Flood hazards have been one of the most destructive and frequent global-wide natural hazards resulting in loss of lives, livestock, and economic assets (Slater and Villarini, 2016; Alfieri et al., 2017; Kreibich et al., 2017; Qiang, 2019; Zhai et al., 2020). Even though lowland and plain areas where 80 % of the world population live can create an easy way for urbanization, they also vulnerable to flood risk, and the hazardous effects of floods will increase in the future due to the changing hydrological cycle in recent years (Lamond et al., 2011). As the number of flood hazards increases, the amount of flood losses are going to follow a parallel trend, accordingly. Hemmati et al. (2020) stated that both the number of floods and destructive economic results have been drastically increased since the 1990s. Munich RE (2020) has compiled a natural catastrophe loss database on natural disasters since 1980s for analyzing and assessing losses resulting from natural disasters. The database reveals that number of floods and their destructive economic results have an upward trend at a global scale.

Independently of flood hazard, the tsunami, which can be a long- or short-term event, is rare but can cause catastrophic damage to economic and social assets and activities (Wolfgang, 2005; Kundzewicz et al., 2017; Subyani et al., 2017; Fukao, 1979). Devastating economic losses and loss of lives have been recorded for the countries that experienced tsunami events, especially for the last 2 decades (Nadim and Glade, 2006; Carreño et al., 2007; Cardona et al., 2010; Sørensen et al., 2012; Lane et al., 2013; Horspool et al., 2014; Goda and Abilova, 2016). Scientists have revealed significant and reliable hazard evaluation methods for tsunami hazard assessment according to adverse consequences of the experienced tsunamis (Jelínek et al., 2012).

Multi-hazard assessment of floods with different natural hazards can be found in the literature. For instance, climate-change-related flood hazard assessment has been widely investigated (Blöschl et al., 2017; Skougaard Kaspersen et al., 2017; Szewrański et al., 2018; Carter et al., 2018; Barkey et al., 2019; Yavuz et. al., 2020b). However, the investigations covering simultaneous assessment of flood and tsunami events have been limited. Even if the coincidence of flood and tsunami hazards may be experienced once in a blue moon, it should also be investigated due to the uncertainty in the time of occurrence of these natural hazards. The objective of this study is to reveal a statistical methodology to evaluate the aggregate potential hazard levels due to flood hazards with the presence of earthquake-triggered tsunamis.

As commonly used issues in stochastic hazard analysis of any kind of hazard in the literature (Bommer, 2003; Helton et al., 2010), aleatory and epistemic uncertainties are considered to generate multi-hazard analysis in this study. The exceedance of flood hazard is strongly likely depending on geological and meteorological circumstances; the hazard is included in the stochastic analyses conducted in this study as epistemic uncertainty. Since the occurrence of the tsunami is generally rare compared with flood hazards, tsunami events are inspected by considering aleatory variability in this study. Additionally, hypothetical earthquake magnitudes Mw are generated using Monte Carlo simulations to obtain a required number of random earthquake sources in the bathymetry.

The proposed methodology is applied to Fethiye city center, which is one of the most popular touristic destinations on the Western Mediterranean coast of Turkey. The selection of this site is based on the documented seven tsunami events throughout history and evidence of tsunami deposits found by researchers (Cita and Rimoldi, 1997; Papadopoulos, 2009; Altinok et al., 2011) around Fethiye Bay. The Fethiye coastline was hit several times with destructive tsunami waves reaching up to 1.8 m, and significant inundation distances were recorded (Papadopoulos, 2009). The location of the study area for the case study is shown in Fig. 1.

The probabilistic multi-hazard assessment approach is applied in this study. In doing so, the aim is to evaluate the two dynamic natural hazards one by one and simultaneously. In total, 523 historical earthquakes recorded between 1900–2013 are retrieved from the European Union-funded “Tsunami risk and strategies for the European region” (TRANSFER Project, 2022) project catalogue. The Gutenberg–Richter relationship is used to determine the best-fitted distribution for the historical earthquake magnitudes. The Gutenberg–Richter relationship is a mathematical expression of the relationship between a number of earthquakes and the Richter magnitudes (Mw) of these earthquakes that occurred in a specific region (Gutenberg and Richter, 1954). They proposed a widely accepted and commonly used empirical equation that explains the relationship between the occurrence probability of an earthquake depending on two seismic constants (i.e., a and b values) which define the frequency-magnitude distribution and the Richter magnitudes experienced in a particular region. The equation is defined as follows: 1log⁡N=-bMw+a, where N is the number of earthquakes experienced in the selected region, and a and b are the constants that are defined specifically for the selected region.

Tsunami hazard curves are generated based on the hypothetical earthquake magnitudes (Mw) produced from 100 000 Monte Carlo simulations. The NAMI DANCE software is used to simulate hypothetical earthquakes having Mw≥6.5 (USGS, 2022) and the resulting tsunami wave heights are computed at the coast of Fethiye city center.

Flood hazards having recurrence periods of 10, 50, and 100 years (Q10, Q50, and Q100) on the other hand is modeled by MIKE 11, MIKE 21 FM, and MIKE Flood considered with and without tsunami wave existence at the coasts (DHI, 2016a, b). As a more frequent flood period, Q10 is evaluated in detail and hazard maps are generated for the flood events having return periods of 50 years (Q50) and 100 years (Q100). Additionally, tsunami-drifted flood hazard levels are also provided for all three flood events to satisfy the multi-hazard assessment procedure presented in this study. Thus, hazard levels considering both flood, earthquake-triggered tsunami, and tsunami-drifted flood hazards can be compared for the selected region. The inundation levels presented in this study have resulted from the numerical analysis of both hazards. Potential hazard that can result from seismicity are not within the scope of this study. The flowchart of the methodology used in this study is illustrated in Fig. 2.

In this study, potential multi-hazard assessment because of the fluvial flood hazard (Q10, Q50, and Q100), with and without the presence of earthquake-triggered tsunamis, is analyzed for Fethiye city center. Inundated areas due to flood only, earthquake-triggered tsunami only, and multi-hazard conditions (i.e., flood + earthquake-triggered tsunami) are determined by numerical computations, and corresponding inundation levels are revealed for each hazard circumstance.

For all flood hazard events considered in this study, maximum water levels are observed within the riverbed. The inundated area due to flood is limited along the streamlines for inland sections. There are also small inundated sections that can be observed due to flood in some parts of the coast of the study area. A large portion of the coastal region is not affected by the flood waves and the inundated area is limited in the coastal parts. A sample inundation map of the study area is given in Fig. 10 for the flood of Q10, Q50, and Q100 obtained from the numerical computations.

Although a water inundation level exceeding 6 m is observed on some parts of the Kurtbeli_1 stream, the effect of the Q10 flood is limited at the coastline. Depending on the computation results, the other streams also have small inundations around the riverbeds.

For the flood hazard having a 50-year return period, maximum water levels are observed only within the riverbed again, but the inundated area is slightly extended compared with the flood event having a return period of 10 years as expected. A large portion of the coastal region is not affected by the flood waves, and the inundated area is limited in the coastal parts.

Depending on the simulation results, a flood hazard having a 100-year return period generates maximum water levels within the riverbeds in the study area. The inundated area due to flood is limited along the streamlines for inland sections. There are also small inundated sections that can be observed due to flood in some parts of the coast of the study area. A large portion of the coastal region is not affected by the flood waves, and the inundated area is limited in the coastal parts.

For the earthquake-triggered tsunami hazard conditions on the other hand, a significant portion of the coastline is estimated to be inundated with 3.5 m tsunami wave heights (see Fig. 11). Compared to the flood hazard level, earthquake-triggered tsunamis might have considerable inundation levels at the coastline. Up to 1 km of land inwards from the coastline is estimated to be inundated due to tsunami waves depending on the hypothetical earthquake-triggered tsunami analysis.

On the other hand, the coastline of the study area is severely inundated due to flood (Q10) which takes place slightly before tsunami peak waves hit the coastal parts of the city. Although the maximum tsunami wave height obtained from the simulations is around 3.50 m, the inundation level for the multi-hazard conditions reaches up to 7.00 m for some parts of the low-lying sections of the study area (see Fig. 12).

Even for the contribution of flood hazard having the shortest return period (i.e., Q10) in the study, the multi-hazard inundation level reaches up to 7.00 m. It will not be surprising that higher inundation levels are definitely observed for multi-hazard assessment with Q50 and Q100 flood hazards.

Quantification of the flood hazard is also carried out for all three case studies by considering the threshold values and classes given in Table 3 (Smith et al., 2014). The results of hazard quantification are presented for all return periods in Fig. 13.

According to the hazard vulnerability classification proposed by Smith et al. (2014), all three flood events having different recurrence intervals (i.e., 10, 50, and 100 years) fall into the H1 hazard class, which can generate negligible adverse effects in some coastal parts of the city center. On the other hand, the fully developed flood events for all three recurrence intervals just after peak tsunami waves reach the coast resulted in varying hazard classes of H1 to H6. It should be noted that the major portion of the hazard is caused by the tsunami.

It can be seen from Fig. 13 that the inundated area is slightly enhanced due to the rate of change in flood discharges coming from the rivers. However, a huge portion of the hazard resulted from the effect of earthquake-triggered tsunamis. It can also be estimated that the rate of change in discharges coming from the rivers may also have some positive effect on the reduction in the additional adverse effect of multi-hazard conditions, due to encountered flows at the coastline.

Fluvial flood hazards having different recurrence intervals and potential earthquake-triggered tsunami hazards are simultaneously analyzed to evaluate the height of inundation levels at the coastline of Fethiye Bay and Fethiye city center. Results demonstrate that the majority of the increase in inundation levels is due to tsunami hazard. However, it should be emphasized that inundation levels are almost doubled in the presence of all flood hazard events at the same time. In the analyses, it is assumed that a fully developed fluvial flood takes place just after the peak tsunami waves hit the coastal region. Therefore, sea levels are determined accordingly for the hydraulic models.

Floods with 10-, 50-, and 100-year recurrence periods were taken into consideration in the study and potential hazards are calculated. Although it is more sophisticated to reduce the effects of tsunamis, the prevention of floods as well as their consequences are a more common procedure. Thus, combined risk analyses of multiple hazards should be taken into consideration in order to reduce risks due to natural disasters.

In conclusion, the coincidence of flood and tsunami events might be very unlikely. But the combination of these two hazards definitely increases the inundation levels and corresponding disaster levels in the selected region. Some other factors, such as seasonal changes in economic and social aspects, the expansion of residential sites, proximity to the fault zones, and climate change effects, should be taken into consideration in combined risk analysis for future years.