1 Introduction

The Eastern Mediterranean Basin has encountered numerous tsunamis even though they have a low frequency of occurrence in comparison to the Pacific and Indian Oceans. Historical documents and geological field investigations reveal that tsunamis occurred due to high seismicity, volcanic eruptions, and submarine/sub aerial landslides in the eastern Mediterranean region in the last 3,500 years. These events have been documented by many scholars.Footnote 1 Cita and Rimoldi (1997), Yalciner et al. (2007a), and Papadopoulos (2009) found evidence of tsunami deposits in Fethiye Bay as a result of geological investigations. Altinok et al. (2011) presents the most recent tsunami catalogue which contains 134 previously recorded tsunamis between the years 1410 BC and 2000. Forty-two of these tsunamis caused loss of property and damage in the Fethiye. In Minoura et al. (2000), the tsunami deposits found in Fethiye Calis beach (North of Sovalye Island) related to the 1630 BC Santorini event have been documented. Historical accounts of events and geological investigations warn that waves would be more damaging today due to increased coastal population and utilization in the area. Therefore, detailed assessment of the probable tsunami risk based on very high-resolution topographic and bathymetric datasets and reliable tsunami sources are necessary to be prepared for the next probable tsunami.

Numerical studies (Dilmen, 2009 and Yalciner et al., 2011) to estimate the maximum inundation limit in Fethiye and surrounding bays were conducted. These studies indicate that tsunami waves reach run-up values up to 3 m at the northern coast of Fethiye Bay. These results do not take into account the detailed risk assessment of the coastal infrastructures due to the limited resolution (15 m) of bathymetric/topographic datasets for the region.

In the present study, a high resolution bathymetry and coastal topography dataset (3 m) collected for Fethiye is used to determine the tsunami risk estimated from 14 tsunami scenarios at the Gulf of Fethiye. We use a numerical model of tsunami generation, propagation, and inundation (NAMIDANCE) to investigate and estimate the effects of a range of probable tsunami sources on the coast of Fethiye. The goals of the present research are to analyse the probable tsunami risk at Fethiye Bay and to develop tsunami inundation maps showing flow depths quantitatively.

The remainder of the paper is organized as follows: Sect. 2 introduces study area. Section 3 describes tsunami numerical model and simulations. Discussion and conclusions are presented in Sect. 4.

2 Study Area

Fethiye is a city and district of the Mugla Province of Turkey situated in the eastern part of the Mediterranean Basin (see Fig. 1). Fethiye and the surrounding region consist of a very narrow shelf and a large canyon connecting the deep bottom of Rhodes Basin with Fethiye Bay. The shelf is wider along the inner margins of the Fethiye and Marmaris Bays. The shelf’s maximum depth is 350 m (Ocakoglu 2012). This area is highly prone to earthquakes, volcanic activities and submarine landslides due to its proximity to active plate boundary zone between the Eurasian and African Plates that could be an initiative source of a probable tsunami.

Fig. 1
figure 1

Plate boundaries of the study region are given as contour maps (left inset). Location of Fethiye Bay, Fethiye Marine, and Calis Beach are depicted at the right inset

Full size image

3 Tsunami Numerical Model and Simulations

The tsunami numerical simulation and visualization code NAMIDANCE is used to compute the tsunami parameters of extreme, but probable tsunamis. The model computes tsunami wave generation, propagation, and inundation by solving non-linear, shallow-water wave Eqs. (1–3). These equations are solved numerically using the finite difference technique and leap-frog scheme for basins of irregular shape and bathymetry with four layers of telescoping grids (Goto and Ogawa 1991; Yalciner et al. 2006; Imamura et al. 1989). The smallest grid has the highest resolution.

$$\frac{\partial \eta }{\partial t} + \frac{\partial M}{\partial x} + \frac{\partial N}{\partial y} = 0 \,$$
(1)
$$\frac{\partial M}{\partial t} + \frac{\partial \eta }{\partial x}\left( {\frac{{M^{2} }}{D}} \right) + \frac{\partial }{\partial y}\left( \frac{MN}{D} \right) + gD\frac{\partial \eta }{\partial x} + \frac{k}{{2gD^{2} }}M\sqrt {(M^{2} + N^{2} )} = 0$$
(2)
$$\frac{\partial N}{\partial t} + \frac{\partial \eta }{\partial y}\left( {\frac{{N^{2} }}{D}} \right) + \frac{\partial }{\partial x}\left( \frac{MN}{D} \right) + gD\frac{\partial \eta }{\partial y} + \frac{k}{{2gD^{2} }}N\sqrt {(M^{2} + N^{2} )} = 0 \,$$
(3)

Here, η is water surface elevation, M and N are discharge fluxes in the x and y directions, respectively, D is total water depth, H is the undisturbed basin depth, and k is the bottom friction coefficient.

The model has been extensively tested against the analytical and experimental benchmark problems (Briggs et al. 1995; Kanoglu 2004; Kanoglu et al. 2013) and verified by field data of the 2004 Indian Ocean tsunami event and 2011 Great East Japan Earthquake and Tsunami (Okal and Synolakis 2004; Yalciner et al. 2007a, b, Yalciner et al. 2011). The model is successfully applied to historical tsunami simulations such as 1867 Virgin Islands (Zahibo et al. 2003), 1883 Krakatau (Choi et al. 2003), 2004 Indian Ocean (Choi et al. 2005; Yalciner et al. 2006), and tsunami scenarios in the Mediterranean (Yalciner et al. 2002, 2004) and Caribbean Seas (Kurkin et al. 2003).

3.1 Tsunami Model Set-Up

Maximum tsunami wave amplitudes and flow depths are computed to estimate the extreme, but plausible tsunami risk and to create inundation maps of the Fethiye Bay. The simulation described hereafter is initialized with a highly accurate bathymetry/topography and a worst-case tsunami source. The model setup of four layers of nested B, C, D, and E grids are created from 3 m resolution of topography/bathymetry of Fethiye (see Fig. 2). Table 1 includes the list of nested grid boundaries. The resolution of the innermost grid for the simulations presented in this study is 3 m (1/10 s). Manning's roughness coefficient is taken as n = 0.025. The simulation takes 5 h in real time.

Fig. 2
figure 2

Nested B–C–D–E grids are overlaid with Google Earth Image. The innermost grid E has the highest resolution of 3 m (1/10 s)

Full size image
Table 1 NAMIDANCE model grid boundary coverage and cell size for Fethiye
Full size table

3.1.1 Bathymetry/Topography Dataset

Bathymetric and topographic datasets for the Gulf of Fethiye are obtained by numerous methods in different terrestrial environments and at various scales and resolutions (see Table 2). It expresses geomorphologic features and the shoreline detailed enough so that it takes into account of the local effects for accurate tsunami propagation and run-up results.

Table 2 Detailed information of collected bathymetric and topographic data
Full size table

The coordinates and elevation of Fethiye shoreline are measured with differential GPS. It has the horizontal and vertical accuracy of 20 cm. A highly accurate digital elevation model at near-shore bathymetry and topography is generated with high-resolution Airborne Quickbird Multispectral Satellite Imagery. General Bathymetric Chart of the Oceans (GEBCO) data of 180 m (1 min) resolution is used for the location and elevation data of the offshore region. All data are assessed for quality and accuracy within each dataset to ensure consistency of the data and gradual topographic transitions along the edges. The final created DEM’s horizontal datum is World Geodetic System 1984 (WGS-1984) and vertical datum is taken as mean sea level due to the maximum tide difference of 30 cm.

3.1.2 Tsunami Source Estimation

Fourteen extreme, but likely sea floor deformations between 7.5 and 8.4 (Table 3) are initialized to be the worst case tsunami source scenarios based on the deliverables of Tsunami Risk and Strategies for the European Region (TRANSFER) Project partners.Footnote 2 Parameters of these scenarios are given in Table 3. These parameters are used to estimate probabilistically the tsunami risk at Fethiye Bay. Return periods for the earthquake sources are estimated as 500 and 1,000 years by FORTH(2). METU-OERC(2) and NOAGI(2) follow a deterministic approach for initial source estimation. According to seafloor deformation results, METU-OERC(2) earthquake source 10 generates highest uplift and subsidence of initial tsunami source and highest maximum wave amplitudes inside the harbor at selected virtual synthetic gauges. The uplift and subsidence map of this source is given in Fig. 3.

Table 3 Probable Tsunami source parameters
Full size table
Fig. 3
figure 3

METU-OERC tsunami scenario 10. Red and blue colors represent the uplift and subsidence of scenario 10, respectively

Full size image

3.2 Simulations

Tsunami simulation of Fethiye Bay is performed with and without inclusion of Sovalye Island to the bathymetric and topographic datasets. Virtual gauges are placed to compare synthetic tsunami waveforms for the worst-case tsunami scenario at the entrance of the bay and inside the bay. The time histories of virtual gauges Offshore 01 and D-1 are shown and plotted at Fig. 4. In the simulations while the wave motion and possible amplifications inside Fethiye Bay are evaluated, the wave attenuation effect of the Sovalye Island in front of the bay was also examined by performing additional simulation with the absence of Sovalye Island. The time histories of water surface elevations at two virtual gauge points outside the bay (Offshore-1) and inside the bay (D-1) are plotted in Fig. 4 with or without Sovalye Island cases.

Fig. 4
figure 4

Location of the selected virtual gauges (left inset) and modeled water levels (m) at the virtual gauges of OFFSHORE-01 (40 m depth) and D-1 (10 m depth) for scenario 10 (right inset)

Full size image

Time histories of the selected gauges show that tsunami waves arrive to Fethiye Bay at less than 10 min. The waves reach 2.1 m elevation inside the bay at numerical gauge point D-1. The maximum water elevation at numerical gauge point D-1 increases to 2.6 m at the case where Sovalye Island is not included at the bathymetry (see Fig. 4).

3.3 Tsunami Inundation Map for the Worst-Case Scenario: 10

A map of quantified tsunami maximum flow depths and inundation results overlaid with Google Earth Image for Fethiye Bay are shown in Fig. 5. Computed tsunami flow depths and inundation are shown on the map in three color-coded depth ranges: 0–0.5, 0.5–2, and >2 m. These depth ranges are chosen because they are approximately knee-high or less, knee-high to head-high, and more than head-high to represent the degree of hazard for life safety. The greatest tsunami flooding is expected to occur on low lands close to Calis Beach on the northeastern shores of Fethiye Bay.

Fig. 5
figure 5

Tsunami inundation map shows the maximum flow depths due to maximum probable tsunami source METU-OERC 10 in Fethiye Bay

Full size image

Elsewhere, tsunami flooding occurs in areas of low topography at the shoreline that have maximum flow depths exceeding 3 m due to the existing conditions. The protection level of Sovalye Island has been investigated by performing simulations with and without the existence of the island. We assert that Sovalye Island, located at the entrance of the bay, partly reduces the impact of the tsunami inside the bay.

4 Discussion and Conclusions

We present the spatial distribution of maximum tsunami flow depths and inundation areas as a result of the tsunami model for the extreme, but plausible tsunami. There was uncertainty regarding the likely earthquake epicenter and the fault rupture mechanism that could cause the worst-case tsunami scenario. Therefore, 14 probable earthquakes were selected, and the rupture parameters of these earthquakes were chosen from project deliverables of the EU funded research, Tsunami Risk and Strategies for European Union (TRANSFER), as potential sources of earthquakes and tsunamis. The earthquake that gives the highest amplitudes in front of Fethiye bay and maximum tsunami flow depths on the Fethiye coast was selected as worst-case tsunami scenario. Simulation of worst-case scenario gave us the quantitative map of inundated areas showing, maximum possible tsunami flow depths, and inundation boundaries.

Inundation is higher in shallow and low-land near-shore regions with flat onshore topography, like the northeastern part of Fethiye Bay. Other areas along the lowland also displayed significant flow depths of more than 3 m, thus showing that in the case of a tsunami, significant inundation is likely to occur. These areas correspond to high-density populated areas including recreational facilities and other infrastructure such as housing, hotels, parks, and restaurants. The existence of these structures in regions with flat topography increases the potential damage of a tsunami.

Sovalye Island located in front of the Fethiye Bay partly prevents entering the tsunami energy inside the bay. On the other hand, it prevents the outgoing of the tsunami energy and causes long duration agitation inside the inner bay. Therefore, the coastal amplification and inundation inside the bay remain significant. The sandy beach at the north coast outside the bay will receive the direct impact of tsunami waves, and hence more inundation distances and flow depths will be observed there.

Tsunami flow depths and inundation limits presented here are computed without the inclusion of the bottom friction due to infrastructures on land. This fact might result in slight overestimation of inundation and flow depths. In spite of this limitation, the results presented here allow the identification of the most vulnerable areas quantitatively that should be evacuated in case of a tsunami in the region.

Incorporation of the information generated from the numerical models provided to determine the areas that could be potentially affected by tsunamis affords policy makers an additional level of preparedness. Because of increasing population and city development throughout the northeastern coasts of Fethiye Bay, it is necessary to disseminate the results to inform locals of the level of threat in their region. The dissemination of this information will provide an additional level of confidence for visitors and residents of coastal locales. Therefore, our research can serve as a guideline for the development of quantitative tsunami risk maps for the eastern Mediterranean region.