1 Introduction

The Adriatic Sea Coast in Albania is the country’s most densely populated, urbanised, and developed region, prominently featuring the significant Durres City. Various key socio-economic sectors, including tourism, industry, ports, fisheries, and energy facilities, are concentrated along this coastal area (AKPT 2016a). Despite the abundance of opportunities and resources, the region faces challenges associated with a variety of natural hazards, including floods, windstorms, droughts, earthquakes, and possible tsunamis (UNDP 1981, 2003; Muco et al. 2012, Sutton et al. 2013). For instance, Muco et al. (2012) state that in Albania, intervals of relatively low seismicity last 31–43 years, and relatively high seismicity last 17–21 years. Consequently, our study is centred on scenario-based tsunami modelling and exposure analysis specifically for the Durres coastal area. This area has been identified by the Intersectoral Integrated Plan (i.e. the official regional landuse plan) as one of the most crucial economic zones in Albania and the broader Balkan region (AKPT 2016b). The primary objective of the regional plan is to outline key strategies for advancing coastal development, focussing on mitigating the adverse effects of unregulated spontaneous growth in the area. It aims to guide sustainable development initiatives to optimise assets over the next 15 years. The Integrated Intersectoral Plan ensures the alignment of territorial policies with the government’s transformative agenda. By converting the coastal generation into a significant driver of economic impact, the plan seeks to enhance competitiveness compared to other similar coastal regions. While directing the trajectory of generational development, this plan prioritises robust environmental protection to safeguard the region’s natural, scenic and asset-based (AKPT 2016b). The plan is crucial as it outlines a comprehensive approach to coastal development, addressing the immediate challenges of unregulated growth and the long-term goal of sustainability.

Large tsunamis in the Mediterranean basin are rare; several decades can elapse between two events, and the memory of tsunamis is short-lived (Papadopoulos 2009). In the Mediterranean basin, tsunami hazard assessments have so far been carried out based on individual credible worst-case scenarios (Petricca and Babeyko 2019; Mori et al. 2022; Hassan et al. 2023; Kamhawy et al. 2023) or probabilistic tsunami hazard assessment based on the concept of area sources (Lorito et al. 2008; Papadopoulos et al. 2014). Tsunami preparedness is improving in the Mediterranean Region, as the “Tsunami Early Warning and Mitigation System in the North-eastern Atlantic, the Mediterranean, and Connected Seas” Intergovernmental Coordination Group (NEAMTWS; IOC 2017) has been established. It is based on five national monitoring centres in France, Greece, Italy, Portugal, and Turkey. In 2018, the first long-term Probabilistic Tsunami Hazard Assessment (PTHA) from earthquake-induced tsunamis for the North-east Atlantic, Mediterranean and Connected Seas region was released, and it is referred to as the NEAM Tsunami Hazard Model 2018 (NEAMTHM18, Basili et al. 2021).

So far, several studies have investigated tsunamis in the Mediterranean–Adriatic Sea generated by inland and offshore sources (e.g. Amato et al. 2021; Denamiel et al. 2019; Maramai et al. 2019; Papadopoulos and Papageorgiou 2014; Pasarić et al. 2012; Paulatto et al. 2007; Peresain and Hassan 2024). According to the relative tsunami potential classification in the Mediterranean Sea, the East Adriatic zone, which covers the region under study, is classified as an intermediate level (Papadopoulos 2016). Overall, these studies highlight that a tsunami with a maximum amplitude of up to a few metres can be expected in the Adriatic Sea, in agreement with several historical events reported in the catalogues (e.g. Maramai et al. 2019; NCEI 2023; Pasarić et al. 2012; Pinat et al. 2005). Currently, however, there are no tsunami risk assessment studies despite the presence of seismogenic sources that can generate severe tsunamis for the Albanian–Adriatic coast or western Balkan countries, which is known as one of the most seismic active regions in Europe (BSHAP 2011; Danciu et al. 2021). Pasarić et al. (2012) identified four true tsunami events along the eastern Adriatic coast, emphasising the Montenegrin and Albanian coast’s potential to generate tsunamis due to the high seismicity. Moreover, Vittori et al. (2021), in the study of the geological impacts and tectonic environment of the November 26, 2019, Mw 6.4 Durres earthquake in Albania, highlighted the importance of seismic activity along the western Albanian coast in assessing seismic and tsunami hazards. Both studies underline the need for detailed and targeted tsunami hazard and risk assessment on the Albanian coast (e.g. Hassan et al. 2020), emphasising the importance of expanding investigations beyond the regional scale.

The seismicity of Albania is primarily attributed to the tectonic setting and geodynamics of the region, which further contributes to a notable seismic hazard and risk in Europe (Fig. 1a). Historical reports and seismicity studies indicated that Durres City (an old town known as Dyrrahum) had been strongly impacted by historical earthquakes (Table 1, Fig. 1b) in 334 or 345 AC, 506, 1273, 1869, and 1870 (Shebalin et al. 1974; Sulstarova and Kociu 1975). The report for aftermath damages and losses for the 1273 earthquake indicated that the town of Durres, with 25,000 inhabitants at that time, had been destroyed (Aliaj et al. 2010). Table 1 shows the events occurring offshore or inland near shore, with earthquakes that may have triggered tsunamis in 345 AC, 1270, and 1273 (NCEI 2023). Still, their validity has not been clearly identified nor is their tsunamigenic potential accurately known. Those events are part of the Euro-Mediterranean Tsunami catalogue (EMTC, Maramai et al. 2019), which includes tsunamis reported by instrumental, historical, and archaeological records of a seismic origin in the Mediterranean Sea. Albania’s western coast is suitable for verifying hazard models because tsunami records from historical EMTC earthquakes are available on both sides of the Adriatic Sea. The region was recently struck by the Durres 2019 Mw 6.4 earthquake, the strongest earthquake in Albania for the last 40 years and the worldwide deadliest earthquake for 2019 (Freddi et al. 2021; Vittori et al. 2021).

Fig. 1
figure 1

a Bathymetry of the study area from the ETOPO1 Global Relief Model from NOAA, that integrates topography, bathymetry, and shoreline data. The two selected Composite Seismogenic Source (CSS), given as transparent grey polygons (ALCS002 and ALCS018). Epicentre of Durres Mw 6.4, 2019 mainshock (black stars) from different agencies with the respective focal mechanism solutions (check Data and Resources for detailed information). Blue lines present the GEBCO contours (NOAA NCEI Visualization), derived from the GEBCO2021 gridded bathymetric data set. Brown solid lines for active faults retrieved from European Fault-Source Model 2020 (EFSM20; Basili et al. (2022)). Inset map showing the location of Albania with respect to other countries. b Map zoomed to show the distribution of the historical events (light blue crosses) with the labels (IDs) as a reference for Table 1. Administrative unit (borders) given by grey line, urban areas refer to the built environment (grey patterns) from Open Street Map, affected areas (dashed grey line) from Copernicus Emergency Management Service Monitors, mark the areas significantly affected right after the earthquake. Background seismicity showing events with Mw ≥ 3.5 for the period 2014–2020 with coloured circles according the depth (IGEO 2022). Dashed grey lines indicate the location of topographic profiles detailed in Sect. 2.1, the numbers indicate the profiles as a reference for plots in Fig. 2. Background map is plotted using ArcGISPro (Esri n.d.). Refer the text for the acronyms and data sources

Full size image
Table 1 List of historical earthquakes and tsunamis that occurred in the Durres municipality (Shebalin et al. 1974; Sulstarova and Kociu 1975; NCEI 2023)
Full size table

The spatial distribution of the main event and aftershocks indicate that they overlap well with the seismogenic zones of ALCS002 (Lushnje) and ALCS018 (Shijak) of Mw 7.5 and 6.8, respectively, as reported in the Database of Individual Seismogenic Sources—INGV (DISS–3.3.0). Notably, the fault plane solutions correlate well with the source parameter in DISS-3.3.0 (Fig. 1 and Table 2). The 2019 earthquake raised concerns regarding the potential of tsunami generation by larger magnitude earthquake of given maximum magnitude (Mmax) and their impacts on the coasts in the vicinity and north of the Adriatic, which have a great socio-economic value. This, in turn, calls for investigating the possible impacts in case a tsunamigenic earthquake of Mmax might occur so that civil protection and communities can be informed about the timing and extent of the impacts and eventually can be prepared.

Table 2 The Durres Mw6.4, 2019 earthquake parametric data determined by different agencies (INGV—National Institute of Geophysics and Volcanology, USGS—United States Geological Survey, ISC—International Seismological Centre, EMSC—European Mediterranean Seismological Center, IGEO—Institute of Geosciences Albania, AUTH—Aristotle University of Thessaloniki, GCMT—Global Centroid Moment Tensor, GFZ—German Research Centre for Geosciences)
Full size table

In this work, tsunami hazard computations, in terms of coastal inundation and maximum tsunami wave amplitude, are performed deterministically by the NAMI DANCE numerical code (Velioglu et al. 2016; Yalçiner and Zaytsev 2017; Lynett et al. 2017; Dogan et al. 2021), considering earthquake worst-case scenarios. NAMI DANCE allows accounting for seismic source properties, high-resolution bathymetry and topography, and non-linear effects in wave propagation. Earthquake–tsunami source data for this study were retried from the Database of Individual Seismogenic Sources (DISS Working Group 2021). In contrast, bathymetry and topography data are from the European Marine Observation and Data Network (EMODnet), and the Shuttle Radar Topography Mission (SRTM). To carry out the tsunami modelling, we selected two composite seismogenic sourcesFootnote 1 capable of triggering tsunamis in the study area, namely the potential tsunamigenic sources ALCS002 and ALCS018, shown in Fig. 1, and the historical events that may have triggered the tsunamis. Peresan and Hassan (2022) and Paulatto et al. (2007) have identified them as sources that can produce relevant amplitudes, thus causing significant damage in many coastal urban settlements in the Adriatic region.

The computed inundation maps are combined with information from exposure analysis to identify the areas where damages can be expected. Exposure analyses can range from global and regional scale (e.g. Crowley et al. 2020; Løvholt et al. 2014, 2016; Pittore et al. 2020; Scaini et al. 2024) to national and local scale (Paulik et al. 2019; Gomez-Zapata et al. 2021), with associated spatial resolution and level of detail. In the absence of local scale exposure datasets for the study area, we perform a preliminary exposure analysis to identify the areas where population and built environment are likely to be affected in case of tsunami. Our outcomes can be complemented in the future with vulnerability and risk assessment for the study area and support the definition of disaster risk reduction strategies such as evacuation plans.

The primary aims of this study are to analyse the propagation of tsunami waves generated from the identified sources and to identify the areas in Durres where damages might occur, preparing the ground for subsequent risk assessment based on a scenario-based approach. The findings of this study have several important implications for future practice, seeking to strengthen capacities for prevention, mitigation, and response to improve the resilience of communities and institutions in the coastal municipalities against earthquakes and tsunamis, particularly in the municipality of Durres.

2 Geographic and seismological settings

Albania, classified as one of the most seismic areas in the Mediterranean (Aliaj et al. 2010; Danciu et al. 2021), is characterised by lowlands in the west and hills and mountains in the east. Geologically and seismo-tectonically is a complicated region with combination of a large number of blocks (rigid blocks such as the Adriatic, certain sectors of the Alpine belt, the Alps, Carpathian Mountains, the Balkan Mountains, the Dinarides, Helenides and Hellenic Arch, and the Anatolian belt, as well as interior basins such as the Tyrrhenian, Aegean, Pannonia and the Black Sea), and the release of stress through plastic deformation (McKenzie 1972, 2020; Slejko et al. 1999; Meletti et al. 2000; Muco et al. 2012).

Many studies on the geodynamics and seismicity of the eastern Mediterranean state that the seismicity of Albania is strongly linked to the interaction of Adria with the Albanides orogen, which is part of a larger collision between the Eurasian and African Plates (Aliaj et al. 2010). Note that at the November 2019 Mw 6.4 Durres earthquake location, the African Plate converges with the Eurasia Plate at a rate of 73 mm/year (Alam et al. 2019). The orogen in Albania and its surroundings are divided into two domains with different present-day tectonics: an external compressional domain, constituting the Adriatic collision zone, and an internal extensional domain. The external tectonic zones (Sazani, Ionian, and Kruja), which make up the orogenic fronts in the Adriatic collision zone, constitute deformed sedimentary cover thrust towards the south-west (Aliaj 1997). Our study area is located on an actively seismotectonic belt, the Periadriatic Basin in the western Albanides, where seismic catalogues report several past events with a magnitude higher than 6 (Mancini et al. 2020).

The outer thrusts of the Apennines and External Dinarides propagated into the Adriatic from the coastal areas towards the offshore, resulting in a complex pattern of northwest–southeast trending anticlines, thrust faults, and back-thrusts and involving Late Pleistocene deposits (Kastelic et al. 2013). The geological map of Durres administrative unit (Qark) from the Albanian Geological Service, indicates that Durres City is situated atop deposits originating from the Holocene and Pleistocene periods, predominantly characterised by the presence of sandstones and conglomerates, marshy, clays, alluvial and marine formations (Kociu 2004; Duni and Theodoulidis 2019; Theodoulidis et al. 2022). The Adriatic and western Balkan regions contribute to the behaviour of neotectonics of southern and central Europe (BSHAP 2011). The tectonic evolution and activity in this region are the main reasons for the high seismicity on the continent and the high seismic hazards and risks in Europe.

To better understand earthquakes or discuss seismicity, an essential step in seismotectonic analysis is directly related to the type of faults in a region. According to the fault plane solution, reverse faults along the coastal line are the dominant faults correlated to the thrusting in most parts of the External Dinarides and Albanides. The faults in the continental section are strike–slip, oblique strike–slip, or even reverse faults due to Adria’s counter-clockwise movements and compression against the Dinarides (BSHAP 2011). The border between normal faulting to the east and thrust faulting to the west extends through central Albania in the Albanides (Salic et al. 2018).

Only during this century, earthquakes in Albania and nearby have produced 7.2% of the overall energy generated by all shallow earthquakes in Europe during this time (Sulstarova 1996), or the energy released from the earthquakes of this belt amounts to 15% of the total energy released from the earthquakes of the whole globe (Aliaj et al. 2010). Albania’s earthquake typology includes microseismicity (1.0 ≤ M ≤ 3.0), many small earthquakes (3.0 ≤ M ≤ 5.0), few medium-sized earthquakes (5.0 ≤ M ≤ 7.0), and, very rarely, large earthquakes (M > 7.0) (Muco 1998; Aliaj et al. 2010). Depth distribution of earthquake focii in Albania is generally shallow, with a depth of 10–20 km and, in many cases, near the surface, for both historical and instrumental events (Muco 1998, 2006). The strongest earthquakes of the Adriatic domain occurred in the coastlines of the southern part of the Adriatic domain, while Mw ≥ 5.5 earthquakes are typical for the entire region (Kastelic et al. 2013; Rovida et al. 2011; Shebalin et al. 1974), and almost every year the region is hit by an earthquake with Mw > 6.5 (Papazachos 1988).

Historical records and seismicity studies reveal that Durres City was strongly impacted by earthquakes (Table 1 and Fig. 1b). These seismic events occurred in 334 or 345 AC, 506, 1273, 1869, and 1870 (Shebalin et al. 1974; Sulstarova & Kociu 1975). Particularly noteworthy is the earthquake of 1273, for which the aftermath report on damages and losses indicated the destruction of the town of Durres (Aliaj et al. 2010).

Table 1 shows the events occurring offshore or inland nearshore in Durres City, with earthquakes that may have triggered tsunamis in 345 AC and 1273 (NCEI 2023). Still, their validity has not been clearly identified, nor is their tsunamigenic potential accurately known.

2.1 Geographic location and geomorphologic features

The Durres municipality, located on the western coast of Albania along the Adriatic Sea, occupies the Durres Plain, fertile lowland shaped by sediment deposition from rivers and the sea (Braholli and Menkshi 2021). The district of Durres covers an approximate area of 430 km2, bordered by the Tirana district to the northeast and the Kavaja district to the south (Fig. 1). The city’s western boundary is defined by the shoreline of the Adriatic Sea, stretching for a distance of 30 km. The local topography is heavily influenced by compressional structures, where the western ridge is formed by the steep, east-dip** eastern limb of a reverse fault-related anticline. Conversely, the eastern ridge marks the highest point of a gentle anticline, particularly at Vrinas. Between these ridges lies a broad syncline, which underlies the coastal alluvial plain (Mancini et al. 2021). In terms of urban geomorphology, the city is situated on the eastern lowermost slope and extends towards the eastern coastal zone. This geographical positioning continues into the inner reclaimed plain areas, predominantly towards the north. Figure 2 presents six detailed 2D topographic profiles developed for Durres heading in north–south and east–west directions, which reveal that the city benefits from natural protection provided by the mountain range on its western side, specifically the hill known as Mali i Durresit, which reaches a height of 187 m.

Fig. 2
figure 2

Plots a to f show the detailed 2D topographic profiles developed for Durres heading in north–south and east–west directions, their distribution given in Fig. 1b. Naming of each subplot e.g. Profile 1–1’ corresponds to the topographic direction 1 to 1’

Full size image

3 November 2019, Mw 6.4 Durres earthquake

The municipality of Durres is part of the regional development pole Tirana–Durres and was envisioned as an international “hubFootnote 2” and as a “port cityFootnote 3” (AKPT 2016a). The municipality presents itself with a strong identity in favour of develo** the service economy, industry, logistics, and tourism typologies such as historical–cultural, natural/rural, ecotourism/agritourism, sports, services, and sea-sun tourism. The Tirana–Durres region is identified as the country’s economic engine (AKPT 2016b), an essential pole of regional development based on the principles of global economic competition in creating economic clusters and metabolic development in the territory.

The region was heavily affected by the 2019 Mw 6.4 Durres earthquake, the largest earthquake recorded by the Albanian Seismological Network (ASN) in the past 40 years. The last earthquake of a larger magnitude in the region was on April 15, 1979, with a magnitude of Mw 6.9. In November 2019, the ASN recorded a series of earthquakes, counting 420 earthquakes ranging in local magnitude from ML 0.7 to 6.3 (IGEO 2019). The event reported intensities were up to IX in Modified Mercalli Intensity (MMI) (Fig. 2 in Xhafaj et al. 2024a, b), and, thus, was classified as “destructive” (IGEO 2019; Andonov et al. 2022) by the European Macroseismic Scale (EMS-98; Grunthal 1998). According to official data from national authorities, the event caused severe damage; it was felt as far away as Bosnia and Herzegovina, Bulgaria, Italy, and Serbia (IGEO 2019). As a result of the disaster, a total of 202,291 people were affected in the country, 47,263 directly, and 155,029 indirectly. The earthquake caused 51 fatalities and injured at least 913 people; moreover, up to 17,000 people were displaced due to the loss of their homes (PDNA 2020). Most of the damages are recorded in the Housing sector (78.5%), followed by the Production (8.4%) and the Education (7.5%) sector. Regarding the losses, the Production sector accounts for the highest share (56.4%), followed by Housing (24.1%) and Civil Protection and Disaster Risk Reduction with 9.4% (PDNA 2020). This earthquake highlighted the vulnerability of buildings in Albania to large earthquakes and the need for stricter construction laws and enforcement to mitigate future seismic hazards.

The epicentre location of the November 2019 earthquake was well determined to the coastal area north of Durres City (Vittori et al. 2021), whereas the hypo-central depth of the mainshock remains controversial. According to the National Institute of Geophysics and Volcanology (INGV), the magnitude was estimated at Mw 6.2, with a hypo-central depth of 21 km. However, the Institute of GeoSciences (IGS) estimated a magnitude of Mw 6.4 at 38 km, while the European Mediterranean Seismological Centre (EMSC) fixed the depth at 10 km with a magnitude of Mw 6.4. Table 2 summarises the locations, depths, magnitudes and other parameters of the mainshock determined by various sources.

The co-seismic surface rupture zone of the November 26, 2019 earthquake is approximately estimated to be within a range of 20–30 km in length (Govorčin et al. 2020; Ganas et al. 2020; Papadopoulos et al. 2020; Vittori et al. 2021). The finite-fault solution of this event determined by Govorčin et al. (2020), using Global Navigation Satellite System and Differential Interferometry Synthetic Aperture Radar data suggest the south-west dip** thrust fault with slip distribution concentrated in the range of 20–30 km along the strike, at a shallow depth between 11 and 23 km with the slips of up to 120 cm. This aligns with previous findings that suggest the occurrence of reverse faulting in the outer lowlands of the western Albanian domain (Fig. 1b, Aliaj (2004); Kiratzi et al. (2007)). Papadopoulos et al. (2020) reported a rupture velocity of 2.6 km/s⁠ and total rupture duration of 16 s. A tsunami alert message of the Advisory Alert level was issued by the Tsunami Warning Center of the Italian Geophysics and Volcanology Institute—INGV (CAT-INGV), 6 min from the origin time of the event. This message was issued as advice to government agencies, but only national and local government agencies have the authority to make decisions regarding the official state of alert in their area and any actions to be taken. According to the CAT-INGV decision matrix, the “advisory” level is indicated when the run-up is expected to be less than 1.0 m, while “information” means no tsunami is foreseen.

The Mw 6.4 Durres earthquake has raised the question regarding the maximum potential expected from western Albania/Balkan seismogenic sources or the active coastal structures. Slejko et al. (1999) stated that the Albanian coast has structures adequate to generate Ms 6.0 earthquakes (Ms-surface wave magnitude). Evidence for the presence of great waves was reported a few times in the past centuries on the Adriatic shorelines due to similar events. This suggests investigating the maximum amplitude a potential tsunami event could reach. Therefore, the 2019 earthquake presents a significant case for investigating the region’s sources and exploring, for the first time, the tsunamigenic potential for Albania.

4 Input data and method

Modelling tsunamis generated from earthquakes is one of the most significant steps of the tsunami hazard assessment, which is carried out based on the identified seismogenic sources, bathymetric, and topographic data of adequate resolution, and the selection of plausible scenarios (Aytore et al. 2016).

4.1 Earthquake–tsunami scenarios

In this work, we developed a set of earthquake–tsunami scenarios for future Mmax from DISS-3.3.0, considering variability in seismological parameters. While the updated seismogenic source model DISS-3.3.0 has clearly shown the potential for large earthquakes here, the impacts of future earthquake–tsunami in Durres City are entirely unknown. This study attempts to address this by modelling the potential tsunami hazard associated with a range of plausible earthquake–tsunami scenarios through an aggregated scenario ensemble analysis to inform contingency planning and preparedness.

Table 3 shows the range of estimated earthquake source parameters and focal mechanisms in the case of 2019 Durres’s swarm, constrained and complemented by the updated DISS-3.3.0 seismogenic sources database, which can provide a more comprehensive understanding of the potential variations and uncertainties in the analyses. Recent earthquake parameters estimate the maximum plausible earthquake–tsunami in a scenario-based approach. This approach involves develo** hypothetical scenarios based on the characteristics of past earthquakes and applying them to specific regions or areas of interest. Analysing the recent earthquake data and understanding the seismotectonic characteristics of a region informed investigation about potential maximum earthquake–tsunami can be done.

Table 3 Earthquake–tsunami scenarios developed by combining information about fault orientation from the recent swarm of 2019 earthquakes and DISS-3.3. The results of the tsunami simulations based on the given data are shown in Figs. 4 and 5
Full size table

To model tsunami scenarios that may occur locally in Durres City, it is important to consider the ALCS002 and ALCS018 seismogenic sources delineated by DISS-3.3.0. These sources have Mmax of 7.5 and 6.8, respectively. However, accounting for uncertainties in estimating hypocentre and seismic focal mechanism parameters is crucial, as observed in the 2019 earthquake sequence with the mainshock Mw 6.4. These uncertainties are assumed to resemble those that may occur in future larger earthquakes constrained by the parameters from the seismogenic sources database in DISS-3.3.0. To address this, fault configurations were developed by combining information about fault orientation (i.e. strike, dip, rake) from the recent swarm of 2019 and DISS fault dimensions (i.e. length, width) for Mmax from the Wells and Coppersmith (1994) magnitude scaling relationship, which is constrained by seismotectonic and geologic data. This approach allows for the development of a database of fault orientations for the 2019 earthquake estimation, considering different agencies’ estimations and earthquake and tsunami data services providers (Table 2). Subsequently, a set of tsunami scenarios, namely ALCS002 and ALCS018, were modelled using these fault plane solutions (Table 3). These scenarios aim to encompass the potential variability in future earthquake-induced tsunamis with the Mmax.

By incorporating these fault configurations and considering uncertainties in the earthquake parameters (e.g. uncertainties in the length and width (± standard derivation) in Table 3), we used different tsunami scenarios to identify the worst-case scenario. The tsunami modelling efforts comprehensively assess the potential tsunami impacts on Durres City. Understanding the possible scenarios and their associated impacts is crucial for effective emergency response planning, coastal zone management, and the development of mitigation strategies to enhance the resilience of the city and its coastal communities.

In our study, we lack the necessary data to validate the accuracy and reliability of the tsunami hazard results specifically for the Durres region. In other words, while we have conducted research and obtained results related to tsunami hazard assessment in the Durres region, we could not verify or validate these findings due to the unavailability of appropriate validation data.

4.2 Topographic and bathymetry data

Without high-quality observational data in the study areas, we define the tsunami models based on the highest resolution bathymetric and topographic available datasets. The accuracy of the bathymetry is a critical factor for accurate inundation modelling (Griffin et al. 2015). The seafloor map was taken from the European Marine Observation and Data Network (EMODnet bathymetric dataset) to compute our simulation within the domain 19.0° E to 20.0° E longitude and 41.0° N to 42.0° N latitude. Tsunami propagation can be accurately evaluated using high-resolution bathymetry and topography data, which are prerequisites for simulating tsunami waves and describing tsunami wave interaction with topo-bathymetric features. The two nested domain grids (Figure S1a) are described as follows:

  • Large-domain grid, with 115 m spatial resolution, obtained through the European Marine Observation and Data Network (EMODnet) for the regions of interest.

  • Medium-domain topography grid, with 30 m spatial resolution, obtained through the Shuttle Radar Topography Mission (SRTM) for the regions of interest.

The large-domain grid (115 m) covers the coastal areas of interest, including the nearshore zones and offshore regions where tsunami wave propagation and behaviour are crucial for accurate modelling. The coverage area is selected to ensure a detailed representation of the underwater terrain relevant to the specific geographic location under study. It includes deep sea features and transition zones near the coastline, as these areas play a significant role in tsunami generation, propagation, and inundation dynamics. Offshore, the bathymetry slopes to depths of 250 m at about 20 km and a depth of 25 m along the coastline. Using a 115 m grid resolution enables a detailed and finely delineated representation of the terrain and features within this specific geographic area, as depicted in Figure S1b.

From a physical–geographical perspective, the medium domain (30 m) focussing on the Durres municipality gives a (detailed) view of the topography, shown in Figure S1c. Durres City is situated in the western lowlands, occupying a level alluvial plain topography surrounded by hills. The highest elevations are observed in the northeast of the November 2019 earthquake (reaching ~ 300 m) and the northwest of Durres port (reaching 170 m). Several beaches run parallel to the coast, while the shallow/flat platform behind the ridges drops off at depths of 10 m inland at a distance of 2.5–3.0 km from the coastline. The majority of the region comprises a plain that has taken shape through the reclamation of the Durres swamp over the past century. These flat topographic areas occupy most of the municipality, as shown from the profiles in Fig. 2 (a) to (f). It is worth mentioning that the marshlands are also flooded during heavy rainstorms, and stagnant waters persist locally (Malfitana et al. 2015).

4.3 Tsunami modelling code and setup parameters

A tsunami hazard assessment for Durres has been carried out using a scenario-based approach that incorporated information about two nearby seismogenic sources and available topo-bathymetry data. To accomplish the objectives, we used the NAMI DANCE code manual (Velioglu et al. 2016; Yalçıner et al. 2017; Yalçiner and Zaytsev 2017; Lynett et al. 2017; Dogan et al. 2021). This code is an extension and update of the TUNAMI-N1/N2 model and solves the non-linear form of the long wave equations considering the appropriate initial and boundary conditions. The adopted code has been validated on performance and against real observation in previous works (e.g. Velioglu et al. 2016; Sogut and Yalciner 2018). NAMI DANCE boasts a computational advantage by utilising Graphical Processing Units, significantly reducing processing time for early warning systems. Its user-friendly modelling practice is enhanced by including a graphical user interface. In addition, NAMI DANCE efficiently processes files from popular Geographic Information System (GIS) software, streamlining pre-processing tasks and saving valuable time.

Generally, the most commonly used and suitable approach for tsunami modelling is the numerical solution of the non-linear shallow water approximation (Imamura 1988). In tsunami modelling, non-linear effects consider non-linear wave behaviour during tsunami propagation and inundation simulation. This approximation neglects the vertical velocity variation over water depth due to the significant difference in tsunami wavelengths and ocean depths. For our computations in this study, we employ the non-linear shallow water equations to simulate tsunamis. Non-linear effects become significant when the amplitude of the tsunami wave is relatively large compared to the water depth. The primary non-linear effects in tsunami modelling are wave steepening, wave breaking, and wave dispersion. This consideration is crucial for capturing the complex interactions between different wave components and their impact on coastal areas. Earthquake-induced tsunamis are the only ones that can be systematically forecasted over large domains because seismic waves travel much faster than tsunami waves (Hirshorn and Weinstein 2011).

While our study has made significant contributions, it is important to acknowledge its limitations. First, the availability of topographic data constrained our ability to produce a high-resolution tsunami inundation map. In addition, the Manning coefficient, known to be site-dependent and influenced by factors such as topography, bottom type, and landuse, was not adequately addressed. In calculations, a standard constant value of n = 0.015 m−1 s is often employed, representing typical bottom characteristics such as sand or small pebbles (Zaytsev et al. 2019). The utilisation of fixed values does not account for these variations. Implementing a spatially varying Manning factor requires a thorough understanding of site-specific characteristics, including their resistance to inundation, which we currently lack. This variability is essential for precise zoning of coastal flooding caused by tsunami waves. Therefore, assigning these parameters without proper site knowledge could introduce biases. However, we conducted parametric tests to evaluate the impact of adopting different Manning coefficients ranging from 0.0025 to 0.4 every 0.0025, covering a wide range of average manning values across excavated or dredged channels and constructed domains (Chow 1959). Our analysis revealed that the maximum tsunami wave amplitude of the standard deviation ranged between 2.6 and 2.0 m. The maximum computed tsunami wave amplitude for a Manning coefficient of 0.015 m−1 s is 2.4 m. Future extensions of this work may focus on acquiring high-resolution topographic data, conducting detailed analyses and adopting various Manning coefficients as extensions of our current work. In addition, a time-step of 0.35 s was selected for computation based on sensitivity analysis. We incrementally increased the time-step by 0.1 s intervals until the modelled tsunami current velocity results stabilised, which occurred at around 0.60 to 0.70 s (Zaytsev et al. 2019).

The translation of slip-on fault plane(s) into seafloor deformation requires a method known as the “Okada Model,” which is based on Green’s function solution to the elastic half-space problem, as described in Okada (1985). In this model, a uniform displacement of the solid is applied over a finite rectangular patch with the specified parameters. When this displacement is inserted into a homogeneous elastic half-space below the free surface, it results in a steady-state solution where the free surface is deformed. This deformation is considered as the seafloor deformation (Fig. 3). It is important to note that this approach is an approximation, as the actual seafloor is rarely flat, and the Earth itself is not a perfectly homogeneous and isotropic elastic material, as assumed in the model. However, for tsunami modelling purposes, it is often considered a reasonable approximation. This is especially true since the precise parameters of fault slip are often poorly known, even for historical earthquakes. Therefore, these cases may not justify a more accurate modelling of the resulting seafloor deformation.

Fig. 3
figure 3

Maps a to g show the initial water surface displacement for the selected scenarios. An approximated representation of the initial water vertical displacement. In this model, the initial wave development is based on the Okada (1985) calculations for co-seismic tsunamis. Static finite rectangular fault with length (L) and width (W) for: a ALCS002, b AUTH, c GFZ, d INGV, e USGS, f CPPT, and g GCMT. Plot (h) shows the box plots demonstrating the variability in data through their quartiles in fault orientation and depth for the 2019 Durres earthquake provided by different providers indicated in Table 2 (check Data and Resources, and the related publications for detailed information), earthquakes in the 2019 swarm (Dragomirov and Dimova 2020), and DISS-3.3 database. Background map is plotted using ArcGISPro (Esri n.d.)

Full size image

The preliminary tsunami simulations for accurately estimating the boundary of the inundation zone are performed considering the ALCS002 and ALCS018 seismogenic sources. The fault dimensions are obtained according to the most widely applied scaling laws (e.g. Wells and Coppersmith 1994) for all fault types/styles. Table 3 gives the rupture parameters estimated in two scenarios with average and maximal parameters for rupture initiation at the corner of the source. The simulations from the above estimations are presented in the next section of this study, showing the potential of the ALCS002 and ALCS018 sources.

5 Tsunami modelling results and discussion

Following the definition of fault plane solutions from recent earthquakes, specifically ALCS002 and ALCS018, a set of earthquake–tsunami scenarios was modelled (Table 3). These scenarios were explicitly designed to capture the potential range of variability in earthquake-induced tsunamis that could occur with the Mmax estimated for the region. By employing these fault rupture models/scenarios, the modelling process considered the various parameters and characteristics associated with the fault sources ALCS002 and ALCS018. These fault plane solutions, which consider fault orientation, fault dimensions, and other relevant seismotectonic and geologic data, serve as the foundation for simulating tsunami scenarios. The primary objective of these modelled tsunami scenarios is to comprehensively assess the potential impacts and characteristics of future earthquake-induced tsunamis with the Mmax. By exploring a range of potential tsunami scenarios, researchers and stakeholders can gain insights into the potential variability in tsunami behaviour and its possible consequences for Durres City and its coastal areas.

The initial condition for each tsunami simulation is given by a static displacement of the seafloor, which is calculated assuming a uniform slip of dislocation in an elastic half-space, using the model of Okada (1985) with the relevant source parameters in Table 3 (Fig. 3). The tsunami intensity maps of interest are constructed by aggregating the simulated individual scenarios. These maps provide valuable information about the expected intensity of tsunamis in different areas. In addition, based on the aggregated scenario, various parameters such as the Expected Time of Arrival (ETAs), maximum wave height, average wave height, and standard deviation of wave height are calculated to create coastal tsunami wave height and inundation maps. Furthermore, the simulation of tsunami wave propagation allows for the generation of synthetic mareographs at specific locations. These synthetic mareographs provide detailed and site-specific data that can be used for further analysis.

Individual scenario tsunami hazard maps are integrated into a GIS project to provide a comprehensive overview. From this aggregation, the maximum (global maximum, i.e. the largest value registered at each site across the coastline), average, and standard deviation of coastal tsunami wave height maxima are computed, resulting in an aggregated tsunami hazard scenario map. The results and discussion section presents the outcomes of modelling the selected tsunamigenic earthquake scenarios (Figs. 3, 4, 5 and 6). Figures 4 and 5 present the aggregated simulation results for possible scenarios within the two seismogenic sources (Table 3).

Fig. 4
figure 4

Maps representing the maximum wave amplitudes (HMAX) for ALCS002 source estimated from a average, and b maximum parameters of the source. The HMAX values from the aggregate scenario are given with labels. The simulations are based on a large-domain grid (115 m spatial resolution). The maps are generated using ArcGISPro (Esri n.d.). For detailed information about the estimated rupture parameters for each seismic source used in the simulations, refer to Table 3

Full size image
Fig. 5
figure 5

Maps representing the maximum wave amplitudes (HMAX) for ALCS018 source estimated from a average, and b maximum parameters of the source. The HMAX values from the aggregate scenario are given with labels. The simulations are based on a large-domain grid (115 m spatial resolution). The maps are generated using ArcGISPro (Esri n.d.). For detailed information about the estimated rupture parameters for each seismic source used in the simulations, refer to Table 3

Full size image
Fig. 6
figure 6

Distribution of 34 numerical gauges along the Durres coast marked by circles and positioned at a 1 km distance, colour coded according to a average (Avg), and b maximum (Max) amplitudes recorded, for the ALCS002 source (see Table 3). Label ID as a reference for the summary of the data given in Table 4. The epicentre of the Durres Mw 6.4 earthquake in 2019 is indicated by black stars (USGS). c Plot showing bars corresponding to the average and maximum amplitudes recorded at each of the 34 numerical gauges, from the ALCS002 generated using a medium-domain topography grid with a spatial resolution of 30 m, acquired through the Shuttle Radar Topography Mission (SRTM) for the specified regions of interest

Full size image

The preliminary results are performed considering the ALCS002 source, defined as one of the outermost thrust fronts that characterise the southernmost portion of the continental collision between the African and European Plates (DISS Working Group 2021). Two nested grids of large (115 m) and medium (30 m) spatial resolution were used for the simulations. Since the largest tsunami waves hit the study area in the first 10 min up to 20 min, the 300 min simulation duration is sufficient to obtain the major tsunami impact in Durres, including additional reflections from all boundaries. Rupture parameters were assumed to be instantaneous, and the initial velocity field was assumed to be identically zero. In the numerical modelling of the tsunami propagation, non-linear shallow water equations were used. From each scenario or simulation, the maximum tsunami wave amplitude sea-level (HMAX), ETA, wave propagation maps, and synthetic mareograms along the Adriatic east coasts focussed in Durres were extracted from the output files. The simulation time series were recorded at 34 Points of Interest (POIs) located near the shoreline of the Durres coastline at a distance of ~ 1.0 km from each other, covering 50 km of the Durres municipality coastline. Moreover, the maps of the maximum and the average tsunami hazard intensity measures were developed (Fig. 4). In addition, we extracted the average and maximum HMAX at the POIs (Table 4).

Table 4 List of estimated average and maximum tsunami wave amplitude for the selected points of interest (POIs). Their distribution is given in Fig. 6
Full size table

The combination of earthquake parameters (represented by maximum amplitude tsunami wave height) and local topography and bathymetry appropriately described in Sect. 4 determines the characteristics of tsunami waves as they propagate and inundate the coastline. By understanding these factors, scientists and engineers can assess the potential hazard and predict the extent of inundation for specific coastal areas, enabling effective tsunami hazard assessment, mitigation, and emergency planning. Maps depicting tsunami generation, propagation, and inundation are generated for seismogenic sources (Figs. 3, 4, 5 and 6). It is worth mentioning that the geomorphological characteristics of Durres make it very vulnerable to tsunami inundation, as a large part of the city is characterised by low elevation.

The preliminary analysis results of ALCS002 (Fig. 4) and ALC018 (Fig. 5), conducted on a large-domain grid with a spatial resolution of 115 m, are presented for HMAX, simulated for average and maximum source parameters, and the static displacements. The ALCS018 tsunamigenic source produces lower peak wave amplitude than ALCS002 due to the difference in Mmax and more than half of the fault zone situated inland. The ALCS002 scenario (Fig. 4) gives high amplitudes along the Durres coast, reaching heights up to 2.5 m, while the amplitudes get weak as going far from the coastline (Fig. 5). In Durres City, the HMAX from the medium domain reaches a peak of 2.5 m inland, which poses a severe threat to the city’s lowlands. This is compared to the heights of 2.0 m obtained for the estimation from the ALCS002. Simulations from the scenarios for ALCS018 produce amplitudes of 0.8 m.

Beyond the Durres coast, the entire Adriatic east coast, extending from Albania to Bosnia and Herzegovina, is expected to be impacted by waves reaching up to ~ 0.3 m. The HMAX amplitudes are expected to reach the coast between 20 and 40 min. The computed maps of HMAX (Figs. 4 and 5) provide insights into the potential wave heights across the region, emphasising the significant impact on Durres City and its coastline, which poses a severe land threat.

Aggregated tsunami hazard maps of tsunami wave height for the coastal area of interest are developed based on a multi-scenario approach. Indeed, employing the defined sources, it becomes feasible to generate a collection of pre-calculated scenarios suitable for individual use within the warning system and formulate an “aggregated scenario”. This aggregated scenario is a comprehensive synthesis, encapsulating the key aspects derived from all calculated scenarios. The same approach has been applied by Peresan & Hassan (2022) and Hassan et al. (2023) to generate tsunami scenarios for selected areas along the Northern Adriatic coast. The preliminary analysis results of ALCS002, conducted on a medium-domain grid with a spatial resolution of 30 m, are presented for HMAX, simulated maximal source parameters as illustrated in Fig. 4b. While for estimated time of arrivals up to 300 min, the displayed HMAX reach up to 2.5 m along the Durres and Lezha shoreline, simulated from maximal source parameters.

Figure 6 presents the spatial distribution of the numerical gauges and the recording corresponding to the average and maximum amplitudes recorded at each of the 34 numerical gauges from the ALCS002 simulation generated using a medium-domain topography grid with a spatial resolution of 30 m for the specified regions of interest. Table 4 summarises the numerical gauges’ locations and average and maximum amplitudes.

In Fig. 6, numerical gauges no. 33 consistently registers the highest wave amplitude, peaking at ~ 2.0 m within the initial 20 min following the earthquake. Meanwhile, numerical gauge no. 25 exhibits the lowest amplitude in both simulated cases, considering average and maximal corner parameters. For comparison, the free surface time series for the selected six independent numerical gauges for a time-step of 300 min are shown in Figure S2.

A preliminary set of tsunami parameters and maps were obtained, including HMAX at selected sites, synthetic mareograms (time series of water level variations) at selected locations, tsunami propagation maps, and distributions of the arrival time of maximum wave heights along the coast. In the simulations of ALCS002, considering both average and maximal source parameters, we determined and plotted the maximum flow depth and inundation area (Fig. 7). In Durres City, modelling indicates that medium-sized waves could reach up to 2.5 m inland, posing a significant danger to the city’s low-lying areas. This is compared with estimates of 2.0 m obtained from ALCS002 data. The most substantial tsunami waves are expected to impact the area within the first 10 to 20 min.

Fig. 7
figure 7

Inundation maps are generated from a the average and b the maximum of the ALCS002 source parameters. The epicentre of the Durres Mw 6.4 earthquake in 2019 is marked by black stars (USGS), the landuse of the Durres municipality is represented by coloured polygons, offering additional context to the geographic information and facilitating the assessment of potential impacts or vulnerabilities (AKPT 2016a; OpenStreetMap contributors 2023). The simulation is generated using a medium-domain topography grid with a spatial resolution of 30 m, obtained through the Shuttle Radar Topography Mission (SRTM) for the specified regions of interest

Full size image

By overlap** the inundation zone with the landuse map, we identified the areas potentially affected by the considered tsunami scenarios (Fig. 7) and computed their extent (Table 5), which provides a preliminary measure of the potential impact on different socio-economic sectors. The available information on buildings, landuse, and infrastructure is assembled into a GIS formatted database for the Durres municipality by combining the Durres landuse plan with the dataset from OpenStreetMap contributors (2023). Table 5 provides a summary of the exposed landuse area (measured in square km) and infrastructure (roads measured in km) within the inundation zone for each scenario, offering a comprehensive overview of the potentially impacted landuse and infrastructure, contributing to a thorough assessment of the municipality’s vulnerability and potential consequences. As shown in Table 5, the industry sector is the most impacted, followed by the tourism sector. In particular, areas devoted to commercial (e.g. the port) and industrial activities can be substantially affected. The impact of the residential and infrastructure (roads) sectors varies from 28 to 37% and 10% of the total exposed assets, respectively. The buildings are the most resilient places as they can be used as vertical evacuation, for the residential areas in the inundation zone. Still, Dominey-Howes & Papathoma (2007) point out that losses are inevitable if the evacuation of low buildings is not possible. Considering that ~ 63% of buildings in Durres municipality are single-store structures and the maximal wave height in land is expected to be more than 0.5 m for the worst scenario, buildings are not the best evacuation option. Combining the results from tsunami modelling (so-called inundation layer) with the exposure of selected assets (buildings and roads) allows for assessing the potential damage to buildings and identifying possible evacuation areas to be considered in tsunami disaster mitigation and preparedness. However, the areas of higher elevation, which can be considered safe for evacuation due to tsunami, are not optimal in a multi-hazard context, as the distribution of the damaged buildings after the 2019 earthquake is concentrated in those areas (Xhafaj et al. 2024b). The risk analysis cannot encompass the entire Albanian coastline. Therefore, further work must be done to assess the tsunami hazard for the western Albanian and Balkan regions.

Table 5 Summary of the exposed landuse area (km2) and infrastructure (roads—km) within the inundation zone for each scenario (Avg—average and Max—maximum source parameters) computed for the municipality of Durres from the ALCS002 source (Fig. 7)
Full size table

6 Conclusions and outlook

Our research investigated various tsunami scenarios stemming from seismic activity along the Albanian–Adriatic coast, specifically examining potential tsunamigenic sources associated with the 2019 earthquake in Durres, Albania. The most critical source identified was ALCS002 with maximal parameters, which resulted in a significantly larger inundation area than the other considered source. In addition, our analyses indicate the potential for tsunamis in the Adriatic Sea, with maximum wave amplitudes reaching several metres, which agrees with historical events documented in various catalogues. The preliminary damage assessment results show that a large fraction (between 10 and 30%) of exposed assets might be affected by the considered scenarios, with potential disruption of civil and industrial activities and transportation infrastructure. These findings are pivotal for enhancing our understanding of and develo** mitigation strategies for tsunami hazards along the Albanian–Adriatic coast. Employing a comprehensive approach, we systematically merged theoretical frameworks with practical applications to enrich the evaluation of earthquake and tsunami risks in the future.

The present study is the first to assess the hazard associated with possible earthquake-generated tsunamis along Albania’s coasts. Further detailed tsunami modelling work will have to be conducted, considering more detailed topographic and bathymetry data when available, to accurately assess and delineate maps of tsunami-inundated areas.

This work addresses a significant gap in current research by providing the first extensive examination of urban-scale tsunami hazard assessment for cities along the Albanian coastline. Future work is planned to develop detailed exposure layers for selected assets (e.g. buildings) and estimate damages produced by the scenarios characterised here. Our results and those envisaged by further risk assessments are paramount for disaster risk reduction and should be integrated into preparedness and urban development plans.