1 Introduction

On 26 May 2021, a Mw 5.1 hit the densely populated cities of Gisenyi and Goma. It was one of the highest magnitude earthquakes to hit the area since the 18th century. As reported by the Rwanda National Seismic Network, operated by the Rwanda Mines, Petroleum and Gas Board (RMB), the earthquake occurred at 05:46 am, Kigali time. Several aftershocks were recorded after the Gisenyi earthquake which were concentrated along the eastern boundary of the Kivu rift. The rift is bounded by eastern and western basin-bounding normal faults associated with the rift system (Fig. 1a-b). The Gisenyi earthquake occurred after the Nyiragongo volcanic eruption on 22nd May 2021 (Boudoire et al. 2022). According to the government of Rwanda, this Gisenyi earthquake caused widespread damage in the DR Congo and Rwanda. Around, 1,500 houses were destroyed by the earthquake in the Rubavu District.

Fig. 1
figure 1

a ALOS DEM map showing the distribution of earthquakes associated with the Nyiragongo volcanic eruption on 22 May 2021 and the major structural features around the study area (12.5 m ALOS DEM data were taken from https://search.earthdata.nasa.gov/search), b Elevation profiles across the major faults around the study area. The white solid line indicates the location of the elevation profile shown in Fig. 1b. The blue-dotted box marks the location of Fig. 4, and the red-dotted box denotes the location of Figs. 5 and 7

Full size image

Macroseismic intensity refers to the classification of the severity of ground shaking based on the observed effects during an earthquake (Dengler and McPherson 1993; Serva 1994; Esposito et al. 1997; Grünthal et al. 1998; Michetti et al. 2004; Cua et al. 2010). Macroseismic intensity has been used as an index to describe the effects of damaging earthquakes (Allen et al. 2009; Silva and Horspool 2019; Weixiao et al. 2021; Naik et al. 2020; Naik et al. 2023). Moreover, it has been applied in volcano-tectonic environments like Mt. Etna, Italy to assess surface faulting (Azzaro et al. 1998, 1999, 2000; Ferreli et al. 2002; Tringali et al. 2023). Additionally, it has been utilized for seismic zonation in Ischia Island, Italy (Vassallo et al. 2021). This approach is more suitable for seismic hazard management, with little or no strong motion data. Since macroseismic intensity reflects the degree of damage induced by an earthquake, it is used for rapid loss modeling, post-disaster reconnaissance surveys, and relief support for the affected regions (Grünthal 1998; Porfido et al. 2007; Earle et al. 2009; Kamer et al. 2009; Musson et al. 2010; Trendafiloski et al. 2011; C.S.LL.PP. 2018; Michelini et al. 2019; Gomez-Capera et al. 2020; Naik et al. 2024). Despite the widespread acceptance of the use of site-specific ground motion intensity measures (IMs) to characterize earthquakes in the case of ground response analysis and design of lifeline structures (C.S.LL.PP. 2018), macroseismic intensity is still widely used in those areas where the seismic monitoring network is sparse or even absent. The macroseismic intensity and resulting ShakeMap derived from it can be used for a rapid loss assessment and a vulnerability analysis of existing structures in an area of interest (Wald et al. 1999; Giovinazzi and Lagomarsino 2004; Michelini et al. 2019). Additionally, macroseismic intensity can be used to determine the earthquake parameters such as event magnitude, peak ground accelerations, (PGA), peak ground velocity (PGV), and peak ground displacements (PGD) based on their empirical relationships (Sibol et al. 1987; Giovinazzi and Lagomarsino 2004; Faenza and Michelini 2010; Azzaro et al. 2011; Nappi et al. 2018; Zanini et al. 2019; Gomez-Capera et al. 2020; Naik et al. 2023b, 2024).

Several intensity scales are utilized to qualitatively evaluate the impact and effects of earthquakes on individuals, the built environment, and natural landscapes. These include the Modified Mercalli Intensity (MMI) scale, the Japan Meteorological Agency (JMA) Seismic Intensity Scale, the China Seismic Intensity Scale (CSIS), the Environmental Seismic Intensity (ESI 2007) scale for assessing primary and secondary environmental effects (Michetti et al. 2007), and the European Macroseismic Scale (EMS-98) (Wood and Neumann 1931; Grünthal, 1998; Supino et al. 2019; Naik et al. 2023a, b). Among these, the EMS-98 scale has recently gained widespread international adoption (Silveira et al. 2003; Galea 2007; Tertulliani et al. 2018; Buforn and Udías 2022; Sarabia Gómez et al. 2022; Martin et al. 2022; Triantafyllou et al. 2022; Tertulliani and Graziani 2022; Del Mese et al. 2023) due to its potential for upgrades and the inclusion of new building types, particularly those designed to be earthquake-resistant, which were not considered during the development of other traditional scales (Del Mese et al. 2023). The EMS-98 scale was designed not only with seismologists in mind but also civil engineers and other potential users (Del Mese et al. 2023). For our current study, we applied the EMS-98 intensity scale to the most damaging earthquake in recent Rwandan history, considering the types of buildings that closely match the macroseismic criteria described in the EMS-98 scale. Although the earthquake did not have a high magnitude, it produced considerable damage as a result of having a very shallow hypocenter. This region is highly populated and is located along the western flank of the East African Rift System, which is a seismically active branch of the African Rift System. Therefore, we systematically surveyed the building damage pattern to understand the macroseismic intensity distribution as well as other seismic parameters, that contributed to the assessment of seismic hazard.

In this study, we have constructed a macroseismic intensity map based on the damage survey collected in the field and on information derived from standard questionnaires in the macroseismic field survey. We have assigned EMS-98 values for 300 locations, according to the classic definitions, in part to facilitate comparisons between the present earthquake damage pattern observations in the field and those determined for other earthquakes. The intensity pattern observed for the 2021 Gisenyi earthquake highlights the importance of heterogeneous building types and variations of local geology for resultant structural damage.

2 Seismotectonic setting of the area

Seismicity in the East African Rift System (EARS) is mainly controlled by two factors: (1) extension associated with continental rifting, and (2) volcanism. Previous studies (Yang and Chen 2010) revealed that the western branch of the EARS is more seismically active, while the eastern branch is more volcanically active (Fig. 2a). Gisenyi, northwestern Rwanda, lies along the Kivu Rift, which is part of the western branch of the East African Rift System (Fig. 2a-b). The Kivu rift is an extensional area with an extension rate of ̴ 2.3–2.8 mm/yr (Stamps et al. 2008; Saria et al. 2014; Geirsson et al. 2017).

Fig. 2
figure 2

Map showing a the distribution of major faults along the East African Rift System (EARS), b simplified geological map of the western part of the EARS focusing on Rwanda and DR Congo (modified after Delvaux et al. 2017)

Full size image

The Kivu Rift has been recognized as seismically active since the first half of the 20th century (Cornet 1910; Passau 1911, 1912; Krenkel 1922; Sieberg 1932; Figs. 1 and 2). Earthquakes above magnitude 5 are rare in the Kivu basin (Barth et al. 2007; Mavonga 2007; Mavonga and Durrheim 2009; Delvaux and Barth 2010). According to the USGS catalog (1973-present) (Fig. 2b), the 2021 Mw 5.3 Gisenyi earthquake was one of the largest earthquakes in the region in addition to the 2002 Mw 6.2 Kalehe earthquake (Wauthier et al. 2015), the Mw 5.9 Bukavu earthquake (d’Oreye et al. 2011) and the 2015 Mw 5.8 Lwiro earthquake (Geirsson et al. 2017) which was followed by a 5.5 Mw aftershock. The 2015 Mw 5.8 Lwiro earthquake led to the death of 3 people with many injured. Seismic activity in the Kivu Rift is closely related to volcanic activity. In addition to larger earthquakes, moderate earthquakes were also recorded in 1977 and 2002, which coincided with the only two recorded fissure eruptions of the Nyiragongo volcano. The activities of Nyiragongo and Nyamulagira are probably directly related to the opening of the Western Rift Valley (Kasahara et al. 1991; Wauthier et al. 2015). The 2002 eruption of the Nyiragongo volcano, for example, was associated with regional rifting events (Komorowski et al. 2002/2003; Tedesco et al. 2007a, b). This devastating eruption left more than 120, 000 people homeless (Tedesco et al. 2007a). Moreover, the 1977 eruption of the Nyiragongo volcano occurred four days after a Mw 5.3 event struck the Bukavu area (Hamaguchi et al. 1992). It should be noted that small to moderate earthquakes occur frequently in this region, and many of them are not related to eruptions. This applies both to the Bukavu earthquake itself and to the two M > 4 earthquakes that occurred in October 2008 approximately 50 km north of Goma, DRC (d’Oreye et al. 2011).

3 Methodology of macroseismic survey

Earlier macroseismic intensity scales, such as Medvedev–Sponheuer–Karnik (MSK), and Mercalli-Cancani-Sieberg (MCS), are often ambiguous in defining structural damage levels, grades and vulnerabilities (Musson et al. 2010; Grünthal et al. 1998; Li et al. 2020; Cito et al. 2022; Del Mese et al. 2023). The implementation of the EMS-98 was an important step that substantially raised the quality of traditional intensity assignments (Cito et al. 2022). The EMS-98 has 12 divisions, and each division describes distinct observations or damage patterns due to the earthquake intensity (Table 1). For assigning intensity values higher than V, as per the EMS-98 scale, detailed field reconnaissance observations gathered using strict data collection protocols are required (Contreras et al. 2021). The EMS-98 intensity values help in the characterization of the seismic vulnerability of each building and places them in a vulnerability class range from A–F, where A indicates the most vulnerable structures and F the least. Using the EMS-98 intensity scale, detailed post-earthquake observations can help to assign an intensity level at a particular location considering the percentage of buildings damaged at that location (Grünthal, 1998, Spence and Foulser-Piggott 2014; Abrahamczyk et al. 2017; Cito et al. 2022). It is noted that Rwanda and Congo do not have any specific building codes. The study area possesses heterogeneous building types (Fig. 3) with some buildings made of mud, bricks, or stones without any formal engineering approach. Some buildings are engineered with pillars and beams (Fig. 3). For our current study, we have applied the EMS-98 intensity scale, taking into account the types of buildings that closely match the macroseismic criteria described in the EMS-98 scale (Table 1) which are presented in the subsequent sections.

Fig. 3
figure 3

Main building types found in the Gisenyi and Rubavu areas

Full size image
Table 1 Damage description as per the EMS-98 intensity scale
Full size table

3.1 Macroseismic survey

In this study, we have conducted a detailed field reconnaissance survey around the Rugerero, Gisenyi, and Rubavu Sectors of the Rubavu District, Rwanda, which were highly affected by the earthquake (Fig. 2). The epicentral region of the 26th May 2021 earthquake in Rwanda and DR Congo has no standard criteria for assessing the macroseismic intensity. This is the very first time an attempt has been made to conduct a detailed post-earthquake survey to assess the macroseismic intensity as well as other ground motion parameters leading to the preparation of a seismic intensity map around the Gisenyi and Rubavu sectors of Rwanda. We have examined the effects of the earthquake on the natural and built environments. Due to international border issues, we were unable to conduct a detailed damage survey in the affected areas of DR Congo.

3.1.1 Effect on natural environment

The 2021 Gisenyi earthquake, which was associated with the Mount Nyiragongo eruption, caused several secondary ground effects, including gas emission and the formation of ground cracks. Boudoire et al. (2022) reported numerous ground cracks, which were primarily concentrated around Goma and Gisenyi, may be related to the magmatic processes associated with the 2021 Mount Nyiragongo eruption (Fig. 4). During our post-earthquake field survey, we also mapped several ground cracks, predominantly located along the eastern basin-bounding faults of the Kivu Rift. Most of these ground cracks exhibit vertical displacements of 3–5 cm (Fig. 4b-d) with an N-S/NNE-SSW orientation which may be associated with the movement of the eastern basin boundary fault during the 26th May 2021 earthquake. Due to their limited number, we did not utilize the ground cracks for our macroseismic intensity estimation in this study.

Fig. 4
figure 4

Ground cracks developed during the 26th May 2021 Gisenyi earthquake around Goma and Gisenyi

Full size image

3.1.2 Effects of the 2021 Gisenyi earthquake on built environment

In Gisenyi City, in the western part of Rwanda, the macroseismic data collected to estimate the seismic intensity utilized the traditional methods of field surveys as well as the distribution of questionnaires immediately after the 2021 Gisenyi earthquake (Supplementary Information S1). The evaluation was based on assessing how ground motion affected people, household items, and caused damage to buildings in Gisenyi City. The macroseismic investigation was carried out through field survey questionnaires where the participants were asked to describe personally observed effects. Twenty-three questions were prepared and distributed to residents of the Gisenyi and Rubavu sectors (Supplementary Information S1). Those questions covered three main topics including people’s perception of the strength of the quake, effects on objects (fixed and moveable), and effects on infrastructures.

To determine the macroseismic intensity (Table 1) around the epicentral area, it was mandatory to identify the types of buildings and the effect the earthquakes on them. Considering the responses to the questionnaires and field survey the heterogeneous building types were identified (Fig. 3). They are divided into five categories: (1) Traditional mud huts, (2) mud houses without any brick elements, (3) brick houses with mud reinforcement, (4) brick houses with cement reinforcement, and (5) modern single or multistory brick houses with or without pillars or beams. Almost all the buildings were identified to be in the vulnerability class A or B as per the EMS-98 intensity scale. The post-earthquake reconnaissance survey suggests that several houses experienced wall cracking, wall collapse, and boundary wall collapse or cracks, in addition to complete collapse of brick and mud-brick houses in several places (Fig. 5). Most of the houses in Gisenyi Rubavu City were built with brick with a few wood house. During the earthquake, wooden houses were the least damaged, showing only a little twisting.

Fig. 5
figure 5

Typical damage pattern observed in the buildings or fence walls around Gisenyi and Rubavu a Umuganda Cell, Gisenyi Sector (Latittude-1.68272°S; Longitude-29.26185°E), b Mbugangari Cell, Gisenyi Sector (Latittude-1.67073°S; Longitude-29.26161°E), c Gikombe Cell, Rubavu Sector (Latittude-1.66247°S; Longitude-29.28150°E), d Rukoko Cell, Rubavu Sector (Latittude-1.66570°S; Longitude-29.254583°E), e Amahoro Cell, Gisenyi Sector (Latittude-1.69305°S; Longitude-29.26217°E), f Umuganda Cell, Gisenyi Sector(Latittude-1.68846°S; Longitude-29.26200°E)

Full size image

On the other hand, reinforced concrete buildings presented high levels of damage and large cracks in walls. A detailed analysis to understand the controlling factor for the building damage pattern observed during the earthquake considering the total 300 data sets collected from the field and detailed survey questionnaires was performed (Supplementary Table ST1, Table 2).

Table 2 Distribution of EMS-98 intensity and PGA for the selected sites during the 26th May 2021 Gisenyi (Rwanda) Earthquake (Mw 5.1)
Full size table

The analysis suggests that around 88% of the houses experienced strong shaking, and produced bumping noises. 38% of the houses completely collapsed, around 73% of houses were damaged with 61% of those being heavily damaged and in need of repair before being reoccupied (Fig. 6). During our field survey, people were also asked whether they saw windows, dishes, and other household objects shaking. Approximately 78% of the people responded that they recalled a strong movement of these objects; many objects fell and were broken (Fig. 6). Based on these observations major controlling factors were determined (Fig. 7).

Fig. 6
figure 6

Pie Charts showing the statistical analysis for the 2023 Gisenyi earthquake shaking effects observed around the study area

Full size image
Fig. 7
figure 7

Pie charts showing the major factors controlling the damage observed during the 26th May 2021 Gisenyi earthquake

Full size image

From the statistical analysis, it is observed that 61% of the structural damage and higher reports of shaking were observed at the location situated within the Kivu basin which is composed of volcanic ash/organic-rich soil or basin fill clay-loose sand deposits (Figs. 1b and 7a-b; Ross et al. 2014).

Another crucial factor was observed which had a strong correlation to the amount of damage to buildings was the age of the building (Fig. 7c). Almost 92% of the damaged buildings were constructed several years ago without any earthquake-resistant measures. Considering the shaking and damage pattern observed in the buildings, EMS-98 intensity was evaluated for all the sites. It indicated that the maximum EMS-98 intensity of IX was observed around the Rubavu market and Gisenyi Sector office (Table 2; Supplementary Table ST1; Fig. 8a). Similarly, an EMS-98 intensity of VIII was observed around Gisenyi Hospital, Ecole des Sciences de Gisenyi, Mbugangali and Umuganda Cells, Gisenyi airport, and Rugerero Sector. An EMS-98 intensity of VII was also observed around the University of Tourism, Technology and Business Studies (UTB), as well as the Rugerero Sector. From the intensity distribution map, it was noted that the higher intensity values were clustered along the eastern basin-bounding fault of the Kivu rift (Figs. 1 and 8a; Table 2; Supplementary Table ST1).

Fig. 8
figure 8

Map showing a the EMS-98 Intensity distribution, b Peak Ground Acceleration (PGA) around Gisenyi and Rubavu sectors for the 26th May 2021 Gisenyi earthquake

Full size image

3.2 Peak ground acceleration (PGA) estimation

Peak ground acceleration (PGA) is the maximum ground acceleration that occurred during the earthquake at a location. PGA is the largest amplitude the accelerogram recorded at a site during a particular earthquake by a seismogram (Musson 2000; Douglas 2003; Douglas et al. 2023). PGA is the most common parameter in engineering applications including ground response analysis, liquefaction potential estimation, seismic building codes design, and seismic hazard maps preparation (Shedlock et al. 2000; Jishnu et al. 2013; Nas et al. 2020; Jena et al. 2021; Naik 2022; Sabetta et al. 2023).

Therefore, a significant amount of seismic hazard assessment has been carried out with a focus on PGA (Abrahamson 2006; Wang 2011; Nas et al. 2020; Jena et al. 2021; Sabetta et al. 2023). For areas having a higher density of instrumental facilities, the PGA values are be easily available. However, the PGA values can be estimated from macroseismic intensity for areas with limited or no instrumental facility.

Several studies have been carried out to relate seismic intensity to peak ground acceleration (Gutenberg and Richter 1956; Hershberger 1956; Ambraseys 1974; Murphy and O’Brien 1977; McCann et al. 1980; Krinitzsky and Chang 1988; Gama-Garcia and Gómez‐Bernal, 2008; Worden et al. 2012; Lesueur et al. 2013; Bilal and Askan 2014; Locati et al. 2017; Du et al. 2019) due to its importance in seismic resistant design (understanding the response of the structures) for future disaster management planning (Bilal and Askan 2014; Du et al. 2019). Nevertheless, most of the correlations are focused on the MM intensity or MCS intensity without any studies focusing on EMS-98 intensity. The relationships between the PGA and intensity are region-specific. Only one or two studies have used EMS-98 intensity to estimate the PGA (Zanini et al. 2019). Since damage pattern and damage level rely on the local scale geological conditions and building vulnerability, local scale studies are required in order to be useful for enhancing the empirical relationship on a regional scale. This can be done by adding more data as well as conducting a local scale seismic hazard analysis. The EMS-98 intensity scale was updated to include earthquake-resistant buildings and to include the most recent traditional intensity scale (Cito et al. 2022; Del Mese et al. 2023). Since there is no such relationship available for the African region, we have utilized an empirical relationship between the EMS-98 intensity and PGA (Eq. 1) as provided by Zanini et al. (2019).

IEMS−98= 2.03 + 2.28 ×log PGA (1).

Since there are no recording stations around the epicentral area, this formula was used to estimate the peak ground acceleration (PGA), and a PGA distribution map was prepared (Fig. 8b).

PGA values of 1.16 g and 0.40 g were highly concentrated along the eastern basin-bounding faults of the Kivu rift whereas the PGA values of 0.15 g were highly concentrated around the Kivu basin (Fig. 8b). The higher PGA values of 1.16 g and 0.40 g were observed around Rubavu market, Gisenyi Sector office, Gisenyi Hospital, Ecole des Sciences de Gisenyi, Mbugangali and Umuganda Cells, Gisenyi airport, Rugerero area. The PGA values of 0.15 g were mostly composed of volcanic/organic-rich soil around the University of Tourism, Technology and Business Studies (UTB), Rugerero area. In some places, the EMS-98 intensity and PGA values are highly scattered. This may be due to the amalgamation of heterogeneous building types.

In this study, a ShakeMap was produced using the EMS-98 intensity values for the 2021 Gisenyi earthquake using the Earthworm Software Module (Fig. 9). The working principle of the ShakeMap program is based on receiving intensity values from the input file and saving the value of each location. Therefore, from the stored values, the software performs the spatial interpolation calculation using inverse distance weighting to generate a reasonable ShakeMap (Mittal et al. 2019; Wu et al. 2019). This methodology has been applied in many studies (e.g. Legendre et al. 2017; Mittal et al. 2018; Yang et al. 2018, 2021; Naik et al. 2023a; b; Naik et al. 2024). The Gisenyi earthquake ShakeMap shows a maximum intensity of IX around Rubavu with other affected areas showing an intensity of VIII (Fig. 9). The intensity map is the first ShakeMap made available for the 26th May 2021 Mw 5.1 earthquake. There is no other such map available for the earthquake at the moment. The ShakeMap does not depict the entirety of the damaged area io due to the non-availability of the data from the DR Congo part. However, the ShakeMap presented here (Fig. 9) can be used for to improve seismic hazard estimation, land use, and land cover planning for the affected area, more specifically for the Goma, Gisyenyi, and Rubavu regions, in the future.

Fig. 9
figure 9

EMS-98 ShakeMap for the 26th May 2021 Gisenyi earthquake prepared from the field-based building damage data

Full size image

4 Discussions

Detailed field reconnaissance surveys were performed around the Rugerero, Gisenyi, and Rubavu Sectors of Rubavu District, Rwanda, which were highly affected by the 26th May 2021 Gisenyi earthquake (Mw 5.1). The Gisenyi earthquake (Mw 5.1) caused extensive shaking and damage around the Gisenyi and Rubavu sectors of Rwanda. Although the earthquake caused shaking and building damage in the DR Congo, due to international border issues we were not able to conduct a field survey in DR Congo.

Around 300 data sets were collected through macrosiemsic survey questionnaires and field surveys, then EMS-98 intensity values were assigned for each location. The EMS-98 values suggest a maximum intensity of IX for the MW, 26th May 2021 Gisenyi earthquake (Fig. 8a). Higher intensity values were observed in urban areas with higher population densities in the Rugerero, Gisenyi, and Rubavu Sectors. We have prepared a seismic intensity distribution map which is the first one created for this earthquake. We estimated PGA values using the EMS-98 intensity and also prepared a PGA distribution map (Fig. 8b) which can be used for future seismic hazard assessment and earthquake-resistant building design. Despite the moderate magnitude, the earthquake produced higher seismic intensity effects due to the lack of earthquake resistant design and nonexistent building regulations in Rwanda. This led to the construction of masonry buildings in the city of Gisenyi, which are unstable and very susceptible to earthquakes (Figs. 3 and 5).

In addition to this, several other factors such as the age of the buildings, non-ductile detailing of structural components, strong-beam weak-column conditions, captive-column and short-column effects, soft and weak stories mechanism, irregularity in plan and elevation, local geology, unconfined infill walls, bad workmanship, low quality of construction materials, superficial detailing of building elements, and the lack of engineering services contributed to the severity of the damage (Ademović et al. 2020; Alih and Vafaei 2019; Günaydin et al. 2021). Most of the damaged buildings were located along the volcanic/organic-rich soil or soil having a higher percentage of sand and silt. This might indicate site amplification (Joshi et al. 2023) and suggests a proper geotechnical characterization is required of the soil to understand its seismic response for a better seismic hazard assessment of the area.

The macroseismic intensity values we reported are similar to those observed in several earthquakes of moderate magnitude in various seismotectonic contexts. In volcanic regions, these intensity values are attributed to factors such as the shallow depth of the earthquake’s hypocenter, the superficial nature of the seismogenic source, the inferior quality of surface deposits, and inadequate construction practices, as highlighted by Nappi et al. (2021) and Tringali et al. (2023). Comparable intensity values have also been recorded for several moderate-magnitude earthquakes worldwide. In active tectonic areas, such as Romania which was hit by the 2004 Vracea earthquake (Mw 6.0) (Constantin et al. 2016), India (Haryana Delhi border) which was hit by a 2012 earthquake (Mw 4.9) (Gupta et al. 2013), Albania which was hit by a 2019 Durres earthquake (Mw 5.6) (Rrezart and Rrapo, 2020), Grecia (West Athens) which was hit by a 2019 earthquake (Mw 5.1) (Kouskouna et al. 2021), Italy which was hit by the 2012 Emilia Romagna seismic sequence, (Mw 5.9) (e.g. Tertulliani et al. 2012), the 2017 Casamicciola earthquake (MW 3.9) (Nappi et al. 2021) and the 2018 Fleri earthquake (Mw 4.9) (Tringali et al. 2023).

During the field survey and macroseismic reconnaissance survey, it was observed that most of the wooden buildings resisted the earthquake. Only minor cracks were observed in the wall (Figs. 5 and 10). Although the wooden structures in Rwanda are not properly designed as per the seismic building code, they had the flexible properties of wooden materials, high strength-to-weight ratio, and lightweight the structure, which allow them to withstand the seismic waves passing through them either without breaking or just the development of minor cracks or damage (Kaushik et al. 2006; Buchanan et al. 2011a, b; Alih and Vafaei 2019). The better performance of wooden buildings was also observed during several moderate magnitude earthquakes such as the 2006 Sikkim earthquake in India (Mw 5.3; Kaushik et al. 2006); the 2010 Elazığ-Kovancılar earthquake in Turkey (MW 6.1; Calayır et al. 2012), the 2011 Christchurch earthquake in New Zealand (Mw 6.3; Buchanan et al. 2011a), and the 2015 Mw 6.0 Sabah earthquake in Malaysia (Mw 6.0; Alih and Vafaei 2019). Considering the shaking intensity, building damage pattern observed in different types of buildings, and the macroseismic intensity calculated from our analysis (Table 2; Supplementary Table ST1), it is recommended that seismic design codes be implemented to prevent the reoccurrence of such damage in the future.

Fig. 10
figure 10

Field photo graphs showing (a) extensional graben with 5 cm vertical displacement along the Eastern Basin Boundary fault, (b) extensional cracks along the boundary wall and its adjacent ground with 3 cm vertical displacement, (c) extensional ground cracks observed on road with 3 cm vertical displacement, (d) extensional cracks and opening in the pedestrians, (e) and (f) Stereo plot showing the orientation of the extensional cracks observed during the earthquakes along the Eastern Basin Boundary fault, Kivu Rift

Full size image

The ShakeMap for the 2021 Gisenyi earthquake (Fig. 9) shows a concentration of higher intensity (IX) around the probable surface rupture and lower intensity (EMS-98-VIII) in other areas. A Higher intensity might also be predicted close to the epicenter (Naik et al. 2023b), however unlike the basin, the affected area is hilly terrain and consists of Meso-Proterozoic rocks. This is indicative of the effect that local geology has on shaking intensity and the macroseismic intensity calculated from our analysis (Table 2, Supplementary Table ST1).

The ShakeMap prepared from the present study can be used for better seismic hazard estimation of Gisenyi City. This is the first-time damaged-based field data were used for the ShakeMap generation of the affected area. The field data-based ShakeMap represents a more accurate hazard scenario than a ShakeMap prepared from the instrumental records for the areas due to having limited or no near source seismic monitoring stations (Musson 2000; Silva et al. 2017; Cito et al. 2022; Naik et al. 2023a, b; Bhochhibhoya and Maharjan 2022; Trevlopoulos et al. 2023).

In addition, the clustering of higher intensity values, higher PGA values, and higher grades of damaged buildings along the eastern boundary of the Kivu basin might be related to the reactivation of the eastern basin-bounding fault. Furthermore, it could be associated with the acceleration of ground motion close to the fault, source directivity, and site effects (Figs. 1 and 10a-b; Mollaioli et al. 2006; Pacor et al. 2018). The reactivation could be associated with the diking process or regional extension. Although a more detailed structural analysis is required for testing these two hypothetical causes, our field survey results suggested it was the reactivation of the eastern boundary fault based on the structural parameters collected from the earthquake-induced fracture pattern observed along the eastern boundary fault (Figs. 1a-b and 10). Most of the extensional fractures show a vertical displacement of 3–5 cm (Figs. 4b-d and 10a-d), and they predominantly indicate N-S/NNE-SSW strikes similar to the trend along the eastern basin-bounding fault (Fig. 10e-f).

Similar phenomena of higher seismic intensity due to the fault reactivation were observed during the 1999 Chi-Chi earthquake (Xie 2019), the 2003 Bingöl earthquake in Turkey (Akkar et al. 2005), the 2008 Wenchuan earthquake (Xie et al. 2010), the 2011 Tohoku earthquake in Japan (Pavlenko 2022), 2013 Cook Strait earthquake in New Zealand (Holden et al. 2013), and the 2011 Christchurch earthquake in New Zealand (Bradley 2016).

The present study can be utilized as an initial step toward a better understanding of the seismic hazard potential of the area. The presented EMS-98 intensity and PGA distribution map is vital for better seismic hazard estimation, land use, and land cover planning for the affected area in the future.

5 Conclusions

EMS-98 intensity and PGA values were estimated for the 26th May 2021 Gisenyi earthquake (Mw 5.1). We determined a maximum intensity Imax = IX to the 2021 Gisenyi earthquake. From the seismic intensity survey, it can be inferred that the buildings located in volcanic soil/organic-rich soil are prone to higher seismic damage. In addition, despite of similar construction method material, older buildings suffered a higher degree of damage. Therefore, there is an urgent requirement for the implementation of a seismic design code in the construction of both modern as well as traditional buildings in Rwanda.

The higher intensity and PGA values were concentrated along N-S/NNE-SSW, which is a similar trend to the eastern boundary fault of the Kivu rift, indicating the direction of rupture propagation.

The prepared EMS-98 and PGA map can be applicable for future seismic hazard assessment and implementation of building codes in the affected region.

Finally, this research concludes that analyses of earthquake-induced damage data can help to identify hazard scenarios, shaking intensity and develop knowledge that is useful to formulate new disaster risk reduction policies in Rwanda which does not have any previous seismic hazard zonation or risk studies.