On 30 October 2020 at 1:51 p.m. Eastern European time (11:51 UTC), an earthquake of M_w 6.9 (KOERI, 2020; NOA, 2020) hit the Aegean coasts of Turkey and Greece. The epicentre was some 14 km northeast of the Greek town Avlakia on Samos Island, and 25 km southwest of Seferihisar, Turkey, at a 10–21 km reported depth. The reported intensity of the event was as high as VIII MMI (USGS, 2020). The event has been followed by >4,000 aftershocks up to M_w 5.2. The event also triggered a tsunami which affected the coastline from Alaçatı to Gümüldür in Turkey, and from Karlovasi and Vathy in northern Samos with wave heights reaching up to 1.9 m (Dogan et al., 2021). The LastQuake App developed by the European-Mediterranean Seismological Centre (EMSC) suggests that the event was felt from as far as Croatia and Romania and north Macedonia (Contreras et al., 2022).

The fault plane solutions and previous field observations point out a pure north-dipping normal faulting motion (Akinci et al., 2021; Kiratzi et al., 2021). The surface projection of the fault plane crosses just the North of the island of Samos. The strong ground vibrations from the mainshock triggered the seismic stations of both the Greek and Turkish networks operated by the Ministry of Interior Disaster and Emergency Management Presidency (AFAD), Boğaziçi University’s Kandilli Observatory and Earthquake Research Institute (KOERI), National Observatory of Athens Seismic Network (NOA) and ITSAK Strong Ground Motion Network. The mainshock was only recorded by two stations in the east of Samos Island, which are the closest to the earthquake epicentre. As expected, the highest peak ground acceleration (PGA) was measured in NS component of one of these stations at about 0.24 g, which is slightly smaller than the expected PGA value of the Greece zonation map (2003) for a 475-years return period. In Turkish territories, while a station in Gümüldür produced a maximum PGA (≈0.21 g), the softest soil station in İzmir Basin-the most damaged area—recorded 0.15 g, which is below the expected PGAs of the recent Turkish earthquake hazard map (2019) for the design basis earthquake.

The most immediate direct economic losses due to the event were officially estimated as ₺415 M and €81 M for Turkey [based on the payout figures reported by the DASK (2021), which are rather conservative compared to the reported insured losses by Doğan (2020), in the range of ₺1.8B] and Greece [based on figures reported at Governmental Gazette 5293/1 December 2020, which are similarly much lower than the estimated €120 M when the expenditure reported by samos24 (2021) are accounted for]. For more details see Aktas et al. (2021), respectively. The building and infrastructure damage concentrated in two districts of Izmir, Bayraklı and Bornova, located around 70 km away from the epicentre of the event and built on alluvial sediments and active normal faults [Izmir Fault (IF) and Karşıyaka Fault Zone (KFZ)]. This area suffered from 13 building collapses and around 200 heavy damage cases (TMoEUCC, 2020). Out of the total of 119 causalities, 116 took place in these districts, while one lost their life in Seferihisar due to the tsunami. In Samos, the event caused two casualties, due to the collapse of an abandoned traditional masonry building. That neither Turkey nor Greece had specific protocols during an ongoing pandemic made the response and recovery process following the earthquake challenging.

The Earthquake Engineering Field Investigation Team (EEFIT) of the United Kingdom Institute of Structural Engineers (IStructE) decided on 2 November 2020 to deploy a mission for reasons including 1) the dramatic damage levels in Turkey, accompanied by a high death toll; 2) the tsunami affecting the area; and 3) the then ongoing COVID-19 pandemic affecting the impact of the event and subsequent recovery efforts. On 5 November a team leader was appointed, and the call for expression of interest was issued. The team was finalised by 13 November and started the mission work on the 16th, during the second national lockdown in England.

It is due to the pandemic and the lockdown that the team adopted an unconventional strategy and coordinated the mission remotely. The EEFIT mission to the 30 October 2020 Aegean Earthquake and Tsunami had the following specific objectives pertinent to its special circumstances:

- Combine field and remote survey strategies for a hybrid mission for damage assessment in buildings, critical infrastructure and geotechnical structures.

The mission lasted approximately 1 month roughly until the 17th December, when the mission lecture took place (accessible from: https://www.istructe.org/resources/case-study/2020-eefit-aegean-earthquake-mission/). During that time, the team gathered information on the event and its impact on buildings and infrastructure from initial reports by other research groups and news items, scanned social media (Contreras et al., 2022) interviewed 12 key stakeholders in Turkish and Greek governmental and non-governmental organisations, launched a public survey to gauge people’s response to and perception of the event, remotely coordinated three sets of field work in Turkey and Greece, ran a mission blog, identified other suitable data sources for damage assessment, and completed remote (secondary) assessment of all damage data collected. The details of all this work can be seen in the mission report (Aktas et al., 2021), which was published in May 2021. The mission timeline is shown in Figure 1.

This paper aims to summarise the main findings of the mission, as well as the key lessons learnt from this experimental reconnaissance strategy with specific emphasis on the 30 October 2020 event, in order to draw conclusions regarding the feasibility of similar approaches in future disasters.

For the aim of damage data collection, the field crews used a specialised tool developed by UCL researchers and academics as part of the Learning from Earthquake project (2017–2022, EP/P025641/1). Details regarding this tool as well as the process of recruiting and training the teams to be deployed in the field and the routes they should follow were determined are presented below. The main challenges regarding the data quality as well as how they were addressed are presented with an overall summary of the collected data.

The hybrid nature of EEFIT’s reconnaissance mission meant that several team members worked in parallel to fieldwork on identifying alternative sources of information which can be used for remote damage assessment or other purposes to complement the field data and enrich our understanding of the impact of the 2020 Aegean even in the affect area. The teams explored and extracted information from a wide variety of online sources including news outlets, social media and official channels and CCTV imagery. A summary of the key sources explored is depicted Table 2. The team also processed photographs sent by the field crew or other local contacts, without using the app. The data focused mainly on the impact of the earthquake and tsunami on buildings but data of seismotectonic and geotechnical interest were also collected.

As part of this approach, a total of 263 buildings were surveyed remotely using the same forms as used in EEFIT’s mobile app and with a focus on the building characteristic and damage description, from sources depicted in Table 2 and mapped in Figure 5. Most of the remotely surveyed buildings (i.e., 89%) were located in Samos. The main challenge with the remote assessment was that the buildings on which alternative data sources were available were more likely to be severely damaged, or their exact location could not be established. Despite these shortcomings, they added to the understanding of how buildings behaved in the villages in Samos.

The evaluation of regional seismotectonic context, local geology and local site conditions together with the recorded strong ground motions may give an idea for the spatial variability of earthquake-induced structural damages. The complex tectonism of the region, in which strike-slip and mostly extensional motion exist, makes the Aegean one of the most seismically active areas in the world. The east-west-trending extensional system in the region is formed by the northeastward subduction of the African plate beneath the Aegean plate (Taymaz et al., 2007), within which the M_W 6.9 shallow normal faulting (East-West striking and north-dipping) Aegean Sea earthquake occurred in the 30 October 2020 (Akinci et al., 2021; Kiratzi et al., 2021).

Mountainous areas and Neogene sedimentary basins separating them topographically dominate on the island of Samos, which is the closest terrestrial location to the epicentre. In addition to Samos Island, one of the other most affected areas from the main shock are İzmir, situated on Miocene sedimentary units and Plio-Quaternary alluvial deposits. Also, we would like to raise particular attention to the districts of Bayraklı and Bornova, which are settled in the sedimentary basin at the mouth of the İzmir Bay, due to heavy earthquake-induced destructiveness in these areas.

The well-distributed seismic ground motion stations -except for the west of the epicentre-of both the Greek and Turkish networks [the Ministry of Interior Disaster and Emergency Management Presidency (AFAD), Boğaziçi University’s Kandilli Observatory and Earthquake Research Institute (KOERI), National Observatory of Athens Seismic Network (NOA) and ITSAK Strong Ground Motion Network] have yielded a great number of recordings of the main shock. Ground motions have been recorded by 100 stations in a 200-km radius around the epicenter (Figure 6A). The numerous stations to the east and north of the epicentre are mostly (76%) operated by the AFAD while the closest station (GMLD) to the epicentre in the Turkish side belongs to the KOERI network. SMG1 and SAMA of the ITSAK and NOA, respectively, present excellent opportunities to assess strong ground motion characteristics to the near-south of the epicentre, however, the western section of the epicentre is devoid of stations. This circumstance makes it difficult to evaluate the inspected damaged buildings in terms of ground motions, particularly in the west of Samos Island. The stations around Bayraklı-Bornova districts of İzmir, where the mainshock causes the heavy structural damage, on the other hand, make a substantial contribution to the assessments.

Figure 6 also incorporates the USGS topography–based V_S30 map of the region to provide an insight into prevailing soil conditions. Since the networks provide V_S30 values for the majority of stations, these network-based V_S30 values have been considered in the histogram plot (Figure 6D) except for several stations. Based on the soil classification of EC8, there is a preponderance of the stations (53% of all stations) within soil class B, whose V_S30 values range from 360 to 800 m/s. Then, soil classes C and D, which comprise 33% of all stations, come after ground class B. The rest of the stations are deployed on the stiff soil class (V_S30 > 800 m/s).

The top panel of Figure 7 illustrates the spatial distribution of peak ground accelerations (PGA) and peak ground velocities (PGV). As expected, the highest peak ground acceleration (PGA) has been measured in NS component of one (SMG1) of two stations in the east of Samos Island, which are the closest territory to the earthquake epicentre, at about 0.24 g (232 cm/s²). This value is slightly smaller than the expected PGA value of the Greece zonation map (2003) for a 475- year return period. In Turkish territories, while a station in Gümüldür produced a maximum PGA (≈0.21 g), the softest soil station in İzmir Basin-the most damaged area—recorded 0.15 g, which is below the expected PGAs of the recent Turkish earthquake hazard map (TSHM, 2018) for the design basis earthquake. However, PGVs of soft soil stations (3513 and 3519) in the İzmir Basin is about as high as those of the nearest stations.

Some prominent recordings are evaluated by comparing them with corresponding design spectra that are generated based on the current national seismic building codes of Turkey and Greece. The seismic demand proposed in the Greek code (EAK, 2003) is exceeded in both horizontal directions of the closest stations (SAMA and SMG1), which are approximately 22 km away from the mainshock epicentre, but more significantly in fault-normal direction, i.e., the North-South (Figure 7). Fault-normal counterparts of each of the two stations result in the highest spectral accelerations, especially in SMG1; the value is almost twice the maximum demand of the design spectrum. If the epicentral distance is neglected since both stations are roughly at the same distance, local soil conditions may govern the response in terms of the capability of amplification; station SAMA operates on a relatively stiffer soil (SC-A in EC8 soil classification) while the other is classified as SC-B based on EC8 (Figure 7). Conversely, the station GMLD, another closest station at 23rd km of the north of the source, is mostly in the design limitations set by the Turkish Seismic Code (TBSC, 2018) in relatively short period range. However, design accelerations levels are exceeded in the period portion of 0.5–1.5 s for SC-B of TBSC-2018 soil classification.

In the aftermath of the earthquake, the most damaged regions are Samos, which is as expected considering the building stock very near to the source, and Bayraklı/Izmir where a significant number of multi-storey RC buildings are either significantly damaged or collapsed. The latter of which necessitates a closer look to enlighten the reasons behind since the theoretical aspect is fulfilled in that the response spectra of two stations (3513 and 3519) in Bayraklı do not drastically exceed the spectral acceleration levels by design spectra of TBSC (2018). However, it should be emphasized herein the flat portion of the response spectrum of 3513 covers longer period range varying from 0.4 to 1.4 s when compared the design spectrum. Despite the lower spectral accelerations than the design demand, the highest damage distribution in the region implies that technical requirements during or after the construction were not fulfilled or were violated (for further discussion see Section 5.1).

The building stocks in the affected areas by the 30 October 2020 Aegean Earthquake are primarily reinforced concrete (RC), masonry, and timber-masonry. The breakdown of these main construction typologies and others (primarily confined masonry and other masonry-concrete hybrids, as well as steel-based construction systems) from our damage assessment activities (only remote assessment of field data) based on location can be found in Figure 8A. As you can see reinforced concrete (RC) is the predominant structural typology in Turkey with 75% (226/302), as opposed to 14% in Samos (31/217). In lack of a systematic building census including construction type information, we are unable to understand the actual portion of RC structures within the whole building stock for the area under consideration in Turkey, however the field observations support that the majority of the buildings are indeed RC. Further our data shows that the RC buildings in Turkey are overwhelmingly mid-/high-rise with 2-storey buildings comprising 16% of all surveyed RC buildings, 3-storey 19%, 4-storey 15%, 5-storey 25% and + 6-storey another 20%, while in Samos low-/mid-rise are more common - 58% of all surveyed RC buildings are 2-storey, and another 29% are 3-/4-storey (Figure 8B). The predominant typology for Greece is masonry with 69% (150/217), which is only 16% of the sample set from Turkey (47/302). The 2011 building census results available for Greece reports the percentage for stone masonry structures for Voreio Aigaio (North Aegean Region inclusive of Samos) as 47.7%, which is very comparable to our estimations (Hellenic Statistical Authority, 2015). Assuming that the values are reflective of Samos specifically, the census however, suggests that we undersampled concrete structures and oversampled timber structures, as it reports 32.1% for the former and 0.2% for the latter, as opposed to our 10 and 8%, respectively (ibid). In both countries, more than 70% of all masonry buildings are low rise traditional typologies, built with local stone or brick and unreinforced (Aktas et al., 2022). The rest has a substantial portion of historic monumental structures, such as public buildings, libraries, churches, mosques, as well as more recent period masonry structures built with hollow clay and concrete blocks, some of which is very low quality. All the timber-masonry buildings are traditionally built hybrid structures with a masonry ground floor and timber framed upper stories, which are common over a wide geographic area, including Turkey and Greece.

The damage classification was made based on EMS-98 (European Macroseismic Scale), where DG0 delineates no damage whatsoever, DG1 no structural damage and negligible to slight damage to non-structural components (such as hairline cracks in all primary structural typologies, limited to mortar joints or plaster in masonry), DG2 moderate damage (including cracks, plaster fall and partial collapse of non-structural components such as chimneys), DG3 substantial to heavy damage (including large and extensive cracking, failure of individual non-structural components), DG4 very heavy damage (including serious failure of walls and partial structural failure of roofs and floors) and DG5 total collapse or near total collapse (ESC, 1998). While the no-/low-damage cases are generally the highest for any given primary structural system, gradually decreasing in number with increasing damage levels, the masonry and timber-masonry buildings in Samos demonstrate a spike at moderate damage levels (Figure 9). As mentioned above, these are predominantly traditional buildings, often old and in poor condition, which is considered to be the main reason behind elevated portion of moderate-to-high damage cases. Further, while the epicentral distance does not appear to be a relevant parameter to explain diverse damage levels (Figure 10A), the PGA_max values are slightly higher for the soft soil station on the east of Samos than for Turkey must be another reason for this discrepancy. Also, although there is no recorded PGA in the west of Samos, the proximity of the surface projection of the fault plane to Samos than Turkey can be associated with these moderate-to-high damage cases (Figures 6A, 10B, see Malcioglu et al., 2022 for further details). Despite the few dramatic collapse/heavy damage cases which have been the most memorable outcome of the 30 October event, the RC seems to be the primary structural system proportionally with the highest number of no-/low-damage cases.

The performance of each of these three main primary structural systems is discussed in more detail in the subsequent sections. Please note that due to the failure bias intrinsic to the alternative data sources we used in this mission, the reported statistics are based on the field-data alone.

The 30 October 2020 Aegean earthquake was accompanied by a tsunami, which is not an uncommon phenomenon in the region (Figure 16A). The tsunami waves reached up to 2.0 m in this particular event (Figure 16B). Therefore, the EEFIT mission also systematically investigated the tsunamic impact on buildings and infrastructure, through a set of questions embedded in the data acquisition system specifically to reflect typical damage mechanisms due to tsunami hydrodynamic effects on buildings. These being: the number and types of openings, particularly at ground floor level; the presence of a basement; assessment of surroundings, including perimeter walls and presence of debris; details of foundation type and damage (if that is visible); and evidence of water marks on the building. Some of these attributes were included in the damage assessment form developed in the 2018 EEFIT mission in Palu (Rossetto et al., 2019), which has been ad opted for the damage assessment carried out in this mission. By the time our field crews visited site however, it was already 1 month after the event, and all debris was cleared and no substantial structural and non-structural damage was found despite significant inundation in some localities. In certain cases, water marks were visible on buildings (Figure 17).

The EEFIT Mission to the 30 October 2020 Aegean Sea Earthquake and Tsunami showed that the building stocks in both countries generally performed well, with the exception of a few RC structures exclusively in Turkey, which suffered from a high degree of damage despite the lower spectral accelerations than the design demand–a situation which can be attributed to a very unique combination of non-compliant design, poor construction practices, and in certain cases, post-occupancy amendments, flagging a lack of more operational quality assurance protocols. Additionally, the traditional building stock in Samos received mostly moderate- and high-damage, as a result of both the comparatively poor upkeep of these buildings and slightly higher recorded pga values here than in Turkey. It is noteworthy to state that the recorded pga values in Greece were also lower expected demand values for a 475-years return period, however the scarcity in the deployed stations in the Island of Samos hampers the reliable evaluation through studying accelerograms. If there were enough strong ground motion stations on Samos, it would probably record higher pgas in the central and north-western parts due to the proximity of the northern edge of Samos to the surface projection of the fault.

Importantly, one of the limitations of this study is that all buildings were assessed from outside due to COVID-19 related lockdown conditions, and the damage levels might have been underestimated.

This was a hybrid reconnaissance activity bringing together the field and remote survey strategies. The team partnered up with local organisations to recruit, train and deploy field crews for an in-situ examination of damage in diverse building stocks, which proved an efficient and speedy way of accessing the site under the COVID-19 pandemic conditions where travelling was not an option for the team. This strategy also brought ease in the overall damage assessment process due to the knowledge of the local field crews of the native languages, relevant national authorities and building stocks, and can be adopted in the future with diligent coordination by the remote team even when there are no extraordinary circumstances hampering international travels.

The mission also explored alternative data sources to gain an understanding of the extent and nature of the damage in buildings and infrastructures to assess their usability. Based on the experience from 30 October event it can be concluded that data availability and content for a remote assessment of the impact of the event on buildings and other structures is strongly country dependent. This may be down to different cultures of engaging with social media in a given context, or how informed citizens are regarding the disaster risk. In the case of the Aegean Mission we were able to find a higher amount of data from a wider set of sources for Greece than for Turkey thanks to a few comprehensive data sources on Youtube and Facebook. Another important factor governing the availability of data, representative of the overall damage levels, seems to be how dispersed damage is. In the case of localised dramatic damage (Turkey), the damage bias in remote data sources is much higher than when the damage levels are more even and more scattered across the affected area (Samos), making it harder to gain a reliable understanding of the damage levels overall within the affected areas.

The EEFIT mission to the 30 October 2020 Aegean Sea Earthquake and Tsunami showed that while relying only on remote investigation methods does not (yet) emerge as an ideal way forward, the hybrid model offers many critical advantages, which should be exploited in the future events.