Pseudo-Global Warming Projections of Extreme Wave Storms in Complex Coastal Regions - The Case of The Adriatic Sea

Climate Dynamics (2020) 55:2483–2509
https://doi.org/10.1007/s00382-020-05397-x
Pseudo‑global warming projections of extreme wave storms

in complex coastal regions: the case of the Adriatic Sea
Cléa Denamiel1 · Petra Pranić1 · Florent Quentin2 · Hrvoje Mihanović1 · Ivica Vilibić1
Received: 4 February 2020 / Accepted: 24 July 2020 / Published online: 3 August 2020
© Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract
This numerical work aims to better understand the behavior of extreme Adriatic Sea wave storms under projected climate
change. In this spirit, 36 characteristic events—22 bora and 14 sirocco storms occurring between 1979 and 2019, were
selected and ran in evaluation mode in order to estimate the skill of the kilometer-scale Adriatic Sea and Coast (AdriSC)
modelling suite used in this study and to provide baseline conditions for the climate change impact. The pseudo-global
warming (PGW) methodology—which imposes an additional climatological change to the forcing used in the evaluation
simulations, was implemented, for the very first time, for a coupled ocean–wave–atmosphere model and used to assess the
behavior of the selected storms under Representative Concentration Pathway (RCP) 4.5 and RCP 8.5 greenhouse gas projec-
tions. The findings of this experiment are that, on the one hand, the AdriSC model is found capable of reproducing both the
Adriatic waves associated with the 36 storms and the northern Adriatic surges occurring during the sirocco events and, on
the other hand, the significant wave heights and peak periods are likely to decrease during all future extreme events but most
particularly during bora storms. The northern Adriatic storm surges are in consequence also likely to decrease during sirocco
events. As it was previously demonstrated that the Adriatic extreme wind-wave events are likely to be less intense in a future
warmer climate, this study also proved the validity of applying the PGW methodology to coupled ocean–wave–atmosphere
models at the coastal and nearshore scales.
Keywords Adriatic Sea · Extreme storms · Pseudo-global warming · Extreme waves · Storm surges
1 Introduction Adriatic (e.g. Rizzi et al. 2017; Torresan et al. 2019). In

particular, Benetazzo et al. (2012) demonstrated that, in
In the past decade, atmospheric regional climate projec- the Adriatic Sea, the observed seasonal wave characteris-
tions—with typical resolutions of 12 to 50-km, have been tics could be numerically reproduced at the regional scale
used in the Mediterranean Sea and most particularly in the and the intensity of the strongest winds (i.e. the so-called
Adriatic Sea (Fig. 1a) in a wide range of impact studies, bora and sirocco events; Brzovíć and Strelec Mahović 1999;
such as the assessment of future wind (e.g. Bellafiore et al. Grisogono and Belušić 2009), and in consequence of the
2012; Belušić Vozila et al. 2019), wave (e.g. Lionello et al. waves, was likely to decrease by the end of the twenty-first
2012a; Benetazzo et al. 2012; Bonaldo et al. 2017) and storm century. The study also highlighted that additional research
surge (e.g. Lionello et al. 2012b; Androulidakis et al. 2015; should focus on distinguishing the effect of climate change
Mel et al. 2013) climates or the characterization of climate- on extreme bora and sirocco events. which lead to respec-
related hazards (i.e. flooding, extreme wave conditions, tively strong air–sea interactions (e.g. Pullen et al. 2006;
coastal vulnerability and erosion processes) in the northern Janeković et al. 2014; Ličer et al. 2016) impacting the
Adriatic thermohaline circulation (Vilibić et al. 2013) and
* Cléa Denamiel high waves potentially associated with storm surges in the
[email protected] northern Adriatic (Vilibić et al. 2017; Bajo et al. 2019)—
more particularly flooding in the Venice Lagoon (Trigo and
1
Institute of Oceanography and Fisheries, Šetalište I. Davies 2002; Cavaleri et al. 2010), or capable of moving
Meštrovića 63, 21000 Split, Croatia
large boulders (weighting up to a ton) along the Croatian
2
SeaTech, Ecole d’ingénieurs, Université de Toulon, coastline (Biolchi et al. 2019a, b).
CS 60584 83041 Toulon Cedex 9, France
13
Vol.:(0123456789)
2484 C. Denamiel et al.
Fig. 1 a Location of the differ-

ent Adriatic Sea wave measure-
ments along the Italian and Cro-
atian coastline and b selected
36 one-day extreme wave
events during the 1979–2019
period—depending on the wind
conditions (bora or sirocco)
schematized by arrows in the
map, used for the SWAN model
evaluation. The notation + 1 (b)
means that the 01/01/2015 event
is also selected
However, due to the complex orography of the elongated 2014; Tolle et al. 2014; Argueso et al. 2014; Rasmussen
semi-enclosed Adriatic basin surrounded by mountains and et al. 2014; Ban et al. 2014; Prein et al. 2015; Fosser et al.
associated with bathymetries evolving from a really shal- 2016; Kendon et al. 2017). Though, because of their tremen-
low and wide shelf (300 km in length with less than 80 m in dous computational costs, such high-resolution applications
depth) in the north to a deep pit (about 1200 m depth) in the have not yet been developed for coupled ocean–atmosphere
south (Fig. 1a), the evolution of the extreme bora and sirocco models.
wind patterns, and their impact on extreme waves and storm The present paper thus aims at both (1) implementing
surges, can only be achieved via high-resolution limited-area and testing the PGW methodology—already used in kilo-
atmospheric models (e.g. Pasarić et al. 2007; Klaić et al. meter-scale atmospheric climate studies, for a complex kil-
2009; Prtenjak and Belušić 2009; Prtenjak et al. 2010; Ric- ometer-scale atmosphere-wave-ocean modelling suite and
chi et al. 2016; Cavaleri et al. 2018) forcing sea-state and (2) quantifying the changes of the Adriatic wave extremes,
surge models (e.g. Cavaleri et al. 2010, 2019). Additionally, and their associated storm surges during sirocco events,
in the recent years, the use of very high-resolution (i.e. kilo- between present-day (for the 1979–2019 period) conditions
meter-scale, also known as convection-permitting) regional and future climate projections (for the 2060–2100 period)
climate models in atmospheric studies—in particular via for two greenhouse gas scenarios: Representative Concentra-
the so-called pseudo-global warming (PGW) downscaling tion Pathway 4.5 (RCP 4.5) and Representative Concentra-
method (Schär et al. 1996; Rasmussen et al. 2011), has been tion Pathway 8.5 (RCP 8.5). To this purpose, Sect. 2 first
proven to greatly improve the future projection of precipita- describes in detail the newly developed Adriatic Sea and
tions and convective storms (Pan et al. 2011; Kendon et al. Coast (AdriSC) modelling suite (Denamiel et al. 2019) used
13
Pseudo-global warming projections of extreme wave storms in complex coastal regions: the case… 2485
in this study, the set of sirocco and bora storms as well as the Mediterranean area and the Adriatic-Ionian region, as well
wave and sea-level measurements selected in the Adriatic as two different nested grids of 3-km and 1-km resolu-
Sea for the 1979–2019 period and, finally, the implementation used for both ROMS and SWAN models and cover-
tion of the PGW methodology for the ocean–wave–atmos- ing respectively the Adriatic-Ionian region (similarly to the
phere AdriSC model. Then, the evaluation of the AdriSC WRF 3-km grid) and the Adriatic Sea only.
model and the statistical analysis of the climate change In the nearshore module, the fully coupled ADCIRC-
impact on the Adriatic extreme waves and storm surges SWAN unstructured model (Dietrich et al. 2012)—cover-
under RCP 4.5 and RCP 8.5 scenarios are presented in ing the entire Adriatic Sea with resolutions ranging from
Sect. 3. Finally, the validity and limitations of the meth- 5-km in the deepest part of the domain to 10 m at the coast
odology and results derived in this numerical experiment (Table 1), is forced every minute with the off-line atmos-
are discussed in Sect. 4 and some conclusions about the pheric results of a dedicated high-resolution WRF 1.5-km
feasibility of using such an approach for long-term climate grid. In more details, the hourly results from the WRF 3-km
studies are presented. grid obtained with the basic module are first downscaled to
a WRF 1.5-km grid covering the Adriatic Sea and the hourly
sea surface elevation from the ROMS 1-km grid, the 10-min
2 Model, data and methods spectral wave results from the SWAN 1-km grid and finally
the 1-min results from the WRF 1.5-km grid are then used to
2.1 The AdriSC modelling suite force the unstructured mesh of the ADCIRC-SWAN model.
The AdriSC modelling suite is installed and fully tested
The Adriatic Sea and Coast (AdriSC) modelling suite on the European Centre for Middle-range Weather Forecast
(Denamiel et al. 2019) has been recently developed with (ECMWF) high-performance computing facilities. Table 1
the aim to accurately represent the processes driving the provides a summary of the AdriSC set-up while more details
atmospheric and oceanic circulation at different temporal can be found in Denamiel et al. (2019).
and spatial scales over the Adriatic and northern Ionian Sea. In this study, in order to reproduce the strongest histori-
In this spirit, the AdriSC modelling suite is based on two dif- cal wave storms which took place in the Adriatic Sea during
ferent modules: a basic module which provides atmospheric the 1979–2019 period and to assess their behavior under
and oceanic baroclinic circulation at the deep sea and coastal climate change projections (RCP 4.5 and RCP 8.5 scenar-
scales, and a dedicated nearshore module which is used to ios), the SWAN model—originally unused in the AdriSC
better reproduce atmospherically-driven extreme events. modelling suite, was set-up in both modules to be coupled
The basic module of the AdriSC suite rely on the with the ocean and atmosphere models (i.e. WRF, ROMS,
use and development of the Coupled Ocean–Atmos- ADCIRC). In the actual configuration, the third generation
phere–Wave–Sediment Transport (COAWST) modelling SWAN model is used with backward space and time propa-
system (Warner et al. 2010). It is built around the Model gation schemes, default initial condition, dissipation from
Coupling Toolkit (MCT) which exchanges data fields and whitecapping by Komen et al. (1984) and Madsen bottom
dynamically couples the Weather Research and Forecast- friction (Madsen et al. 1988). The wave model receives forc-
ing (WRF) atmospheric model, the Regional Ocean Mod- ing from WRF 3-km (wind fields) and ROMS 3-km/1-km
eling System (ROMS) and the Simulating WAves Nearshore (ocean surface currents, sea-level and friction) every 10 min
(SWAN) model. The basic module (Table 1) is set-up with in the basic module and from WRF 1.5-km (wind fields)
two different nested grids of 15-km and 3-km resolution and ADCIRC (ocean barotropic currents, sea-level and
used in the WRF model and covering respectively the central friction) every minute in the nearshore module. In addition,
Table 1 Summary of the AdriSC modelling suite main features

Models Basic module Nearshore module
Atmosphere Ocean Atmosphere Ocean
WRF ROMS-SWAN WRF ADCIRC-SWAN
Number of domains 2 2 1 1
Resolution 15 km 3 km 3 km 1 km 1.5 km 5 km to 10 m
Initial and boundary conditions ERA-Interim MEDSEA ERA-Interim WRF 3-km ROMS-SWAN 1-km
Duration of run (with d0 the day of 72 h 36 h
the event at 0 h) from d0 − 48 h to d0 + 24 h from d0 − 12 h to d0 + 24 h
Frequency of outputs Hourly 1-min
13
the computation of the bottom stress of the ocean models usually lasting 10–12 h—with rare occurrences as long as
(respectively ROMS and ADCIRC) was updated in order to 36 h, and often bring rain—sometimes mixed with Saha-
account for the spatial distribution of the sediment grain size ran dust (Cushman-Roisin et al. 2001). Both sirocco and
at the bottom of the Adriatic Sea extracted from the Adriatic bora episodes may vary in intensity and spatial coverage,
Seabed database (Jenkins et al. 2005) and the wave effects. extending either over the whole Adriatic or just a part or
For the evaluation runs, during the 1979–2019 period, in being conjoined, with sirocco blowing in the southern and
order to reproduce the historical storms as accurately as pos- bora in the northern Adriatic. Finally, in terms of extreme
sible, the basic module was set-up to run for 3 days. Initial conditions, as sirocco winds can produce extended fetch,
conditions and boundary forcing were provided the 6-hourly contrarily to the bora winds which are fetch-limited, the
ERA-Interim re-analysis fields (Dee et al. 2011; Balsamo largest wave heights were recorded in the northern Adriatic
et al. 2015), either the monthly or the daily re-analysis during extreme sirocco events (Leder et al. 1998; Bertotti
MEDSEA-Ocean fields (Pinardi et al. 2003), depending on et al. 2011; Pomaro et al. 2017). These waves can be asso-
whether the storm took place before or after the 1st of Janu- ciated with extreme storm surges in the Venice Lagoon,
ary 1987, and either the 6-hourly ERA-Interim wave fields the Gulf of Trieste (Fig. 1a) and the whole northern Adri-
or the hourly MEDSEA-Wave fields (Ravdas et al. 2018), atic (Lionello et al. 2012a; Međugorac et al. 2015).
depending on whether the storms took place before or after In this study, in order to perform the evaluation of the
the 1st January 2006. The nearshore module, forced by the AdriSC nearshore module, the choice of the studied extreme
results of the basic module, was set-up to run for the last day events was mostly driven by the available information and
and half of the basic module simulations. measurements recorded during the 1979–2019 period. For
As an in-depth sensitivity study of the impact of model the sirocco events, the 14 selected storms (Fig. 1b) were
resolution on the wave and storm surge representation is extracted from the long-term record of the Venice extreme
out of scope in this work, the advantages of using kilom- flooding (https: //www.comune .venezi a.it/it/conten t/le-acque
eter scale model in the atmosphere and unstructured meshes -alte-eccezionali). For the bora events, only 22 of the most
in the ocean is briefly discussed in Appendix 1. However, recent extreme storms were selected (Fig. 1b) as more wave
hereafter, only the last 24-h 1-min wave and sea-level results measurements became available in the Adriatic Sea at the
of the nearshore module extracted from respectively the end of the twentieth century (more details on the selected
unstructured SWAN model (referred as AdriSC unSWAN bora events are provided in Appendix 2). The majority of
in this study) and the ADCIRC model are analyzed. the selected bora events peaked in the northern Adriatic,
where bora wind is the strongest (Grisogono and Belušić
2.2 Documented historical extreme wave storms 2009). The set of wave measurements (Table 2)—used
in the Adriatic Sea to evaluate the skills of the AdriSC nearshore module to
reproduce the 36 selected storms, spans between 1979 and
In the Adriatic Sea, only two most frequent winds—bora 2019 and consists in 6 stations along the Italian coast—
and sirocco (Fig. 1a), can produce fetches large enough Acqua Alta tower from Pomaro et al. (2018), Venice 1
to drive extreme wave storms (Pomaro et al. 2017). The (only for the significant wave height measurements) from
bora is a cold east-northeast wind which flows through the Copernicus Marine Environment Monitoring Service
eastern Adriatic mountain passes, being particularly severe (ftp://my.cmems-du.eu/Core/INSITU_GLO_WAVE_REP_
along the Croatian coastline where its intensity sometimes OBSERVATIONS_013_045/history/mooring/) and Venice
surpasses 30 m s −1 and its gust reaches up to 70 m s −1 2, Ortona, Ancona, Monopoli from the Italian Data Buoy
(e.g. Jiang and Doyle 2005; Kuzmić et al. 2005; Belušić Network managed by ISPRA (Bencivenga et al. 2012), 4
and Klaić 2006; Gohm et al. 2008; Grisogono and Belušić stations along the Croatian coastline—Rovinj, Split, Ploče
2009; Trošić 2015). Continuous gale force (> 15 m s−1) and Dubrovnik, from the Croatian Hydrographic Institute
bora winds are most common during the cold season (Hrvatski hidrografski institut—HHI), and one station in
(November through March) and have an average dura- the middle of the northern Adriatic shelf—IVANA-A also
tion of 12 h with rare events that can last up to 2 days. from HHI. However, it should be noticed that storm cover-
The sirocco is a warm southeast wind originating from age is about 5 times higher from the Italian than the Croatian
North Africa, blowing over the Mediterranean Sea and measurements (Table 2). In addition, concerning the extreme
sometimes affecting the Adriatic Sea, being channelized storm surges associated with the sirocco events, two long-
by the surrounding mountains, with gust reaching more term hourly sea-level measurements extracted between 1979
than 30 m s −1 (e.g. Poje 1992; Jurčec et al. 1996; Pen- and 2019 from tide gauges located respectively in the Venice
zar et al. 2001; Pasarić and Orlić 2004). Although sirocco Lagoon (at Punta Della Salute, 45.4310° N and 12.3364° E,
winds are not as strong as the bora, continuous gale force maintained by ISMAR Venezia) and in the Gulf of Trieste
events occur more frequently between October and March, (at 45.6544° N and 13.7561° E, maintained by ISMAR
13
Table 2 Adriatic Sea wave buoy data available along the Italian and eu/)—part of the Coordinated Regional Climate Downs-
Croatian coastlines between 1979 and 2019 caling Experiment (CORDEX) initiative (https://esg-dn1.
Name Location Period of measure- Sampling Eventsa nsc.liu.se/search/cordex/) which coordinates the production
ments of climate change projections at the regional scale (Giorgi
et al. 2009; Giorgi and Gutowski 2015), is to provide cou-
Acqua Alta 12.5088° E 01.01.1979– 3h 33
45.3138° N 01.01.2018 pled ocean–atmosphere regional model results. The RCMs of
Venice 1 12.6627° E 01.06.2013– 1h 3 the Med-CORDEX ensemble are based on several numerical
44.9735° N 01.01.2015 models running in coupled or uncoupled mode and forced by
Venice 2 12.8330° E 01.06.2002– 1h 9 different Global Climate Models (GCMs). However, at the
44.9717° N 01.01.2015 time of this study, due to a reported issue with the CNRM-
Ortona 14.5056° E 01.07.1989– 1h 17 CM5 CMIP5 GCM forcing for the historical run (that removes
42.4150° N 19.05.2011
reliability of this product, https://www.medcordex.eu/warni
Ancona 13.7144° E 10.03.1999– 1h 13
ngs/Communication-Issue-Files_CNRM-CM5_histor ical
43.8297° N 07.11.2014
_6hLev_en.pdf), the only coupled results publicly avail-
Monopoli 17.3767° E 01.07.1989– 1h 19
40.9750° N 01.01.2015 able—with high enough temporal and spatial resolutions for
IVANA-A 13.2997° E 01.11.2007– 30 min 6 the historical period (1950–2005) and the two climate sce-
44.7262° N 13.05.2010 narios RCP 4.5 and RCP 8.5 (2006–2100), were those of the
Rovinj 13.4599° E 11.05.2018– 30 min 1 LMDZ4-NEMOMED8 RCM model (Hourdin et al. 2006;
45.0655° N 01.01.2019 Beuvier et al. 2010) forced by the IPSL-CM5A-MR GCM
Split 16.4650° E 29.10.2007– 30 min 5 model (simulations r1i1p1). These results—defined as two
43.4883° N 01.01.2019
continuous LMDZ4-NEMOMED8 simulations (1950–2100)
Ploče 17.3913° E 09.12.2016– 30 min 4
extending the historical run with either the RCP 4.5 or the RCP
43.0096° N 01.01.2019
8.5 runs, are referred as SCEN 4.5 and SCEN 8.5, respectively,
Dubrovnik 17.9667° E 06.04.2017– 30 min 3
42.6460° N 01.01.2019 and used hereafter to force the PGW simulations. The PGW
climatological changes derived from SCEN 4.5 and SCEN 8.5
a
Number of the selected 36 storms covered by the measurements between the 1979–2019 and the 2060–2100 periods are thus
tested in this study.
Trieste) were also used to evaluate the skill of the AdriSC Finally, the key development of this work is the extension
nearshore module. of the PGW method—which had till now only been used in
atmospheric models, to the ocean models and more particu-
2.3 Pseudo‑global warming methodology larly to the AdriSC modelling suite. For the atmosphere, as
described in many previous studies (Pan et al. 2011; Kendon
The two major challenges posed by performing kilometer- et al. 2014; Tolle et al. 2014; Argueso et al. 2014; Rasmussen
scale climate projection simulations are, on the one hand, the et al. 2014; Ban et al. 2015; Prein et al. 2015; Fosser et al.
relative slowness of the AdriSC modelling suite (a month of 2016; Kendon et al. 2017), the ERA-Interim air tempera-
results produced per day with the basic module alone), and ture (T ERAI ), relative humidity (RH ERAI ) and horizontal wind
on the other hand, the low temporal and spatial resolutions velocities 𝐕ERAI = (VxERAI , VyERAI ) defined on 37 atmospheric
(only few vertical levels for daily or monthly data) of the pressure levels (p) are modified between 1000 and 70 hPa
coupled regional climate model (RCM) results available to with respectively ΔT(tclim , x, y, p) , ΔRH(tclim , x, y, p)
provide boundary conditions to the WRF 15-km and ROMS and Δ𝐕 = (ΔVx (tclim , x, y, p), ΔVy (tclim , x, y, p)) derived
3-km models. To address these concerns, the projection of from SCEN 4.5 and SCEN 8.5 by subtracting the atmos-
the extreme Adriatic Sea wave events for the RCP 4.5 and pheric results from the 1979–2019 period to those of the
RCP 8.5 scenarios is performed, in this study, via a pseudo- 2060–2100 period and producing 6-hourly three-dimen-
global warming (PGW) method. The principle of the PGW sional climatologic changes for the 366 days of the year
simulations—as first introduced by Schär et al. (1996) and (tclim ). The WRF 15-km boundary and initial conditions of
described in details by Rasmussen et al. (2011), Kröner et al. the PGW simulations (T SCEN , RH SCEN , VxSCEN and VySCEN )
(2017) and Brogli et al. (2019a, b), is to impose an addi- are thus given by:
tional climatological change (e.g. a temperature change ΔT
representative of the increase in temperature between past T SCEN (t, x, y, p) = T ERAI (t, x, y, p) + ΔT(tclim , x, y, p)
and future climate) to the forcing used to produce the evalu- RH SCEN (t, x, y, p) = RH ERAI (t, x, y, p) + ΔRH(tclim , x, y, p)
ation runs. VxSCEN (t, x, y, p) = VxERAI (t, x, y, p) + ΔVx (tclim , x, y, p)
In the Mediterranean Sea, one of the specific aim of VySCEN (t, x, y, p) = VyERAI (t, x, y, p) + ΔVy (tclim , x, y, p)
the Med-CORDEX experiment (https://www.medcordex. (1)
13
In order to adjust the height of the surfaces of constant time- averaged ΔT presented in Fig. 2b clearly show that,
pressure to the temperature and relative humidity changes, near the surface of the earth, the differences in tempera-
the geopotential—depending on the virtual temperature ture between scenarios RCP 4.5 and RCP 8.5 reach more
TvSCEN , the ERA-Interim geopotential 𝜙ERAI at the reference than 1.5 °C (for both the ocean and the atmosphere). For the
pressure pref = 1000 hPa and the gas constant R, is recalcu- ocean, no significant difference between the two scenarios is
lated as follow: seen below depth of 1000 m. For the atmosphere, this differ-
ence only starts to decrease above 400 hPa and is minimized
p RTvSCEN above 100 hPa. In addition, the time variations of the spa-
∫pref
𝜙SCEN (t, x, y, p) = 𝜙ERAI (t, x, y, pref ) − dp
p tially-averaged ΔT for scenario RCP 8.5 (Fig. 2c) highlights
(2) that the temperature change imposed to the atmosphere at
Finally, the 2-m air temperature change ΔTS derived from 2 m height is, most of a year, at least 0.5 °C higher than the
SCEN 4.5 and SCEN 8.5 runs is used to adjust the ERA- one imposed to the sea surface temperature. Finally, Fig. 2d,
Interim surface (ground and 2-m air) temperatures (TSERAI ) e present the vertical variations of the temporally-averaged
such as: ΔT along the southern and western boundaries of both the
atmosphere and ocean models and illustrate the importance
TSSCEN (t, x, y) = TSERAI (t, x, y) + ΔTS (tclim , x, y) (3) of using spatially varying temperature changes for realis-
tic climate simulations. In Fig. 3, the surface distribution
The developed methodology for the ocean follows the of the temporally-averaged RCP 8.5 changes show that: for
principles of the PGW for the atmosphere. In this study, the atmosphere, the orography plays a major role in terms
the MEDSEA ocean temperature (T MEDSEA ) , salinity of the intensity of the changes (i.e. the strongest increase
(SMEDSEA ) and currents (VxMEDSEA , VyMEDSEA ) defined on in temperature, decrease in relative humidity and change
72 unevenly spaced vertical levels (z) , are thus modified in wind speed are generally found at the highest altitudes),
with respectively ΔT(tclim , x, y, z) , ΔS(tclim , x, y, z) and and, for the ocean, the changes imposed to the Adriatic and
Δ𝐕 = (ΔVx (tclim , x, y, z), ΔVy (tclim , x, y, z)) derived from northern Ionian Seas (i.e. strongest increase in temperature
SCEN 4.5 and SCEN 8.5 ocean results to produce daily cli- and salinity) do not correspond to the changes imposed in
matologic changes (tclim ) for the 366 days of the year. The the western side of the domain where the strongest changes
ROMS 3-km boundary and initial conditions of the PGW in current speed occur. Concerning the sea surface elevation,
simulations (T SCEN and SSCEN ) are thus given by: the RCP 8.5 changes are mostly negative and only of the
order of a few centimeters (with a maximum of 8 cm). Given
T SCEN (t, x, y, z) = T MEDSEA (t, x, y, z) + ΔT(tclim , x, y, z)
that on the one hand, the open boundary of the LMDZ4-
SSCEN (t, x, y, z) = SMEDSEA (t, x, y, z) + ΔS(tclim , x, y, z) NEMOMED8 model (similarly to all the Med-CORDEX
VxSCEN (t, x, y, z) = VxMEDSEA (t, x, y, z) + ΔVx (tclim , x, y, z) simulations, Adloff et al. 2018) does not properly include
VySCEN (t, x, y, z) = VyMEDSEA (t, x, y, z) + ΔVy (tclim , x, y, z) the projected Atlantic sea-level changes, but just takes into
(4) account the thermosteric effects and, on the other hand, the
thermal stretching is balanced by the haline shrinking, these
In the ocean, the static stability depends on the density
results are in accordance with the estimated − 7 cm to 13 cm
(𝜌) and the vertical variations of the local potential density
expected in the Mediterranean Sea (Tsimplis et al. 2008;
(𝜎n ) such as:
Jordà and Gomis 2013; Gualdi et al. 2013). Thus, for real-
1 𝛿𝜎n istic sea-level projections, mass change-induced sea-level
E(x, y, z) = − (5) increase—approximated to 50–60 cm in the Mediterranean
𝜌(x, y, z) 𝛿z
till 2100 (Jordà and Gomis 2013), should be added to the
The stability of the ocean forcing (at the boundaries presented PGW sea-level estimates.
and for the initial condition) is thus ensured by imposing In addition to the changes imposed to the ERA-I and
ESCEN ≥ 0 at all vertical levels. Finally, the sea surface eleva- MEDSEA forcing presented in the previous paragraphs, the
tion change Δssh derived from SCEN 4.5 and SCEN 8.5 volume mixing ratio of five atmospheric gases (carbon diox-
runs is used to adjust the MEDSEA surface layer (sshMEDSEA ) ide, methane, nitrous oxide and chlorofluorocarbons 11 and
such as: 12) used in the evaluation runs is modified in the scenario
runs using projected values (Bernstein et al. 2008) averaged
sshSCEN (t, x, y) = sshMEDSEA (t, x, y) + Δssh(tclim , x, y)
between 2060 and 2100 (Table 3). Further, the historical
(6)
monthly Adriatic Sea river discharges are climatologically
The temperature changes (ΔT) imposed at the bounda- changed for the RCP 4.5 and RCP 8.5 scenarios (Fig. 4)
ries of both the ocean and atmosphere models are illus- following the study of Macias et al. (2018). Concerning the
trated in Fig. 2. The vertical variations of the spatially- and waves, the forcing used in the evaluation simulations were
13
Fig. 2 a Spatial domain and boundaries of the WRF 15-km model day of a year (DOY) of the spatially-averaged 2-m air (in green) and
and, within the red box, the ROMS 3-km model. b Vertical variations sea-level (in blue) climatologic temperature changes ΔT for scenario
of the spatially- and temporally-averaged temperature changes ΔT for RCP 8.5. Vertical structure of the temporally-averaged temperature
scenarios RCP 4.5 and RCP 8.5 following pressure level in the atmos- changes ΔT (RCP 8.5) imposed at the southern and western bounda-
phere and depth in the ocean. c Time evolution depending on the ries of d the WRF 15-km model and e the ROMS 3-km model
kept unchanged for the scenario runs as the required data in climate change mode, for both RCP 4.5 and RCP 8.5
needed to apply the PGW methodology to the waves was not scenarios, imposing the PGW methodology (presented in
available. However, since the open boundary of the ROMS Sect. 2.3).
3-km grid is located at least 400-km south of the Strait of
Otranto, the wave field within the Adriatic basin is not con-
sidered to be highly affected by the propagation of these 3 Results
forcing. Finally, as this study aims to estimate the impact of
climate change on atmospherically-driven extreme events 3.1 Evaluation of the AdriSC nearshore module
and not to forecast future storms, the tidal forcing imposed wave component
for the evaluation runs was also kept unchanged for the sce-
nario runs. To assess the skill of the AdriSC unSWAN model, the last
To summarize, the set of 108 runs used in this study con- 24-h of the 1-min significant wave height, peak wave period
sists in 36 Adriatic Sea wave storm simulations (selected and mean wave direction results are extracted from the 36
in Sect. 2.2) carried out with the AdriSC modelling suite simulations carried out in evaluation mode at the 11 loca-
(described in Sect. 2.1)—for 3 days within the general tions of the wave stations presented in Sect. 2.2. The data are
module (i.e. coupled WRF-ROMS-SWAN) and 1.5 day analyzed in three steps (see Figs. 5, 6, 7). First, the overall
within the nearshore module (i.e. WRF 1.5-km and cou- behavior of the model is presented as a scatter plot (Figs. 5a,
pled ADCIRC-unSWAN), in evaluation mode first and then 6a, 7a) for the entire set of simulations and measurements.
13
Fig. 3 Surface distribution of the temporally-averaged RCP 8.5 the ROMS 3-km domain. The variations of the sea surface elevation
changes of a temperature (ΔT ), relative humidity (ΔRH ) and wind (Δssh) RCP 8.5 changes are presented in c as temporally-averaged
speed (ΔV ) in the atmosphere for the WRF 15-km domain and b tem- surface distributions and time-varying open boundary conditions
perature (ΔT ), salinity (ΔS ) and current speed (ΔV ) in the ocean for
Then, the quantile–quantile distributions of the wave param- On the whole (Figs. 5, 6, 7), for the 36 studied storm
eters are displayed separately for the Italian and Croatian events, the unSWAN model is in good agreement with the
wave stations (Figs. 5b, 6b, 7b) and, finally, the performance available wave measurements (significant height, peak
of the unSWAN model wave distributions during bora and period and mean direction): (1) in the scatter plots, the
sirocco events (Figs. 5c, 6c, 7c) is illustrated with violin points with higher density (in red) are mostly located along
plots (Hintze and Nelson 1998). the reference lines, (2) the quantile–quantile distributions for
13
Table 3 Atmospheric gas Carbon dioxide Methane Nitrous oxide Chlorofluorocarbon

volume mixing ratios used for CO2 (ppmv) CH4 (ppbv) N2O (ppbv)
the evaluation and scenario runs CFC-11 (pptv) CFC-12 (pptv)
Evaluation 379 1774 319 251 538

RCP 4.5 528 1680 365 105 242
RCP 8.5 762 3470 408 99 231
ppmv parts-per-million volume, ppbv parts-per-billion volume, pptv parts-per-trillion volume
Fig. 4 Monthly climatologic

changes (in percentage)
imposed to the Adriatic Sea
river discharges for scenario
runs RCP 4.5 and RCP 8.5
Fig. 5 Evaluation of the AdriSC

unSWAN significant wave
height distributions against
measurements a for all the
available stations and selected
storm events as a scatter plot
showing the density (number of
occurrences #) with hexago-
nal bins, b separately for the
Italian and Croatian stations as
a quantile–quantile plot, and
c separately for the bora and
sirocco events as violin plot
distributions
13
Fig. 6 As in Fig. 5, but for the

peak wave period
both the Croatian and Italian stations also follow the refer- result from the fact that some measurements were provided
ence lines, except for the mean direction which is not well as integer values), (2) the model slightly underestimates (up
reproduced for the Croatian stations, and (3) the shapes of to 1 s) the values below 5 s and above 9 s for the Italian sta-
the violin plots for the unSWAN model results are similar to tions (Fig. 6b), and (3) the mean and median of the model
those obtained for the measurements during both bora and distribution for the bora events (Fig. 6c) are overestimated
sirocco events. (6.88 s vs. 6.61 s and 7.67 s vs. 7.10 s, respectively). Finally,
However, in more detail, some discrepancies between for the mean wave direction, the major problem is the very
the measurements and the unSWAN model results can be large underestimation (up to 200°) of the values between 0°
noticed. For the significant wave height (Fig. 5), (1) the and 200° for the Croatian stations (Fig. 7b).
spread of the scatter plot (Fig. 5a) increases from about In a nutshell, the unSWAN model seems to have more
0.25 m up to 2 m when reaching the highest values, (2) difficulties to represent the wave conditions during bora
the model slightly overestimates (up to 0.25 m) the values events than during sirocco events, which means that the
between 2 and 4 m for both the Croatian and Italian stations WRF 1.5-km model is most probably overestimating the
and considerably overestimates (up to 1 m) the values above intensity of the bora winds. Further, the model is capable of
4 m for the Croatian stations (Fig. 5b), and (3) the mean and reproducing the intensity of the extreme wave events (see
median of the model distributions (Fig. 5c) are also over- quantile–quantile distributions) but not their timing (see
estimated for the bora (2.41 m vs. 2.17 m and 2.52 m vs. spread of the scatter plots), and have better agreement with
2.13 m, respectively) and for the sirocco (2.13 m vs. 2.05 m measurements along the Italian coast than along the Croatian
and 1.94 m vs. 1.67 m, respectively) events. coast. For the last point, the analysis of each Croatian station
For the peak wave period (Fig. 6), (1) similarly to the sig- (not shown here) reveals that the mismatching of the model
nificant wave height, the spread of the scatter plot (Fig. 6a) for the wave directions between 0° and 200° principally
increases from about 1 s up to 6 s when reaching the high- occurs at the IVANA-A and Dubrovnik locations which, as
est values (to be noted: the discontinuities seen in the plot the model is in good agreement with the data for the other
13

mean wave direction
nine stations, likely results from a problem with the meas- and mean wave directions—calculated for each storm over
urements at these two locations. Beside these limitations, the the last 24-h results of the AdriSC nearshore module. Then,
evaluation of the unSWAN model has shown that the newly the climate change impact on the wave extremes is given by
added wave component of the AdriSC modelling suite can the differences (referred hereafter as climate adjustments)
be used to reproduce the historical Adriatic wave storms in maximum significant wave heights, maximum peak wave
with a good level of accuracy. periods and mean wave directions between the climate
change simulations (with the RCP 4.5 and RCP 8.5 sce-
3.2 Impact of climate change on Adriatic extreme narios treated separately) and the evaluation runs. Finally,
waves the median and root-mean-square (RMS) of these climate
adjustments are calculated for the ensembles of the 22 bora
With the aim to quantify the climate change impact on (Fig. 9) and the 14 sirocco (Fig. 10) events. The analysis
the Adriatic extreme wave events under both RCP 4.5 and of the baseline conditions (Fig. 8) shows that the typical
RCP 8.5 projections for the 2060–2100 period, two kind of significant wave heights and peak wave periods are above
results are statistically analyzed: the spatial variations of the 3.5 m and 8 s, respectively. For the bora events, this particu-
extreme wave conditions (Figs. 8, 9, 10), and the temporal larly applies to the Italian coast between 42° N and 45° N
variations of the wave parameters at chosen locations along of latitude, peaking between 44° N and 45° N latitude with
the Adriatic Sea (Figs. 11, 12, 13, 14). the respective values of 5 m and 10 s. This is the result of
The spatial analysis of the extreme wave conditions con- the maximum in both bora speed and outreach, coming off
sists first in defining the baseline conditions (Fig. 8), which the Croatian city of Senj at latitude 44.99° N (the Senj Jet,
are presented as the median over the ensembles of the 22 Grisogono and Belušić 2009). The typical wave propagation
bora and the 14 sirocco storm simulations in evaluation (pre- for the analyzed bora episodes is mostly towards south-west
sent climate) mode. The considered parameters are the maxi- and west in the northern Adriatic and north-westward in
mum significant wave heights, maximum peak wave periods the south. The latter indicates that majority of the selected
13
Fig. 8 Baseline (present climate) plots defined as the median, over the entire Adriatic Sea, of the maximum significant wave heights, the maxi-
mum peak wave periods and the mean wave directions of the 22 bora (top panels) and 14 sirocco (bottom panels) events
severe bora events peaked in the northern Adriatic, while For the sirocco events (Fig. 10), the typical significant
being mostly conjoined with sirocco conditions in the heights and peak periods are also mostly decreased in the
southern Adriatic. For the sirocco events, wave heights are northern Adriatic for both RCP 4.5 and RCP 8.5 scenarios
substantial in the open Adriatic Sea for the entire domain (i.e. small or negative median of the climate adjustments)
(except close to the Italian shoreline and in the coastal Croa- but increased (i.e. positive median of the climate adjust-
tian area), peaking with values up to 6 m associated with ments) for the rest of the domain.
wave periods of 10 s between 44° N and 45° N latitude off For the RCP 8.5 scenario, the northern Adriatic
the Croatian islands and coast. The typical wave propagation decreases can reach up to 0.5 m in significant height
for sirocco events is towards north and north-west. and 0.8 s in peak period and are associated with large
Typical significant wave heights and peak wave periods RMS above 0.6 m and 0.5 s, respectively. Furthermore,
are foreseen, in the future climate, to overall decrease during the direction of the sirocco waves seems to be totally
bora events (Fig. 9), for both RCP 4.5 and RCP 8.5 scenar- unchanged (both median and RMS of the climate adjust-
ios. Negative median of the climate adjustment is projected ments are low) for both RCP 4.5 and RCP 8.5 scenarios.
over the entire Adriatic domain, except at its southern part The analysis of the temporal variations of the wave
in the RCP 4.5 scenario. For the RCP 8.5 scenario, the nega- parameters during the selected Adriatic wave storms is
tive median of the climate adjustments reaches up to 1 m based on the 1-min unSWAN series extracted at 5 open
in significant height and 1 s in peak period off the Italian Adriatic Sea locations ( O1–O5, Figs. 11, 12, 13, 14). A
coast between 41°N and 43°N latitude and is associated with comparison between the distributions obtained in the
larger RMS values surpassing 0.8 m and 1 s, respectively. evaluation and future climate (for RCP 4.5 and RCP 8.5
The direction of the bora winds is also affected over the scenarios) simulations is performed, separately for the
entire domain (up to 50° change of direction for the RCP ensemble of the 22 bora and 14 sirocco events. The results
8.5 scenario), but mostly along the Italian coastline between are presented as a combination of scatter and probabil-
42° N and 43° N latitude and regions where strong wind ity density function (PDF) plots for the significant wave
shear occur in the bora jets. height and peak wave period parameters (Figs. 11, 12, 13),
13
Fig. 9 Climate change (RCP

4.5 and RCP 8.5) impact on the
waves defined as the median
and root-mean-square (RMS)
of the climate adjustments (sce-
nario minus evaluation results)
of the maximum significant
heights, the maximum peak
periods and the mean wave
directions of the 22 bora events
and as polar histogram plots for the mean wave direction wave periods and significant wave heights during bora events
(Figs. 12, 13, 14). are likely to decrease under climate change projections:
For the bora events (Figs. 11, 12), the scatter plots at
the five chosen locations (Fig. 11) reveal that the overall • concerning the significant wave height distributions, the
distribution of the peak wave period vs. the significant wave values are consistently lowered in the future climate,
height is not significantly modified under climate change with a minimum of 0.25 m at location O 5 (northern most
projections and is presenting a strong linear relationship at part of the Adriatic Sea) for the RCP 4.5 scenario and a
all locations (i.e. the peak periods tend to increase linearly maximum of 2 m at location O4 (where strongest bora
with the significant wave heights), with a little spread in the wind are likely to blow) for the RCP 8.5 scenario while
northern Adriatic only (locations O 4 and O
5). However, the the tail is generally becoming less heavy under RCP 8.5
analysis of the PDF distributions confirms that both peak scenario at locations O 3, O4 and O5 (i.e. the probability
13

14 sirocco events
of significant wave heights above 3 m is largely reduced), ods above 7 s is largely reduced), but does not present
but does not significantly change in the southern Adri- major changes for RCP 4.5 scenario, except at location
atic (locations O1 and O2), at locations under the sirocco O5 (where the tail is clearly less heavy), and at location
influence, for both RCP 4.5 and RCP 8.5 scenarios; O1 (where the tail is more heavy).
• concerning the peak wave period distributions, the values
are also lowered at locations O 2, O3, O4 and O
5, with a Regarding the mean wave direction distributions
minimum of 0.5 s at location O 2 and a maximum of 1.5 (Fig. 12), the most significant changes appeared at location
s at location O3, both obtained for the RCP 4.5 scenario O3 where the waves primarily propagated westward in the
while, as for the significant wave height, the tail is gen- evaluation mode, while they shift southward and south-east-
erally becoming less heavy under RCP 8.5 scenario at ward for both RCP 4.5 and RCP 8.5 scenarios. This behavior
all locations (i.e. the probability of the peak wave peri- can also be noticed, in a smaller measure, at locations O2,
13
Fig. 11 Peak wave period (Tp)

vs. significant wave height (Hs)
distributions derived from the
1-min AdriSC unSWAN evalu-
ation and future climate projec-
tion (RCP 4.5 and RCP 8.5)
results of the 22 bora events and
extracted at 5 open Adriatic Sea
locations (O1 to O5). The results
are presented as a combination
of scatter plots displaying the
distributions of the peak period
vs. the significant height
O4 and O5 (Fig. 12), where however the main direction of changes are to be expected concerning the sirocco wave
propagation is unchanged. Finally, at location O1, the most parameters under RCP 4.5 and RCP 8.5 scenarios:
noticeable changes in direction occur for the RCP 4.5 sce-
nario, where the south-westward waves are shifted to north- • concerning the significant wave height distributions, the
westward and south-eastward, but the main direction of values are mostly unchanged, even though slightly (about
propagation (i.e. north north-westward) is also unchanged. 0.25 m in average) increased in the southern Adriatic Sea
For the sirocco events (Figs. 13, 14), as for the bora and decreased in the northern Adriatic Sea while the tail
events, the scatter plots at the five chosen locations (Fig. 13) is, however, generally becoming less heavy under RCP
reveal that the overall distribution of the peak wave period 8.5 scenario in the northern Adriatic Sea at locations O4
vs. the significant wave height is not substantially modi- and O5 (i.e. the probability of significant wave heights
fied under climate change projections. However, these dis- above 3 m is reduced) but slightly heavier or unchanged
tributions do not present a strong linear relationship in the for the remaining locations under RCP 8.5 scenario and
northern Adriatic and show that, at locations O3, O4 and for all locations under RCP 4.5 scenario;
O5, for peak periods above 7 s, significant wave heights can • concerning the peak wave period distributions, the
vary between 1 and 7 m. The analysis of the PDF distribu- values are also mostly unchanged at all locations and
tions reveals that, compared to the bora events, less dramatic for both climate change scenarios, but the respective
13
Fig. 12 Mean wave direction

distributions derived from
the 1-min AdriSC unSWAN
evaluation and climate projec-
results of the 22 bora events
and presented as rose plots at 5
open Adriatic Sea locations ( O1
to O5)
probabilities are increased in the southern Adriatic in evaluation mode remains unchanged under both RCP
(locations O 1, O 2 and O
3) while the tail is generally 4.5 and RCP 8.5 scenarios.
becoming slightly less heavy under RCP 8.5 scenario To summarize, the spatial variations of the extremes and
in the northern Adriatic (i.e. the probability of the peak the 1-min time series—extracted along the open Adriatic
wave periods above 9 s is reduced) but does not present Sea, reveal that, under warming climate change (for both
major changes for RCP 4.5 scenario. RCP 4.5 and RCP 8.5 scenarios), significant wave heights
and peak wave periods are likely to, on the one hand,
Finally, the mean wave direction distributions (Fig. 14) strongly decrease over the entire domain with a south-
show that the north-westward main direction of the waves eastward shift of direction in the central Adriatic during the
13
Fig. 13 As in Fig. 11, but for

the 14 sirocco events
extreme bora events, and, on the other hand, decrease in presented results are in good agreement with other wave cli-
the northern Adriatic with no significant change in direc- mate studies, which all envisage a decrease of wave heights
tion during the extreme sirocco events. These results are in the Adriatic Sea, in particular concerning sirocco events
in good agreement with the study of Belušić Vozila et al. (Benetazzo et al. 2012; Lionello et al. 2012a; Bonaldo et al.
(2019) who estimated the possible future changes in wind 2017; Pomaro et al. 2017).
speed over the Adriatic region, for the 2041–2070 period,
from an ensemble of 19 high-resolution (0.11°) CORDEX 3.3 Impact of climate change on northern Adriatic
simulations and found that overall the mean wind speed as storm surges
well as the number of storms is reduced under the RCP 8.5
scenario for both bora and sirocco conditions. However, In terms of the climate change impact on storm surges, the
Belušić Vozila et al. (2019) also highlights an increase of flooding of the coastal cities along the Adriatic coast—and
the bora mean wind speed in the northern Adriatic, which most particularly in the northern Adriatic, is known to be
is not in accordance with the presented results (Figs. 9, 11). driven by extreme sirocco conditions (e.g. Robinson et al.
Although selection of bora events differs between the two 1973; Cavaleri 2000; Cavaleri et al. 2010; Raicich 2015;
studies, this result may indicate the limitation of the PGW Medugorac et al. 2015), such as those selected in Sect. 2.2
methodology which can only be used to assess how past which led to the highest water levels recorded in Venice
storms would behave under climate change. Additionally, the lagoon between 1979 and 2019. The atmospherically-driven
13
Fig. 14 As in Fig. 12, but for

the 14 sirocco events
extreme sea-level changes in the northern Adriatic under the ADCIRC model results—extracted at the locations of
RCP 4.5 and RCP 8.5 scenarios and following the PGW the two tide gauges (Sect. 2.2) from the last 24-h results of
methodology can thus be assessed in this study with the the 14 sirocco simulations carried out in evaluation mode.
ensemble of the selected 14 sirocco events. To be noted, (1) as the bathymetry used in the ADCIRC
The AdriSC ADCIRC model capability to reproduce model may be imprecise and may use a different vertical ref-
the storm surges in the northern Adriatic and more specifi- erence level than the tide gauges, the local mean sea-levels
cally in the Venice Lagoon and the Gulf of Trieste is first of the model results at Venice and Trieste locations were
assessed with a sea-level distribution quantile–quantile anal- adjusted by adding the difference between the measured and
ysis (Fig. 15a) of the available hourly measurements and the modelled mean values calculated for the ensemble of
13
Fig. 15 Analysis of the northern

Adriatic storm surge distribu-
tions during the 14 sirocco
events: a quantile–quantile anal-
ysis of the AdriSC ADCIRC
results and the measurements at
Venice and Trieste tide-gauge
stations, b baseline sea-level
plot defined as the median of
the maximum sea-levels gener-
ated by each storm, c, d sea-
level distributions derived from
the 1-min AdriSC ADCIRC
evaluation and climate projec-
results and extracted respec-
tively at Venice and Trieste tide
gauge stations
the 14 sirocco events; (2) as tides play an important role is given by the differences (referred hereafter as climate
on extreme storm surges, they were not removed from the adjustments) in maximum sea-levels between the climate
sea-level signals, and (3) as this study only aims to present change simulations (with the RCP 4.5 and RCP 8.5 scenarios
the climate change impact on the sea-level distributions dur- treated separately) and the evaluation runs. The median and
ing sirocco events in the northern Adriatic Sea and not to root-mean-square (RMS) of these climate adjustments are
reproduce individual storm surges, the model evaluation can calculated for the ensemble of 14 sirocco events (Fig. 16). In
be performed with the presented quantile–quantile analysis. terms of the results, the baseline condition (Fig. 15b) shows
Similarly to the analysis performed for the waves in that typical storm surges during sirocco events reach 1.4 m
Sect. 3.2, the impact of climate change for the 2060–2100 in the Venice Lagoon and 0.9 m in the Gulf of Trieste and,
period on the northern Adriatic storm surges under both under climate change conditions (Fig. 16), are decreased (i.e.
RCP 4.5 and RCP 8.5 projections, is then estimated via the negative median of the climate adjustments) by more than
statistical analysis of two kind of results: the spatial varia- 0.25 m over the entire northern Adriatic domain for both
tions of the maximum sea-levels obtained for each storm RCP 4.5 and RCP 8.5 scenarios, and by more than 0.35 m in
(Figs. 15b, 16) and the 1-min sea-level temporal variations the Venice Lagoon for the RCP 8.5 scenario. The decrease
at both the Venice Lagoon and the Gulf of Trieste tide gauge of the storm surges in the future climate is also associated
locations (Fig. 15c, d). with an important variability (i.e. RMS) of about 0.35 m in
For the spatial variations, the sea-level baseline condi- average for both RCP 4.5 and RCP 8.5 scenarios and reach-
tion (Fig. 15b) is defined as the median over the ensemble ing more than 0.45 m in the Venice Lagoon for RCP 8.5
of the 14 selected storm simulations in evaluation mode and scenario.
their maximum sea-levels—calculated for each storm over These results are confirmed by the temporal analysis of
the last 24-h results of the AdriSC nearshore module. As the sea-level distributions (for the evaluation runs and the
in Sect. 3.2, the climate change impact on the storm surges RCP 4.5 and RCP 8.5 projections) at Venice and Trieste
13
Fig. 16 Climate change (RCP

4.5 and RCP 8.5) impact on the
northern Adriatic extreme sea-
levels defined as the median and
root-mean-square (RMS) of the
climate adjustments (scenario
minus evaluation results) of the
maximum sea-levels for the 14
sirocco events
tide gauge locations (Fig. 15c, d). It appears that, at both these results do not imply that flooding of the Venice city
locations under RCP 4.5 and RCP 8.5 climate change, the will be less likely in the future.
modes of the sea-level distributions are decreased by about
0.25 m while the tails of the distributions are less heavy
(i.e. the probability of storm surges above 0.75 m is greatly
decreased) and the maximum surges are reduced by about 4 Discussion and conclusions
0.25 m. Finally, the probability of sea-levels below -0.5 m
(which was the minimum reached in evaluation mode) is Understanding how climate change could impact extreme
largely increased which reveals that, under the climate wave storms—one of the most devastating natural hazards
change projections, some of the strong sirocco events simu- occurring along the littoral, is of crucial importance for the
lated in evaluation mode might take more time to develop (or future of coastal communities. However, in order to prop-
never developed) as storms of lower intensity in the RCP 4.5 erly capture such atmospherically-driven extreme events
and RCP 8.5 simulations. Thus, for these events, the wind- and their repercussions on the coast (e.g. extreme wave
wave set-up—building up in the northern Adriatic during and storm surges, coastal erosion, etc.), the implementa-
the 3 days of simulation in evaluation mode, was decreased tion of computationally expansive kilometer-scale coupled
as it built up over a smaller period with weaker winds (or ocean–wave–atmosphere models is required for each spe-
did not built up at all). cific coastal region considered. Consequently, the classical
To summarize, under climate change projections with approach used in climate studies (i.e. 30 years of evaluation
imposed PGW methodology, the northern Adriatic sirocco run, 50 years of historical run and 100 years of scenario
storm surges are not only likely to decrease but also to be runs) is too impractical and costly to be applied for this
less frequent, which is in accordance with the analysis of kind of investigations. As an alternative, the pseudo-global
the sirocco wind and storm surge projections (Lionello warming (PGW) methodology—originally developed for
et al. 2012a; Androulidakis et al. 2015; Belušić Vozila et al. kilometer-scale atmospheric studies, presents the advantage
2019). However, as the subsidence of the Venice Lagoon of generating at a limited computational cost (i.e. once the
is not considered in this study, and the sea-level change is PGW climatological forcing is created, each event can be
imposed as a mean effect for the entire 2060–2100 period in simulated over a short period of time) an ensemble of storms
the PGW methodology ignoring the Atlantic global sea-level used to statistically assess the impact of climate change on
rise (see discussion about sea surface elevation in Sect. 2.3), extreme events.
13
The principal novelty of this study was thus to imple- the statistical approach consisting in running ensembles
ment the PGW method within the AdriSC modelling suite— of short simulations for extreme events seems to provide
a multi-model chain dedicated to the study of the Adriatic robust results. However, it should be noticed that, (1) as
Sea (Denamiel et al. 2019), which couples the atmospheric only a small ensemble of storms was selected, the simulated
model WRF at 15-km, 3-km and 1.5-km of resolution with wave and storm surge distributions may not be fully repre-
the ocean and wave models ROMS and SWAN at 3-km sentative of neither the historical Adriatic extreme events
and 1-km of resolution and the unstructured ADCIRC and between 1979 and 2019 nor their future projections for the
SWAN models with up to 10 m resolution along the coast. 2060–2100 period, (2) as the simulations were performed
In particular, a lot of attention was paid to the best way to over a three-day period, the last 24-h results analyzed in
represent not only the ocean forcing (salinity, temperature, this study might be influenced by the imposed initial con-
currents) but also the sea-levels, the rivers, the waves and ditions and finally, (3) as the same ensemble of storms is
the tides. As the Adriatic Sea collects up to a third of the used in evaluation and climate projection modes, the fre-
Mediterranean Sea fresh water budget (Ludwig et al. 2009), quency of the extreme events cannot be analyzed with the
the river discharge forcing—which is projected to be highly PGW method. Additionally, due to the lack of reliable
impacted by climate change, was modified following the ocean–atmosphere Med-CORDEX runs at the time of this
study of Macias et al. (2018). However, due to the known study, the climatological forcing used in the presented PGW
uncertainties and/or lack of data linked to sea-level rise and simulations were derived from a single model instead of an
wave climate projections over the entire Mediterranean Sea ensemble of Regional Climate Models (RCMs) which would
(Ruti et al. 2016; Adloff et al. 2018), these forcing were have provide more robust climate change projections. The
kept unchanged in this work. Additionally, in order to only generation of such a forcing, or potentially of an ensemble of
account for the impact of global warming in the storm surge PGW simulations—as already adopted by the atmospheric
analysis, the tidal forcing also remained untouched. If these community (e.g. Li et al. 2019), could be achieved in a near
approximations have little consequences on the extreme future when more ocean–atmosphere RCMs will become
wave results (as the boundary of the SWAN 3-km model is available in the Mediterranean Sea.
far enough from the studied area), they can impact the pre- Notwithstanding the limitations of the PGW approach
sented storm surge distributions due to the non-linear nature implemented in the AdriSC modelling suite, this study dem-
of their interactions with the nearshore water depths— onstrates that such a method has the potential to be applied
including local sea-levels and tides (Johns et al. 1985; Speer to kilometer-scale ocean–atmosphere models for long- or
and Aubrey 1985; Parker 1991; Zhang et al. 2017; Yang short-term simulations which can be embedded in the tra-
et al. 2019). The impact of sea-level rise was thus ignored in ditional CORDEX sub-domains including an oceanic com-
this work and only the atmospherically-driven storm surge ponent (e.g. Ruti et al. 2016; Tinker et al. 2016; Zou and
distributions were analyzed. Zhou 2016; Han et al. 2019). For coastal areas such as the
The other important component of this numerical work Adriatic Sea, this can open the door to a better understand-
was to provide a thorough evaluation of the AdriSC modeling of the climate change impacts on various processes (e.g.
ling suite skill to reproduce historical extreme events and dense water formation, ocean sea surface thermal interac-
to provide meaningful climate projections via the PGW tions during storms and hurricanes, etc.) closely depending
method. To achieve these goals an ensemble of 22 bora and on the air–sea feedback mechanism.
14 sirocco extreme wind-wave events were selected between
1979 and 2019 and ran in both evaluation and climate pro- Acknowledgements The contribution of all the organisations that
kindly provided the observations used in this study—Copernicus
jection (for RCP 4.5 and RCP 8.5 scenarios) modes. The Marine Environment Monitoring Service, data buoy network of the
evaluation of the distributions of both the wave parameters Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA,
(significant height, peak period and mean direction) against Italy), Hrvatski hidrografski institut (HHI, Croatia) and Istituto di
11 stations located along the Adriatic coast, and the storm Scienze Marine (ISMAR, Italy), is acknowledged. Special thanks to
Roman Brogli from the Eidgenössische Technische Hochschule (ETH
surges against the Venice and Trieste tide gauges, revealed Zürich, Switzerland), Laurent Li from the Université Pierre et Marie
that overall the AdriSC model is capable of reproducing Curie (Paris, France) and Srđan Čupić from HHI (Croatia) who pro-
the selected 36 historical extreme events. Concerning the vided respectively insights concerning the pseudo-global warming
climate simulations with the PGW method, the wave and (PGW) method, the regional climate model ocean–atmosphere results
(from LMDZ4-NEMOMED8) used in this study and the wave measure-
storm surge distributions—showing a general decrease of ments along the Croatian coastline. Acknowledgement is also made for
the extreme bora and sirocco intensity for both RCP 4.5 and the support of the European Centre for Middle-range Weather Forecast
RCP 8.5 scenarios, follow the previous studies published (ECMWF) staff, in particular Xavier Abellan and Carsten Maass, as
in the Adriatic Sea (Benetazzo et al. 2012; Lionello et al. well as for ECMWF’s computing and archive facilities used in this
research. Finally, the authors would like to thank the two anonymous
2012a; Androulidakis et al. 2015; Bonaldo et al. 2017; reviewers for their valuable comments. This work has been supported
Pomaro et al. 2017; Belušić Vozila et al. 2019) and thus
13
by projects ADIOS (Croatian Science Foundation Grant IP-2016-06- study, the quantile–quantile plots (Fig. 17) displaying the
1955) and ECMWF Special Project (The Adriatic decadal and inter- performance of the unSWAN and SWAN 1-km models for
annual oscillations: modelling component).
an ensemble of 11 stations and 36 extreme events (presented
Author contributions IV and CD contributed to the study conception in Sect. 2) illustrate the impact of the model resolution on
and design. Material preparation and data collection were done by HM the significant wave height and peak wave period. They
and CD. Set-up of the models and simulations were performed by FQ show that for extreme conditions the unSWAN model pro-
and CD. Analysis of the results and production of the figures were vides better results than the SWAN 1-km model, particularly
performed by PP, IV and CD. The first draft of the manuscript was
written by CD and all authors commented on previous versions of the concerning the peak wave period, even though wave heights
manuscript. All authors read and approved the final manuscript. are slightly overestimated by the unSWAN model and per-
fectly represented by the SWAN 1-km between 1 and 3 m.
Funding ADIOS project: Croatian Science Foundation Grant The second argument presented here is that the inclusion
IP-2016-06-1955. European Centre for Middle-range Weather Fore- of the precise coastline and bathymetry (e.g. the precise geo-
cast (ECMWF) Special Project: The Adriatic decadal and inter-annual
oscillations: modelling component). morphology of the Venice Lagoon) as well as the numerous
small islands (more than 1000 islands, isles, islets, rocks are
Availability of data and material The model results and the measure- identified along the Croatian coast) acting as barriers during
ments used to produce this article can be obtained under the Open extreme storm events, is crucial to properly simulate wave
Science Framework (OSF) FAIR data repository https://osf.io/7d6jq/ transformation and consequently storm surges in the north-
(https://doi.org/10.17605/osf.io/7d6jq).
ern Adriatic. This geomorphological impact is illustrated
Code availability Codes used to produce this article can be obtained (Fig. 17) with spatial plots of maximum significant wave
under the Open Science Framework (OSF) FAIR data repository https height and maximum peak wave period for the unSWAN
://osf.io/7d6jq/ (https://doi.org/10.17605/osf.io/7d6jq). model and for the difference between the unSWAN and
SWAN 1-km models during the sirocco storm of the 22nd
Compliance with ethical standards of December 1979 and the extreme bora event of the 7th of
February 2012. The effect of the geomorphology and the
Conflict of interest The authors declare that they have no conflict of islands located along the eastern Adriatic Sea on the wave
interest.
propagation and transformation is particularly seen on the
unSWAN results which show that both maximum significant
wave height and maximum peak wave period are smaller
Appendix 1 along the Croatian coasts than in the rest of the domain.
The difference plots highlight that, on the one hand, the
Despite an in-depth sensitivity study of the model resolution overall increase in maximum significant wave heights in the
impact on extreme event representation being out of scope unSWAN simulations (between 0.5 and 1.5 m compared to
of this paper, the reasons why kilometer-scale resolution in SWAN 1-km) mostly results from the increase in resolution
the atmosphere and meter-scale resolution in the ocean are (and thus in intensity) of the atmospheric wind forcing and,
a pre-requisite to meaningful wave storm modelling in the on the other hand, for the presented bora event, the proper
Adriatic region are briefly discussed hereafter. representation of the islands in the unSWAN simulations
The first argument is based on recently published studies dramatically influenced the reproduction of the maximum
supporting the idea that Adriatic extreme events can only be peak periods (up to 5 s difference with the SWAN 1-km
captured with resolutions higher than those generally used in model).
Regional Climate Models (RCMs). For the atmosphere, the Finally, the use of unstructured meshes is not only useful
capability of climate models to reproduce mesoscale wind for pure oceanographic purposes but also for the study of
phenomena has been tested and atmospheric resolution—via other processes driven by extreme events such as, for exam-
a better reproduction of the orography and consequently the ple, the motion of boulders during sirocco storms which
enhancement of jet flows on finer grids, has been found to requires to perfectly reproduce the wave transformation at a
be one of the most important model characteristics known to meter-scale resolution to derive the transport of these blocks
impact wind speed driving extreme waves and storm surges weighting up to a ton (see Biolchi et al. 2019b).
in the Adriatic region (e.g. Belušić et al. 2017; Josipović
et al. 2018). Similarly, for the ocean, extreme wave or flood
hazard assessments have been found to only be achieved Appendix 2
with models capable to reproduce the proper geomorphol-
ogy (including complex coastline and bathymetry) of the As no catalogue of extreme historical bora events exists in
Adriatic Sea coastal regions (e.g. Cavaleri et al. 2010, 2019; the Adriatic region, an in-depth bibliographical research—
Torresan et al. 2019). Finally, based on the results of this presented in Table 4 and including meteorological bulletins
13
Fig. 17 Analysis of the sensitivity of the wave results to model reso- butions of unSWAN results and differences between unSWAN and
lution: with quantile–quantile distributions of the AdriSC SWAN SWAN 1-km results for the maximum significant wave height and
1-km and unSWAN results vs. measurements for the ensemble of 9 the maximum peak wave period during one extreme sirocco event (22
stations and 36 extreme events (left panels) and with spatial distri- December 1979) and one extreme bora event (7 February 2012)
(in Croatian), local newspapers, photographs and videos, was

thus undertaken to generate the ensemble of 22 historical bora
storms used in this study.
13
Table 4 References used to select the date of the 22 representative extreme bora events of the ensemble
Event Reference: research articles, meteorological bulletins, news reports and videos
24/11/1991 Brzović and Benković (1994)

28/12/1992 Ivančan-Picek and Tutiš (1996)
01/02/1994 Bilten iz područja meteorologije, hidrologije, primijenjene meteorologije i zaštite čovjekova
16/02/1997 okoliša, https://meteo.hr/proizvodi.php?section=publikacije&param=publikacije_publikacij
09/03/1998 e_dhmz&el=bilteni
28/03/1995 Brzović (1999)
30/03/1995
21/12/1998 Bajić and Glasnović (1999)
07/11/1999 Klaić et al. (2003)
14/11/2004 Belušić and Klaić (2006)
15/12/2007 https://www.istramet.hr/vijesti/15122006-snijeg-se-zadrzao-i-tokom-dana/
10/03/2010 https://www.tportal.hr/vijesti/clanak/orkanska-bura-otkinula-komad-krova-skole-20100310
02/03/2011 https://www.tportal.hr/vijesti/clanak/orkanska-bura-nosi-sve-pred-sobom-20110302
07/02/2012 Ličer et al. (2016)
11/02/2012
11/11/2013 https://rijeka.meteoadriatic.net/vremeplov-orkanska-bura-u-rijeci-i-na-kvarneru-11-11-2013/
31/12/2014 https://www.youtube.com/watch?v=0kNlrL_diFY
01/01/2015
05/03/2015 https://www.crometeo.hr/orkanska-bura-na-jadranu-foto-video/
17/01/2017 https://www.tportal.hr/vijesti/clanak/snazna-bura-u-istri-stvara-velike-teskoce-20170117
13/11/2017 https://www.tportal.hr/vijesti/clanak/bura-otrgnula-novi-brod-i-zabila-ga-u-mol-u-puli-foto-20171113
24/09/2018 https://www.tportal.hr/vijesti/clanak/orkanska-bura-srucila-se-na-split-kontejneri-setaju-ulicama-
problemi-u-pomorskom-prometu-ali-i-na-nebu-foto-20180924
References Geophys Res Lett 42:1165–1172. https: //doi.org/10.1002/2014G

L062588
Bellafiore D, Bucchignani E, Gualdi S, Carniel S, Djurdjević V,
Adloff F, Jordà G, Somot S, Sevault F, Arsouze T, Meyssignac B, Li
Umgiesser G (2012) Assessment of meteorological climate mod-
L, Planton S (2018) Improving sea-level simulation in Mediter-
els as inputs for coastal studies. Ocean Dyn 62:555–568. https://
ranean regional climate models. Clim Dyn 51:1167–1178. https
doi.org/10.1007/s10236-011-0508-2
://doi.org/10.1007/s00382-017-3842-3
Belušić D, Klaić ZB (2006) Mesoscale dynamics, structure and predict-
Androulidakis YS, Kombiadou KD, Makris CV, Baltikas VN, Kresteni-
ability of a severe Adriatic bora case. Meteorol Z 15:157–168
tis YN (2015) Storm surges in the Mediterranean Sea: variability
Belušić A, Prtenjak MT, Güttler I, Ban N, Leutwyler D, Schär C (2017)
and trends under future climatic conditions. Dyn Atmos Oceans
Near-surface wind variability over the broader Adriatic region:
71:56–82. https://doi.org/10.1016/j.dynatmoce.2015.06.001
insights from an ensemble of regional climate models. Clim Dyn
Argueso D, Evans JP, Fita L, Bormann KJ (2014) Temperature
50:4455–4480. https://doi.org/10.1007/s00382-017-3885-5
response to future urbanization and climate change. Clim Dyn
Belušić Vozila A, Güttler I, Ahrens B, Obermann-Hellhund A,
42:2183–2199. https://doi.org/10.1007/s00382-013-1789-6
Telišman Prtenjak M (2019) Wind over the Adriatic region
Bajić A, Glasnović D (1999) Impact of Adritic Bora on traffic. In:
in CORDEX climate change scenarios. J Geophys Res Atmos
Proceedings from the 4th European conference on applied mete-
124:110–130. https://doi.org/10.1029/2018JD028552
orology, Norrkoping, SMHI
Bencivenga M, Nardone G, Ruggiero F, Calore D (2012) The Italian
Bajo M, Međugorac I, Umgiesser G, Orlić M (2019) Storm surge and
Data Buoy Network (RON). Adv Fluid Mech IX, WIT Trans
seiche modelling in the Adriatic Sea and the impact of data
Eng Sci 74:321–332
assimilation. Q J R Meteorol Soc 145:2070–2084. https://doi.
Benetazzo A, Fedele F, Carniel S, Ricchi A, Bucchignani E, Sclavo
org/10.1002/qj.3544
M (2012) Wave climate of the Adriatic Sea: a future scenario
Balsamo G, Albergel C, Beljaars A, Boussetta S, Brun E, Cloke H,
simulation. Nat Hazards Earth Syst Sci 12:2065–2076. https://
Dee D, Dutra E, Muñoz-Sabater J, Pappenberger F, de Rosnay
doi.org/10.5194/nhess-12-2065-2012
P, Stockdale T, Vitart F (2015) ERA-Interim/Land: a global land
Bernstein L, Bosch P, Canziani O, Chen Z, Christ R, Riahi K (2008)
surface reanalysis data set. Hydrol Earth Syst Sci 19:389–407.
IPCC, 2007: climate change 2007: synthesis report. IPCC,
https://doi.org/10.5194/hess-19-389-2015
Geneva. ISBN 2-9169-122-4
Ban N, Schmidli J, Schär C (2014) Evaluation of the convection-
Bertotti L, Bidlot J-R, Buizza R, Cavaleri L, Janousek M (2011) Deter-
resolving regional climate modeling approach in decade-long
ministic and ensemble-based prediction of Adriatic Sea sirocco
simulations. J Geophys Res Atmos 119:7889–7907. https://doi.
storms leading to ‘acqua alta’ in Venice. Q J R Meteorol Soc
org/10.1002/2014JD021478
137:1446–1466. https://doi.org/10.1002/qj.861
Ban N, Schmidli J, Schär C (2015) Heavy precipitation in a changing
Beuvier J, Sevault F, Herrmann M, Kontoyiannis H, Ludwig W,
climate: does short-term summer precipitation increase faster?
Rixen M, Stanev E, Béranger K, Somot S (2010) Modeling the
13
Mediterranean Sea interannual variability during 1961–2000: Dietrich JC, Tanaka S, Westerink JJ, Dawson CN, Luettich RA Jr,
focus on the Eastern Mediterranean Transient. J Geophys Res Zijlema M, Holthuijsen LH, Smith JM, Westerink JG, Wes-
Atmos 115:C08017. https://doi.org/10.1029/2009JC005950 terink HJ (2012) Performance of the Unstructured-Mesh,
Biolchi S, Furlani S, Devoto S, Scicchitano G, Korbar T, Vilibić I, SWAN+ADCIRC Model in computing hurricane waves and
Šepić J (2019a) The origin and dynamics of coastal boulders surge. J Sci Comput 52:468–497. https://doi.org/10.1007/s1091
in a semi-enclosed shallow basin: a northern Adriatic case 5-011-9555-6
study. Mar Geol 411:62–77. https://doi.org/10.1016/j.marge Fosser G, Khodayar S, Berg P (2016) Climate change in the next
o.2019.01.008 30 years: what can a convection-permitting model tell us that
Biolchi S, Denamiel C, Devoto S, Korbar T, Macovaz V, Scicchitano we did not already know? Clim Dyn 48:1987–2003. https://doi.
G, Vilibić I, Furlani S (2019b) Impact of the October 2018 Storm org/10.1007/s00382-016-3186-4
Vaia on coastal boulders in the northern Adriatic Sea. Water Giorgi F, Gutowski WJ (2015) Regional dynamical downscaling and
11:2229. https://doi.org/10.3390/w11112229 the CORDEX initiative. Ann Rev Environ Resour 40(1):467–
Bonaldo D, Bucchignani E, Ricchi A, Carniel S (2017) Wind stormi- 490. https://doi.org/10.1146/annurev-environ-102014-021217
ness in the Adriatic Sea in a climate change scenario. Acta Adriat Giorgi F, Jones C, Asrar G (2009) Addressing climate information
58(2):195–208 needs at the regional level: the CORDEX framework. WMO Bull
Brogli R, Sørland SL, Kröner N, Schär C (2019a) Causes of future 58(3):175–183
Mediterranean precipitation decline depend on the season. Envi- Gohm A, Mayr GJ, Fix A, Giez A (2008) On the onset of bora and the
ron Res Lett 14:114017. https://doi.org/10.1088/1748-9326/ formation of rotors and jumps near a mountain gap. Q J R Mete-
ab4438 orol Soc 134:21–46. https://doi.org/10.1002/qj.206
Brogli R, Kröner N, Sørland SL, Lüthi D, Schär C (2019b) The role of Grisogono B, Belušić D (2009) A review of recent advances in
Hadley circulation and lapse-rate changes for the future European understanding the meso- and microscale properties of the
summer climate. J Clim 32:385–404. https://doi.org/10.1175/ severe Bora wind. Tellus A 61:1–16. https://doi.org/10.111
JCLI-D-18-0431.1 1/j.1600-0870.2008.00369.x
Brzović N (1999) Factors affecting the Adriatic cyclone and associated Gualdi S, Somot S, Li L, Artale V, Adani M, Bellucci A, Braun A,
windstorms. Contrib Atmos Phys 72:51–65 Calmanti S, Carillo A, Dell’Aquila A, Déqué M, Dubois C, Eli-
Brzović N, Benković M (1994) Severe Adriatic bora storms 1987– zalde A, Harzallah A, Jacob D, L’Hévéder B, May W, Oddo P,
1993. Croat Meteorol J 29:65–74. https://hrcak.srce.hr/69265 Ruti P, Sanna A, Sannino G, Scoccimarro E, Sevault F, Nav-
Brzovíć N, Strelec Mahović N (1999) Cyclonic activity and severe Jugo arra A (2013) The CIRCE simulations: regional climate change
in the Adriatic. Phys Chem Earth B 24(6):653–657. https://doi. projections with realistic representation of the Mediterranean
org/10.1016/S1464-1909(99)00061-1 Sea. Bull Am Meteorol Soc 94:65–81. https://doi.org/10.1175/
Cavaleri L (2000) The oceanographic tower Acqua Alta activity and BAMS-D-11-00136.1
prediction of sea states at Venice. Coast Eng 39(1):29–70. https Han G, Ma Z, Long Z, Perrie W, Chassé J (2019) Climate change
://doi.org/10.1016/S0378-3839(99)00053-8 on Newfoundland and Labrador shelves: results from a regional
Cavaleri L, Bertotti L, Buizza R, Buzzi A, Masato V, Umgiesser G, downscaled ocean and sea-ice model under an A1B forcing
Zampieri M (2010) Predictability of extreme meteo-oceano- scenario 2011–2069. Atmos Ocean 57(1):3–17. https://doi.
graphic events in the Adriatic Sea. Q J R Meteorol Soc 136:400– org/10.1080/07055900.2017.1417110
413. https://doi.org/10.1002/qj.567 Hintze HL, Nelson RD (1998) Violin plots: a box plot-density trace syn-
Cavaleri L, Abdalla S, Benetazzo A, Bertotti L, Bidlot J-R, Breivik Ø, ergism. Am Stat 52(2):181–184. https://doi.org/10.1080/00031
Carniel S, Jensen RE, Portilla-Yandun J, Rogers WE, Roland A, 305.1998.10480559
Sanchez-Arcilla A, Smith JM, Staneva J, Toledo Y, van Vled- Hourdin F, Musat I, Bony S, Braconnot P, Codron F, Dufresne JL, Fair-
der GPh, van der Westhuysen AJ (2018) Wave modelling in head L, Filiberti MA, Friedlingstein P, Grandpeix JY, Krinner
coastal and inner seas. Prog Oceanogr 167:164–233. https://doi. G, LeVan P, Li ZX, Lott F (2006) The LMDZ4 general circula-
org/10.1016/j.pocean.2018.03.010 tion model: climate performance and sensitivity to parametrized
Cavaleri L, Bajo M, Barbariol F, Bastianini M, Benetazzo A, Bertotti physics with emphasis on tropical convection. Clim Dyn 27:787–
L, Chiggiato J, Davolio S, Ferrarin C, Magnusson L, Papa A, 813. https://doi.org/10.1007/s00382-006-0158-0
Pezzutto P, Pomaro A, Umgiesser G (2019) The October 29, Ivančan-Picek B, Tutiš V (1996) A case study of a severe Adriatic
2018 storm in Northern Italy—an exceptional event and its mod- bora on 28 December 1992. Tellus A 48:357–367. https://doi.
eling. Prog Oceanogr 178:102178. https://doi.org/10.1016/j. org/10.3402/tellusa.v48i3.12065
pocean.2019.102178 Janeković I, Mihanović H, Vilibić I, Tudor M (2014) Extreme cooling
Cushman-Roisin B, Gačić M, Poulain P-M, Artegiani A (2001) Physi- and dense water formation estimates in open and coastal regions
cal oceanography of the Adriatic Sea: past, present and future. of the Adriatic Sea during the winter of 2012. J Geophys Res
Springer, New York Oceans 119:3200–3218. https://doi.org/10.1002/2014JC009865
Dee DP, Uppala SM, Simmons AJ, Berrisford P, Poli P, Kobayashi Jenkins C, Trincardi F, Hatchett L, Niedoroda A, Goff J, Signell R,
S, Andrae U, Balmaseda MA, Balsamo G, Bauer P, Bechtold McKinney K (2005) http://instaar.colorado.edu/~jenkinsc/dbsea
P, Beljaars ACM, van de Berg L, Bidlot J, Bormann N, Delsol bed/coverage/adriaticsea/adriatico.htm
C, Dragani R, Fuentes M, Geer AJ, Haimberger L, Healy SB, Jiang Q, Doyle JD (2005) Wave breaking induced surface wakes and
Hersbach H, Hólm EV, Isaksen L, Kållberg P, Köhler M, Matri- jets observed during a bora event. Geophys Res Lett 32:L17807.
cardi M, McNally AP, Monge-Sanz BM, Morcrette JJ, Park BK, https://doi.org/10.1029/2005GL022398
Peubey C, de Rosnay P, Tavolato C, Thépaut JN, Vitart F (2011) Johns B, Rao AD, Dubinsky Z, Sinha PC (1985) Numerical modelling
The ERA-Interim reanalysis: configuration and performance of of tide-surge interaction in the Bay of Bengal. Philos Trans R
the data assimilation system. Q J R Meteorol Soc 137:553–597. Soc Lond Ser A Math Phys Sci 313(1526):507–535. https://doi.
https://doi.org/10.1002/qj.828 org/10.1098/rsta.1985.0002
Denamiel C, Šepić J, Ivanković D, Vilibić I (2019) The Adriatic Sea Jordà G, Gomis D (2013) On the interpretation of the steric and mass
and Coast modelling suite: evaluation of the meteotsunami components of sea-level variability: the case of the Mediter-
forecast component. Ocean Model 135:71–93. https: //doi. ranean basin. J Geophys Res Oceans 118:953–963. https://doi.
org/10.1016/j.ocemod.2019.02.003 org/10.1002/jgrc.20060
13
Josipović L, Obermann-Hellhund A, Brisson E, Ahrens B (2018) Međugorac I, Pasarić M, Orlić M (2015) Severe flooding along the
Bora in regional climate models: impact of model resolution eastern Adriatic coast: the case of 1 December 2008. Ocean Dyn
on simulations of gap wind and wave breaking. Croat Meteorol 65:817–830. https://doi.org/10.1007/s10236-015-0835-9
J 53:31–42. https://hrcak.srce.hr/231266 Mel R, Sterl A, Lionello P (2013) High resolution climate projection
Jurčec V, Ivančan-Picek B, Tutiš V, Vukičević V (1996) Severe Adri- of storm surge at the Venetian coast. Nat Hazards Earth Syst
atic jugo wind. Meteorol Z 5:67–75 Sci 13:1135–1142. https://doi.org/10.5194/nhess-13-1135-2013
Kendon EJ, Fowler HJ, Roberts MJ, Chan SC, Senior CA (2014) Pan L-L, Chen S-H, Cayan D, Lin M-Y, Hart Q, Zhang M-H, Liu Y,
Heavier summer downpours with climate change revealed by Wang J (2011) Influences of climate change on California and
weather forecast resolution model. Nat Clim Change 4:570– Nevada regions revealed by a high-resolution dynamical down-
576. https://doi.org/10.1038/nclimate2258 scaling study. Clim Dyn 37:2005–2020. https://doi.org/10.1007/
Kendon EJ, Ban N, Roberts NM, Fowler HJ, Roberts MJ, Chan SC, s00382-010-0961-5
Evans JP, Fosser G, Wilkinson JM (2017) Do convection-per- Parker BB (1991) The Relative Importance of the various nonlinear
mitting regional climate models improve projections of future mechanisms in a wide range of tidal interaction (review). Tidal
precipitation change? Bull Am Meteorol Soc 98:79–93. https hydrodynamics. Wiley, New York, pp 237–268
://doi.org/10.1175/BAMS-D-15-0004.1 Pasarić M, Orlić M (2004) Meteorological forcing of the Adriatic:
Klaić ZB, Belušić D, Grubišić V, Gabela L, Ćoso L (2003) Mes- present vs. projected climate conditions. Geofizika 21:69–86
oscale airflow structure over the northern Croatian coast during Pasarić Z, Belušić D, Klaić ZB (2007) Orographic influences on the
MAP IOP15—a major bora event. Geofizika 20:23–60 Adriatic sirocco wind. Ann Geophys 25:1263–1267
Klaić ZB, Prodanov AD, Belušić D (2009) Wind measurements in Penzar B, Penzar I, Orlić M (2001) Vrijeme i klima hrvatskog Jadrana.
Senj: underestimation of true bora flows. Geofizika 26:245–252 Nakladna kuća. ‘Dr. Feletar’, Zagreb
Komen GJ, Hasselmann S, Hasselmann K (1984) On the exist- Pinardi N, Allen I, Demirov E, De Mey P, Korres G, Lascaratos A,
ence of a fully developed wind-sea spectrum. J Phys Le Traon P-Y, Maillard C, Manzella G, Tziavos C (2003) The
Oceanogr 14:1271–1285. https : //doi.org/10.1175/1520- Mediterranean ocean Forecasting System: first phase of imple-
0485(1984)014%3c1271:OTEOAF%3e2.0.CO;2 mentation (1998–2001). Ann Geophys 21:3–20. https://doi.
Kröner N, Kotlarski S, Fischer E, Lüthi D, Zubler E, Schär C (2017) org/10.5194/angeo-21-3-2003
Separating climate change signals into thermodynamic, lapse- Poje D (1992) Wind persistence in Croatia. Int J Climatol 12:569–586
rate and circulation effects: theory and application to the Euro- Pomaro A, Cavaleri L, Lionello P (2017) Climatology and trends of the
pean summer climate. Clim Dyn 48:3425–3440. https://doi. Adriatic Sea wind waves: analysis of a 37-year long instrumental
org/10.1007/s00382-016-3276-3 data set. Int J Climatol 37:4237–4250. https://doi.org/10.1002/
Kuzmić M, Janeković I, Ivančan-Picek B, Trošić T, Tomažić I (2005) joc.5066
Severe northeastern Adriatic bura events and circulation in Pomaro A, Cavaleri L, Papa A, Lionello P (2018) 39 years of direc-
greater Kvarner region. Croat Meteorol J 40:320–323 tional wave recorded data and relative problems, climatological
Leder N, Smirčić A, Vilibić I (1998) Extreme values of surface wave implications and use. Sci Data 5:180139. https: //doi.org/10.1038/
heights in the Northern Adriatic. Geofizika 15:1–13 sdata.2018.139
Li Y, Li Z, Zhang Z, Chen L, Kurkute S, Scaff L, Pan X (2019) Prein AF, Langhans W, Fosser G, Ferrone A, Ban N, Goergen K, Keller
High-resolution regional climate modeling and projection over M, Tölle M, Gutjahr O, Feser F, Brisson E, Kollet S, Schmidli
western Canada using a weather research forecasting model J, van Lipzig NPM, Leung R (2015) A review on regional
with a pseudo-global warming approach. Hydrol Earth Syst Sci convection-permitting climate modeling: demonstrations, pros-
23(11):4635–4659. https://doi.org/10.5194/hess-23-4635-2019 pects and challenges. Rev Geophys 53:323–361. https://doi.
Ličer M, Smerkol P, Fettich A, Ravdas M, Papapostolou A, Mantzi- org/10.1002/2014RG000475
afou A, Strajnar B, Cedilnik J, Jeromel M, Jerman J, Petan S, Prtenjak MT, Belušić D (2009) Formation of reversed lee flow over
Malačič V, Sofianos S (2016) Modeling the ocean and atmos- the north-eastern Adriatic during bora. Geofizika 26:145–155
phere during an extreme bora event in northern Adriatic using Prtenjak MT, Viher M, Jurković J (2010) Sea-land breeze develop-
one-way and two-way atmosphere–ocean coupling. Ocean Sci ment during a summer bora event along the north-eastern Adri-
12:71–86. https://doi.org/10.5194/os-12-71-2016 atic coast. Q J R Meteorol Soc 136:1554–1571. https://doi.
Lionello P, Cavaleri L, Nissen KM, Pino C, Raicich F, Ulbrich U org/10.1002/qj.649
(2012a) Severe marine storms in the Northern Adriatic: char- Pullen J, Doyle JD, Signell RP (2006) Two-way air–sea coupling: a
acteristics and trends. Phys Chem Earth 40(41):93–105. https study of the Adriatic. Mon Weather Rev 134:1465–1483. https
://doi.org/10.1016/j.pce.2010.10.002 ://doi.org/10.1175/MWR3137.1
Lionello P, Galati MB, Elvini E (2012b) Extreme storm surge and Raicich F (2015) Long-term variability of storm surge frequency in
wind wave climate scenario simulations at the Venetian litto- the Venice Lagoon: an update thanks to 18th century sea-level
ral. Phys Chem Earth 40(41):86–92. https://doi.org/10.1016/j. observations. Nat Hazards Earth Syst Sci 15:527–535. https://
pce.2010.04.001 doi.org/10.5194/nhess-15-527-2015
Ludwig W, Dumont E, Meybeck M, Heussner S (2009) River dis- Rasmussen R, Liu C, Ikeda K, Gochis D, Yates D, Chen F, Tewari M,
charges of water and nutrients to the Mediterranean and Black Barlage M, Dudhia J, Yu W, Miller K, Arsenault K, Grubišić
Sea: major drivers for ecosystem changes during past and V, Thompson G, Gutmann E (2011) High-resolution coupled
future decades? Prog Oceanogr 80(3–4):199–217. https://doi. climate runoff simulations of seasonal snowfall over Colorado:
org/10.1016/j.pocean.2009.02.001 a process study of current and warmer climate. J Clim 24:3015–
Macias D, Stips A, Garcia-Gorriz E, Dosio A (2018) Hydrologi- 3048. https://doi.org/10.1175/2010JCLI3985.1
cal and biogeochemical response of the Mediterranean Sea Rasmussen R, Ikeda K, Liu C, Gochis D, Clark M, Dai A, Gutmann E,
to freshwater flow changes for the end of the 21st century. Dudhia J, Chen F, Barlage M, Yates D, Zhang G (2014) Climate
PLoS One 13(2):e0192174. https: //doi.org/10.1371/journ change impacts on the water balance of the Colorado headwaters:
al.pone.0192174 High-resolution regional climate model simulations. J Hydrome-
Madsen OS, Poon Y-K, Graber HC (1988) Spectral wave attenuation by teorol 15:1091–1116. https: //doi.org/10.1175/JHM-D-13-0118.1
bottom friction: theory. In: Proceedings of the 21st international Ravdas M, Zacharioudaki A, Korres G (2018) Implementation and
conference on coastal engineering, ASCE, pp 492–504 validation of a new operational wave forecasting system of
13
the Mediterranean Monitoring and Forecasting Centre in the I: a multi-model chain for the definition of climate change hazard
framework of the Copernicus Marine Environment Monitoring scenarios. Water 11:1157. https://doi.org/10.3390/w11061157
Service. Nat Hazards Earth Syst Sci 18:2675–2695. https://doi. Trigo IF, Davies TD (2002) Meteorological conditions associated
org/10.5194/nhess-18-2675-2018 with sea surges in Venice: a 40 year climatology. Int J Climatol
Ricchi A, Miglietta MM, Falco PP, Benetazzo A, Bonaldo D, Berga- 22:787–803. https://doi.org/10.1002/joc.719
masco A, Sclavo M, Carniel S (2016) On the use of a coupled Trošić T (2015) The onset of a severe summer bora episode near
ocean–atmosphere–wave model during an extreme cold air out- Oštarijska Vrata Pass in the Northern Adriatic. Meteorol Atmos
break over the Adriatic Sea. Atmos Res 172–173:48–65. https:// Phys 127:649–658. https://doi.org/10.1007/s00703-015-0393-1
doi.org/10.1016/j.atmosres.2015.12.023 Tsimplis M, Marcos M, Somot S, Barnier B (2008) Sea-level forc-
Rizzi J, Torresan S, Zabeo A, Critto A, Tosoni A, Tomasin A, Mar- ing in the Mediterranean Sea between 1960 and 2000. Glob
comini A (2017) Assessing storm surge risk under future sea- Planet Change 63(4):325–332. https://doi.org/10.1016/j.glopl
level rise scenarios: a case study in the North Adriatic coast. acha.2008.07.004
J Coast Conserv 21:453–471. https://doi.org/10.1007/s1185 Vilibić I, Šepić J, Proust N (2013) Weakening thermohaline circula-
2-017-0517-5 tion in the Adriatic Sea. Clim Res 55(3):217–225. https://doi.
Robinson AR, Tomasin A, Artegiani A (1973) Flooding of Ven- org/10.3354/cr01128
ice: phenomenology and prediction of the Adriatic Sea storm Vilibić I, Šepić J, Pasarić M, Orlić M (2017) The Adriatic Sea: a long-
surge. Q J R Meteorol Soc 99:688–692. https://doi.org/10.1002/ standing laboratory for sea-level studies. Pure Appl Geophys
qj.49709942210 174:3765–3811. https://doi.org/10.1007/s00024-017-1625-8
Ruti PM, Somot S, Giorgi F, Dubois C, Flaounas E, Obermann A, Warner JC, Armstrong B, He R, Zambon JB (2010) Development
Dell’Aquila A, Pisacane G, Harzallah A et al (2016) Med- of a Coupled Ocean-Atmosphere-Wave-Sediment Transport
CORDEX initiative for Mediterranean climate studies. Bull Am (COAWST) modeling system. Ocean Model 35:230–244. https
Meteorol Soc 97:1187–1208. https://doi.org/10.1175/BAMS- ://doi.org/10.1016/j.ocemod.2010.07.010
D-14-00176.1 Yang W, Yin B, Feng X, Yang D, Gao G, Chen H (2019) The effect
Schär C, Frei C, Luthi D, Davies HC (1996) Surrogate climate-change of nonlinear factors on tide-surge interaction: a case study of
scenarios for regional climate models. Geophys Res Lett 23:669– Typhoon Rammasun in Tieshan Bay, China. Estuar Coast Shelf
672. https://doi.org/10.1029/96GL00265 Sci 219:420–428. https://doi.org/10.1016/j.ecss.2019.01.024
Speer PE, Aubrey DG (1985) A study of non-linear tidal propaga- Zhang H, Cheng W, Qiu X, Feng X, Gong W (2017) Tide-surge
tion in shallow inlet/estuarine systems. Part II: theory. Estuar interaction along the east coast of the Leizhou Peninsula, south
Coast Shelf Sci 21(2):207–224. https://doi.org/10.1016/0272- China sea. Cont Shelf Res 142:32–49. https://doi.org/10.1016/j.
7714(85)90097-6 csr.2017.05.015
Tinker J, Lowe J, Pardaens A, Holt J, Rosa Barciela (2016) Uncertainty Zou L, Zhou T (2016) A regional ocean–atmosphere coupled model
in climate projections for the 21st century northwest European developed for CORDEX East Asia: assessment of Asian sum-
shelf seas. Prog Oceanogr 148:56–73. https://doi.org/10.1016/j. mer monsoon simulation. Clim Dyn 47:3627–3640. https://doi.
pocean.2016.09.003 org/10.1007/s00382-016-3032-8
Tolle MH, Gutjahr O, Busch G, Thiele JC (2014) Increasing bioen-
ergy production on arable land: does the regional and local cli- Publisher’s Note Springer Nature remains neutral with regard to
mate respond? Germany as a case study. J Geophys Res Atmos jurisdictional claims in published maps and institutional affiliations.
119:2711–2724. https://doi.org/10.1002/2013JD020877
Torresan S, Gallina V, Gualdi S, Bellafiore D, Umgiesser G, Carniel S,
Sclavo M, Benetazzo A, Giubilato E, Critto A (2019) Assessment
of climate change impacts in the North Adriatic coastal area. Part
13

Pseudo-Global Warming Projections of Extreme Wave Storms in Complex Coastal Regions - The Case of The Adriatic Sea

Uploaded by

Copyright:

Available Formats

Pseudo-Global Warming Projections of Extreme Wave Storms in Complex Coastal Regions - The Case of The Adriatic Sea

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pseudo-Global Warming Projections of Extreme Wave Storms in Complex Coastal Regions - The Case of The Adriatic Sea

Uploaded by

Copyright:

Available Formats

Climate Dynamics (2020) 55:2483–2509

Pseudo‑global warming projections of extreme wave storms

1 Introduction Adriatic (e.g. Rizzi et al. 2017; Torresan et al. 2019). In

Fig. 1 a Location of the differ-

Table 1 Summary of the AdriSC modelling suite main features

Table 3 Atmospheric gas Carbon dioxide Methane Nitrous oxide Chlorofluorocarbon

Evaluation 379 1774 319 251 538

ppmv parts-per-million volume, ppbv parts-per-billion volume, pptv parts-per-trillion volume

Fig. 4 Monthly climatologic

Fig. 5 Evaluation of the AdriSC

Fig. 6 As in Fig. 5, but for the

Fig. 7 As in Fig. 5, but for the

Fig. 9 Climate change (RCP

Fig. 10 As in Fig. 9, but for the

Fig. 11 Peak wave period (Tp)

Fig. 12 Mean wave direction

Fig. 13 As in Fig. 11, but for

Fig. 14 As in Fig. 12, but for

Fig. 15 Analysis of the northern

Fig. 16 Climate change (RCP

(in Croatian), local newspapers, photographs and videos, was

24/11/1991 Brzović and Benković (1994)

References Geophys Res Lett 42:1165–1172. https:​ //doi.org/10.1002/2014G​

You might also like

References Geophys Res Lett 42:1165–1172. https: //doi.org/10.1002/2014G