J Coast Conserv (2011) 15:353–368 DOI 10.1007/s11852-010-0090-7

Coastal inundation in the north-eastern mediterranean coastal zone due to storm surge events Yannis N. Krestenitis & Yannis S. Androulidakis & Yannis N. Kontos & George Georgakopoulos

Received: 24 August 2009 / Revised: 19 November 2009 / Accepted: 27 January 2010 / Published online: 23 February 2010 # Springer Science+Business Media B.V. 2010

Abstract Low-elevation coastal areas and their populations are at risk during and after the appearance of a storm surge event. Coastal flooding as a result of storm surge events is investigated in this paper for a number of areas around the north-eastern (NE) Mediterranean coastal zone (Adriatic, Aegean and north Levantine seas). The sea level rise (SLR) due to storm surge events is examined for the period 2000– 2004. Wind data, atmospheric pressure and wave data for this period as well as in situ sea elevation measurements (from stations around the Mediterranean coasts) were used. Potential inundation zones were then identified using a 90-m horizontal resolution digital elevation model (DEM). At these zones, the sea surface elevations were calculated for the study period, using the collected data and a 2D storm surge simulation model (1/10o×1/10o) output, examining the sea level alteration in specific coastal areas, where in situ measurements are absent and are characterised as “risky” in inundation areas, due to their topography. In order to determine the level of storm track implication on major SLR incidents, the trajectories of the respective storm events were computed. The aim of this paper is to investigate the major storm surge events that appeared during the study period, identify the major “risky” costal regions along the north-eastern Mediterranean coast and determine their hazard level due to inundation caused by storm surge phenomena. The combination of the risk level determination of an area and the calculation of sea level alteration is an important tool in terms of predicting and protecting the coastal area from extreme meteorological incidents. Y. N. Krestenitis (*) : Y. S. Androulidakis : Y. N. Kontos : G. Georgakopoulos Laboratory of Maritime Engineering & Maritime Works, School of Civil Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece e-mail: [email protected]

Keywords Storm surge . Modelling . Inundation . Sea level rise . Mediterranean Sea

Introduction Storm surge is a rise of water associated with a moving low pressure weather system (cyclone). Storm surges are a result of low pressure meteorological systems creating suction over the covered area of ocean and of the heavy winds usually present in storms. Several processes are important during a storm surge event: (1) the Ekman transport by winds parallel to the coast may move water towards the coast, depending on the coast-wind orientation, causing a rise in sea level, (2) winds blowing toward the coast push water directly toward the coast, (3) wave set-up and other wave interactions transport water towards the coast adding to the first two processes, (4) edge waves generated by the wind travel along the coast, (5) the low pressure inside the storm raises sea level by one centimetre for each mbar decrease in pressure through the invertedbarometer effect, and (6) the storm surge adds to the tides, and high tides can change a relative weak surge into a much more dangerous one. Low-elevation coastal areas and their populations are at risk during and after the appearance of a storm surge event. Any rise in sea level will have adverse impacts (e.g. coastal erosion and flooding) which depend on the time scale and the magnitude of the rise, as well as the associated human response (Paskoff 1993). High sea levels and the strong forces exerted by accompanying waves, impact directly or by over-topping sea defences on humans, property and habitats. They may even cause loss of life, damage (through inundation and waves), loss of habitat and useful land, property, infrastructure, services and so forth.

354

Coastal flooding as a result of storm surge events is a possible threat in Eastern Mediterranean coasts. Travelling mid-latitude low pressure systems act to raise the sea level directly below them, but this effect alone is quite weak in semi enclosed basins such as the Mediterranean Sea (Pirazzoli 2000), where the wind tends to be the dominant force. Although, the Mediterranean Sea is not on the main storm track of the European and North Atlantic area (Rogers 1997), storm track events originated mainly from Africa, with a direction from south-west (SW) to north-east (NE) affect significantly local sea-surface elevation at north-east Mediterranean coasts. In the 2000–2004 period several storm surge events appeared in the Mediterranean, where despite their birth area, they affected particularly the east (E) and NE region due to their major trajectory to the north-east. The main objectives of this paper are (a) to determine significant storm surge events that appeared during the study period and examine their major tracks and magnitude, (b) to correlate these meteorological events to the respective sea level alterations along the NE Mediterranean coast, (c) to determine the low elevation coastal areas and their risk level to inundation due to storm surge events and (d) to examine each storm surge process importance to the sea level alterations in each sub-domain of the study area.

Methods and data Atmospheric and sea elevation data Sea level pressure (SLP), wind fields (POSEIDON forecasting system, http://www.poseidon.hcmr.gr) and wave data (wave height, direction and wave period), based on the wave model SWAN—Simulating Waves Nearshore (Soukissian et al. 2007) on a 1/10 × 1/10 grid covering the entire Mediterranean basin, were used for the period 2000–2004. In order to verify and calibrate the storm surge model sea level height (SLH) data from gauge stations around the Mediterranean were collected from the websites of the Med-GLOSS program (Monitoring Network System for Systematic Sea Level Measurements in the Mediterranean and Black Sea, http://medgloss.ocean.org.il/), the European Sea-level Service (http://www.eseas.org/) and the Greek Hydrographic Service (http://www.hnhs.gr/ portal/page/portal/HNHS) (Fig. 1). Additional sea-level data were collected by using satellite data concerning the study period and area. The altimetry data were collected from the French space agency (Aviso/Altimetry project, http://www.jason.oceanobs.com/). These data derived from the composition of various scans of different satellite projects (Topex/Poseidon, Jason-1, ERS-1 and ERS-2, EnviSat, and Doris). The satellite data were mostly used to

Y.N. Krestenitis et al.

evaluate the model’s performance at open sea, because the accuracy of satellite data close to the coast is not high enough. The astronomical tide surge ranges in low levels (few centimetres) in the Mediterranean region because it is nearly landlocked, with the only major opening to open seas being the Strait of Gibraltar. According to Pugh (1996), basins of oceanic depths such as the Mediterranean Sea, which connect to the oceans through narrow entrances, have small tidal ranges. The same author (Pugh 1996), claims that the dimensions of these basins are too restricting for the direct tidal forces to have much effect and the areas of the entrances are too small for sufficient oceanic tidal energy to enter to compensate for the energy losses which would be associated with large tidal amplitudes. Because the connection with the Atlantic Ocean through the Straits of Gibraltar is so restricted, the influence of direct gravitational forcing within the Mediterranean is probably of comparable importance to the external forcing. In spite of this, in the storm surge modelling section, that is described in chapter 2.3, the Atlantic tide contribution was included in the simulations as a boundary condition in the open boundary at the Gibraltar Straits. Therefore, the astronomical tide was not removed from the in-situ data, because it is believed that it does not severely bias the analysis (Moron & Ullmann, 2005). The storm surge contribution (meteorological tide) is the most important factor to sea level height alteration measured at the above East Mediterranean stations with the exception of the north Adriatic region, where according to Pugh (1996), the diurnal tides of the area are relatively large (in the vicinity of Venice) because a natural oscillation is excited by the Mediterranean tides at the southern Adriatic entrance. So, especially in this area, the combination of tidal and storm surge, as described in chapter 3.1, can produce significantly high SLRs. Storm surge detection and storm-track computation A method for automated detection and tracking of storms or cyclones was developed by the “National Aeronautics and Space Administration—Goddard Institute for Space Studies” (NASA-GISS, http://data.giss.nasa.gov/stormtracks). A version of the NASA’s methodology is used in this study, as described below. Storm tracks can be identified by following low pressure centres on synoptic charts and plotting their trajectories on maps, thereby producing ‘‘cyclone tracks’’ in the pure sense. The major objectives of this analysis are a) the identification of the major incidents during the study period and b) the investigation of their atmospheric characteristics such are their direction and pressure magnitude in order to relate them with the respective SLR incidents.

Coastal inundation in the north-eastern mediterranean coastal zone Fig. 1 Gauge stations in the NE Mediterranean, providing SLH timeseries for years 2000–2004 (black) and coastal inundation high risk areas (grey)

45

355

Trieste Venice Zadar Rovinj Split Ancona Dubrovnik Alexandroupolis Otranto

40

Lefkada

Chios Antalya

Catania 35

Gauge Stations Low Elevation Areas

10

15

The computation and plotting of storm tracks is accomplished by using an algorithm that identifies and tracks sea level pressure (SLP) minima (Pmin). The algorithm searches for and identifies absolute minima from the gridded field of every 12-hour period of the desired year. The minimum for the ensuing 12-hour SLP grid is searched and its position is located. Any Pmin within a critical radius of 1440 km are joined by a segment, representing the path of the low pressure center during that 12-hour period (a cyclone"s center can travel at a mean speed of no more than 120 km/hr or 1440 km / 12 hr) (NASA-GISS, http://data.giss.nasa.gov/stormtracks). Any two associated minima identify one storm track segment, as long as the storm lasts at least 36 h. If at any time two segments on the same track are found to define an acute angle of less than 85° (NASA-GISS, http://data.giss.nasa. gov/stormtracks), the low pressure centers are considered to represent separate storms (extra tropical cyclones are not found to “double back” on themselves over 12 h). Finally, Fig. 2 Maximum (a)
and  Mean (b) SLP difference ΔP and the direction of the storm surge events that appeared in the period 2000–2004

20

25

30

35

throughout the full duration of a storm track, the low pressure center of the track must be less than 1015 hPa. In this way, storm tracks are computed and plotted for the Mediterranean region. The use of absolute minimum SLPs instead of local minima (like NASA’s method) produces the risk of not identifying some storm tracks generated by the passage of secondary low pressure centers. For that, but also for presentation reasons and visual confirmation, a SLP contour map for every 12-hour period is produced, creating a slideshow of 730 maps (732 for leap years) for every year, presenting low pressure regions (SLP<=1015mBars) and their development. From the analysis of the whole series of storm events the following resulted: a) The majority of the storm surges in the Mediterranean region are directed from West to East (Fig. 1. and b) the highest depression ΔP (ΔP=PR—Pmin, where PR is the reference atmospheric pressure equals to 1015 hPa) varies mainly between 10–20 hPa, (Fig. 2a). In Fig. 2b, events from the whole study period (2000–2004)

356

Y.N. Krestenitis et al.

Table 1 The characteristics of the main applications of the storm surge model

Expr1 Expr2 Expr3 Expr4 Expr5 Expr6

CS

Bathymetry

CD

Smith & Banke 4.5 10−3 Smith & Banke Amorocho & DeVries Smith & Banke Smith & Banke

h ≤ 250 h ≤ 250 Realistic h ≤ 250 h ≤ 250 Realistic

Wang Wang Wang Wang 5 10−3 5 10−3

m m m m

are presented, showing the mean ΔP value of each event. The directions are the same presented in Fig. 2a as far as the minimum pressure values are concerned, but the ΔP now is lower and it varies around 10 hPa. Storm surge modelling A 2-dimensional hydrodynamic model solving the depthaveraged shallow-water equations is used to provide the sea level for entire Mediterranean basin on a 1/10o × 1/10o spatial grid. The 2D model is an updated version of the AUT storm surge model (DeVries et al. 1995), with the following forms of the momentum Eqs. 1 and 2 and the continuity Eq. 3: @U @U @V @z þU þV  fV þ g @t @x @y @x   1 @P t sx  t bx 1 @Sxx @Sxy þ ¼  þ r @x rH @x rH @y

Table 2 The RMSE and r values between measured and computed SLH timeseries

Model application

@V @V @V @z þU þV þ fU þ g @t @x @y @y   1 @P t sy  t by 1 @Sxy @Syy ¼ þ  þ r @y rH @x rH @y

@z @ ðHU Þ @ ðHV Þ þ þ ¼0 @t @x @y

where t is the time, x and y are the spatial coordinates, ζ is the water level elevation above the mean sea level, U and V are the x–y components of the depth-mean current, H=(h+ζ) is the total water depth (h is the undisturbed water depth and ζ is the water level), g is the acceleration of gravity, f is the Coriolis parameter, τsx and τsy are the x and y components of the wind stress, τbx and τby are the x and y components of sea bottom stress, Sxx, Sxy and Syy are the wave radiation stresses, ρ is the density of the water and P is the atmospheric pressure at sea level. The calculation of the wind stress is based on the wind velocity data at 10 m, according to the formula of Eq. 4: ð4Þ

where ρΑ is the air density, W=(Wx, Wy) is the wind velocity and CS is the surface friction coefficient. Two deferent formulas of the surface friction coefficient were

Chios

Alexandroupolis

RMSE

r

expt1

0.103

0.47

expt2 expt3 expt4 expt5 expt6

0.131 0.098 0.103 0.102 0.098 Lefkada RMSE 0.098 0.096 0.100 0.098 0.098 0.100

0.46 0.49 0.47 0.47 0.49

expt1 expt2 expt3 expt4 expt5 expt6

ð3Þ

t sx ¼ rA CS jW jWx & t sy ¼ rA CS jW jWy

ð1Þ

ð2Þ

r 0.49 0.47 0.52 0.49 0.49 0.51

RMSE

Dubrovnik r

RMSE

r

0.183

0.31

0.127

0.42

0.184 0.186 0.182 0.183 0.186 Otranto RMSE 0.111 0.118 0.110 0.111 0.111 0.110

0.33 0.34 0.32 0.31 0.34

0.128 0.130 0.130 0.127 0.130 Split RMSE 0.135 0.138 0.136 0.135 0.135 0.136

0.44 0.45 0.42 0.42 0.41

r 0.47 0.46 0.49 0.47 0.47 0.49

r 0.43 0.45 0.45 0.43 0.43 0.45

Coastal inundation in the north-eastern mediterranean coastal zone Fig. 3 Model-insitu SLH scatter diagrams for Dubrovnic (Adriatic Sea) and Alexandroupolis (Aegean Sea)

357 0.5

0.5

0.3

Alexandroupolis 2004

0.1

-0.5

-0.3

-0.1 -0.1

0.1

0.3

in-situ (m)

in-situ (m)

Alexandroupolis 2002

0.5

0.1 -0.5

-0.3

a

-0.5 model (m)

0.1

0.3

0.5

in-situ (m)

insitu (m)

Dubrovnik 2004

0.3

-0.1 -0.1

0.3

0.5

b

0.5

0.1

-0.3

0.1

-0.5 model (m)

0.5

-0.5

-0.1 -0.1

-0.3

-0.3

Dubrovnik 2002

0.3

0.3

0.1

-0.5

-0.3

-0.1 -0.1

0.1

0.3

0.5

-0.3

-0.3

d

c -0.5 model (m)

-0.5 model (m)

used in the present model, i.e. the Eq. 5 (Smith and Banke, 1975) and Eq. 6 (Amorocho and DeVries, 1980):

The bottom stress is related to the water velocity by the following expression:

 Cs ¼ ð0:63 þ 0:066W Þ 103

t bx ¼ rCD U

Cs ¼ 1:04 103 for W7:0m=s;

Cs ¼ 2:54 103 for W20:0m=s

ð5Þ and 103 Cs ¼ 0:115 W þ 0:235; 7:0m=s  W  20:0 m=s

ð6Þ

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðU 2 þ V 2 Þ & t by ¼ rCD V ðU 2 þ V 2 Þ

ð7Þ

where CD is the bottom drag coefficient and using the law of the wall in the bottom logarithmic layer (Wang, 2002), and it is computed according the following equation: (  )  2 1 H CD ¼ max ln  1 ; 0:0025 ð8Þ k zo

Table 3 Potential inundation areas in km2 and the respective coastline length (km) and density population of each area (persons / km2)

Venice Lagoon Neretva Delta Gulf of Manfredonia Albanian coasts Alexandroupolis Kavala Patraikos Gulf (North) Patraikos Gulf (South) Thessaloniki Seyhan-Tarsus

Potential inundation area (km2)

Front length (km)

Population density n (persons / km2)

6300 200 200 1600 300 300 200 500 700 800

288 16 63 236 24 113 118 149 69 153

n>500 100>n>25 100>n>50 250>n>100 20>n>10 200>n>25 200>n>25 200>n>25 n>250 100>n>25

358

Y.N. Krestenitis et al.

Fig. 4 North-east Aegean map of coastal vulnerability

where κ is the Von Karman constant and zo is the seabottom roughness here set equal to 0.001. The wave radiation stresses used in the momentum equations were calculated from the wave height and direction, according to the following equations: Sxx ¼ E ð2n  1Þ þ Encos2 a Sxy ¼ Syx ¼ E2 n sinð2aÞ Syy ¼ Eð2n  1Þ þ En sin2 a

ð9Þ

where E is the wave energy density and α is the wave propagation direction. The 2D storm surge model has been implemented in the entire Mediterranean basin on a 1/10o×1/10o grid. The model forced with the data assimilated atmospheric forecast fields, provided by the POSEIDON system, based on the atmospheric SKIRON Limited Area Model (LAM) (Papadopoulos et al., 2002) and with wave data for the period 2000–2004. Tidal boundary conditions where imposed in the Gibraltar open boundary by using specific tidal components measured in the area’s station (Ceuta) and the tidal harmonic analysis Eq. 10. X
 zðtÞ ¼ Aj cos wj t  ϕx ð10Þ j

Table 4 Studied extreme storm surges for the period 2000–2004

where Aj is the amplitude, ωj the frequency and φj the phase of the j tidal constituent respectively. Several applications of the storm surge model have been carried out in order to test and calibrate the model. Some of these applications are indicated in Table 1. The calibration procedure was based on the calculation of the root mean square error (RMSE) and the Pearson product-moment correlation coefficient (r) between the measured and the computed SLH timeseries, for a number of stations in Adriatic and Aegean Sea. In Table 2, the RMSE and r values for the six model applications (indicated in Table 1) and for 6 stations, are presented for the model run with the 2004 year forcing. Also, the scatter diagrams for two stations (one in N. Aegean and the other in Adriatic) are presented in Fig. 3, for two different yearly forcing, i.e. years 2002 and 2004, showing an acceptable performance of the model for the scope of this article, especially for high SLH values, which in any case are the most important in a SLR investigation. Overestimation of the low (below mean sea level) values is observed, where the model results show differences with the respective measured ones. After the calibration procedure, the experiment chosen, as the best one was the 3 rd experiment, with the higher values of the correlation coefficient and lower values of RMSE for most of the testing cases.

Case

Starting date

Duration

Most affected area

1 2 3 4 5

1/03/2001 30/11/2002 03/02/2003 15/12/2003 21/01/2004

3 5 4 6 4

North Adriatic Adriatic Aegean Sea Levantine Sea Aegean Sea & Seyhan-Tarsus Delta

days days days days days

Lowest pressure (hPa) 981.415 999.104 985.242 996.502 980.585

Coastal inundation in the north-eastern mediterranean coastal zone

359

Fig. 5 Storm track trajectories of the 5 studied cases.

Digital elevation model Potential inundation zones were calculated from the 90-m resolution shuttle radar topography mission (SRTM) digital elevation model (DEM) (http://srtm.csi.cgiar.org). SRTMDEM has spatial horizontal resolution of 90 m and absolute vertical accuracy of 15 m in the mainland and less than 1 m over the coastal zone (Sun et al. 2003). According to it, high risk areas in NE Mediterranean are located at the northern coast of Adriatic Sea, at Dalmatian coast, at Albanian coast, at central-eastern Italian coast, at western Greek coast, at the North East Aegean coastal zone, and at south-east Turkey (Fig. 1). The potential inundation area and the coastline length of each detected region is presented in Table 3. All these areas are characterized for their wide coastal zone with a surface elevation less than 1.0 m. Low elevation “risky” areas in the Adriatic and in the Ionian seas are the Venice lagoon (Italy), Neretva Delta (Croatia), the Albanian coasts, the gulf of Manfredonia (southern Adriatic Sea, Italy) and Patraikos gulf (Greece).

a

The Venice Lagoon and its city Venice is characterised by significantly high population (Table 3) and is very sensitive to water level variations, given that the city of Venice is only about 1 m above mean sea level and the average depth of the lagoon is 1.5 m. Especially, according to Bondesan et al. (1995) south of Venice city e.g. around the Po delta, areas with more than 2 m below sea level widespread exist. According to the same authors the total coastal area below sea level between the main river courses in NE Italy is about 2375,48 km2. In this study, the calculated potential inundation area that extends inland even more than 50 km is conclusively 6300 km2. Venice has long been famous for being the city that is partially under water where 1 m of extra water above the expected astronomical high tide results in significant flooding (aqua alta); the severest flood which occurred in November 1966 was caused by an extra 190 cm (Robinson et al. 1973). Water levels significantly higher than the expected astronomical tide level are reported several times a year, especially between October and January (Bargagli et al., 2002).

b

1040 in-situ

1035

SKiron mean SLP (hPa)

1030 1025 1020 1015 1010 1005 1000 995 990 45

50

55

60 Time (days)

65

70

75

Fig. 6 Measured and modeled SLP (a) and satellite SLH (b) in 28 of February 2001

360

Y.N. Krestenitis et al.

0.8 in-situ

0.6

model SLH (m)

0.4 0.2 0 -0.2 -0.4 45

50

55

60 Time (days)

65

70

75

Fig. 7 Model and in-situ SLH variance in Venice during case 1

Neretva is the largest river in the eastern part of the Adriatic basin. Even if the length of the front of the “problematic” area to the sea is only 16 km, the area of the delta and the surrounding territories that could reveal flooding problems is about 200 km2. Although the population density is not high, Neretva delta importance is recognized due to its natural beauty, diversity of its landscape and visual attractiveness. Additionally, the basin is situated between major regional rivers and contains the most significant portion of fresh drinking water. The Albanian coastal zone is characterised by its low land elevation that extends in long length. Major urban centres are situated at this coastal area, such as Vlora, Durres and Shengjin. Possible sea level rise up to 1 m could cause significant problems in economic life of this area based mainly in agriculture. The Gulf of Manfredonia is situated on the Apulian coast, just south of the Gargano Promontory (east Italy). The coastal area “in risk” is about 200 km2, and the coastline is 63 km long. In Western Greece two neighboring geographical regions, one in Peloponnesus (Patra area) and one in Sterea Ellada can be characterised as “in danger” areas, both situated around Patraikos Gulf. The total area of these regions is approximately 500 km2 and 200 km2 respectively and the length of their coastlines are 149 km and 118 km likewise. The most dangerous areas due to storm surge inundation are situated in the northern region of the Aegean Sea, below

the major SW to NE trajectory of the storms coming from the central Mediterranean and Africa (Laskaratos et al., 1991). According to (Vries et al., 1995) these north Aegean areas (Thessaloniki, Kavala, Alexandroupolis) are expected to be dominated by atmospheric pressure, while in the N. Adriatic, storm surges are wind-dominated, as was analysed in the previous section. Two major areas “in danger” can be spotted, one west of Kavala city as far as the lagoon of Porto Lagos, covering an area of 300 km2, with 113 km coastline, and one along Evros Delta, situated exactly in the Greek-Turkish borders, east of the Alexandroupolis city, cover an area of 300 km2, with 24 km width of the delta front (Fig. 4). Especially, the first area is extensively populated and covered with tourist, urban, and industrial structures, while in both areas significant ecosystems exist. In the NW Aegean, the gulf of Thessaloniki is a coastal ecosystem of major importance not only environmentally, but also due to the various socioeconomic activities associated to the heavy populated area of Thessaloniki. Three delta rivers constitute a protected ecosystem, important intermediate stop-point of various emigration birds. The region that may reveal flooding phenomena due to storm surge events covers an area of about 700 km2 and the length of the coastline in front of this area is about 69 km. In south-east Turkey, east of Mersin, an area of 800 km2 has been noticed as possible inundation coastal area, where the Seyhan-Tarsus delta is situated. The front coastline of this area has length of 153 km.

Results Several significant storm surge events were examined for the study period and the area. Five extreme cases in NE Mediterranean region throughout the 5 year period of study are presented in Table 4. These storm surge events affected significantly extended coastal regions. The influence of each case is examined on each one of the three major NE Mediterranean regions (Adriatic, Aegean and North Levantine Seas). The selection of these particular cases is based on a combination of several factors. The most important are a) the maximum SLH that occurred during study

Table 5 Storm track implication in SLH rise for case 1 Gauge Station

Date (Model Day)

insitu SLH(m)

SLP (hPa)

R (annual)

r (6days)

low pressure-station distance (km)

Ancona Trieste Rovinj Zadar Split Dubrovnik

02–03 p.m. (61) 03–03a.m (62) 02–03 a.m. (60.5) 03–03 a.m. (61.5) 03–03p.m (62) 02–03 p.m. (61)

0.44 0.32 0.37 0.43 0.38 0.27

995.894 1002.164 1003.059 1000.622 1007.009 1007.628

−0.43 −0.43 −0.41 −0.40 −0.38 −0.36

−0.59 −0.51 −0.27 −0.27 −0.11 −0.12

1344 1283 1666 968 1578 1726

Coastal inundation in the north-eastern mediterranean coastal zone

361

Adriatic and Ionian Seas In the Adriatic Sea, the sea level alteration due to storm surges is highly related to the tide’s amplitude, which is usually higher than in the rest of the Mediterranean (Tsimplis et al. 1995), especially in the northern Adriatic. Astronomical forcing produces an almost complete “cooscillation” with the Mediterranean, where the continuous driving from the southern inlet is much more important than the negligible local direct forcing from the moon and sun (Bondesan et al. 1995). Additionally, the southeast winds (Sirocco) raise the sea level, especially in the North Adriatic, where a long-lasting Sirocco and low air pressure can also raise the water level up to 1 m (Vilibic et al. 2005). Wind influence is less important in the South Adriatic, where the air pressure influence is dominant giving rise to sea level changes of up to 30 cm (Leder 1988). Specifically, in Venice lagoon, several numerical modelling approaches to predict the sea level alteration due to storm surge events have been applied in the past (Lionello et al., 2006; Bajo et al., 2007). Bajo et al. (2007) examined a low pressure system that appeared during May 2004 above Northern Adriatic and simulated the produced storm surge in Venice (Bajo et al., 2007—their chapter 4.2). This storm surge event is also detected in the simulations of this study (not shown). The calculated SLH in Venice is one of the highest of the year ranging around 0.4 m, 10 cm lower than the simulated values by Bajo et al. (2007). Case 1 corresponds to a storm that occurred firstly over Atlantic ocean, moved to the East and occupied the northcentral Mediterranean (Fig. 5) between the 1st and 3nd of March, 2001, affecting the sensitive lagoon of Venice. This is an Atlantic originated storm and a classic example of a

Fig. 8 N. Adriatic (Zadar−14o 52 , 43ο 44 ) SKIRON wind timeseries March 2001 (simulation days 59–90, case 1)

period, especially in the “risky” areas b) the existence of a storm event when a maximum SLH occurs, c) the sufficient available data, such are in-situ sea level data and satellite images and d) inclusion of storms with different trajectories for each sub-region in order to investigate each area’s different SLH behaviour in relation to the storm’s origination and spatial evolution. The minimum duration of these cases is 3 days and their center’s lowest pressure is lower than 1000 hPa. Some of these events altered the SLH in distant coastal areas away from the major track due to the relative winds. The track of each selected case was computed and plotted (Fig. 5) according to the method presented in chapter 2.2. In most cases each storm is represented by a single trajectory (Cases 1,3,5 highlighted in black, blue and green line respectively). Exceptions are case 2 and 4 (highlighted in red and purple, respectively), where for each storm two successive trajectories are needed in order to represent a more complex form of storms, due to the appearance of secondary low pressure systems.

a

b

0.3

Lefkada

0.25

in-situ model

0.2

SLH (m)

0.15 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2

330

335

340 345 Time (days)

350

355

Fig. 9 (a) Model and in-situ SLH for Lefkada and (b) satellite sea level altimetry (cm) on the 4th of December, 2002 (day 337)

362

Y.N. Krestenitis et al.

0.4

Venice

Patraikos Gulf

Albania

0.3

SLH (m)

0.2

0.1

0

330

340

330

340

330

340

-0.1

Time (days)

-0.2

Fig. 10 Venice,Albania, and Patraikos Gulf model SLH during 4th of December 2002 event (day 337)

NW storm, sweeping the North Adriatic Sea for a couple of days or so. This storm is the stronger second phase of a system that appeared on 27th of February, again originated in the Atlantic and moved to the east, adding up an extra SLH rise in the N.Adriatic Sea. Fig. 6a demonstrates the simulated SLP (Papadopoulos et al. 2002) and the measured respective one (ftp://ftp.ncdc.noaa.gov/pub/data/ gsod) for a period of 30 days starting at 15 of February 2001. The SKIRON simulated SLP that is used as a forcing input in the storm surge simulations occurs significantly high performance. The measured SLH in the Venice gauge station for 2001 is presented in Fig. 7. Another higher SLR occurs a few days later (day 69), connected with another SLP drop as it is observed in the SLP diagram. Therefore, in the SLH simulation (Fig. 7) the higher rises appear during the days of the case 1 storm. This two phase pressure driven surge is also presented in the mathematical simulations where a high SLR was calculated at day 59.5 of 2001 during the first storm phase and another higher SLR was calculated almost one day and a half later (day 61 of 2001). Even if the model determines the SLH extremes in the same dates (day 59.5 and 61) and time with the measured ones, it seems that in the case of Venice, the model overestimates the high values by about 30%. Significant sea level rise at the open N. Adriatic during the first storm phase (28/02/01) was recorded in satellite measurements (Fig. 6b), too.

In Table 5 the highest sea elevations during case 1 (second phase) period for several stations are presented with the respective system-station distance and station’s SLP in the time of the sea level maximum appearance. Although the system’s lowest pressure is approximately 981 hPa (Table 5) the highest sea elevations in Adriatic occurred under much higher SLP. Additionally, the SLPSLH correlation for each station was computed for a 6-day window, starting 6 days before the date that the maximum SLH was recorded. The highest sea level rises are observed in the north end of the region, while in southern areas (e.g. Dubrovnic) the measured SLH is low and the distance of the low pressure center to the gauge station is big. The storm affected mostly the sea level pressure of the western coast (Ancona), where the highest SLHs were measured and the highest 6 day correlation was computed (r=−0.59). One day later the storm moves more to the east and affects the east coast, where e.g. in Zadar SLH of 0.43 m was measured. The 6-day correlation in Zadar is low (lower than the respective annual correlation R) indicating the possible wind-driven surge in Zadar in a contrary to the west (Ancona) and north (Trieste) areas, where the SLR is mainly affected by SLP drops. Strong south winds during the first 5 days of March are obvious in Fig. 8, where the wind time-series of March 2001 in the area of Zadar is presented. In 2002, a very important incident that spread all over Eastern Mediterranean and was affecting not only the Adriatic Sea but also the North Aegean, coasts of Asia Minor and southern Turkey (Case 2, Fig. 5). From the storm track analysis, it is obvious that this is not an Atlantic originated or an extra-tropical cyclone, but a Mediterranean-basin restrained one, generated in it and affecting its interior alone. Furthermore its anti-clockwise rotation around south Italy, follows the typical anti-clockwise spin a cyclone has in the North Hemisphere. Additionally, from the storm track analysis, it is safe to assume that the two storms making up case 2 are, in fact, two phases of the same cyclone system. The duration of the entire event was five days and the centre of the system was located mainly above Italy and Adriatic, where the greatest SLH values were detected (min∼SLP= 999 hPa), affecting though the rest of Mediterranean Sea. Altimetry remote sensing produced maps (Fig. 9b) show

Table 6 Storm track implication in SLH rise for Case 2 Gauge Station

Date (Model Day)

Ancona Rovinj Zadar Split Dubrovnikj Lefkada

03–12 03–12 03–12 03–12 03–12 03–12

p.m.(337) p.m. (337) p.m. (337) p.m. (337) p.m. (337) p.m. (337)

Insitu SLH (m)

SLP (hPa)

R (annual)

r (6days)

Storm distance (km)

0.6 0.53 0.50 0.52 0.39 0.27

1001 1005 1002 1002 1002 1005

−0.61 −0.55 −0.58 −0.52 −0.52 −0.43

−0.63 −0.46 −0.46 −0.47 −0.72 −0.84

344 511 427 414 451 658

Coastal inundation in the north-eastern mediterranean coastal zone Fig. 11 Model SLH (m) during case 3 event in several areas of the Aegean Sea (Thessaloniki-NW, Alexandroupolis-NE, Irakleio-S, Chios-Central, Kavala-NE)

363

0.6 0.5 0.4

SLH (m)

0.3 0.2 0.1 0 -0.1

25

30

35

40 25

30

35

40

25

-0.2 -0.3

Aegean Sea A significant storm surge incident affecting the Aegean Sea, with low pressure around 985 hPa (Table 1, case 3)

Alexand.

40 25

Irakleio

30

35

40 25

30

35

40

Chios

Kavala

occurred between 3 of February 2003 until 6 of February 2003 and its low pressure trajectory passed above the Adriatic and Aegean Seas (Fig. 5). The direction of the storm is first W-E, like the majority of the events observed in this study period and above the Balkans it turns to the south and it enters the Aegean Sea where it almost disappears above the central area. The gauge station of Chios has lots of data missing during year 2003 and Alexandroupolis’ station also. So, the use of mathematical simulations is crucial in cases like this one, where in situ sea level measurements are absent during significant meteorological events. The modeling output (day 35) confirms a significant sea level rise during this storm surge in the Aegean Sea (Fig. 11). This SLR exceeds the level of 0.5 m in the northern areas while lower levels are observed in southern areas. This north to south decrease agrees with the north to south direction of the storm, due to the pressure-driven surge that occurs just below the low pressure system. For the entire Aegean the highest SLR due to meteorological conditions of the entire year (2003) occurred. This spatial distribution of sea level alteration during case 3 is observed in almost all the study period, where the highest values are observed just below the low pressure systems. The wind-driven water pilling up towards to the coasts is limited in comparison to the Adriatic, due to Aegean’s morphological complexity with numerous channels and islands.

a 1040

b 0.4

1030

0.3

1020

in-situ model

0.2 SLH (m)

SLP (hPa)

35

Time (days) Thessaloniki

high sea level elevations near several low land elevation areas, such as the coastal area of Venice (Italy), the Dalmatian Coast (Croatia), the Albanian coasts (Albania) and the gulf of Patra (Greece, Ionion Sea). In Fig. 9a, the modelled and the observed SLH are presented for Lefkada (Ioanian Sea) showing both a high rise during the storm’s appearance, which decreases gradually from day 340 and thereafter. The simulation SLH output can be used to investigate the sea level alteration in areas or stations where observation data are not available or don’t exist for the specific event. So, these observations were confirmed by the mathematical simulation experiments for several Adriatic and Ionian areas where the simulated sea level elevation for the same date (day 337) showed a significantly high value (Fig 10). The south and central Adriatic (Lefkada and Dubrovnik) show high 6-day SLP-SLH correlation even if they show the lowest SLH rise during the case 2 storm. On the contrary, NE Adriatic (Rovinj, Zadar and Split), shows high sea elevations and lower correlation coefficients indicating the low SLP influence to the sea level elevation alteration (Table 6). Propagating SE winds are propably the main factor of the N. Adriatic SLH rise.

Fig. 12 Skiron SLP(a) and insitu and model SLH (b) variance during Case 5 (Chios)

30

1010 1000

0.1 0

990

-0.1

980 970

-0.2 15

18

21 24 Time (days)

27

30

15

18

21 24 Time (days)

27

30

364

Y.N. Krestenitis et al.

Table 7 Chios 2004 max SLH and storm track implication analysis Chios Incidents in 2004

Date (model day)

1 (Case 5) 2 3 4

21/01 29/01 16/11 06/05

In-situ SLH (m)

SLP(hPa)

R=−0.62 r (6days)

storm distance

0.31 0.30 0.25 0.26

983 1010 1009 1009

−0.91 −0.56 −0.57 −0.71

305 787 802 1511

(21) (29.5) (319) (127)

Another storm incident that affected the whole Aegean region appeared in January of 2004 (Case 5), where a low pressure system was generated above Greece and moved from NW to SE and then to NE as presented on Fig. 5. Especially, in the evening of 22nd and in the morning of 23 rd the amplitudes of pressure are extremely low (Fig. 12a), mainly above the coasts of Asia Minor where SLP is about 980 hPA, where one of the lowest pressure values of the entire study period appears. The nearest gauge stations to the high sea level rise areas are Alexandroupolis, Antalya and Chios (central Aegean). Unfortunately, the nearest of all, Antalya, seems to have missing SLH data during those days, but all the others have been able to record this storm surge event. The gauge station’s analysis for Chios is presented in Table 7, where the most important incidents that influenced Chios sea elevation are indicated. On the 21st of January of 2004 there’s no major SLH rise incident recorded in the gauge station of Alexandroupolis but it had a greater impact in the area of Chios, which was recorded by the Chios gauge station and simulated by the model as shown on Fig. 12b. Additionally, an important point is that Chios is situated close to the case 5 storm trajectory. The r=−0.91 correlation (Table 7) indicates a strong relationship between SLP and SLH when the recorded SLP value on that day was the all-year minimum one, and in fact, far minor than any other recorded value (SLP=983 hPa). Additionally, the SLH recorded on that day, is also the yearly maximum (SLH=0.31 m), which is actually the most important SLH-rise incident for Chios during 2004. Additionally the case 5 (incident 1, Table 7)

South Turkey coastal region West originated storms influence significantly the sea level alteration of the Levantine Sea and especially its north coast (South Turkey) and the coasts of Cyprus. According to a 6-year analysis by Cazenave et al. (2002), the sea level is rising at a rate of 25–30 mm/year in the Levantine Basin, while in the Ionian Sea it is falling by 15–20 mm/year, but for the whole Mediterranean Sea the mean rate of level rise is 7 ± 1.5 mm/year. The drop of the Adriatic Sea was studied also by Tsimplis and Baker (2000), who relate it to the general drop of the west Mediterranean region. This renders any extreme sea level rise more dangerous to the the east Mediterranean coastal areas year by year.

b

1.2 Case_4_model

1 0.8

W–E NW–SE SW–NE W–E

r correlation is the highest of all the other 2004 storm surge events, and it is also greater than that of the annual correlation for Chios (R=−0.62). Investigating several other storm surge events for Chios during 2004, it is observed that the west to east tracks show the highest 6-day SLP-SLH correlation, on the contrary to the north-south tracks. The distance between the low pressure system and the gauge station for case 5 was about 305 km, which is the closest than any other incident appeared during 2004. Therefore, this case shows higher correlation (r=−0.91) than the other 2004 incidents and maximum annual SLH value. So, it is obvious that for Aegean Sea the areas situated below or close to the low pressure systems are affected directly by the lower pressure, while winds play a less important role in the sea level alteration during a storm surge event.

a 1.4

1.4 1.2 Case_5_model

1 0.8 SLH (m)

SLH (m)

Fig. 13 Seyhan-Tarsus simulated sea level alteration during case 4 (a) and case 5 (b) storms

storm direction

0.6 0.4 0.2

0.6 0.4 0.2 0

0 -0.2

-0.2

-0.4

-0.4

-0.6

-0.6 335

345 355 time (days)

365

1

11 21 time (days)

31

Coastal inundation in the north-eastern mediterranean coastal zone

365

Fig. 14 Seyhan-Tarsus simulated wind time-series for (a) December 2003 (simulation days 335–365, case 4) and (b) January 2004 (simulation days 1–31, case 5)

During case 4, significant sea level heights were observed along the south coast of Turkey. Generally, this low pressure system swept across eastern Mediterranean and resulted in the SLH-rise incident in the coastal area SE of Antalya. A closer analysis reveals that there are actually two different storms that intersect right above the Aegean (Fig. 5). The first storm, with a 48-hour head-on, is sweeping the African coasts starting from Tunisia and heading East (SW-NE), while the other one comes from the North (N-SE). The two intersect on the afternoon of December 15 of 2003 (model day: 349) above the Aegean and then a stronger storm front heads to the Black Sea, whereas a secondary low pressure system moves towards Southern Turkey and to the inlands of Asia and on the 350.5 day it is located right above the south coast. The storm surge of December 15, 2003 and its impact in the Seyhan-Tarsus Delta area’s sea elevation is presented in Fig. 13a. It is obvious that there is a SLH-rise on 15-12-2003 (day 349.5) but it reaches its maximum value during day 350.5. Similar results arise during case 5, where this storm over the Aegean and Asia Minor storm affected the sea level of the south Turkey coast (Fig. 13b). The difference between the two storms is that in case 5, the trajectory started from Aegean and ended above Asia Minor, without passing near the studied Seyhan Tarsus delta. In both cases though, the strongest winds in NE Levantine Sea of each year, occurred during these storms’ appearance (Fig. 14a, b). The winds in both cases blow

Discussion and conclusions The SLH model output for different coastal areas can be used to estimate the inundation period for low altitude areas. This procedure requires the use of either the measured sea surface elevation (for the coastal areas close to the gauge station and the period of available data), or the SLH values computed by the storm surge model (for any part of the coastal zone). The mean yearly inundation frequency (%) is listed in Table 8, estimated by the model for two specific storm surge events: (a) the sea level rise is greater than 0.30 m (above the mean sea level), and (b) the sea level rise is 0.60 m above the mean. The results indicate

b 1035

1030

1030

1025

1025

mean SLP (hPa)

a 1035 mean SLP (hPA)

Fig. 15 İskenderun measured SLP series for (a) December 2003 (case 4) and (b) January 2004 (case 5)

from the open sea towards the coastal area (SW winds), pile up the water to shallow depths and increase the sea level along the Seyhan-Tarsus delta. From the meteorological station of İskenderun (Turkey) situated near to the delta (ftp://ftp.ncdc.noaa.gov/pub/data/gsod) the sea level atmospheric pressure was collected for the period of each case. Significant SLP decrease is observed during the period of each storm surge case (Fig. 15a, b.). The lowest SLP of the area is higher than the low SLP pressure of the storm system centre (Table 4), but the combination of the decrease in the SLP over the study delta area with the strong SW winds resulted these significant sea level rise, for both storm cases.

1020 1015 1010

1020 1015 1010

1005

1005

1000

1000

995

995 335

345 355 Time (days)

365

1

11 21 Time (days)

31

366 Coastal Area Adriatic & Ionian Sea Rovinj (HR) Dubrovnik (HR) Lefkas (GR) Split (HR) Trieste (IT) Ancona (IT) Otranto (IT) Zadar (HR) Patraikos Gulf(GR) Venice (IT) Manfredonia Gulf (IT) Neretva Delta (HR) Albanian coasts (AL) Aegean Sea Alexandroupoli (GR) Chios (GR) Kavala (GR) Herakleio (GR) Thessaloniki (GR) N. Levantine region Antalya (TR) Seyhan-Tarsus (TR)

that coastal flooding occurs with a frequency which in not neglible. The highest frequencies in both cases are observed in north Adriatic and in south Turkey, where also the highest sea level rises were calculated. The 30 cm overtopping frequency decreases from north to south, in the Adriatic Sea. This shows the higher risk level of the north closed coast of each region (dead end), where the wind dominated surge in combination to the channel shape of the sea piles up the water northwards during storms (solid line, Fig. 16) (Bondesan et al. 1995). In the Aegean sea, the trend is different (dash line, Fig. 16); the values in the central area are higher than in the northern and southern areas due to the sea level alteration that usually occurs right below the low pressure systems. This trend points to a stronger relation between the storm surge and the atmospheric pressure than with the relative winds. Additionally, the maximum sea level values during the study period followed a similar distribution, i.e. the highest values occurring in the N. Adriatic (Fig. 17). The complex morphology of the Aegean, with narrow passes and numerous islands is considered to be an important restraining factor in the wind driven surge events. Generally, the sea level rises, due to storm surges in the Aegean, are significantly lower in magnitude than in the Adriatic (Table 5). In other words, the wind driven storm surges influence more the

maxSLH (m)

date

%>30cm

%>60cm

0.575 0.646 0.442 0.634 0.625 0.594 0.520 0.580 0.398 0.671 0.563 0.674 0.532

26-12-2004 05-02-2003 03-02-2001 05-02-2003 26-12-2004 05-02-2003 05-02-2003 04-02-2003 05-02-2003 07-11-2000 05-02-2003 05-02-2003 05-02-2003

1.41 0.55 0.10 0.90 1.91 1.29 0.16 0.90 0.10 1.49 0.52 0.82 0.14

0.00 0.01 0.00 0.02 0.01 0.00 0.00 0.00 0.00 0.03 0.00 0.03 0.00

0.578 0.476 0.536 0.437 0.535

06-02-2003 02-02-2003 06-02-2003 23-01-2004 06-02-2003

1.21 1.09 0.67 0.50 0.89

0.00 0.00 0.00 0.00 0.00

1.171 1.256

23-01-2004 23-01-2004

3.22 4.43

0.21 0.41

north-ends of the two closed basins and although more pressure-driven surges occur in the Aegean, north to south frequency overtopping reduction, seen in the Adriatic, is not observed there. The sea level fluctuation increases during the study period, in many areas of the NE Mediterranean region. During the last simulation year (2004) the maximum SLH in the Adriatic Sea and especially along the SE Turkish and Cyprus coasts approaches, and in many cases exceeds, the 1 m level (Fig. 18). High overtopping frequencies also occur in the same areas. In the majority of the Greek coasts the sea level rise ranges between 2.5 30 cm overtopping (%)

Table 8 Max SLH and overtopping frequencies (%) for several areas in 2000–2004

Y.N. Krestenitis et al.

2 1.5 1 Adriatic Sea

0.5 0 47

Aegean Sea

45

43

41

39

37

Latitude

Fig. 16 Frequencies of SLH>0.3 cm in various areas of the Adriatic and Aegean Sea in the years 2000–2004 (model hindcasting)

Coastal inundation in the north-eastern mediterranean coastal zone

367

0.7 Max sealLevel rise (m)

0.65 0.6 0.55 0.5 0.45 0.4

Adriatic Sea Aegean Sea

0.35 0.3 47

45

43

41

39

37

Latitude

Fig. 17 Maximum SLH in various areas of the Adriatic and Aegean Sea in the years 2000–2004 (model hindcasting)

0–0.5 m. An exception is the south east region, at the Levantine-Aegean passage, where maxima higher than 0.5 m appear. Good yearly correlation between sea level pressure and sea level height exists mostly in the Aegean Sea and less in the Adriatic and North Levantine Seas. In the latter areas the winds and especially those of southern direction are the dominant factor causing the sea level to reach maximum heights. So, wind driven surges appear more frequent in the NE Mediterranean region, following the general rule for the entire Mediterranean Sea, where, in such enclosed regions, the inverted-barometer effect is quite weak. The combination of the morphological complexity with the storm track directions in the Aegean Sea induces a deviation from this rule. That is, in many cases the sea level pressure factor tends to be dominant. Storm surge events in the Aegean Sea exhibit lower magnitudes and overtopping frequencies than in the other studied areas. The dominant storm surge direction is from West to East. Due to the complexity of the coastal morphology and the high population of coastal areas in the largest part of the study area, storm surge events can be extremely dangerous for the human coastal economy and infrastructure. Areas

with high risk of inundation, due to their low altitude and mild slope, exist in all Mediterranean countries. In situ measurements are absent in many such areas, so the use of alternative tools, like mathematical simulations and satellite images, is imperative. A useful tool, in terms of generating reasonably based scenarios of forecoming flooding events in the coastal areas of the Mediterranean, can arise by combining a method of storm tracks identification and observation of their dominant characteristics together with the identification of coastal areas in risk of inundation. Warning systems of coastal flooding improve authorities’ preparedness and help coastal human society to take proper actions in cases of imminent extreme meteorological incidents. In sustainable and integrated development of the coastal zone, a storm surge event is a crucial factor, affecting significantly the quality of life. Additionally, hindcasting storm surge events and studying previous inundation cases can offer to stakeholders a better view of possible changes and measures that need to be taken in already developed, risky coastal areas. This is more crucial for E.U. countries because the members states should adopt the European Directive 2007/60/ΕC about the assessment and management of flood risks, while by 2011, they should undertake a preliminary flood risk assessment of the river basins and the associated coastal zones, in order to identify areas where potential significant flood risk exists. Furthermore, by 2013 they have to develop flood hazard maps and flood risk maps for such areas and by 2015 flood risk management plans must be drawn up for these zones. Combining atmospheric forecast with storm surge forecast is a next step in this study in order to produce a common prediction method of SLH-rise incidents in the Mediterranean. Thus, the combination of atmospheric forecasting modelling with a storm surge hydrodynamic model may estimate with good accuracy near-future storm surge events and their accompanied sea level rise.

Fig. 18 Frequencies of SLH>40 m (model hindcasting) along the East Mediterranean region in 2000–2004 (a) and maximum heights for the same area in 2004 (b)

368 Acknowledgments This study was carried out with the financial contribution of the E.U. project CORI: «Prevention and management of sea originated risks to the coastal zone», INTERREG IIIB/ ARCHIMED. We thank Gerasimos Korres and the POSEIDON team (Hellenic Center for Marine Research) for providing the POSEIDONSKIRON atmospheric and wave fields. The satellite images were downloaded from the AVISO website: www.aviso.oceanobs.com. We also thank the anonymous reviewer for his helpful suggestions and acknowledge ICPSM of the Municipality of Venice for the kind concession of the ISMAR-CNR data for year 2001.

References Amorocho J, DeVries J (1980) A new evaluation of the wind stress coefficient over water surfaces. J Geophys Res 85:433–442 Bajo M, Zampato L, Umgiesser G, Cucco A, Canestrelli P (2007) A finite element operational model for the storm surge prediction in Venice. Estuar Coast Shelf Sci 75:236–249 Bargagli A, Carillo A, Pisacane G, Ruti PM, Struglia MV, Tartaglione N (2002) An integrated forecast system over the Mediterranean basin: Extreme surge prediction in the Northern Adriatic Sea. Mon Weather Rev 130:1317–1332 Bondesan M, Castiglioni GB, Elmis C, Gabbianellis G, Marocco R, Pirazzoli PA, Tomasin A (1995) Coastal areas at risk from storm surges and sea-sevel rise in Northeastern Italy. J Coast Res 11:1354– 1379 Cazenave A, Bonnefond P, Mercier F, Dominh K, Toumazou V (2002) Sea level variations in the Mediterranean Sea and Black Sea from satellite altimetry and tide gauges. Glob Planet Chang 34:59–86 European Parliament and Council (2007) Directive 2007/60/EC on the assessment and management of flood risks. Off J Eur Union 288:27–34 Laskaratos A, Krestenitis Y, Nittis K (1991) Experimental and numerical studies of storm surges in the Aegean Sea. Temporal and Spatial Variability of Major Floods Around European Coasts. Final Report, University of Athens, Research Project Leder N (1988) Storm surges along the east coast of the Adriatic Sea. Acta Adriat 29:5–20 Lionello P, Sanna A, Elvini E, Mufato R (2006) A data assimilation procedure for operational prediction of the storm surge in the northern Adriatic Sea. Cont Shelf Res 26:539–553

Y.N. Krestenitis et al. Moron V, Ullmann A (2005) Relation between sea-level pressure and sea-level height in the Camargue (French Mediterranean coast). Int J Climatol 25:1531–1540 Paskoff R (1993) Côtes en danger. Pratiques De La Géographie Masson, Paris Papadopoulos A, Kallos G, Katsafados P, Nickovic S (2002) The Poseidon weather forecasting system: An overview. Glob Atmos Ocean Syst 8:219–237 Pirazzoli PA (2000) Surges, atmospheric pressure and wind change and flooding probability on the Atlantic coast of France. Oceanol Acta 23:643–661 Pugh DT (1996) Tides, surges and mean sea-level. John Wiley & Sons, Chichester UK Robinson R, Tomasin A, Artegiani A (1973) Flooding of Venice: Phenomenology and prediction of the Adriatic storm surge. Q J R Meteorol Soc 99:688–692 Rogers JC (1997) North Atlantic storm track variability and its association to both north Atlantic oscillation and climate variability of northern Europe. J Climat 10:1635–1647 Smith SD, Banke EG (1975) Variation of the sea surface drag coefficient with wind speed. Q J R Meteorol Soc 101:665– 673 Soukissian T, Hatzinaki M, Korres G, Papadopoulos A, Kallos G, Anadranistakis E (2007) Wind and wave atlas of the Hellenic seas. Hellenic Centre for Marine Research Publications, HCMR Sun G, Ranson KJ, Kharuk VI, Kovacs K (2003) Validation of surface height from shuttle radar topography mission using shuttle laser altimeter. Remote Sens Environ 88:401–411 Tsimplis MN, Baker TF (2000) Sea level drop in the Mediterranean Sea: An indicator of deep water salinity and temperature changes. Geophysical Res Lett 27:1731–1734 Tsimplis MN, Proctor R, Flather R (1995) A two-dimensional tidal model for the Mediterranean sea. Geophysical Res Lett 100:16223– 16239 Vilibić I, Orlić M, Ćupić S, Domijan N, Leder N, Mihanović H, Pasarić M, Pasarić Z, Srdelić M, Strinić G (2005) A new approach to sea level observations in Croatia. Geofizika 22, 21– 57 Vries H, Breton M, Mulder T, Krestenitis Y, Ozer J, Proctor R, Ruddick K, Salomon JC, Voorrips A (1995) A comparison of 2D storm surge models applied to three shallow European seas. Environ Softw 10:23–42 Wang XH (2002) Tide-induced sediment resuspension and the bottom boundary layer in an idealized estuary. J Phys Oceanogr 32:3113–3131