Characteristics and coastal effects of a destructive marine storm in the Gulf of Naples (Southern Italy)

Destructive marine storm bring large waves and unusually high surges of water to coastal areas, resulting in significant damages and economic loss. In this study it is examined the characteristics of a destructive marine storm on the strongly inhabited coastal area of Naples Gulf, along the Italian coasts of the Tyrrhenian Sea, which is highly vulnerable to marine storms due to the accelerated relative sea level rise trend and the increased anthropogenic impact on the coastal area. The marine storm, occurred on the 28th December 2020, was analysed through an unstructured wind-wave coupled model that 5 takes into account the main weather-marine components of the coastal setup. The model, validated with in-situ data allowed to establish threshold values for the most significant marine and atmospheric parameters (i.e., wind intensity and duration) beyond which an event can produce destructive effects. Finally, a first assessment of the return period of this event was evaluated using local press reports on damage on urban furniture and port infrastructures.


Study area
The Naples Gulf (Figure 1), Tyrrhenian Sea, is one of the most populated Italian areas, with 3,016,762 inhabitants and a 55 medium density of 2,672 inhabitant/km 2 . The urban territory includes 92 municipalities on a surface of 1,171 km 2 , with a 60% of small municipalities (surface <10 km 2 ) and an 11% of big ones (surface >25 km 2 ). The lasts include the coastal cities of Naples, Torre Annunziata and Pozzuoli. The urbanized area in the Gulf occupies only the 32.54% of the total surface, and consequently the population density in this area is more than 8,000 inhabitant/km 2 . Under these circumstances, the urbanized coasts of the Gulf can be certainly considered highly sensitive from both a social and economic point of view to severe marine 60 storms. On the other hand, main cities are located in narrow coastal plains, with commercial activities and infrastructures located only few meters above sea level (Ascione et al., 2020). In fact, the present coastal morphology in the Gulf ( Figure   1) is characterized by an alternation of articulated seacliffs with sheltered pocket beaches, and narrow coastal plains, often strongly urbanized (Ascione et al., 2020;Aucelli et al., 2017). In particular, the high-coastal sectors in the Gulf can be divided in seacliffs made of volcanic deposits typically bordered by wide shore platforms (often of polycyclic origin), and plunging 65 cliffs in hard limestones located along the eastern side of the Gulf (Aucelli et al., 2016a, b;Pappone et al., 2019;Aucelli et al., 2019;Mattei et al., 2020). The main coastal plains in the Gulf, that are Fuorigrotta, Chiaia, Sebeto and Sarno plains (filled by successions of volcanoclastic deposits) host the most populated cities in the Gulf, i.e. Naples and Torre Annunziata (Romano et al., 2013;Vacchi et al., 2016;Cinque, 1991). From a geological point of view, the Gulf of Naples is an active peri-Tyrrhenian basin extending for about 1000 km 2 . It is characterized by physiographic features typical of a passive continental margin sector, with a continental shelf between -140 and -180 m of depth (Milia andTorrente, 1999, 2003). The structure of the Gulf of Naples is controlled by numerous Quaternary fault systems, NE-SW trending SE-dipping and NW-SE trending SW dipping, linked to the last stages of the opening of the Tyrrhenian Sea (Fedele et al., 2015;Milia, 2010). Between the Middle and Upper Pleistocene, the fault systems were responsible for the development of the half-graben of the Gulf of Naples and Sorrento Peninsula fault block ridge (Milia 75 and Torrente, 2003). The landscape of this area is strongly influenced by the presence of two active volcanos: the Campi Flegrei poly-caldera on the west and the Vesuvius stratovolcano on the east (Figure 1), that interfered with its Late Pleistocene -Holocene evolution (Iannace et al., 2015;Isaia et al., 2018;Santacroce et al., 2003).
During the Holocene, the morpho-evolutive trends of the coasts of the Gulf of Naples has been characterized by sudden coastal changes strongly related to the interplay between glacio-isostatic sea level rise and volcanic forcing (Cinque et al., 80 2011;Aucelli et al., , 2019Aucelli et al., , 2018aMattei et al., 2020). The latter was driven by the combined effects of volcanic eruptions with consequent landscape mantling by pyroclastic products, and vertical ground movements of metric entity related to sudden uplift for inflating and subsidence for deflating of the magmatic chamber. Since the historical times, the anthropic impact started interfering with these natural forcing often producing permanent modifications of the original coastal landforms (Aucelli et al., 2021;Pappone et al., 2019;Mattei et al., 2018), through mining activities and construction of port structures 85 and infilling. However, the major forcing factor to be taken into account as main cause of the recent coastal changes in the Gulf certainly is the local wave climate. In detail, main stormy events (Menna et al., 2007) with wave height values up to 4.8 m, are associated with atmospheric low-pressure systems and occur during winter (December -February). According to Saviano et al. (2019), high-frequency radar (HFR) data shows that the highest waves mainly approach from 180 • N to 210 • N this confirming a marked South -West directionality, as expected from the local morphology of the Gulf.

90
On the contrary, in late spring and summer periods the main wind regime is represented by breezes, with SSW direction and maximum speed values of 8 m/s (Menna et al., 2007), that produce low wave height values ranging from 0.4 to 0.6 m (Benassai et al., 1994;Buonocore et al., 2003;Saviano et al., 2019).
Considering the seasonal surface circulation, during winter, cyclonic and anticyclonic circulation systems alternate in the Gulf due to the interaction between the local wind forcing and the large-scale circulation of the Tyrrhenian Sea. In spring, when 95 a shallow and sharp seasonal thermocline is present, coastal upwelling is recorded and generates internal waves that propagate along the coast causing relevant mixing processes (de Ruggiero et al., 2018). In summer, the breeze forcing induces a relatively regular diurnal current oscillation (de Ruggiero et al., 2016). In this last season, surface currents typically rotated clockwise under the effect of land and sea breeze over an entire day (Uttieri et al., 2011). In autumn, the circulation is similar to the one recorded in winter.

Wind observations
Data from different facilities, devoted to meteorological in situ observations, were consulted to provide an historical analysis of the wind event that affected the study area in the last ten year, and to evaluate the accuracy of the simulation results. The following weather stations located in the Naples city were considered (see Figure 1):
-p02 (14 • 16'30.63"E -40 • 50'24.46"N) is part of the National Tidegauge Network, and it is managed by the Italian 110 Institute for Environmental Protection and Research (ISPRA). The weather station is located in the port of Naples at the Diaz pier.
Considering such data bases, an analysis of wind events was applied to 2010-2020 wind records in order to evaluate and classify the storms that approached from SW in the last 10 years. Therefore, the dataset was filtered for events coming from 202 • -242 • directions, highlighting those with velocity higher than 13.9 m/s (the lower limit of the "near gale" class in Beaufort 115 scale) and duration > 6 hours (according to Allen, 1991). Subsequently, for each of the selected events, the magnitude (M) was evaluated according to the following equation: where h is the event duration in hours and v is the wind speed in m/s (modified from Allen, 1991).

Wave observations 120
The in-situ sea waves observation was carried out with the wave recorder b01 (14 • 19'24.60"E -40 • 37'07.82"N), managed by the University of Naples "Parthenope" (Fig. 1a,c). The buoy is located in the Gulf of Naples near the Vervece islet, and it is operative since 2015. It is equipped with a BRIZO-X directional GNSS wave height sensor to record wave statistics as significant wave height (H s ), maximum wave height (H m ), peak wave period (T p ), mean zero upcrossing period (T m ), mean wave direction(D m ), wave spread (D s ), and wave spectra.

125
A subset of the dataset was used to evaluate the accuracy of the offshore wave simulations during the considered 2020-December storm event.

Atmospheric-marine numerical workflow
In order to characterize the meteo-marine scenario during the 2020-December storm event, an high spatial resolution model patterns.
In the present application, the simulated wind speed (W S s ) and direction (W D s ), the sea level pressure (SLP s ), the sig-150 nificant wave height (Hs s ), and the mean wave period (T m s ) and direction (Dm s ) were analyzed to characterize the 2020-December storm along the whole Gulf of Naples, using the maximum resolution dataset available for WRF and WW3 models.

Set-up evaluation
The reconstruction of the main erosive effects of the maximum water level time interval (t max ) occurred during the 2020-December storm in the study area, was performed with the following methodological steps.

155
1) The first step is aimed to detect the most-impacted coastal sectors in the Gulf of Naples during t max . To do this, each weather-marine parameter W S s , Hs s and SLP s was classified taking into account the 5 classes of intensity and/or impact on the coasts summarized in Table 1. Wind intensity was classified according to Beaufort scale, considering also the range of wind variability in the Gulf in the last 10 years. In particular, class 1 corresponds to fresh breeze, class 2 is strong breeze, classes 3 and 4 are near gale and class 5 is gale or higher than gale. class 1 is from calm to slight wave, class 2 is moderate wave, class 3 is rough wave, classes 4 and 5 form rough to very rough.
Atmospheric pressure was classified in equal intervals, starting form class 1 corresponding to high-pressure conditions, up to class 5 corresponding to low-pressure ones.
The total effect of the three classified parameters was evaluated by calculating the average values between the abovementioned classes, and consequently classified in five classes according to Table 1.

165
2) The second step was the calculation of the wind setup in the coastal sketches where the total effect was high, according to the following equations (Reeve et al., 2018): where U w is the wind speed (m/s), h is the water depth, ρ is the density of air (a) or water (w) and C w = air/water friction coefficient.

170
The maximum set-up at the downwind coast is where F is the fetch length in metres and η w is the wind set-up in metres.
3) The last step provided the calculation of the coastal setup as the sum of wind, wave and barometric setup, along the urbanized coastal sketches with maximum values of wind setup. The aim was to evaluate the flooding during t max only where 175 the storm had destructive effects on anthropic structures. In the surf zone, the rise of the mean water level at mean depth d x was calculated as following (Dean and Dalrymple, 1991): whereζ b is the wave setdown at the breaking depth, given by: where H b is breaking wave height and γ b is the breaker index.
For spilling type breakers on dissipative beaches the assumption commonly employed is that γ b remains a fixed ratio throughout the entire surf zone: where d b is the mean depth at breaking. The breaking index was calculated by various Authors. According to Kamphuis

185
(1991), the singificant breaking index is given by: where m is the slope of the seabed.
The barometric setup was calculated as follow: where ∆P a is the pressure variation during the event.

Results
The 2020-December storm came from SW, between 202 • N and 242 • N. As described in the following sections, the event was characterized by anomalous wind conditions that strongly influenced effects on the coast.

Historical wind events analysis 195
An historical analysis of wind events coming from SW between 2010 and 2020 was applied to data recorded in p01 and p02 weather stations to classify the event investigated in this paper. The results reported in Figures 3 and 4 show that the most intense events with a duration higher than six hours (red points in the figures) come from the NW sector during winter months, confirming previous studies (Menna et al., 2007;Saviano et al., 2019Saviano et al., , 2020.

Meteo-marine storm event reconstruction
The scientific workflow described in Section 3.3 supported the reconstruction of the 2020-December storm event. As shown in 205 Fig. 6, the study area was characterized by a low pressure front with a minimum SLP s value equal to 995 hPa at t max ± 1h during 28 December 2020.  On the same day, this intense atmospheric low-pressure system was accompanied by a widespread rainfall and strong winds coming from SW (W S s > 20 m/s) with gusts > 25 m/s, as shown in Figure 7.
Moreover, the sea state at t max ± 1h was characterized by high waves (about 4 m as recorded by the wave recorder b01).  The combination of these and other (eg. tide level) coastal dynamic agents caused a violent storm-surge in Naples that strongly flooded the city's waterfront overnight.
The validation of the accuracy of WRF and WW3 models was done through the comparison between the hourly numerical results in the time interval 27-30 December 2020, and the data recorded by the weather and wave recording stations (p01, p02 and b01). The results evidenced that the numerical models were in good agreement with the observations. In particular, the 215 root mean square error (RMSE) was about 0.23 m for the significant wave height in b01 and about 3.06 m/s and 3.35 m/s in case of wind speed in p01 and p02, respectively.

Discussion of the spatial classification of storm surge effects
The results of the spatial classification of the cumulative storm surge effects show that it was maximum (red class) along three cliffed urbanized sectors (Ischia, Capri and Sorrento Penisula) and three strongly anthropized urban low-coastal areas 220 (Posillipo, Port of Naples and Torre Annunziata), as shown in Figure 9d.
This result can be explained through the analysis of the wind and wave spatial distribution during 28th December event in the Gulf of Naples. As a matter of fact, the wind speed in the western part of the Gulf (Figure 9a) is lower than that in the Central-Eastern part. On the contrary, the significant wave height was maximum in proximity of the islands (Capri and Ischia) with values higher than 5 m and between 4 and 5 m along the other urban exposed sectors. The spatial fetch distribution of 225 the coastal areas, coupled with the wind speed variation, was such that the cumulative effect of the storm surge was maximum among the urbanized areas exposed to NW storms that is Naples coast (Posillipo and Port sectors) and Torre Annunziata.
However, the Naples coast (sectors 5 and 6 in Figure 10) was the most-exposed to the wind effects, as the wind setup calculation demonstrated. Consequently, the mean coastal setup was evaluated only in this sector, according to the procedure described before. In particular, the wave setup was calculated (equation 2) by using the high precision submerged DTM, 230 obtained from a Multibeam survey of the Neapolitan area, from which the slope in shallow waters was measured.  As a results of the calculation (step 3 in Section 3.4), the time-averaged water-level elevation at the coastline (coastal setup) during t max in the most exposed area of Naples waterfront was 1.6 m, resulting from the sum of wind setup (