The Role of Video Cameras and Emerging Technologies in Disaster Response to Increase Sustainability of Societies: Insights on the 2023 Türkiye–Syria Earthquake

Abstract

New technologies are being used to facilitate the recognition process during and after earthquakes. These advanced tools are essential to keep track of what is left from of the destruction suffered by the built stock. Among the new technologies are video recordings captured during seismic events, footage from drones, and satellite imagery acquired before and after the event. This review paper presents a series of examples collected from the 2023 Türkiye–Syria earthquakes to illustrate how these new technologies offer a unique and efficient way to capture, document, and transfer information among experts in seismology, earthquake engineering, and disaster management. Whenever possible, these examples are accompanied by simple qualitative explanations to enhance understanding. To demonstrate the potential of video cameras and drone imagery for quantitative analysis, in addition to the various simple examples provided, two case studies are provided—one on road blockages, and another on intensity assessment and wave attenuation as observed in video cameras. These technologies are critical and merit considerable focus, particularly video cameras, which have not received much attention recently, on helping to understand seismic wave passage and their impact on the built environment. Enhancing our use of video cameras in this context can significantly contribute to the sustainability and resilience of our society. With the rapid advancement of image analysis, we advocate for a collaborative platform for accessing and utilizing imagery materials, aiding current and future generations in analysing the causes of such tragedies.

1. Introduction

1.1. Description of the Earthquakes

On 6 February, 2023, two significant seismic events occurred in the southeastern part of the Anatolian plate (Figure 1a). The first event, measuring M_w 7.8, occurred at 01:17:35 a.m. (UTC), 04:17 a.m. local time, followed by an M_w 6.7 aftershock 11 min later and an M_w 6.5 event 19 min later. The second large event, an M_w 7.6, occurred at 13:24 pm (local time) [1]. The tectonic environment in this region is well understood because of the interaction of three plates that have collided at a rate of 20/30 mm/year (Figure 1a—[2,3], leading historically to significant seismic activity. These interactions have caused large earthquakes in the past, with an average recurrence interval of more than 200 years (Figure 1b) [4], making this occurrence somewhat anticipated by experts despite the uncertainty of the exact timing.

Although these earthquakes occurred along previously identified fault traces (Figure 1b), the level of destruction to humans and to the building stock was far higher than expected. The destructivity started with the fault rupture of 300 km and continued with the high strong motions generated, acting on a large portion of the stock built in a stripe along the faults (Figure 1b in green). Numerous aftershocks ensued, including an M_w 6.2 event on February 20, causing additional damage. This series of events has introduced many new issues that may take years to fully understand and has raised several concerns about current approaches in geology, seismology, and earthquake engineering, highlighting the need for a thorough re-evaluation of many standard procedures.

By 20 March, 2023, estimates of victims indicate more than 53,000 confirmed deaths, 129,000 injuries, and 2 million homeless. The population of the affected region is 13.5 million in Türkiye and 2 million in Syria, and direct losses are 4% of the GDP of Türkiye; total losses may reach 10% of GDP [6] (data from 23 February 2023), contrary to the initial estimates of up to 10,000 victims and an economic impact of 0–1% GDP of Türkiye [7]. The numbers 1 year later might be higher, especially with respect to the victim count. Approximately 35,300 buildings collapsed, 17,500 were demolished, 180,000 were severely damaged, 40,000 were moderately damaged, 431,000 were slightly damaged, and 860,000 were undamaged.

Despite many advancements taking place in recent years, both in the Earth science as well as in the engineering fields, the sequence of events with such large impacts demonstrated, either in the number of victims or in the amount of destruction and consequential effects, that much more is needed to make societies less vulnerable to these disasters. The adoption of more resilient and sustainable urban and rural environments is crucial. New technologies, such as the ones discussed in this paper, were not available a decade or so ago, but they have now advanced enough to accelerate the monitoring of disaster effects. This progress can help alleviate the suffering of impacted populations and provide the knowledge needed to build communities better prepared for future events. Sustainability is now essential for addressing new challenges. The rising impact of earthquakes and other natural hazards underscores the need for precise, affordable, and proactive measures.

1.2. In-Field Instrumentation and Dissemination of Information

The area struck by this massive earthquake was well instrumented in terms of an accelerometric strong motion (SM) and a displacement global navigation satellite system (SNSS) network that recorded not only the two larger events but many aftershocks that followed both events [8,9]. As all these stations are well interconnected and linked to a central management unit, it was simple to gather that information in a few days. This example shows how much the panorama has changed from past large earthquakes when these data required many weeks to handle before being openly shared with the scientific community. Unfortunately, health monitoring of structures (HMS) data was not available, with the exception of data from one base-isolated hospital in Osmaniye [10], and no downhole instrumentation was installed. Like SM, GPS instrumentation (the base of SNSS) was also, for the first time [11], treated with great celerity and made accessible to the community. The European Space Agency, ESA [12], together with Turkish authorities, the United Nations, and other space agencies, provided a set of satellite images and other relevant information on the affected areas to define the extent of the disaster and assist emergency aid organizations. These elements are fundamental for scientists to analyse ground movement to aid recovery, reconstruction, and long-term research.

An experimental thousand-kilometre DAS (Distributed Acoustic Sensing) array made with fibre optics and implanted across Anatolia revealed an unprecedented M5 event in the North Anatolia Fracture Zone (NAFZ) [13], highlighting the importance of this new technology for recording seismic signals in a distributed array configuration.

The initial reports about the earthquakes were disseminated by the media through various channels. In the first weeks following the 6 February, 2023, sequences, numerous national and international field missions were organized. These efforts generated significant documents containing data and pictures. Additionally, workshops were held and scientific meetings were convened [14].

Future generations will certainly analyse the collected information. Data treatment requires all the available information, including video camera footage, drone images, and all the above-mentioned items, which are quite large and need great attention to correctly interpret the data, and new algorithms are needed to facilitate the complex treatment of all available instrumentation. However, the relentless pressure in science and technology to publish in the first place has altered the pace of new paper releases in journals. To illustrate the “speediness” of publications, we compiled a selection of papers relevant to the overarching themes discussed herein (see

Table 1. Examples of recent publications in peer-reviewed journals.

Topic	Authors
Earthquake source	[15,16,17,18,19,20,21,22,23]
Surface deformation in the vicinity of fault ruptures	[24,25,26,27,28]
Ground motion	[29,30]
Geo-hazards	[31,32]
Building performance	[33,34,35,36]
Mosque and minaret performance	[37,38]
Codes	[39,40,41,42]
Post-earthquake clean-up and debris removal	[43]

). This situation is significantly different from that of previous large earthquakes, which entailed much longer times for journal publications.

Furthermore, several special issues of specialized journals dedicated to the earthquakes of 6 February, 2023, are currently being assembled. For instance, one issue, comprising 7 papers, is organized by the Bulletin of Earthquake Engineering [44] under the theme “Reconnaissance Missions and First Observations”. Another, featuring 15 papers, is organized by Engineering Geology (2024) on the topic “Engineering Geological and Geotechnical Aspects of 6 February, 2023, Türkiye Earthquakes”.

It should also be mentioned that in July 2024, the 18th World Conference on Earthquake Engineering [45], the largest international conference that takes place every four years, the Scientific Committee gave great audience to the 6 February 2023, allowing for the presentation of 105 papers structured in three different topics.

The entire problem discussed in this paper can be viewed as a series of interconnected situations from a top-down perspective. Satellite imagery provides a comprehensive view of the territory, accurately analysing changes and anomalies on the Earth’s surface, both vertically and at inclined angles. Drones capture detailed images of hidden or hard-to-reach places, primarily around buildings or infrastructures such as bridges, dams, and geotechnical anomalies. Video cameras, which have not yet been addressed much in publications to the authors’ knowledge, offer a more detailed look at the behaviour of seismic waves in open spaces and their impact on structures and human activity. All of these three inter-related topics are crucial for achieving a comprehensive understanding of the reasons behind the significant destruction that has occurred. We present several examples of new technologies for examining the impacts of seismic waves on nature, populations, buildings, infrastructures, and other relevant cases.

Video cameras have long been used for precise measurements during experimental studies in laboratories. Recently, they have also played important roles in emergency-response workflows. They are an important source of information on the mechanisms of wave propagation and the effects on nature, population behaviour, and the performance of engineering structures. Video footage can be collected from fixed points to observe traffic flow or from surveillance cameras to detect intruders, or it can be captured by individuals via their personal mobile phones. This footage adds valuable context to information gathered through local visual inspection, pre- and post-event photographs, recording instrumentation, drones, satellite imagery, or citizen accounts. Oliveira et al. [46] demonstrated how video cameras could track the motion of an object from its initial vibratory state to its final state by animating images of the moving object. However, the primary drawback of video cameras is that the footage obtained is from random locations, making it impossible to anticipate where they should be before the event.

Please note that this paper does not explore the use of video cameras as measuring devices for monitoring experiments.

Drones equipped with high-resolution video cameras provide another valuable source of information for understanding the extent of destruction observed from close proximity. In Türkiye, authorities began cleaning massive amounts of rubble immediately after it was determined that there were no further survivors. With the removal of this debris, the remaining methods used to study the effects of the earthquake are solely satellite images, photographs, and drone footage. These sources form the most important reservoirs of information, offering both broad overviews (satellites) and detailed views from difficult angles (drones). They provide crucial insights into the predominant direction of building collapse, the volume of debris, and areas obstructing roads, which are essential for post-event analysis.

Satellite imagery, in addition to assisting in the understanding of crustal deformation, plays a pivotal role in enhancing data analysis, particularly in scenarios involving extensive areas of destruction. New technologies that increase data accuracy and utilize artificial intelligence (AI) and machine learning (ML) have made significant progress. However, caution is advised and onsite validations remain imperative. This information complements the building-by-building assessments conducted in the present case with the help of the World Bank [47] for an operation requiring a considerable logistics effort, which was very well organized and planned.

Even in a brief analysis, we cannot overlook the key reasons for the poor performance of part of the building stock. Chief among these are the substandard construction practices of the past 20 years and the complacency of authorities in permitting these deficiencies. Furthermore, the Turkish technical community is equipped with modern construction standards, the most recent from 2018 [48]. We completely concur with David Alexander’s blog [49], where he points out the errors committed by the construction industry and related state organizations, which issued construction licences without providing supervision. Many other world experts [39,43,50,51,52,53,54,55,56,57,58,59] conveyed in the media, right after the events, similar opinions. Many research institutions, as already mentioned, are now working on the instrumental gathered data and using the information described by several field missions to produce scientific publications.

In summary, the key points highlighted in this paper are as follows:

(A) New technologies play a crucial role in facilitating rescue operations and rapidly assessing the most severely damaged areas. They are also very important in the reconstruction phase as continuous monitoring tools trying to make the newly reconstructed areas more sustainable in the future.

(B) Such technologies were successfully utilized during the Türkiye–Syria earthquake, including satellite and drone imagery. The importance of video camera footage per se is emphasized in this paper.

(C) The significance of these technologies in the understanding of various phenomena is demonstrated through qualitative and quantitative analysis. This analysis includes determining the earthquake’s origin, studying the behaviour of various structures subjected to static and dynamic ground motion, and assisting in defining macroseismic intensities.

(D) Examples of quantitative analyses are provided to show the importance of these technologies. The insights gathered from video cameras are already corroborated by analytical modelling ([46,60,61,62,63], among others), which facilitates the reconstruction of ground motion, impacting support devices based on significant visible displacements.

Furthermore, the analysis of building behaviour through drone imagery is briefly reported in Section 2.2.1 and Section 2.2.2 and partly forms the basis for the quantitative study of road blockages caused by debris from collapses, which is addressed in Section 3.

Finally, another important aspect of disseminating short videos and sharing other data is their potential to be utilized by journalists to reach a broader audience beyond the scientific community, thereby providing a public service. When journalists and researchers collaborate, the resulting data can have profound benefits, becoming a vital vehicle for science and risk communication, not only in the impacted area but also in other earthquake-prone regions worldwide. Following this catastrophic event, numerous interviews, breaking news segments, webinars, conferences, and social media posts emerged globally, addressing concerns and expectations specific to various cities and countries. In Portugal, news coverage and videos in the media, along with awareness of the Portuguese post-earthquake mission to Türkiye [55], were pivotal in the decision to construct a new large hospital in Lisbon with base isolation [64].

1.3. Organization of the Paper

The present paper is structured into three parts:

Part I explains the merits of new technologies and provides a large collection of qualitative examples. These examples illustrate various observations that play a crucial role in rescue operations, assessing heavily affected regions, guiding reconstruction policies, and advancing scientific understanding. The merits and limitations of each one of them are discussed.

Part II presents several quantitative examples to demonstrate the effectiveness of drone imagery in exploring road blockages and contributing to the assessment of debris amounts, as well as the use of video cameras to analyse the behaviour of simple objects. This analysis, based on a theoretical framework, contributes to a better understanding of wave propagation and its effects.

Part III is dedicated to analysing future initiatives that these new technologies, combined with data science, may offer to enhance the sustainability of societies.

All the Supplementary Materials, which complement the data shown in different areas, can be found at: https://www.mdpi.com/article/10.3390/su16177618/s1.

• PART I—Qualitative Insights Into New Technologies For Disaster Response And Scientific Progress

2. New Technology’s Advances

Social media platforms serve as valuable tools for analysing earthquake impacts from various perspectives, yielding novel insights. The data sources range from YouTube videos and films produced by international news agencies such as the BBC, NBC, NTV, and Aljazeera to information collected by international scientific bodies such as the EMSC and USGS. Additionally, videos shared on social media platforms such as Instagram, LinkedIn, and X contribute to this wealth of data. However, the most reliable source remains the AFAD [65,66], a Turkish agency under the Ministry of Interior, known as the Disaster and Emergency Management Authority.

Despite the rapid and accurate access to information provided by these platforms, the insight gained from field missions to affected sites is unparalleled. In the 19th century, Mallet [67] and Milne [68] conducted their studies based on local observations. Ambraseys [69] summarized the importance of such field missions:

“The site of a damaging earthquake is a full-scale laboratory from which significant discoveries may be made by seismologists, geologists, engineers, sociologists, or economists, not to mention politicians”.

Similarly, several international field missions have occurred since the onset of this series of earthquakes, and most of them have prepared several reports with updated versions of what they observed and made preliminary interpretations of the facts, including the AFPS [70], GEER-EERI [71], KOERI [72], and Portuguese post-earthquake missions [55,64], among others. The last report made by EEFIT [10], published one year after the event, retained the same ideas and was supported by long and detailed descriptions.

2.1. Satellites

Satellites play a vital role in detecting crustal deformation both before and after an earthquake event, providing valuable insights into the geophysical changes associated with seismic events. These changes can indicate the stress field created in the region, which can help pre-establish the crustal zones with greater deformation. Detection can be performed via various methods, such as interferometric synthetic aperture radar (InSAR), differential GPS, and gravity field measurements. These methods help to observe and monitor crustal deformation over time. Spatial resolution has increased with advancements in satellite technology. A few years ago, we achieved a resolution of 10/20 m, but the new generation of satellites and data treatment can easily reach a magnification greater than 1 m. In some cases, the detail can be captured at the centimetre scale.

Furthermore, satellites play a crucial role in conducting post-earthquake damage assessments from a global perspective. They provide extensive aerial coverage, enabling rapid response, pre- and post-event comparisons, damage mapping, and severity assessment via inclined imagery and other long-term recovery methods. However, importantly, their spatial resolution may not always capture fine-grained details. Hence, ground-based assessments, complemented by drones and other onsite data-collection techniques, are often required for a more detailed understanding of damage at the local level.

2.1.1. Crustal Deformation

Knowledge of the stress field created in the region before the crustal zones showing greater deformation are established is a major contribution of satellites. The density of the fringes indicates deformation. This technology has been recognized for more than a decade, but it provides information only if the region is being monitored.

With increased accuracy, on the order of millimetres of GPS vertical measurements, we can clearly identify the traces that have moved during an earthquake. Furthermore, horizontal displacements are currently measurable, and plate movement along fault traces can be observed. “Images from ESA [12], Sentinel-1A satellite captured on 9/10 February clearly showed the physical effects of the earthquake on the ground, including deformation of up to 6 m along a 300 km section of the fault, and the second event causing a second ~125 km rupture” (Copernicus/NERC/COMET [73]). This information (satellite radar data) was compared with optical images from Sentinel-2 to double-check the expressed values. Together with ground deformation, it is possible to determine how much the stress field changed with the fault rupture. Since we delve into kinematic variables, we cannot anticipate the stress field prior to an event or, consequently, the proximity of a rupture. We lack constitutive laws to close the loop of stresses and strains [74]. Only other physical/chemical variables (gravimetric, electromagnetic, ionospheric, radon emission fields, etc.) may help us to predict the proximity of an event. The numbers shown by the different satellite technologies need to be made compatible with the long-term rates obtained from other sources.

Based on the provided information, it is possible to identify the following new issues for study:

-

Compatibility between strain rates obtained via GPS and rates of occurrence, especially for cases of complex tectonics, such as the Anatolian plate.

-

Kinematics of rupture: the first event occurred along 300 km, approximately 150 km on either side of the fault from the epicentre; the second event had a different mechanism.

-

Wave propagation is clearly influenced by the path (orogeny, topography) and geotechnical 3D site effects.

-

Directional effects on rupture (fling, pulse-like) and Doppler effects. During the 4:17 a.m. event, damage appears concentrated along the fault trace, and it would be very insightful to determine if collapses essentially occurred during the fling passage.

-

Uncertainties in the process of defining seismic action. Moreover, the hazard and the most updated recommendations [48] were below the observed seismic field, especially over long periods.

-

The possibility of stress transfer among faults in the same tectonic environment, as pointed out by Stein et al. [75], causing interdependences or, in the present case, causing “migration of earthquakes” from M_w 7.8 to M_w 7.6, which must be considered in future hazard modelling in terms of intra-main rupture and intercutting for nearby faults.

2.1.2. Damage Assessment

For the first time, satellite imagery was used to assess the building’s damage status. Previously, only collapsed buildings (D5—collapse) could be identified, but other classes of damage (D1—negligible to D4—near collapse) can also be detected. During the 2010 Haiti earthquake, initial steps were taken towards this goal by having experts from around the world manually identify collapsed buildings via vertical images (World Bank, [76]). With the advent of inclined observations and higher-resolution images, it is now possible to assign a degree of damage (D1 to D5) to buildings, as shown in Figure 2. Satellite-based remote sensing can quickly provide an overview of affected areas and complement in-field missions when safety and accessibility limit post-earthquake operations on the ground.

Note*: Damage thresholds (Dk = 0, 1, 2, … 5) are defined according to the EMS-98 damage scale [77]: D0 (no damage—to structural and nonstructural elements), D1 (negligible—no structural damage, slight damage to nonstructural elements), D2 (moderate—slight structural damage, moderate nonstructural damage), D3 (substantial—moderate structural damage, heavy nonstructural damage), D4 (near collapse—heavy structural damage, very heavy nonstructural damage), and D5 (collapse—very heavy structural damage).

While initial satellite images might differ from reality, they still provide a good indication of where problems lie. Miura et al. [78]) used data from the 2016 Kumamoto and 1995 Kobe, Japan earthquakes to initiate the identification of collapsed and non-collapsed buildings from post-disaster aerial images, using Neural Networks. Matsuoka et al. [79] further developed the method, stating: “Although the spatial resolution is low, ScanSAR images are capable of detecting changes in the ground surface, and if they had been available immediately after the earthquakes, a more efficient disaster response could have been achieved. With the help of AI, it may be possible to quickly produce images like those in Figure 2, showing the concentration of damage”.

It is valuable to compare the data shown in Figure 2 with the reality obtained through a building-by-building survey conducted by 7300 Turkish engineers and 400 professors from more than 20 universities in collaboration with the World Bank [47]. This campaign began soon after the events and focused first on the worst-affected cities in the 11 affected provinces. According to OCHA [80], 115,000 people were recovered alive from buildings in bad shape. Counting the buildings that did not sustain any damage (D0, see note in Section 2.1.2 above) is crucial for understanding why they performed well, whether due to lower seismic action, good construction practices, or both. Resolving this issue would be a significant achievement.

In summary, we want to emphasize the accuracy of satellite data obtained just a few days after the events. These data not only help pinpoint zones of greater destruction and the extent of damage but also provide estimates of the number of buildings that are slightly damaged, collapsed, or in need of demolition.

Notably, regarding the Al Haouz M_w 6.8 earthquake that occurred on September 8, 2023, due to the restrictions imposed by the Moroccan government, for an extended period of time, satellite imagery served as a primary source of scientific information for the international technical and scientific communities.

2.2. Drones

Unmanned aerial vehicles (UAVs), commonly known as drones, immediately covered the skies to ascertain from close distances the state of damage inflicted on the building stock. Owing to significant advancements in drone technology in recent years, drones can serve as fundamental tools for visualizing the consequences of disasters.

Several images are shown below to demonstrate how valuable they are. Their close-range observations provide a detailed aerial perspective, complementing and enhancing the information gathered from satellite images. One of the significant advantages of using drones in post-earthquake damage assessments is their ability to cover large areas in great detail. Traditional ground-based evaluations can be time-consuming and challenging, especially in situations where extensive damage has occurred across vast regions. Drones, on the other hand, can swiftly navigate the affected areas, capturing high-resolution imagery and video footage from various altitudes and angles. This comprehensive aerial perspective allows experts to analyse the extent and patterns of damage, identify areas of severe destruction, and assess the overall impact of the earthquake.

Furthermore, drones are instrumental in accessing difficult or unsafe locations. After an earthquake, certain areas may be structurally compromised or present hazards such as collapsed buildings, debris, or unstable terrain. Sending human inspectors into these areas can be risky. Drones can fly over hazardous zones and collect critical visual data without risking the safety of personnel. This capability enhances the evaluation process and allows for a thorough assessment of all areas, even those that are difficult to access.

The precision of drone images is practically of the same order as that of individual inspections. Drones can observe from above, sidewise, and different angles details that no one can make easily and at short notice. Even some of the interiors can be observed with small drones. Equipped with a highly sensitive camera, it is possible to perform a laser scan and recover many details to be stored for later analysis. In Türkiye, the removal of debris started immediately after the short time provided by authorities to recover people who were still alive. All images of collapsed buildings are now in the “hands of drone pictures and films” [81]. Researchers need these images to recover the status of the building stock damaged by the shaking. An individual inspection of the interior is even more effective, and video cameras can provide a live description of events. Additionally, experts can review and analyse the imagery to assess the severity of the damage, identify specific vulnerabilities or failure patterns in structures, and make informed decisions regarding the reconstruction process. Drones equipped with specialized sensors, such as thermal imaging cameras, can detect heat signatures, assess the structural integrity of buildings, and alert authorities to urban fires. While this capability was not a significant issue in Türkiye, where most buildings have seismic natural gas shut-off valves, it is crucial for identifying hidden dangers such as potential gas leaks or hot spots within structures. This information is thus vital for efficient rescue operations and ensuring the safety of all involved individuals as used by Mishra et al. [82], who developed a methodology of drone surveillance for search and rescue in natural disasters with great accuracy, based on the identification of a large number of images of people for human detection and action recognition.

The importance of these technologies is highlighted by the Disaster Risk Management Knowledge Centre [83], an initiative by the EC European Union dedicated to its latest Flash News (13 November 2023) and an entire section titled “Drones and Planes: Unprecedented Imagery Resolution Supports Disaster Assessment”. This initiative underscores these technologies’ pivotal role in “disaster response”, aiding in accurate damage assessment and recovery tracking, and promising substantial contributions to future disaster management endeavours.

2.2.1. Observations from Drones (Captured during Daylight Only)

Based on a large amount of drone footage captured in the initial days following the two main events, some of which were produced by international media outlets, it is possible to gain valuable insights into the overall impact of earthquakes on a wide range of structures. Compilations that combine not only drone imagery but also video camera animations constitute essential resources for studying and comprehending these events.

Based on the analysis of Videos #01–#05 (Video References are given in the Supplementary Materials) and Figure 3, Figure 4 and Figure 5, several insightful comments can be made regarding each observed situation. Drone footage, in particular, has significant potential for identifying individuals who are trapped within collapsed structures and need urgent assistance. For instance, in Kahramanmaras, a few days post-event, drone observations revealed a notable scene around the collapsed area near the football stadium. Here, makeshift tents had been erected to shelter the homeless, and there was substantial traffic congestion on the open roads, likely because residents attempted to salvage whatever belongings they could. This area appears to have suffered extensive devastation during the daylight M_w 7.6 earthquake, although this assertion warrants further verification.

Figure 3, Figure 4 and Figure 5 illustrate various modes of building collapse, highlighting the complexity and variability of structural failures. Some buildings have experienced total collapse, while others show partial collapse affecting specific sections, such as front or back walls, stairways, and other structural elements. In certain cases, only the ground floors of buildings are impacted, whereas in others, the first two or three stories have collapsed, leaving the upper stories largely intact but occasionally tilted. Moreover, some buildings have collapsed in clusters, either at street corners or within the central areas of blocks, indicating localized patterns of structural failure.

Figure 3a presents a striking example of the variability in the extent of damage, showing both collapsed and non-collapsed buildings within short distances. This variability underscores the complexity of the impact of earthquakes on different structures. Figure 3b depicts a different pattern of structural behaviour in buildings ranging from 4 to 6 stories high. These structures exhibit a “house of cards” dismantling effect, characterized by a variety of collapse scenarios, including random tilting, partial collapses, torsion, sliding roofs, and the displacement of artefacts on terraces.

Complementary observations of the built stock reveal several notable aspects that contribute to a deeper understanding of structural behaviour during seismic events. One prominent form of collapse observed involved the loss of the first and/or second floor, often accompanied by the failure of a corner column. This structural failure caused the building to lean slightly before pancaking almost vertically. This phenomenon is reminiscent of the structural behaviour observed following the 2017 earthquake in Mexico.

In other instances, structures lost the first two floors due to the out-of-plane failure of brick walls. However, the collapse halted at that point, preventing further structural failure. Unfortunately, no footage capturing the exact mechanism of overturning was obtained, leaving some aspects of these failures to be further investigated.

In some cases, parts of buildings collapsed completely, while other sections remained standing. This partial collapse scenario highlights the varied impacts of seismic forces on different parts of the same structure. Additionally, some buildings began to collapse at higher stories, particularly at elevations where the plan section was increased. This observation suggests a correlation between structural design changes and points of failure, which will be further elaborated upon.

The second earthquake exacerbated the number of collapses, particularly in buildings already compromised by the first event. Many buildings that withstood the initial quake succumbed to subsequent failures, with tragic consequences. While numerous individuals survived the first earthquake, some returned to their homes only to die during the second event, underscoring the devastating impact of sequential seismic activities.

Industrial facilities also suffered damage in various locations, although the extent of damage varied. Notably, the majority of collapses occurred in buildings ranging from 5 to 10 stories high and were predominantly constructed in the 1970s or 1980s. This pattern is likely related to the frequency of ground shaking and the strong pulse-like ground motion generated by the fault rupture, as discussed by Baltzopoulos et al. [84]. Interestingly, smaller buildings and taller structures did not experience significant collapse, suggesting that their structural resilience was influenced by different seismic response characteristics.

Site-specific factors such as basin effects, soil liquefaction, and other local geological conditions undoubtedly aggravated the situation, contributing to the variability in structural damage. On a positive note, gas stations largely escaped significant damage, which can be attributed to their steel structures being designed to withstand substantial forces. This example of resilience underscores the importance of robust structural design in mitigating earthquake damage.

Furthermore, equipment on roads, intersections, poles, and streetlamps sustained minimal damage, although some exhibited significant oscillations, as evidenced in video footage. These observations highlight the varied impacts of seismic forces on different types of infrastructure and the importance of considering these factors in future urban planning and disaster preparedness efforts looking for better sustainability.

In summary, the combination of drone footage, detailed visual analysis from Figure 3, Figure 4 and Figure 5 and complementary observations of the built stock provide a comprehensive understanding of the diverse and intricate presentation of building collapses in the affected areas. This information is crucial for informing future building codes, improving structural resilience, and enhancing disaster response strategies.

Figure 3. Variability of collapse and non collapse: (a) Collapses in Iskenderum on both sides of the road; the amount of obstruction is much larger than a 45° angle may suggest. (b) Partial building dismantled in Hatay, requiring total demolition (with copyright permission from PEMA [85]).

Figure 3. Variability of collapse and non collapse: (a) Collapses in Iskenderum on both sides of the road; the amount of obstruction is much larger than a 45° angle may suggest. (b) Partial building dismantled in Hatay, requiring total demolition (with copyright permission from PEMA [85]).

New forms of “pancaking” appear more like “slide dragging”, a mechanism not commonly observed before these earthquakes. One of the most striking images from this earthquake sequence is the collapse of large, reinforced concrete (R/C) buildings constructed post-2000, some of which were built in the last 10–15 years that completely tilted. This incident is a consequence of insufficient inspection, leading to poor concrete quality and engineering design of longitudinal and transverse steel bars, which fail to meet the required standards [86]. Additionally, the absence of basements (just one level below the surface in almost all buildings) to balance upper stories (as observed by the authors) and the presence of large velocity pulses in the records [87] contributed to these failures.

A tentative explanation for the “slide-dragging” mechanism may consider that columns rotate at both the top and bottom ends, remaining intact along their heights, resulting in the frame being articulated at those nodes. The slabs then move laterally due to gravity, landing on top of each other with a separation equal to the inter-story height. In contrast, during a pancake collapse, insufficient story strength causes columns to explode or rotate in more locations than just the two nodes due to flexural or shear behaviour, leading to slabs falling almost vertically.

Figure 4. Extent of damage in several cities of Türkiye: (a) Mid-rise RC residential buildings that have collapsed in a side-sway manner) (capture from Video #03); (b) overturning of foundation (@adamaxoi).

Figure 4. Extent of damage in several cities of Türkiye: (a) Mid-rise RC residential buildings that have collapsed in a side-sway manner) (capture from Video #03); (b) overturning of foundation (@adamaxoi).

Figure 5. Several modes of partial and full collapse as captured by drones: (a) First story on the right-hand side; (b) right-lateral movement following the 1999 earthquake [88], leading to an almost vertical pancake; (c) first story on the bottom side; (d) overturned building in Golbasi due to a lack of anchorage of foundations together with liquefaction [89].

Figure 5. Several modes of partial and full collapse as captured by drones: (a) First story on the right-hand side; (b) right-lateral movement following the 1999 earthquake [88], leading to an almost vertical pancake; (c) first story on the bottom side; (d) overturned building in Golbasi due to a lack of anchorage of foundations together with liquefaction [89].

2.2.2. Comparing Prior and Post-Earthquake Images

Comparing prior and post-earthquake images (see Video #06) can yield valuable scientific insights into the effects of seismic events and the associated changes in the affected areas. Some potential scientific findings can result from such comparisons: surface deformations, fault rupture mapping, landslide and mass movements, subsidence and uplift, damage assessment and vulnerability analysis, surface faulting, and rupture length. A detailed analysis of image comparisons can contribute to determining the surface fault trace, rupture length, and associated parameters that are characteristic of this earthquake. This information can aid in fault characterization, seismic hazard assessment, and modelling of future earthquake scenarios.

Importantly, the scientific findings resulting from image comparisons depend on image resolution and quality, the availability of pre-earthquake data, and the expertise of the researchers conducting the analysis. These findings contribute to the overall understanding of earthquake processes, hazard assessment, and mitigation strategies, ultimately enhancing our ability to prepare for and respond to future seismic events. Video #06 presents a comparison before and after the events, which shows the impact produced by the earthquake events, revealing a large number of collapsed buildings.

2.3. Video Cameras

As previously mentioned, video cameras have been used in laboratory research for many years to collect data during experimental tests [90], particularly in situations where direct contact with the structure under analysis may be challenging or when significant displacements are expected. Numerous databases now exist that store information from past tests [91]. However, using video cameras to observe the behaviour of structures impacted by natural hazards (such as earthquakes, tsunamis, landslides, etc.) is a relatively new field with limited research. The primary reason for this is the recent availability of such data. It is only in the last two decades that people with video cameras have become willing to share a wide range of footage on the internet and other media during disasters. These videos may contain highly valuable data for understanding the behaviour of various objects, and video cameras have become indispensable tools for elucidating the dynamics of these events, effectively serving as real-world research laboratories. However, they cannot, by any means, replace analytical and experimental studies. In fact, observations made with video cameras are already being supported by analytical modelling, which can recover the ground motion applied to structures during events where large visible displacements occur.

2.3.1. Brief Note on the Evolution of Video Cameras to Study Disasters

The methodology presented in this paper, which employs advanced technologies such as video camera footage and drone imagery to analyse both man-made and natural disasters, is scarcely represented in the literature. Conversely, the use of video data for health monitoring of structures (HMS) has undergone considerable advancements in recent years, becoming a crucial component of measurement devices in laboratories and in the field. To further discuss this topic, Wang et al. [91] conducted a comprehensive review, covering various aspects related to video data, ranging from disaster video capture to methodologies for image processing during laboratory testing, focusing on extracting information regarding nonlinear structural behaviour. The use of drone imagery has become increasingly prevalent since its introduction to the market in approximately 2016. Bychkov [92] is an example of a criticism of the well-set theory of fault rupture based on video camera observations. Although few international agencies have gathered data for further studies, there has been limited utilization of these specific technologies in this context.

An exception to this trend is the Euro–Mediterranean Seismological Centre [1], which features a section called “Medias” on its website that presents unexplained pictures and video-camera footage submitted by witnesses. In PART III of the paper, we employed these resources as examples to analyse several characteristics of the wave passage.

Examining the evolution of video cameras in more detail, we see that in the last decade of the 20th century and early 2000, Japanese researchers developed technology to use videotaped pictures from security cameras in convenience stores and ATM machines to calibrate the seismic intensities obtained from their questionnaire surveys [93,94]. They were able to recover ground motion during large earthquakes, “encouraging readers to collect and make use of videotaped pictures in earthquake engineering and seismology studies”. However, with the advent of strong motion instruments, especially in Japan [95], video camera technology no longer plays such an important role.

The 11 March, 2011, Tohoku M_w 9.2 earthquake was one of the first natural events where video data provided substantial insights crucial for studying such events. Ngo et al. [96] used this information and Google Earth^® to determine with great precision both the height of tsunami inundation in several coastal cities of Japan and the flow velocity of the waters. Expanding the work of Ngo et al. [96] and Mcdonough–Margison et al. [97] developed an Imagery Catalogue to address all available data.

Yang et al. [98] published a paper in which 68 videos recorded during the 12 May 2008, Wenchuan, China, earthquake were used to estimate the macroseismic intensity in a region of 1000 km around the fault area via visual inspection.

Oliveira et al. [46] are among the authors who emphasize the importance of these technological devices, highlighting numerous compelling situations that warrant in-depth analysis.

Vinnell et al. [99] studied earthquake shaking via video footage to analyse human behaviour during an earthquake in New Zealand and obtained very interesting results from the observations of 68 individuals subjected to two types of shaking.

The quantification of movement produced during an earthquake, tsunami, or other geohazard event (landslides, lateral spreading, etc.) relies on the fact that it is possible to measure dynamic displacements if our sight can discriminate small movements at a distance in situations where a line or point (corner of a building) or another landscape feature is measurable. Oliveira et al. [46] used the motion of a tall building as a target 5 km away from the video camera to easily estimate the frequency of vibration of the first mode and, with reasonable accuracy, the amplitude of motion.

Nevertheless, we should be mindful of the uncertainties in extracting information from video camera footage, especially if we aim to conduct quantitative analyses. These uncertainties can be summarized as follows: the location of the camera is often not well known (e.g., the building story and the number of stories), and the assessment of physical parameter values needed to characterize the observed motion is difficult to perform (e.g., evaluating angles of oscillating lamps). Therefore, caution should be exercised to mitigate these uncertainties.

Finally, we would like to note that mixed solutions are available, such as drones equipped with video cameras and inspections that use video cameras to record observations. However, in this analysis, we focus on video cameras that capture the wave passage and its consequences, excluding other types of observations.

Myagmar–Ochir et al. [100] make a sound review of video surveillance systems in Smart City. They introduce the complexity of multiple systems observing strategic locations and using advanced technologies of AI to provide insights to enhance safety, security, and, in a word, quality of life. They also look to key issues such as authentication, authorization, data integrity, storage, and management, as well as the limitations and future developments in the field. Unfortunately, in this area of disasters, the application of these technologies is still in its early stages, with video footage obtained at random and not prepared as a scientific experiment.

2.3.2. Examples of Video Camera Observations during the Türkiye–Syria Events

Regarding video footage, the initial earthquake, which occurred at 04:17:34 a.m. local time, lacks the richness seen in events such as the 2015 Nepal earthquake or the 2017 Puebla earthquake in Mexico. The reasons might be as follows: (i) there were few installed video cameras; (ii) the earthquake occurred at night, resulting in a loss of electricity supporting the cameras just 20 s after the onset of the first event; and (iii) numerous collapses occurred in areas that experienced higher intensities of shaking, which potentially led to damage or disruption in the transmission of images from the cameras. As we move further from the epicentre, the number of instances captured by video cameras increases. Nevertheless, several cases deserve mention near the ruptured fault.

Information about the second event at 13:24:50 p.m. local time is available essentially from personal mobile devices. Over time, an increasing number of videos began to emerge on the internet, some with interesting topics, others with repeated topics, and others with some redundancies and compilations of shorter videos. It is important to be cautious when assessing the authenticity of images or video footage. For example, we encountered instances where videos from unrelated events were intermixed with footage from the current event. Some companies working in the area, sometimes to promote their products, create these fake images. These images may be of other earthquakes or may be due to demolitions (Figure 6a–c). We, therefore, advise users to meticulously verify the validity of the captured material before proceeding with analysis. One crucial point of validation is the presence of a timestamp embedded in most video data, which not only indicates the time of capture but also governs the playback speed.

As time passes following the Türkiye–Syria event, it is likely that the links providing access to the videos may change, and the referenced links could undergo slight alterations. We cannot predict whether future data-protection regulations might impact this initiative. Nevertheless, up to this point, YouTube has been demonstrated to be a reliable platform for accessing footage depicting such occurrences. We have adhered to the recommended format for citing this type of information [101].

2.3.3. Seismic Observation Cases

A series of three videos lasting approximately 23 to 35 min (Video #07a, Video #07b, Video #7c) encompass compilations of recorded footage from video cameras. These recordings, essentially the first two videos, were captured both at night but also during the day following the M_w 7.8 event. These videos exhibit a sequence of captivating wave propagation phenomena occurring both in the natural surroundings and inside buildings. As previously mentioned, the analyses of these videos yield crucial insights.

Observations from the first earthquake, M_w 7.8 (at night)

The videos presented in the video references show many cases where most of the points described below can be observed (Video #08). Several examples presented here have already been observed in other earthquakes [46], but the coverage of this earthquake is much richer, especially for RC building collapses. Please note that we are not marking all the videos where we observed “the described phenomenon” but have selected only a few to highlight their relevance and to avoid excessive information.

The first earthquake, with a magnitude of M_w 7.8, occurred at night and has been thoroughly documented by various YouTube videos. The precise timing of the earthquake, 04:17:41 h (local time), is captured in numerous videos, with the initial shaking commencing mere seconds after the onset of the earthquake (see Figure 7). Notably, the initial shock lasted for 25 s, followed by a second shock 30 s later, as observed in Video #08. Multiple records indicate two significant onsets, likely pointing to different “fault asperities”.

Within the first 20 s of shaking, a power outage occurred, plunging the area into darkness. The arrival of primary (P) and secondary (S) seismic waves was distinctly noticeable at locations further from the epicentre, as evidenced in Video #9a and Video #9b. The suspended lamps swayed approximately 30°, while chairs without wheels remained stationary. In contrast, chairs with wheels moved several centimetres, even as objects and monitors on tables stayed still. Pendulum lamps in mosques exhibited low-frequency swinging (T = 5/6 s), whereas computer monitors oscillated at a higher frequency (T = 0.9 s).

In supermarkets, items fell from shelves, and in offices, paper descended from top shelves, while pendulum lamps swung at large angles (Video EMSC #13, Video EMSC #17). Hanging shirts moved synchronously, and curtains swayed laterally. Birds displayed unusual behaviour prior to the seismic vibrations, showing their heightened sensitivity to seismic waves, which surpass human detection.

Suspended lamps at significant distances, over 600 km from the epicentre (or fault trace), registered the motion, as shown in Video EMSC #12. Although few swimming pools exist in the area, one video recorded large waves (Video EMSC #21). In Diyarbakir, a small aquarium (50 cm long) exhibited sloshing modes with frequencies of 3 Hz at a 5 cm water height and 2 Hz at a 10 cm height. Various aquariums showed water oscillations, and water in a basket located 610 km from the epicentre displayed high-frequency ripples (Video EMSC #11).

Façade collapses in buildings, similar to those observed in previous events such as the 2017 earthquake in Mexico, often led to the collapse of other structural parts. Illumination poles exhibited out-of-phase movements and cracks in roads opened and closed, as noted by Oliveira et al. [46]. Interior and exterior doors jolted open and shut, and snow fell from rooftops. Frames for X-rays (walkthrough metal detector) balanced like a two-leg frame (Figure 8, Video #08).

People struggled to maintain their balance, swaying when standing alone, often needing to gather in groups or cling to handrails for support (Figure 9). Climbing stairs became challenging, and cars moved laterally and longitudinally, appearing as if their brakes were not engaged (Video #07b) (Figure 10a,b). Cars were rocked laterally at a frequency of 1.2 s. Observations from various videos indicate that object responses were predominantly unidirectional, except for lamps, which oscillated in both directions (umbrella-type) (Figure 11). During the oscillations, people were seen praying, and most lamps in a lamp shop oscillated in phase.

A short explanation for the oscillation of a car in the transverse direction is as follows: a harmonic force of 300 N was applied to the car at A (by two people), as shown in Figure 12, causing the vehicle to oscillate. If the vehicle weighs 1000 kg, the ground acceleration (PGA) that causes small oscillations is approximately 3 m/s² (the centre of gravity is considered to be 0.9 m from the ground) (PGA × 0.9 × 1000 kg = 300 N × 1.0 m).

Observations from the second earthquake, M_w 7.6 (during the day)

The second earthquake, which registered a magnitude of M_w 7.6 and occurred during the day, resulted in a series of significant observations. One notable phenomenon was the occurrence of a small tsunami near fishing nets in the harbour, which was likely exacerbated by the strong vertical component of seismic activity. This vertical force may have contributed to the severity of the structural collapse observed during the event.

Numerous instances of building collapse were captured on film, primarily showing the initial moments of these catastrophic failures. The resulting dust clouds, which are common in all earthquakes and have been documented since ancient times, significantly obscured visibility. These clouds not only prevent people from seeing the landscape but also pose critical health risks, especially for individuals suffering from lung illnesses (Figure 13a).

A particularly important mode of building collapse, frequently observed in various video recordings, involves the failure of a solitary corner column. This failure often initiates the collapse of the entire structure. The corner column was typically displaced from its vertical position, a likely consequence of the preceding M_w 7.8 shock, which weakened the structural integrity (Figure 13b). This type of failure has been documented in other earthquakes where camera footage was available, providing valuable insights into the mechanics of such collapses [46] (see Video #10a, Video #10b).

The visual documentation from this earthquake highlights the critical importance of understanding the sequence of structural failures and the role of vertical seismic components in exacerbating damage. These observations not only contribute to our knowledge of earthquake dynamics but also underscore the need for improved building designs that can withstand such complex forces.

Figure 14 offers an intriguing perspective by presenting the same scene (from the same angle) before and after the event, allowing for an immediate understanding of what has occurred. This comparison can be achieved via photographs and a few video cameras. Various agencies have utilized these comparisons for public dissemination.

Videos #11 to #13 depict numerous instances of various collapsed buildings, which is crucial for understanding the progression from initiation to collapse. Further examples could be included here, all of which will be thoroughly examined once the relevant information becomes accessible.

Figure 15 clearly demonstrates that the collapse of certain buildings within a block is almost vertical (Video #14), while others are left standing, contradicting the values of the so-called “behaviour modifiers” used in fragility curves for buildings situated at the ends or corners of blocks. This observation shows that earthquakes present numerous unprecedented challenges that deviate from traditional concepts.

Other situations

Regrettably, only a limited number of health monitoring instruments have been implemented in buildings within the region. Nonetheless, by using video cameras, it becomes feasible to analyse dynamic images in detail for a few cases. One such case features an almost square reinforced concrete (RC) building built prior to 2000 (refer to Video #07b, timestamp 00:25:53). This 10-story structure exhibits oscillations at a frequency of approximately 0.55 Hz over a span of 20 s, indicating a combination of translational and torsional movements. Notably, this frequency is remarkably low for a structure of this size and type. These observations were captured by a fixed camera designed to record displacement motion, with a fixed light pole serving as a reference line between the camera and the building. The estimated amplitude at the building’s top is 0.4 m (equivalent to a peak-to-peak amplitude of 0.8 m), which is probably due to nonlinear behaviour. Based on these observations, modelling this RC structure might be a challenge for even an expert in nonlinear studies.

In the present situation, it is crucial to differentiate between the damage caused by the M_w 7.8 event and that caused by the M_w 7.6 event (occurring during nighttime and daytime, respectively) to comprehend the extent of damage incurred during the initial event and, consequently, the amount attributable to accumulated damage. This situation represents a new area of research—the occurrence of one event followed by another that may not be classified as an aftershock, a phenomenon that, to our knowledge, has not been addressed before. Video cameras are invaluable tools for monitoring these effects, which are being observed for the first time. Understanding how people reacted after the initial event—whether they left their houses or stayed inside—is crucial, although it remains a challenging topic. In this context, as previously mentioned, video cameras can provide important information about the timing of a collapse. Figure 13 is a good example of this discrimination, showing damage to the ground floor column prior to the M_w 7.6 event and the collapse during this event.

In another case, we clearly observed the vibration of two Ottoman minarets with two balconies in the Yaygın village of Muş, 400 km from the epicentre, oscillating at a frequency of 0.2 Hz for 15 s (Video #07c). According to studies of similar structures Bayraktar et al. [103], this frequency is very low, and these structures should be studied in detail to explain the observed pattern, which is not common.

To show that video camera footage can enhance the understanding of historical events, we refer to two cases where the absence of video footage would have hindered a comprehensive analysis of the behaviours exhibited by these structures. The first case is a retrospective examination of the collapse of the Tetrastyle Canopy, while the second case is the collapse of the Dharahara Tower [104], both of which are located in Kathmandu. These incidents occurred during the Nepal earthquake on 25 December, 2015. Without video footage, it would have been impossible to effectively recover the collapse of these two historical structures. In fact, in the second case, a nonlinear study that did not use video footage led to the incorrect identification of the type of collapse [62]. Another example deserving the attention of experts is the collapse of a stone lantern in Shinto Temple, which was video recorded from the initial motion to full collapse during the recent earthquake on 1 January, 2024, in Noto, Japan [105]. In this case, all the necessary information is available to perform a nonlinear analysis of a standing column made of superposed solid blocks and to attempt to explain the observed behaviour. This stone lantern has experienced many collapses throughout history, and numerous analytical and experimental studies have been conducted. However, a study with a complete sequence of images such as these has never been captured, or at least not reported in the literature.

Numerous other subjects were omitted from this study, primarily because no visual animation was accessible. Despite the advantageous perspective provided by drones, certain aspects, such as liquefaction, landslides, lateral spreading, critical infrastructure, industrial facilities (Video #15), and historical landmarks, were not covered. We handpicked specific drone videos to highlight significant phenomena witnessed during these earthquakes that warrant acknowledgement: far-reaching fault ruptures (Video #16, Video #17) and the destruction of mosques and minarets (Video #18 to Video #20).

While the subject of base isolation is not explicitly addressed in this review paper, it undoubtedly represents a significant stride in the realm of science and technology. This progress is well documented through observations in video recordings, as shown in Video #21a and Video #21b. These videos vividly depict the actions of medical personnel in the neonatal unit during critical moments within hospitals equipped with such a system. Another interesting example of a video camera as a monitoring device during an earthquake is the capture of the relative displacement of an isolated support bearing in a hospital during the Hualien, Taiwan, M_w 7.4 earthquake, 3 April 2024 (Video #22).

2.3.4. Additional Examples Captured with Video Cameras to Be Used in Conjunction with Macroseismic Intensity Scales

The following examples highlight observations exclusively captured through the assistance of new technologies, a central topic of this review paper. The various points discussed earlier can seamlessly integrate into the information required for assigning intensity measures according to scales such as the EMS-98 [77] or the ESI-2007 [106], the latter more oriented towards an environmental context. In fact, video camera footage can help in assigning macroseismic intensities, encompassing both intensities lower than V-VI (EMS-98), which focus on the behaviour of objects inside buildings or in nature, and intensities higher than VII, which assess the extent of damage to the building stock [95].

Examples with nature

The significant fault fissure near Hatay (Antakya) (refer to Video #16 and Video #17) demonstrates the importance of image observation. Lateral displacements exceeding an offset of 7–8 m along the fault line are discernible with various methods, including drone imagery and satellite pictures. This phenomenon is also evident in the offset of railway tracks and road alignments, as depicted in drone imagery.

While small bushes balance during the passage of seismic waves, trees exhibit minimal oscillation due to their near-critical damping. The observed motion was probably in low-intensity shaking.

Examples with the population

Video cameras are excellent tools for illustrating how people behave during earthquakes. They reveal individuals’ initial reactions, their adherence to rules set by authorities, and their evacuation from buildings. Eleven minutes elapsed from the first shock at 04:17 a.m. to the first strong aftershock at 04:28 a.m. This duration provided ample time for occupants to safely evacuate a 12-story building, potentially saving many lives. We should be able to track people’s behaviour under these circumstances. Here, again, caution should be exercised when consuming media reports. As stated in a BBC News article [107], the Google Android Earthquake Alert System for early warning (EEWS), which was implemented in the region in 2021, reportedly failed to send alerts to the local population. However, a subsequent clarification was provided during a “Fibre Optics Workshop” held on 6 July, 2023, in Plouzane/Brest. The system was actually installed as a pilot experiment in the Istanbul area, a considerable distance from the affected region and not in the affected area.

In a recent retrospective analysis of the approximate hundred accelerograms that recorded the first event, according to Rea et al. [108], “it would take 10–20 s to issue the alert with leading times 10–60 s for a threshold IMM = IV. For larger thresholds, the times would be larger and smaller, respectively. The complexity of a bilateral rupture would condition these numbers”. These findings are helpful for assessing the reliability of the EEWS.

• PART II—Quantitative Insights: Drone Imagery and Video Camera Capabilities in Understanding Road Blockages and Wave Propagation

In Part II, we present a selection of quantitative examples highlighting the advantages of drone imagery in assessing road blockage. Additionally, we explore the capabilities of a variety of video cameras, which enable the observation of the behaviour of simple objects and contribute to a better understanding of wave propagation and its effects.

3. Blockage with Debris: A Model Suggested from Field Observation (from Pictures and Drones)

In Section 2, we present a comprehensive overview of how several buildings collapsed during the Türkiye–Syria earthquakes. Figure 3, Figure 4, Figure 5, Figure 14, Figure 15 and Figure 16 depict various common collapse patterns observed during these events, with numerous other instances documented across the affected regions. The collapses captured through aerial imagery (from drones, planes, satellites, and airborne laser scanning) are vital for developing new proposals to analyse road blockages. We briefly summarize the findings from previous earthquakes and from analytical and experimental studies. Using the data collected during this event, we compare our proposal with earlier studies.

3.1. Amount of Debris Generated by Destructive Earthquakes

The debris caused by the collapse or partial collapse of buildings is a critical issue for the functionality of roads after an earthquake, especially in urban areas. The blockage or reduction in the capacity of a road for evacuation operations is as vital as the amount of debris to be removed. The Haiti earthquake, in 2010, produced 10 million m³ of debris, while Kobe (1995) and Tohoku (2011) caused 20 million tons each [109]. New technologies such as aerial imagery from satellites and drones [110], together with artificial intelligence (AI), machine learning (ML), etc., have been used for the estimation of debris amounts [111] during the Türkiye–Syria earthquakes, which produced 210 million tons [43]. This volume is an enormous amount of debris for treating and finding suitable disposal sites. Six months later, the mountains of rubble (debris), asbestos, and heavy metals from the collapsed building pose a health risk to the communities, especially at Samandağ, one of at least 18 locations where authorities have dumped rubble across Hatay Province.

3.2. Prior Studies on Road Obstructions

Schweier et al. [112] classify collapses into five categories: “Inclined layers”, “Pancake collapses”, “Debris heaps”, “Overturn collapses”, and “Overhanging elements”, each accompanied by a set of drawings. The classification is based primarily on the building typology, height, and plant geometry, as well as the degree of damage (ranging from D1 to D5), with corresponding post-earthquake geometry and volume values. They compiled a catalogue using laser-scanning images taken before and after an earthquake, incorporating data for all these elements. This catalogue shares many common ideas with those presented below.

Report 4 of SYNER-G [113] proposes dealing with “road obstruction” by building a model to estimate the extension of roads obstructed by the total or partial collapse of buildings after most conditioning variables controlling the phenomenon are set. These variables are the “type of buildings”, the number of stories, and the width of the road, counting the distance between the building face and the road. The report establishes a simple rule for an RC building with a D5 damage degree:

X = 2/3 × n  (unit: m)

(1)

where X is the distance occupied by the debris perpendicular to the road´s longitudinal axis and n is the number of stories. For masonry buildings, the reasoning is more complex and is based on “geotechnical principles”. The road is considered blocked if half of its width is obstructed.

Moreover, one collapse is sufficient for a road to be considered blocked. Therefore, urban tissue (perpendicular intersections, skewed, corners, etc.) also contributes to the overall obstruction.

Osaragi et al. [114] and Ravari [115] proposed other models, summarized in Figure 17, for masonry and RC buildings, respectively. In this case, (a = X) is provided by Equation (2):

X = H × tan(20°) × M   (unit: m)

(2)

where H is the height and M = 1.3 for RC buildings and M = 1.1 for other materials.

Sediek et al. [109] developed a model based on analytical and shaking table experiments for RC structures of various heights (n = 5 to n = 12) with and without infilled walls. Based on neural networks, they formulated an expression that captures the extent of the debris field as a function of the building base. The exterior a (Figure 17) is 0.8 to 0.7 of the bases for 5 to 12 stories, respectively, for aligned collapse and 0.5 to 0.4 for skewed collapse.

3.3. Proposal of a Model for Assessing Road Blockages after Earthquakes and Comparative Analyses

By examining the road blockages and considering the various instances of partial and total collapse of the reinforced concrete (RC) buildings in Türkiye and Syria, as depicted in Figure 3, Figure 4, Figure 5, Figure 14, Figure 15 and Figure 16, we can refine the previously presented models for the cases observed frequently in Türkiye. The abundance of examples in various affected cities makes the classification we propose below both evident and deserving of our attention.

Specifically, we can enhance the model proposed by Osaragi et al. [114] and Ravari et al. [115], as shown in Figure 17, by incorporating the schematic representations of the full (D5) and partial collapse (D4) of RC buildings illustrated in Figure 18. In this context, the critical parameter is primarily the base dimension b. In one case, the number of stories (n) also counts. This observation is of most interest to estimate the extension of blockage of a road due to the collapse (partial or total) of buildings.

Partial collapses occurred frequently, including instances such as the collapse of entire façades and groups of exterior stairways, which obstructed space in front of existing buildings.

From a practical viewpoint, the following considerations arise for full collapses:

-

When buildings collapse between adjacent structures, the collapse tends to be nearly vertical, resulting in debris being confined to a limited area. ((Figure 18a)-2)

-

When isolated buildings without infills collapse and the structural situation is a weak column–strong beam, the hinge is most likely at the connection node. Consequently, the collapsed building often topples to the side, occupying a large amount of space ((Figure 18a)-3, 4).

-

In the case of foundation failures (just one floor below ground, or coupled with liquefaction), the complete overturning of the structure is critical ((Figure 18a)-5). This phenomenon can result in the obstruction of roads, either partially or entirely. Similar collapse modes have been constructed for past events, namely for the Kobe earthquake in Japan (1995) and the earthquake in Chile (2010).

For partial collapses, we have essentially two situations:

-

Façade collapses as a whole ((Figure 18b)-1)

-

Façade slides down and debris pulls over the base ((Figure 18b)-2).

Figure 18 shows that all of the proposed values are much larger than previously believed: a = X easily reaches 1.0 × b to 1.5 × b for full collapse and 0.5 × b to 1.0 × b for partial collapse. Furthermore, for complete overturning, the numbers are equal to the total building height. More specifically, in the first four cases of Figure 18, X varies from 0.5 × b to 1.5 × b, and in the last two cases, X varies from 0.5 × b to 1 × b. In the cases of lateral “sliding-dragging” and overturning, the values can reach n × b and the maximum h. For partial collapse, the values vary from 0.5 × b to 1 × b.

For masonry structures, such as those presented in Figure 3b, the obstruction is very different from that in the above and previous studies, and Equations (1) or (2) lead to reasonable values.

A final consideration is the randomness in the buildings that undergo complete collapse. In this instance, we examine reinforced concrete (RC) buildings ranging from 7 to 10 stories high, likely constructed before 2000 and located in a long valley. Assuming that construction practices are similar throughout the area, we observe in Figure 19a three zones of total collapse interspersed with clusters of standing buildings, which may require demolition but remain intact. Could standing waves traversing the valley with crests and troughs cause differential amplification, as suggested by the SAR deformation field (Wang et al. [116]?

However, the collapse of single structures, while many other structures remain nearby, may be an exception to the general collapse pattern, probably because of incorrect alterations in the structural system throughout the life of that building and the lack of randomness. A comparison of prior and post-earthquake imagery may help to explain those failures.

In contrast with the previous observations, small buildings do not show damage to the roof (Figure 19b).

In other areas of poor old masonry buildings located after crossing the mountain ridges to the east (Figure 1b), no damage is observed, even though they are at similar distances from the fault trace, with heightened intensities near the fault trace and attenuated intensities further from it, especially to the east of the mountains.

4. Quantitative Analyses of Ground Motion with Video Cameras

Video camera footage represents important source material that can be used for understanding the behaviour of nature and built objects under seismic input. This material has great potential for qualitative interpretation and is crucial for generating new ideas for research. In the previous sections of this review paper, we presented several examples to illustrate this remarkable tool. However, in a few cases, we can use video cameras to perform quantitative analyses. We have done so in the past [46] and in several parts of this paper, as in Section 2.3.3. In Lemos et al. [104], we present suggestions for many more exercises possible only with video footage, correcting earlier interpretations made without using video footage of the collapse. Video cameras have proven instrumental in capturing structural movement during seismic events, as demonstrated in the case of the Hualien, Taiwan, M_w 7.4 earthquake on 3 April 2024 [Video #22], which shows the relative movement between the foundation and the base isolation system of a hospital. Furthermore, experiments can be devised for future events, utilizing specially designed video cameras tailored to observe such movements with precision.

This is relevant to scenarios such as the ones reported in the Annex Table, whose values were derived using these theoretical and experimental models.

4.1. Wave Propagation Observed in Objects (Oscillation of Lamps, Water in Tanks, Falling of Objects, etc.)

In the present situation regarding the Türkiye–Syria earthquakes, we started collecting as much information as possible from video cameras. In addition to our own search, the majority of films were obtained by EMSC [1] under the rubric DYFI-EMSC [117] and correspond to the 4:17 a.m. event. The list, organized by distance to the epicentre, is presented in the Annex in Supplementary Materials. As we can observe, there is only one data point within the first 100 km from the epicentre; all others are at larger distances, extending to values over 600 km. The reasons are explained in Section 2 and correspond to shaking intensities not higher than VI/VII (EMS-98). We visualized all videos and classified the types of phenomena under scrutiny. Most cases involved hanging lamps, curtains, poles, and sloshing water. Based on the descriptions and previous studies (e.g., the maximum angle of rotation of a particular long suspended lamp), we estimated the input ground motion.

The values obtained are from single observations and, therefore, do not reflect the number of cases with descriptions similar to those recommended in the EMS-98 scale. Additionally, the size of the building and the story height where the observation was made were not considered. Nevertheless, we assumed that the values provide an insightful representation of local input motion, despite the significant uncertainties.

The large values of inputs were not considered in the first analysis because of the difficulty in determining the exact locations of the recordings and the large uncertainties in assessing their values from large angles of rotation, objects falling from shelves, etc., in cases of D1/D2 damage.

We studied several aspects related to this sample of observations: (i) the correlation between intensities (EMS-98) and PGAs, (ii) the attenuation of PGAs with epicentral distances (not JB distances) up to large distances, as much as 650 km, and (iii) a map showing the geographical locations of observations. The results of the correlations are presented in Figure 20, which shows quite low values of PGA for particular intensities compared with those reported in other studies [118].

The results for the attenuation of waves are plotted as PGA vs. epicentral distance (JB distance would probably yield better results) up to large distances, as high as 650 km (Figure 21). The values for points close to the fault are less reliable because of difficulties in analysing the angles of lamp oscillation. If we include in Figure 21 the values referred to in (Section 4.1) above for Kahramanmaras (i.e., PGAs are 320 to 500 and 700 cm/s² for epicentral distances of 35–45 km), the curve rises significantly towards smaller distances.

The results exhibit a comparable trend and great scatter to the outcomes derived from SM stations, as depicted in Figure 22a,b. The attenuation obtained from video camera footage of oscillating lamps is more similar to the attenuation for a Sa (period T = 1.5 to 2 s), which corresponds to the period of the lamps rather than the PGAs.

Note: The 1755 earthquake M_w 8.5⁺ was experienced in cathedrals in Barcelona 1200 km away from the presumable epicentre where the chandeliers swayed [46]. Another interesting example is the observation of oscillating lamps, such as those in Figure 11, of a few degrees on the 9th floor of a 12-floor RC building founded in a Miocene formation, 850 km away from the Morocco earthquake on 8 September, 2023, M_w 6.8 [119].

Figure 22. Comparison with values taken from video camera footage: (a) Wan et al. [120]; (b) Baltzopoulos et al. [84], following Bommer et al. [121] proposal for “rock sites”.

Figure 22. Comparison with values taken from video camera footage: (a) Wan et al. [120]; (b) Baltzopoulos et al. [84], following Bommer et al. [121] proposal for “rock sites”.

Figure 23a presents the PGA values observed from different video camera footages throughout the perception area, complementing the maps of DYFI EMS [117] in terms of the PGA, and Figure 23b compares the strong motion data(SM) recorded in Türkiye.

Interestingly, the distribution of video camera points follows the fault line (blue line in (a), black in (b)). Video cameras show good geographical coverage with more observations towards the south but with much less points. The SM data are concentrated in the fault line with very similar results. Further research may add more data, especially near the fault, which is the zone of higher acceleration. Unfortunately, the sample is not large enough to allow interpretations such as anomalies in the PGA in relation to fault distances. However, it is clear that future events providing this information will afford quantitative data beyond simple qualitative observations.

The large number of video recordings collected by the authors of this paper, combined with an additional set of 30 recordings obtained from EMSC Media videos [1] and a substantial selection of others sourced from the internet and stored in-house, form a solid foundation for an initial video footage database.

• PART III—Future Initiatives

In Part III, we briefly summarize the main achievements made with this review paper, emphasizing the innovations brought up, the creation of a collaborative platform to “storage” data, and suggestions and recommendations for future actions.

5. Final Considerations and Future Research

5.1. Final Considerations and Innovations

Compared with previous work, this paper makes the following innovative contributions:

We recognize that “new technologies” are not merely future prospects but also vital tools, as evidenced by their crucial role in addressing the challenges posed by recent earthquakes. In future seismic events, these technologies can be leveraged extensively to provide significant advantages for both the population and the advancement of scientific understanding.

Satellite imagery, which is traditionally focused on large-scale views, can now be employed to examine details at the human scale and disaggregate the damage information into various levels (D0 to D5).

Drone imagery has emerged as a primary tool for damage assessment at the local scale. These images play a critical role in preserving information for future reference, particularly in cases of demolition. We demonstrate how such imagery enables the retrieval of otherwise lost information.

Finally, video cameras, whether stationed at fixed points or operated by citizens, can offer valuable insights into the passage of seismic waves and the behaviour of structures in both near-field and far-field locations. We curated a selection of videos to illustrate these methodologies, anchored in analytical and experimental studies.

As far as the use of video surveillance cameras for monitoring various spaces, including urban settings and building interiors, the amount of available information on the internet or gathered by earthquake agencies has significantly increased in recent years. This trend is evident, as observed in the aftermath of the recent Türkiye–Syria earthquakes.

In brief, this is the main contribution of the paper, introducing clear innovation and highlighting the importance of video camera images. Although these images were already used in the late 20th century, they have only been utilized in a few cases since then.

In the case of these earthquakes for future events requiring urgent demolition, we recommend that immediately after the event and before demolition commences, an agency should be tasked with extensively documenting the use of drones in all buildings slated for demolition. This documentation would create a record for future studies to understand the events, akin to forensic studies. We can ascertain that such an effort has not been systematically pursued, as the Portuguese Emergency Mission to Antakya [85] produced several drone films that remain largely unknown. At the time of this publication, Ersoz et al. [122] announced the production of a “SiteEye Disaster Plugin” with 28,000 images and videos compiled specifically to facilitate the study of disaster responses, as mentioned above.

The main results of the quantitative analyses suggest a new proposal for defining the distances occupied by debris obstructing roads. These distances are significantly larger in some situations than those indicated by previous studies, highlighting the need for updated models. Additionally, the analysis of video camera footage offers complementary insights into ground motion characteristics, especially in areas where the built park was not destroyed, as well as for collapsed structures. Numerous examples are cited as good cases for linear and nonlinear analyses of destroyed structures.

5.2. Creation of a Collaborative Platform for Satellite, Drone and Video Camera Footage

To consider the demand and potential of aerial imagery and videos to be used as tools for research, creating a repository of information, such as a data bank for a video camera with classified and organized information that other experts can use, is vital. The example of EMSC “Medias” is a good starting point for storing video camera footage [1]. It is imperative to establish data banks (or collaborative platforms) for storing all this material, thus contributing significantly to the future memory and understanding of such disasters, as was performed in the 1980s with photos from damaged cases [123].

A taxonomy for video camera topics can be performed according to several criteria: following the macroseismic scales EMS-98 [77] and ESI-2007 [106], by topics such as buildings, bridges, dams, and geotechnical issues, or by damage state, etc.

An exemplary approach to organizing an imagery database is illustrated in [97] who devised an Oriented Imagery Catalogue (OIC) to effectively manage all available data pertaining to the 2011 Tohoku tsunami imagery.

The rapid advancement of image analysis, which results in large volumes of data, underscores the need for a platform such as Google Earth^®. Such a platform can enable users to collaboratively access and utilize imagery materials. We advocate for the development of such a platform to cater to the needs of both current and future generations interested in understanding seismic wave propagation and its impact on the urban environment.

5.3. Suggestions for Future Study

As short-term suggestions for future work, we can point to the following:

-

As there are no dense SM arrays for considering the intra- and inter-variability of ground motion (GM), further studies should be conducted based on building behaviour. Both drone images and/or video camera footage are essential for these types of studies. The randomness observed in Figure 19a,b should be understood. Additionally, the influence of transverse-longitudinal predominant motion is crucial in determining the predominant orientation of ground motion.

-

The attenuation of ground motion is very much frequency- and azimuth-dependent, as suggested in Figure 21 and Figure 22b. Much work should be done to clarify this issue.

-

Road blockages have already been observed. A more quantitative analysis can be performed to provide robust evidence for practical applications such as amount of debris, using drones and satellite images.

-

Among the various cases we suggested for analytical studies, we focused on the movement of minarets, as observed in Video #07c. The detected frequency of 0.2 Hz may not correspond to the first mode but is likely related to the ground motion displacement orbit. Further study is needed to clarify this issue.

For the long term:

Future innovations should equip video cameras with universal time and coordinate systems provided by GPS, incorporating altimetry, orientation, and accelerometric monitoring. AI techniques should be employed to prepare drone paths according to pre-established plans, thereby avoiding situations with poor visibility. Additionally, satellite imagery with higher accuracy for both vertical and horizontal components should be utilized to enhance the detection of various degrees of damage.