shoaling (e.g., Goda, 2000), or detailed 25 numerical approaches, which provide a more complete wave characterization in the nearshore (e.g., Carniel et al., 2011). Then, the actual water depth over the crest during surge may be estimated as h cr,s = h cr +η s , where h cr is the alongshore-averaged still water depth over the crest and η s the surge contribution, which is different depending on the dominating wave direction, but in agreement with both the data collected at the Ancona harbor (Rete Mareografica Nazionale, ISPRA) and previous literature studies (Orlić et al., 1994;Villatoro et al., 2014).

30
The above-introduced terms and the relative wave height estimated using the local root-mean-square wave height H rms,l (US Army, 1977), are summarized in Tab. 3. The relative wave height, especially H rms,l /h cr,s , which is larger in 2010-2011 and 2012-2013 and smaller in 2011-2012, suggests, respectively, a landward and seaward bar migration, which has been actually observed (e.g., see Fig. 4). The estimated local wave angles suggest an almost orthogonal-to-shore direction during the NNE-wave-dominated period. Our observations are supported by the numerical results of Dubarbier et al. (2015), who found that the variability in sandbar migration is sensitive to water level over bar crest, this being consistent with storm-surge variations occurring in our site. On the other hand, wave obliquity mainly affects the rates of bar growth and migration, but not their migration direction. This suggests that the difference between Bora and Scirocco waves, in terms of wave incidence, does not influence the bar direction, but eventually their propagation speed. The outer bar variation, i.e. the change of the alongshore-5 averaged outer bar height (∆H bar /h cr,s ) and cross-shore area (∆Ω/(x cr h cr,s )), has also been analyzed. As illustrated in Tab

Discussion
Recent studies on the dynamics of barred beaches suggested us to correlate wave-climate and bathymetric surveys of an unprotected beach of the Adriatic Sea. In fact, though some results on sandbar migration along the Tyrrhenian Sea were recently illustrated (e.g., Parlagreco et al., 2011), the bar dynamics of typical Adriatic sandy beaches have not been already 15 investigated. Further, the correct understanding of the bar migration is important when dealing with beach management and tourism. To this aim, the coast of Senigallia has been here investigated since, similarly to many Adriatic sandy beaches, this is characterized by a significant flow of tourism, especially in the summertime (see Sect. 2).
Hence, the bathymetric surveys of the area South of the harbor, which has been seen to be stable in the long term, enabled us to analyze a multiple-bar array typical of the sandy beaches of the Middle Adriatic. Such a part of the basin is subject to 20 sea storms mainly due to NNE (Bora) and ESE (Levante-Scirocco) winds, which are characterized by significantly different surges.
The seabed-depth variation and the wave climate between consecutive surveys, as well as the bar features (height, width, location) analyzed for each survey, enabled us to couple the beach/bar dynamics with the wave forcing.
In the studied area the tidal excursion (∼ 40 cm) is small and only subtidal bars exist. Since the analyzed beach slope ranges between 1 : 35 ∼ 0.03 (swash zone) and 1 : 200 ∼ 0.005 (offshore area), such bars fall into the group of two-dimensional longshore bars (Wijnberg and Kroon, 2002). Further, the wave energy in such a microtidal environment is quite high.
In the analyzed region and during the investigated time periods, the beach experienced many sea storms that enabled us to give an overall interpretation to the bar migration process as a function of the wave climate. Coupling wave steepness and 5 the Dean number (i.e. the ratio of wave height to sand fall velocity and wave period), both ESE and NNE are associated with erosive wave conditions (e.g., see Dean and Dalrymple, 2004). However, during the time periods dominated by ESE forcing, waves are characterized by a reduced steepness H m0 /L p0 = 0.213 (exactly the same in 2010-2011 and 2012-2013), while this is about 1/3 larger during the NNE-forcing-dominated period (H m0 /L p0 = 0.316). Such a behavior is also confirmed if we do not account for the most energetic waves (see Sect. 4.1), but directly estimate the most frequent combination (H m0 , T p ).

10
Further, an increase of the bar steepness H bar /W bar is associated to a decrease of H m0 /L p0 (e.g., compare the bar geometry in Fig. 2 with the associated wave steepness). reach the shore with an almost perpendicular incidence, this improving the intense smoothing of the bars.
Hence, it has been seen that the relative wave height can be properly applied for the prediction of bar migration in an environment different from those already proposed in the literature (e.g., Ruessink and Terwindt, 2000;Houser and Greenwood, 2005), i.e. a nearshore area characterized by a reduced tidal excursion, and partially influenced by the presence of rigid structures. This allows the application of such a predictive parameter for similar nearshore environments, and also for a medium-term predic-25 tion. Hence, such a parameter is valid for different environments, characterized by tidal excursions of some centimeters (e.g., Lake Huron, Houser and Greenwood, 2005) to decimeters (Adriatic Sea, present study) to meters (e.g., North Sea, Ruessink and Terwindt, 2000). Assuming that the bar migration mainly occurs during sea storms, the involved sediment transport mainly depends on the incoming short waves (especially when the bars move landward, i.e. ESE waves dominating) and the undertow (especially for seaward motion, associated with NNE waves), with the infragravity waves probably being of some importance 30 in such a dissipative beach (e.g., see Wright and Short, 1984;Ruessink et al., 1998).
While the correlation between bar width and bar height is clear only for some cases, the former increasing with the latter, an overview of the available data enable further conclusions. Between 2010 and 2011, the largest waves, mainly propagating from ESE, provided a height increase of the outer bar (in agreement with Houser and Greenwood, 2005) only North of the "Rotonda", and, at the same time, a width increase and a steepness reduction of both outer and intermediate bars (blue and 35 green symbols in Fig. 7a). While between 2011 and 2012 the bars are largely smoothed due to the NNE dominating waves (purple symbols), the ESE stormy conditions occurred between 2012 and 2013 gave rise to geometric features of the bars similar to those observed in 2011 (orange symbols).
The cross-shore bar area increases moving southward, especially from the Senigallia harbor to the "Rotonda", which partially disturbs the growth of the middle bar. This could also be analyzed in view of the equilibrium-profile theory, described by Eq. 1.

5
The analysis of the shape parameter A (see Fig. 7b) suggests that d 50 slightly decreases moving southward. Some important oscillations of A characterize the region between profiles 1 and 11, this underlining the influence of the rigid structures, while a generally decreasing trend can be observed South of the "Rotonda" (notice that the larger values referring to 2006 may be due to the lower resolution of the surveyed bathymetry in the nearshore , i.e. up to a depth of 1.5−2m, with respect to the following surveys). Such a decrease is in agreement with the sediment-size distributions observed in 1989 and 1990 by Lorenzoni et al. 10 (1998a). This is probably due to: i) the river jetty (Fig. 1a), which induces a complex flow field, i.e. a mix of refraction, diffraction and reflection, that generates wave-wave interactions, crossing waves and intense vorticity, especially when sea storms come from ESE (e.g., see ; ii) the river discharge, especially during severe weather conditions, which gives rise to an intense plume that both propagates southeastward and promotes sediment deposition along its path (e.g., see Brocchini

Conclusions
The nearshore dynamics is characterized by different levels of analysis: i) the long-period beach stability is of the order of decades, ii) the medium-term evolution of the beach forms (e.g., submerged bars, artificial nourishments) is of the order of 30 years or seasons, and iii) the short-term erosion of the beach profile is of the order of days or hours. While i) and iii) have been widely investigated, the medium-term beach variability has not been sufficiently analyzed. Hence, recent findings suggested us to investigate the medium-term morphodynamics of the sandy barred beach of Senigallia, located in the Middle Adriatic Sea.
The present work both illustrates how a proper buoy-data handling leads to the prediction of the morphological changes of a barred beach and offers a useful tool, for coastal engineers and managers, to: i) properly predict the emerged beach stability (e.g., shoreline retreat, erosion), ii) accurately design nourishments for submerged beach recovery, iii) estimate the sediment transport flux through the entrance of nearby harbors, iv) choose the best place to drop the dredged sediment coming from nearby harbors, eventually with nourishment purposes.

5
A more detailed analysis could be achieved through use of either data collected by another waverider (e.g., that of Cesenatico, FC, which is ∼ 80 km North of Senigallia) or a reconstructed climate (e.g., Mentaschi et al., 2015), with the aim to characterize the wave forcing in the period 2006-2010. Although global reanalysis or numerical modeling may provide a more detailed wave characterization, use of available regional climate models (e.g., The Medatlas Group, 2004) is easier and may represent a valid alternative. Further, the dynamics of the nearshore area before, during and after storm events could also be inspected by means 10 of novel devices like: i) Lagrangian drifters, able at measuring both three-dimensional hydrodynamics and seabed depth (e.g., Postacchini et al., 2016a), ii) video-monitoring systems, like that available at the Senigallia harbor since 2015, to reconstruct the coastline (e.g., Archetti, 2009;Vousdoukas et al., 2011;Archetti et al., 2016), as well as wave field and bed morphology (e.g., Palmsten et al., 2015), iii) radar images, like those used for the reconstruction of both wave field and bathymetry, through the depth inversion technique (e.g., Ludeno et al., 2015). 15