| 16 Jun 2026
The marine and coastal hazards of Mediterranean cyclones
Abstract. This study aims to contribute to the understanding of marine and coastal hazard due to cyclones in the Mediterranean region, through the reproduction of sea level and wind waves with the use of an unstructured and coupled 2D hydrodynamic-wave modelling system. The modelling system is used to produce a numerical hindcast for the period 1994–2020 which is validated against tide gauges, wave buoys and satellite-borne instruments. The sea level and wave results are analysed for 2483 Mediterranean cyclones detected through a meteorological approach based on a composite cyclone detection and tracking method. We study which portion of extreme values (i.e. values exceeding the 99th percentile) of storm surge and significant wave height is associated to Mediterranean cyclones and we find that in the Mediterranean basin, extreme waves are more frequently associated to cyclones than extreme storm surge. Indeed, more than 50 % of extreme significant wave heights of the hindcast is associated to waves generated by cyclones in most of the Mediterranean basin. The percentage is higher than 75 % in the Western Adriatic basin and in some sub-regions of the Ionian-Meridional basin, as well as in the northwestern part of the Aegean Sea. For storm surge we find that more than 50 % of extremes values are linked to cyclones in the Western Mediterranean basin, in most of the Thyrrenian and the Ionian-Meridional basin, and in the Western Adriatic Sea. On the contrary, less than 50 % of extreme values of storm surge are associated to cyclones in the Northern and Eastern Adriatic basin, in the Ligurian Sea and in the Levantine-Aegean basin. The cyclone hazard is then evaluated through selected indicators, including peak values of storm surge and significant wave height, as well as the storm erosion potential index and the total storm wave energy which are integrated indices suitable to indicate cyclones intensity. The spatial distribution of the 99th percentile of all hazard parameters shows that the regions characterized by high hazard values vary depending on the parameter considered. In case of maximum significant wave height and total storm wave energy, the highest values are reached in the Western Mediterranean Sea, in the open sea area between the Balearic Islands and Sardinia. In case of maximum storm surge and storm erosion power index, the highest values are observed in the Northern Adriatic basin while the Western Mediterranean is characterized by moderate values. Despite some common patterns between hazard parameters in the Mediterranean basin, the analysis of their relationships stress that there is not an high correlation among most of them. Through some case studies we show that these parameters provide complementary information enabling a more comprehensive overview of cyclone hazard.
This paper by Pavan et al. presents and analyses a new numerical hindcast to better understand and quantify marine and coastal hazards in the Mediterranean Basins. The paper is overall well written and presented and the topic addressed is suitable for NHESS. However, in addition to a relatively long list of minor issues that needs to be addressed, I think that the authors need to include a discussion on the limitations of their numerical hindcast. Indeed, due to the 500 m spatial resolution along the coast combined to their 2DH approach relying on a radiation stress coupling, the contribution of wave setup to the storm surge should be strongly underestimated along the coast (please consider recent studies discussing these issues such as Lavaud et al., 2020; Martins et al., 2022 or Toomey et al., 2024). Also, although I understand that all the major events of the last 25 years cannot be analysed in details, I am surprised that storm Gloria (January 2020) is not mentioned in this study. Indeed, this event combined some of the largest storm surges recorded along the Eastern Coast of Spain with some of the largest waves ever recorded in the Mediterranean Basin(Hs >8m) and contributed to the substantial flooding of the Ebro Delta (e.g. Amores et al., 2020). Figure 5 and 6 show maximum values for storm surges and Hs much weaker than these values, possibly illustrating the underestimation of extreme values discussed above. In the same vein, the extreme surge of ~1.60 m observed in Venice on the 12/11/2019 is not mentioned in this study. Again, the maximum hindcasted surge value <1 m for this period (figure 5-A), questions the accuracy of the hindcast for the most severe events.
For these different reasons, I thing the paper deserves moderate to major revisions before being considered further for publication to NHESS.
Minor issues:
-L34: “for the state of coastal area” is not clear to me.
-L56: retrieve ?
-L57: please consider as well papers by Toomey et al. (2024).
-L76: which formulation?
-L83-85 and 99: the computation of surface stress and its dependence to the sea state is a fundamental problem for storm surge modelling, please explain how it is computed and/or provide some references.
-L98: please explain how radiations stresses are computed: based on bulk parameters or integrated on the whole spectra?
-L128: a time step of 300 s implies a CFL condition on the spatial advection >>1 and therefore suggests that the fully implicit mode of WWIII is employed, could you please confirm and provide and adequate ref?
-Section 2.2 is very technical and I wonder if it could not be shortened?
-L164: please check the ref John A. Knaff, shouldn’t it read “Knaff and Molenar”?
-L182: “the analysis”.
-L196: which sea-levels?
-L237: TSWE also uses the energetic wave period Te, what if the model accuracy on this parameter?
-L271: please explain that the problem originates from the seasonal cycle of mean sea-level (e.g. Marcos and Tsimplis, 2007), which can only partly be reproduced with a 2DH barotropic model.
-L278-280: the sentence is very vague and its relevance is very limited. As said in my general comments, please rather provide a detailed discussion on the model limitations.
-Table 3: please add a normalised RMSE so that the predictive skills of this hindcast can be compared against other hindcast.
-L313: Italian Peninsula.
-L326-327: to 41% of the …than 73% of the.
-Figure 5: I am not sure to understand “the spatial distribution of 99th percentile of the maximum surge”, is this the 99th percentile of the maximum?
-L417: 9 March of which year?
-L418: “Athos was”
-Figures 11 to 14: it is not clear to me why showing TSWE for some storms and Hs or surge for others. If this is because these quantities were particular for each storm, please explain.
-L426: “a significative surge” is not clear to me, please rather give the value.
-L429: the Adriatic Basin.
-Discussion: as said in my general comments, I would start by a discussion of the hindcast predictive skills (possibly compared to previously published studies) and limitations.
-L456: complement rather than complete.
-L484: “Marcos et al. (2019) investigated …”
Cited references:
Amores A., Marcos M., Carrió D.S., Gomez-Pujol L., 2020. Coastal impacts of Storm Gloria over the north-western Mediterranean. Natural Hazards and Earth System Sciences, 20 (7), 1955 – 1968
Lavaud L., Bertin X., Martins K., Arnaud G., Bouin M.-N., 2020. The contribution of short-wave breaking to storm surges: The case Klaus in the Southern Bay of Biscay. Ocean Modelling, 156, art. no. 101710.
Marcos, M., and M. N. Tsimplis, 2007. Variations of the seasonal sea level cycle in southern Europe. Journal of Geophysical Research 112, C12011, doi:10.1029/2006JC004049.
Martins, K., Bertin, X., Mengual, B., Pezerat, M., Lavaud, L., Guérin, T. and Zhang, Y.J., 2022. Wave-induced mean currents and setup over barred and steep sandy beaches. Ocean Modelling 179, 102110.
Toomey T., Marcos M., Wahl T., Agulles M., Enríquez A.R., Amores A., Orfila A., 2024. Wave setup estimation at regional scale: Empirical and modeling-based multi-approach analysis in the Mediterranean Sea. Weather and Climate Extremes, 44, art. no. 100685