Exploring traditional flood-resilience strategies at historical shrine spaces: evidence from high flood-exposed potential areas in Nagoya

1 Introduction

In recent years, “socio-hydrology,” which emphasizes interactions between human activities and the water cycle, has been proposed (Sivapalan et al., 2012). Research that links social determinants of exposure and vulnerability, such as strategies for maintaining resilience to flood events with hydrological processes, and elucidates the feedback loops between them, has attracted significant attention (Di Baldassarre et al., 2013; Kittipongvises et al., 2020; Xia et al., 2022; Pande and Sivapalan, 2017).

In the existing literature, socio-hydrological frameworks clearly capture the interactions between humans and floods at the urban scale. For example, Liao (2012) examined how technical defenses, such as levees and breakwaters, generate the “levees-effect.” Ciullo et al. (2017) analyzed how social systems respond to evolving flood threats through technological change. Furthermore, the study by Di Baldassarre et al. (2015), which focused on urbanized areas, suggested that long-term reliance on modern engineering defenses may lead to a decline in communities' social adaptive capacity and an increase in latent vulnerability. However, the existing literature primarily focuses on modern civil engineering structures, such as dykes and breakwaters, in the study areas, resulting in a lack of a comprehensive understanding of the non-structural adaptation strategies that traditional communities have developed over centuries of practice.

A key insight into the long-term interaction between humans and disasters including floods is that historical structures built over many years often embody the physical characteristics of extreme disasters in their respective regions, as well as the disaster-resilience strategies developed to address them. For example, the cluster of temples in the Himalayan valleys is strategically positioned to avoid avalanches (Sain et al., 2023). Similarly, in China, as temples were originally constructed in accordance with principles such as “adapting to the terrain and following watercourses,” they have been situated in relatively safe locations on mountains, thereby enhancing their long-term resilience to natural disasters (Li et al., 2012; Yu et al., 2023; World Heritage Centre, 2026). Dai et al. (2025) investigated the disaster resilience of traditional temple sites in the Longmen Mountain region of Sichuan Province against seismic hazards and found that they tend to be situated in towns, suburbs, and on flat terrain on mountain slopes—that is, in areas far removed from disaster risks. In Japan, Suppasri et al. (2013) documented tsunami breakwaters and “tsunami stones” indicating past tsunami inundation in coastal areas, noting that these remains serve as records of historical extreme events and as embodied records of ancestral disaster awareness. In the 2011 Great East Japan Earthquake and Tsunami, many shrines that had stood in the area for a long time escaped tsunami damage (Takada and Kuwako, 2016). Additionally, through an examination of the locational characteristics of shrines in areas frequently affected by flooding, Hattori and Nii (2012) noted that shrines may be situated in positions that protect settlements from flooding. Tone et al. (2005) used landscape model parameters to quantitatively analyze the distribution and spatial changes in the role that Japanese temples have played as places of refuge since the Edo period. In this way, historical faith-based facilities such as temples and shrines have played diverse roles within traditional local communities, serving as venues for rituals, spaces for social interaction, and places of refuge in times of risk. Takada and Kuwako (2016) propose the hypothesis that “shrines that play an important role in the faith of a particular locality possess locational characteristics that can ensure safety even when natural disasters occur.” In line with this hypothesis, faith-based facilities such as temples and shrines, which have endured over the long term, can be regarded as specific socio-hydrological markers.

Furthermore, historical and cultural heritage is vulnerable to natural disasters (United Nations Educational, Scientific, and Cultural Organization: https://whc.unesco.org/en/disaster-risk-reduction/). From the perspective of the conservation of such heritage, numerous studies have been conducted on specific temples and shrines, focusing on the impact of climate change on natural disaster risks and on disaster risk reduction through investment in disaster control facilities (e.g., Bajracharya, 2025; Ravan et al., 2023; Dimabayao et al., 2025). In Japan, analyses of actual disaster damage at specific temples and shrines (e.g., Takada and Kuwako, 2016) and studies on the spatial relationships between sacred structures, such as temples and shrines, and hazard maps as indicators of natural disaster risk (e.g., Fujita and Kumagai, 2007) have been conducted. However, most previous studies have been limited to analyzing the safety of specific temples and shrines under current or future climate conditions, and have not focused on the fact that the current safety of temples and shrines with a history of several 100 years or more is based on the natural disaster risks at the time of their construction and subsequent disaster experiences. In particular, it is inferred that the construction of temples and shrines before the advent of modern technology was empirically designed to maximize returns on investment and reduce disaster risks, based on the natural disaster risks and experiences of the time. For example, Taniguchi and Lee (2020) inferred from a case study of the restoration work (1789–1796) of the Phra Chetuphon Vimolmangklararm Temple in Bangkok, Thailand, one of three respected temples, that a 1.5-meter elevation was implemented to protect the temple from flooding. However, in Japan, the hydraulic forces of flooding in fast-flowing rivers are significant, posing a high risk that wooden buildings, such as historical faith-based facilities, will be easily swept away. Furthermore, as the flood patterns of rivers with unregulated channels—dating from before the advent of modern technology—vary from one event to the next, it has been difficult to quantitatively assess the safety of Japanese historical faith-based facilities spaces, which have a history spanning several centuries, based solely on elevation data or flood analysis.

To regard such historical faith-based facilities as socio-hydrological markers, it is first necessary to verify the hypothesis put forward by Takada and Kuwako (2016). However, quantitative studies examining the relationship between historical faith-based facilities and flood risk at the time of their construction remain limited in the existing literature. The authors have been conducting flood and socio-hydrological analyses (Komori et al., 2022) focusing on Nagoya City, Japan, which has vast low-lying areas. Komori et al. (2022) demonstrated that in the estuary region of Nagoya, where the river is wide and hydraulic forces are relatively low, flood exposure is high; in other words, in areas with lower elevations than the surrounding terrain, flood inundation patterns remain consistent across events. Consequently, this study focuses on shrines situated in areas with high flood-exposure potential in Nagoya, a city with numerous historically long-standing shrines. It aims to clarify the safety of shrine spaces established in such areas over the course of several centuries. Furthermore, this study specifically seeks to quantitatively identify the shrine characteristics that ensure the safety of shrine spaces in the face of flood-exposure potential, whilst also examining the local communities' perceptions of flood risk at the time of the shrines' construction. As the actual state of local communities' awareness and interest is crucial to promoting disaster prevention and mitigation measures (D'Souza et al., 2021), this study is considered cutting-edge research within the socio-hydrological processes of disaster prevention and mitigation science.

2 Materials and methodology

2.1 Key parameters relating to the interaction between humans and floods

To investigate the natural disaster risks at the time of the shrines' establishment and the characteristics of the shrines' spatial resilience in the study area, this study employed the following two flood-related parameters.

2.1.1 Flood-exposure potential

In this study, the Rainfall-Runoff-Inundation (RRI) model (Sayama et al., 2012) was used to simulate flooding under extreme rainfall conditions in a natural state without flood control facilities such as embankments, and to calculate the potential inundation depth within the shrine space.

2.1.2 Safety of shrine spaces against flood-exposure potential

This study treated the height design of the main hall, which maintains functional integrity by preventing water ingress into this key faith-based facility during flooding, as an indicator of shrine space safety. This adaptive architectural design is not merely a physical parameter but an embodiment of local disaster consciousness, providing empirical evidence for evaluating how local communities have perceived and avoided flood risks over hundreds of years.

2.2 Flood-exposure potential estimation using the RRI model

The RRI model is a two-dimensional distributed hydrological model that simulates rainfall runoff, river flow, and floodplain inundation within a single framework. By combining topographic, land-cover, and rainfall information, it represents the exchange between hillslopes, river channels, and inundated floodplains. This structure makes the model suitable for examining inundation patterns in low-lying urban areas. The use of RRI is also supported by work in the Shonai River system, which drains the lowland area around Nagoya. Sayama et al. (2012) showed that RRI can integrate rainfall-runoff and inundation processes in large-scale flood analysis. In addition, the Chubu Regional Development Bureau of the Ministry of Land, Infrastructure, Transport, and Tourism reported that flood forecasting for the Toki-Shonai River system is being shifted from the former PWRI model to the RRI model, because RRI is expected to improve prediction accuracy and will be tested using large flood events (Chubu Regional Development Bureau, Ministry of Land, Infrastructure, Transport and Tourism, 2025; Seemanta and Kawasaki, 2017).

In this study, the model was not used to reproduce a specific historical flood. Instead, it was used to compare the relative flood-exposure potential characteristics of shrine locations under the same hydrological assumptions. The input data included digital elevation data, river network information, land cover data, and rainfall data for four return period scenarios. The digital elevation model represented the topographic gradient and low-lying terrain of Nagoya City, and land cover information was used to assign surface roughness parameters. Watershed boundaries and river channels were delineated from the topographic and river-network datasets. Rainfall inputs were prepared for 10-, 20-, 50-, and 100-year return-period precipitation scenarios. The main simulation scenario assumed a natural-state flood condition without modern flood control facilities, including embankments, detention reservoirs, and intensive river channel modifications. This scenario was adopted because the study focuses on the long-term relationship between historical shrine location, safety of historical shrine space, and historical flood-exposure potential, rather than on present-day flood risk under the engineered river system. Many target shrines were established before modern flood control infrastructure was fully developed. Present-day flood conditions, strongly affected by embankments and drainage systems, therefore, do not necessarily reflect the hydrological environment that shaped historical shrine site selection. This distinction is important in Nagoya City. The local hydrological environment has been extensively altered by modern river engineering, especially after the 1959 Isewan Typhoon, which accelerated flood-control works and altered flood dynamics in and around Nagoya (Nakasu et al., 2011). The natural state simulation was therefore used as a common counterfactual baseline for evaluating the spatial flood susceptibility of shrine locations before the dominant influence of modern flood control infrastructure.

Because this natural-state scenario is counterfactual, direct calibration against event-based water-level or inundation-depth records was not feasible. The results should therefore not be read as exact historical flood depths. In this study, the RRI outputs were used to extract shrine locations in areas of high flood-exposure potential and to compare relative flood-exposure potential among shrine sites under model simulations.

2.3 Target shrines selection

To construct the initial sampling frame, we first compiled a list of shrine candidates in Nagoya City using the Nagoya Shrine Guide. Because this source includes a wide range of shrine-related entries, including small shrines, halls, abolished shrines, and subordinate shrines located within other shrine precincts, these entries were excluded from the initial dataset. We retained only independent shrine sites with identifiable worship spaces and shrine precincts. As shown in Figure 1, this screening resulted in 129 shrine sites in Nagoya City. These 129 shrines served as the candidate pool for subsequent target-shrine selection.

The target shrine selection was conducted from two complementary perspectives: flood-exposure potential and historical continuity. The first screening procedure was a potential flood-exposure-first approach, in which shrines located in areas of high and recurrent flood-exposure potential were first identified using the RRI-simulated inundation results, and their historical continuity was then examined. The second screening procedure was a historical-continuity-first cross-check, in which shrines with sufficient historical continuity were first identified from historical records, and their flood-exposure potential was then evaluated. The purpose of using these two complementary procedures was to confirm that the final target shrines were the most suitable for the research.

2.4 Field survey and definition of physical parameters on safety of shrine spaces

We conducted a field survey from 1 to 4 December 2024, measuring physical parameters at target shrines. As shown in Figure 2, the three vertical physical parameters (H1, H2, and H3) defined in this study are grounded in distinct adaptation strategies corresponding to the spatial structure and functional roles of the shrines:

• H1 (Overall spatial safety): the floor height of the main hall relative to the surrounding terrain (H2 + H3). This is considered an indicator of the shrine space's flood safety.
• H2 (Structural defense/functional adaptation): the floor height by which the main hall has been lifted from the shrine precinct. This is regarded as a targeted, cost-effective resilience strategy that prioritizes the core faith-based facility (the main hall) without altering the broader landform, reflecting localized protection.
• H3 (Site-wide defense/environmental adaptation): the elevation of the shrine precinct relative to the surrounding terrain. This parameter is regarded as a holistic socio-spatial strategy designed to protect not only the main hall but the entire site to preserve community assembly grounds during flood events.

2.5 Analysis of the H3 contribution to the safety of shrine spaces against flood-exposure potential

The core methodological element of this study is to explore traditional flood-resilience strategies by assessing the extent to which H3 contributes to H1 in shrines constructed in areas with high flood-exposure potential. The rationale for this analysis lies in the ability to clearly identify whether, when faced with a specific flood level, builders tended to raise the entire foundation to protect the entire shrine site (H3) comprehensively, or whether they tended to focus on the primary defense of the shrine's main hall, lifting only the significant faith-based facilities (H2). This will enable us to quantitatively assess the relative importance of H3 as a traditional flood-resilience strategy within the shrine space and to clarify whether this relative importance differed across shrines.

3 Results and discussion

3.1 Abstracting historical shrines in areas with high flood-exposure potential

3.1.1 Flood-exposure-first screening

We first overlaid the locations of the 129 shrine candidates with the RRI-simulated inundation areas under the four return-period scenarios. As shown in Figure 3, the number of shrines located within the simulated inundation area was 45 under the 100-year return-period precipitation scenario, 35 under the 50-year return-period precipitation scenario, 33 under the 20-year return-period precipitation scenario, and 17 under the 10-year return-period precipitation scenario. The decrease from the 100-year to the 20-year return-period precipitation scenarios was relatively gradual, particularly between the 50-year and 20-year return-period precipitation scenarios, where the number decreased only from 35 to 33. In contrast, when the return period was changed from 20 to 10 years, the number of inundated shrines decreased sharply from 33 to 17. This marked reduction suggests that many shrines included in the 20-, 50-, and 100-year return-period precipitation scenarios were located in areas inundated only under relatively larger flood events. In contrast, the 17 shrines inundated even under the 10-year return-period precipitation scenario were located in areas with more frequent flood risk. Therefore, these 17 shrines were considered the core group in areas of high and recurrent flood-exposure potential. This threshold was adopted because the study focused on analyzing the safety of shrine spaces that have historically existed in areas where flood-exposure potential was not merely an exceptional hazard but a recurrent environmental condition.

We then examined the historical records of these 17 shrines. As shown in Table 1, six shrines were excluded from the final sample. Three shrines were excluded because their construction histories were unclear, disputed, or unavailable: Rokusho Shrine, Oi Shrine, and Oaza-jinja. Another three shrines were excluded because they were substantially destroyed during World War II and subsequently rebuilt: Enoki Hakusan Shrine, Hachioji Kasuga Shrine, and Nishifuruwatari Shrine. These shrines could not be used to evaluate historical architectural and topographical flood-resilience characteristics because their current structures do not necessarily preserve the spatial or architectural conditions that existed before modern reconstruction.

Table 1. Historical screening of shrines located in areas of high and recurrent flood exposure.

This table summarizes the historical information used to assess whether the 17 shrines within the 10-year simulated inundation area retained sufficient historical continuity to evaluate long-term flood-resilience characteristics. Shrines with unclear, disputed, or unavailable construction histories, as well as shrines substantially destroyed during World War II and subsequently rebuilt, were excluded from the final sample (Shaded items).

Consequently, the remaining 11 shrines were selected as the target shrines of this study (Figure 4). These shrines satisfied both criteria: they are located in areas of high and recurrent flood-exposure potential, and they retain sufficient historical continuity to allow the evaluation of traditional flood-resilience strategies through shrine location, landform characteristics, and physical parameters related to the safety of shrine space.

3.1.2 Historical-continuity-first cross-check

To examine the robustness of the target selection, we also conducted a historical-continuity-first cross-check. In this procedure, the 129 shrine candidates were first screened based on historical records. Shrines were retained if the Nagoya Shrine Guide provided identifiable information on foundation, reconstruction, renovation, relocation, or major architectural modification, and if their foundation or major architectural/spatial modification history could be traced to more than 100 years before the present study. Shrines were excluded if their historical descriptions indicated unclear construction dates or large-scale post-1926 destruction or reconstruction that disrupted historical continuity. However, reconstruction, renovation, or relocation that occurred more than 100 years ago was not treated as an exclusion criterion, because such cases can still represent shrine spaces with long-term historical continuity.

This historical-continuity-first screening identified 26 shrines with sufficient historical records and historical continuity. We then overlaid these 26 historically continuous shrines with the RRI-simulated inundation areas under the same four return-period scenarios. As shown in Table 2, 15 of the 26 shrines were located within the inundation area under the 100-year return-period precipitation scenario, 13 under the 50-year return-period precipitation scenario, 12 under the 20-year return-period precipitation scenario, and 11 under the 10-year return-period precipitation scenario. The 11 shrines identified under the 10-year return-period precipitation scenario corresponded to the same 11 final target shrines selected by the potential flood-exposure-first procedure.

Table 2. Number of historically continuous shrine candidates located within simulated flood-inundation areas under four return-period scenarios.

This table summarizes the potential flood-exposure screening of the 26 shrines identified through the historical-continuity-first procedure. These 26 shrines had identifiable historical records and sufficient historical continuity, defined as foundation, reconstruction, renovation, relocation, or major architectural/spatial modification more than 100 years before the present study, without records of large-scale post-1926 destruction or reconstruction.

The analyzable historic shrines accounted for only 20% of all shrines, and 42% of these were located in the inundation area under the 10-year return-period precipitation scenario. Given that adopting a 20-year return-period precipitation scenario would increase the sample size by only one shrine (4%), the final 11 shrines should be understood as a purposive sample of information-rich shrines that best fit the purpose of this study: examining long-term socio-hydrological interactions embodied in shrines located in areas of high flood-exposure potential. All of these shrines are located in areas with high flood-exposure potential, and the safety of these shrine spaces against flood-exposure potential provides valuable empirical evidence for studying the interaction between humans and floods. It should be noted that a report indicates that only Narumi Hachiman Shrine suffered flood damage from the Ise Bay Typhoon (1959), and that a record exists of the main hall being lifted on stilts in 1968. There were no records of flood damage at any of the other target shrines.

3.2 Flood safety at the historical shrine spaces in the high flood-exposure potential

3.2.1 Assessment of flood safety at the historical shrine spaces

Figure 5 shows the comparison of H1 with flood inundation levels simulated using RRI under 10-, 20-, 50-, and 100-year return-period precipitation scenarios at all target shrines. H1 exceeded the simulated inundation level for a 100-year return-period precipitation scenario, excluding Toyokuni Shrine. These results suggested that, for the majority of historic shrines located in areas with high flood-exposure, the safety of their spaces was ensured against a 100-year return-period precipitation scenario at their respective sites. Furthermore, a significant positive correlation was observed between the simulated inundation level and H1 (Figure 5). This correlation is considered sufficient evidence that local flood characteristics were a key factor in the physical structural design of the shrines. Consequently, historical shrines located in areas with high flood-exposure potential were likely constructed on sites with safer characteristics than those of the surrounding areas.

Table 3 shows the historical information and H3 for the 11 historic shrines identified as analysable in the “historical continuity first” screening, but with no flood-exposure potential. H3 for almost 70% of these shrines was 0 m or 0 to 0.5 m, confirming that flood-resilience strategies do not appear to have been considered for them.

Table 3. Historical information and H3 for the 11 historic shrines identified as analyzable in the “historical continuity first” screening, but with no flood-exposure potential.

These H3 values are preliminary estimates based on visual inspection from available photographs. Shaded rows indicate shrines that were excluded from the final target sample due to lack of historical continuity or high flood exposure potential.

3.2.2 Landform classification of the target shrines

Table 4 shows the landform classification of the shrines, based on the Landform Classification Map for flood control. It was found that the landform of shrines established in flood-prone areas can be broadly classified into the following three categories.

Table 4. Landform classification located target shrines based on the Landform Classification Map for Flood Control.

Floodplains indicate low and nearly flat depositional areas with potential flood exposure. Uplands/terraces represent relatively higher, step-like flat surfaces. Slightly elevated lands (natural levees) formed along present or former river channels, and dunes/sandbars formed by sandy or gravelly deposits along coastal, lacustrine, or fluvial margins (Geospatial Information Authority of Japan, Disaster Prevention Geography Division, 2015).

Firstly, it was found that the surrounding areas of Ajima Shrine, Shichisho Shrine, Narumi Hachiman Shrine, Otakacho Hachiman Shrine, Manbacho Hachiman Shrine, and Jaike Shrine possessed locational characteristics that enable them to avoid flood damage. These areas include slightly elevated land, such as sandbars, dunes, and natural levees with low flood exposure, as well as uplands and terraces. As an example, Figure 6 shows an elevation map of the area surrounding Ajima Shrine. The shrine is situated on the edge of the Nagoya Plateau and borders a lowland formed by the Shonai River; it is believed to have been established at this location because of the significant elevation difference with the surrounding area, which provides a relatively safe site.

Secondly, it was found that Ueda Hachiman Shrine, Yamada Tenman Shrine, and Hiyoshi Shrine, although built on floodplains with a high risk of flooding, are situated at higher elevations than the surrounding area and thus possess site characteristics that allow them to avoid flood damage (Figure 7). As an example, Figure 8 shows the elevation map of Ueda Hachiman Shrine. Although the shrine is situated on a floodplain formed by the Tenpaku River and might be considered at high risk of flooding, it was built on a slight elevation, atop a keyhole-shaped tumulus from the middle Kofun period. It is therefore thought that the shrine was established in a relatively safe location compared to the surrounding area.

As a third classification, Sengen Shrine and Toyokuni Shrine are situated on floodplains, where the safety of the shrine precincts was ensured by raising the main hall to a sufficient height (Figure 7). As an example, the elevation map of Sengen Shrine is shown in Figure 9. It is a small shrine located in a high-risk flood area, sandwiched between the Shonai and Shin Rivers. In this area, a plain dotted with lowlands and slight elevations has been formed by sedimentation resulting from the flooding of the Shonai and Shin rivers; the shrine is built on one of these slight elevations. Although the elevation difference was smaller than at other shrines, the main hall had been sufficiently raised by 1.7 meters. It is therefore considered that the shrine was built on a site with relatively safer characteristics than the surrounding area.

3.3 Heterogeneity in the traditional flood-resilience strategy at the historical shrine spaces in the high flood-exposure potential

Figure 10 shows the comparison of H3 for each target shrine with RRI-simulated inundation level under the 100-year return-period precipitation scenario at all target shrines. It was found that Ajima Shrine, Shichisho Shrine, Otakamachi Hachiman Shrine, Ueda Hachiman Shrine, Yamada Tenman Shrine, and Hiyoshi Shrine can ensure sufficient safety due to the elevation of the shrine precinct relative to the surrounding terrain (H3). All of these shrines were established before the Meiji era (1868). Until the Meiji era, Japan was a feudal state characterized by a decentralized governance structure based on land and lord-vassal relationships. This structure included the “shoen system” (from the late 11th to the late 12th century), which managed private lands owned by the aristocracy and temples and shrines, and the “Bakuhan system” (from the late 12th century to the mid-19th century), governed by the shogun and daimyo. Within Japan's feudal system, shrines played a vital role that extended beyond being mere places of worship; they served as centers of land control, community cohesion, and a foundation for political authority. As the heart of local communities, they united the local inhabitants spiritually and socially by enshrining the “Ujigami”—the guardian deity of the region. Furthermore, a privileged group responsible for managing rituals was formed, which became closely linked to village decision-making and the social hierarchy. It was inferred that this social context led to a traditional flood-resilience strategy designed to ensure not only the functional safety of the main hall, a key faith-based facility, but also the safety of the entire shrine space, a venue for community life, particularly when constructing shrines in areas highly susceptible to flooding.

It was clear that the Sengen Shrine, Manbacho Hachiman Shrine, Toyokuni Shrine, and Jaike Shrine, whose H3 values were 0 m, have ensured safety by lifting the main hall (H2). Except for Narumi Hachiman Shrine, these shrines were established during or after the Meiji era (from 1868 onwards). Furthermore, although Narumi Hachiman-gu, established before the Meiji era, was originally built on a slight elevation with low flood-exposure potential, the floor height of the main hall (H2) was lifted in 1968 following the 1959 flood damage from the Ise Bay Typhoon. Following the Meiji Restoration (1868), the feudal system came to an end with the abolition of the “Bakuhan system” and establishment of prefectures (1871), and Japan transformed into a centralized, modern sovereign state. As Japan established itself as a modern sovereign state, shrines played a vital role in symbolizing national unity. The government of the time redefined “Shinto” as a spiritual pillar for the people, in opposition to Western Christianity, and proceeded to institutionalize it. Consequently, shrines nationwide were placed under state administration; following the principle of “one shrine per village,” small shrines were consolidated or abolished, and their management was closely integrated with local administration.

It was inferred that this shift in the management system of shrines from local communities to the administration led to a traditional flood-resilience strategy shift in the safety of newly established shrine spaces in flood-prone areas: simply lifting the floor height of a main hall, a key structure of faith-based facility (H2), as the focus shifted to prioritizing major defenses. Raising the entire shrine site (H3) is far more costly than lifting only the main hall of key faith-based facilities (H2); furthermore, as this shift in management was implemented nationwide, it is presumed to have been a cost-effective solution. In addition, from a construction technology perspective, this could be achieved using the traditional Japanese technique of “Kake-Zukuri” (suspended construction), which involves assembling timber to support structures without any nails on uneven or low-lying sites (Nishi and Hozumi, 1996). This technique can ensure that the shrines' critical parts are elevated cost-effectively; for example, the stage at Kiyomizu Temple in Kyoto, a World Heritage Site established in 778, was built on a 13-meter cliff (Larsen, 1994).

Regarding the influence of landform characteristics on the traditional flood-resilience strategy, three shrines were constructed on floodplains before the Meiji era, and three after. Whereas, the three pre-Meiji shrines exhibited H3 corresponding to RRI-simulated inundation level under the 100-year return-period precipitation scenario, the three post-Meiji shrines had an H3 value of 0 m. This result indicated that, independent of the high flood-exposure potential inherent in the floodplain, the traditional flood-resilience strategy at target historical shrine spaces changed before and after the Meiji era (Figure 7). Similarly, on slightly elevated land and upland/terrace with low flood-exposure potential, three shrines were constructed before the Meiji era, whilst two shrines were constructed in similar locations after the Meiji era. Whilst the three pre-Meiji shrines achieved H3 corresponding to RRI-simulated inundation level under the 100-year return-period precipitation scenario, the two post-Meiji shrines did not. This also confirmed that, despite the landform characteristics, traditional flood-resilience strategies for the shrine spaces were changed before and after the Meiji era.

Despite target shrines managed by different organizations by local communities and local government before/after the Meiji era, the fact that the safety of the shrine spaces against a 100-year return-period precipitation scenario was ensured across the 11 shrines under study suggested that the administration inherited local communities' awareness of flood risk. Ultimately, this study provided empirical evidence that faith-based facilities serve as important physical repositories of data on the long-term interaction between humans and floods.

4 Conclusion

This study provided empirical evidence demonstrating that historic shrines, dating back hundreds of years and situated in flood-prone low-lying areas of Nagoya City, serve as vital physical repositories of data on the long-term interaction between humans and floods. A significant positive correlation was observed between potential flood depth derived from several extreme precipitation events and total floor height of the main hall relative to the surrounding terrain (H1). Furthermore, the height of the shrine precincts (H3) relative to the surrounding topography was generally 0 m for shrines built in areas without flood risk at the time.

These results indicated that local hydrological characteristics were a key factor for these shrines. This confirmed that, even in the absence of measurement records, the structure of the shrines serves as a reliable physical indicator for deciphering ancestral perceptions of flooding and reconstructing historical flood patterns. These results supported the hypothesis put forward by Takada and Kuwako (2016) that “shrines playing a significant role in the religious life of a particular locality possess site characteristics that can ensure safety even during natural disasters.” We demonstrated that historical faith-based facilities situated in areas with high flood-exposure potential at the time can be utilized as socio-hydrological markers regarding the long-term interaction between humans and disasters. In addition, the analytical framework in this study combining historical continuity, flood-exposure modeling, and field measurements might also be applicable to other types of cultural heritage, such as historical temples, castles, or traditional settlements, or even to historical sites in other countries.

Using historical faith-based facilities as socio-hydrological markers, we revealed that traditional flood-resilience strategies differed between the periods before and after the Meiji era (1868). Furthermore, despite target shrines managed by different organizations by local communities and local government before/after the Meiji era, the fact that the safety of the shrine spaces against a 100-year return-period precipitation scenario was ensured across the 11 shrines under study suggested that the administration inherited local communities' awareness of flood risk. It also suggests that the local community's awareness of flood risk has been passed down from as far back as the 10th century (Ajima Shrine) to the present day (Jaike Shrine). However, these conclusions were based solely on the relationship between flood risk and the total floor height of the main hall relative to the surrounding terrain (H1), based on hypotheses proposed by Takada and Kuwako (2016). It is, therefore, essential to validate how the local community have perceived flood risk. Unfortunately, due to limited historical and documentary evidence, this study was unable to verify the local community's awareness of flood risk for each shrine. Future socio-hydrological research, including historical investigations into shrine renovations, disaster damage at each shrine, and historical flood event records, is expected to deepen our understanding of how local communities developed, maintained, and transmitted flood-risk awareness.

In addition, the elevation of historical faith-based spaces is influenced not only by disaster risk but also by the religious beliefs of the local community, the enshrined deities, and local culture and environment. To examine the hypothesis proposed by Takada and Kuwako (2016), this study established strict selection criteria to identify only those shrines in which safety from flood risk was considered to be the primary factor contributing to the elevation of the shrine space. Future studies aim at extending the approach to historical faith-based facilities built in areas with moderate and low flood risk at the time. By investigating the relationship between their spatial height and RRI-simulated inundation levels, and by validating the religious beliefs of the local community, the enshrined deities, and local culture and environment, it will be possible to examine spatial variations in local communities' awareness of flood risk and other influencing factors.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Jenia Mukherjee, Indian Institute of Technology Kharagpur, India

Reviewed by: Shanjia Zhang, Lanzhou University, China

Sihan Zhang, Chiba Daigaku—Matsudo Campus, Japan