Skip to main content

Pseudo-global warming projections of extreme wave storms in complex coastal regions: the case of the Adriatic Sea


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 projections. 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.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15
Fig. 16

Availability of data and material

The model results and the measurements used to produce this article can be obtained under the Open Science Framework (OSF) FAIR data repository (

Code availability

Codes used to produce this article can be obtained under the Open Science Framework (OSF) FAIR data repository (


Download references


The contribution of all the organisations that kindly provided the observations used in this study—Copernicus Marine Environment Monitoring Service, data buoy network of the Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA, Italy), Hrvatski hidrografski institut (HHI, Croatia) and Istituto di Scienze Marine (ISMAR, Italy), is acknowledged. Special thanks to Roman Brogli from the Eidgenössische Technische Hochschule (ETH Zürich, Switzerland), Laurent Li from the Université Pierre et Marie Curie (Paris, France) and Srđan Čupić from HHI (Croatia) who provided respectively insights concerning the pseudo-global warming (PGW) method, the regional climate model ocean–atmosphere results (from LMDZ4-NEMOMED8) used in this study and the wave measurements along the Croatian coastline. Acknowledgement is also made for the support of the European Centre for Middle-range Weather Forecast (ECMWF) staff, in particular Xavier Abellan and Carsten Maass, as well as for ECMWF’s computing and archive facilities used in this research. Finally, the authors would like to thank the two anonymous reviewers for their valuable comments. This work has been supported by projects ADIOS (Croatian Science Foundation Grant IP-2016-06-1955) and ECMWF Special Project (The Adriatic decadal and inter-annual oscillations: modelling component).


ADIOS project: Croatian Science Foundation Grant IP-2016-06-1955. European Centre for Middle-range Weather Forecast (ECMWF) Special Project: The Adriatic decadal and inter-annual oscillations: modelling component).

Author information

Authors and Affiliations



IV and CD contributed to the study conception and design. Material preparation and data collection were done by HM and CD. Set-up of the models and simulations were performed by FQ and CD. Analysis of the results and production of the figures were 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 manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Cléa Denamiel.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.


Appendix 1

Despite an in-depth sensitivity study of the model resolution impact on extreme event representation being out of scope of this paper, the reasons why kilometer-scale resolution in the atmosphere and meter-scale resolution in the ocean are a pre-requisite to meaningful wave storm modelling in the Adriatic region are briefly discussed hereafter.

The first argument is based on recently published studies supporting the idea that Adriatic extreme events can only be captured with resolutions higher than those generally used in Regional Climate Models (RCMs). For the atmosphere, the capability of climate models to reproduce mesoscale wind phenomena has been tested and atmospheric resolution—via a better reproduction of the orography and consequently the enhancement of jet flows on finer grids, has been found to be one of the most important model characteristics known to impact wind speed driving extreme waves and storm surges 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 with models capable to reproduce the proper geomorphology (including complex coastline and bathymetry) of the Adriatic Sea coastal regions (e.g. Cavaleri et al. 2010, 2019; Torresan et al. 2019). Finally, based on the results of this study, the quantile–quantile plots (Fig. 17) displaying the performance of the unSWAN and SWAN 1-km models for an ensemble of 11 stations and 36 extreme events (presented in Sect. 2) illustrate the impact of the model resolution on the significant wave height and peak wave period. They show that for extreme conditions the unSWAN model provides better results than the SWAN 1-km model, particularly concerning the peak wave period, even though wave heights are slightly overestimated by the unSWAN model and perfectly represented by the SWAN 1-km between 1 and 3 m.

Fig. 17
figure 17

Analysis of the sensitivity of the wave results to model resolution: with quantile–quantile distributions of the AdriSC SWAN 1-km and unSWAN results vs. measurements for the ensemble of 9 stations and 36 extreme events (left panels) and with spatial distributions of unSWAN results and differences between unSWAN and SWAN 1-km results for the maximum significant wave height and the maximum peak wave period during one extreme sirocco event (22 December 1979) and one extreme bora event (7 February 2012)

The second argument presented here is that the inclusion of the precise coastline and bathymetry (e.g. the precise geomorphology of the Venice Lagoon) as well as the numerous small islands (more than 1000 islands, isles, islets, rocks are identified along the Croatian coast) acting as barriers during extreme storm events, is crucial to properly simulate wave transformation and consequently storm surges in the northern Adriatic. This geomorphological impact is illustrated (Fig. 17) with spatial plots of maximum significant wave height and maximum peak wave period for the unSWAN model and for the difference between the unSWAN and SWAN 1-km models during the sirocco storm of the 22nd of December 1979 and the extreme bora event of the 7th of February 2012. The effect of the geomorphology and the islands located along the eastern Adriatic Sea on the wave 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 along the Croatian coasts than in the rest of the domain. The difference plots highlight that, on the one hand, the overall increase in maximum significant wave heights in the unSWAN simulations (between 0.5 and 1.5 m compared to SWAN 1-km) mostly results from the increase in resolution (and thus in intensity) of the atmospheric wind forcing and, on the other hand, for the presented bora event, the proper representation of the islands in the unSWAN simulations dramatically influenced the reproduction of the maximum peak periods (up to 5 s difference with the SWAN 1-km model).

Finally, the use of unstructured meshes is not only useful for pure oceanographic purposes but also for the study of other processes driven by extreme events such as, for example, the motion of boulders during sirocco storms which requires to perfectly reproduce the wave transformation at a meter-scale resolution to derive the transport of these blocks weighting up to a ton (see Biolchi et al. 2019b).

Appendix 2

As no catalogue of extreme historical bora events exists in the Adriatic region, an in-depth bibliographical research—presented in Table 4 and including meteorological bulletins (in Croatian), local newspapers, photographs and videos, was thus undertaken to generate the ensemble of 22 historical bora storms used in this study.

Table 4 References used to select the date of the 22 representative extreme bora events of the ensemble

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Denamiel, C., Pranić, P., Quentin, F. et al. Pseudo-global warming projections of extreme wave storms in complex coastal regions: the case of the Adriatic Sea. Clim Dyn 55, 2483–2509 (2020).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Adriatic Sea
  • Extreme storms
  • Pseudo-global warming
  • Extreme waves
  • Storm surges