1 Introduction

Sea level rise (SLR) due to climate change poses a serious risk to human life in the world’s coastal areas (IPCC, 2019, 2022). Japan is surrounded by ocean and is extremely sensitive to SLR. Over the period from 1902 to 2010, the global mean sea level increased by 16 [12–21] cm. SLR increases have been observed in the waters around Japan since the 1980s, with a rise of 4.1 [0.1–8.2] mm/year during the period 2006–2015 (MEXT, JMA, 2020). Global mean SLR will continue to increase beyond 2100 at a rate that will depend on future greenhouse gas emissions (IPCC, 2021).

Assessing the coastal impacts of SLR and the effectiveness of adaptations is crucial for reducing adverse impacts of climate change in coastal areas. Several previous studies in this area have focused on Japan, including Suzuki (2014), which estimated the damage of SLR and storm surge based on three general circulation models (GCMs) and three representative concentration pathways (RCPs), and Oba et al. (2021), which also estimated the impacts of SLR. However, the present study has several distinctive features:

First, this study evaluates SLR and its inundation effects in the coastal areas of Japan at a resolution of 1 km, using the latest climate scenarios, tidal data, and shared socioeconomic pathways (SSPs). It also offers a consistent and detailed impact assessment of SLR in Japan, including the potentially inundated areas and affected population.

Second, the present study evaluates economic damages based on the “Manual for economic evaluation of flood control investment (draft)” (hereafter, “flood control manual”) (MLIT, 2020), while the authors’ previous study (Oba et al., 2021) used the econometric relationship between past hydrological disasters, similar to Tamura et al. (2019) on the global scale. Other efforts to estimate the global economic damage of SLR have also appeared, including Fankhauser (1995), the FUND model (Darwin & Tol, 2001), and the DIVA model (e.g., Hinkel et al., 2014), which used wet and dry land losses associated with the per capita income and population density of the affected areas. One of the unique contributions of this study is its use of the most current Japanese SSP population and land use projections, which were released in 2021 (NIES, 2021a, 2021b; Yoshikawa et al., 2022) and make it possible to evaluate the socioeconomic impacts in detail. Although it is difficult to apply the “flood control manual” method on a global scale, the currently available data make it possible to make projections for Japan. Accordingly, the study reevaluates the economic damage due to SLR in Japanese coastal areas.

Third, this study compares the cost of the adaptations of protection and relocation in a consistent manner and discusses appropriate adaptation strategies for Japan’s coastal areas. Coastal adaptations to reduce vulnerability are generally classified as protection, accommodation, and retreat (or relocation) (Hino et al., 2017; IPCC, 2019, 2022; Klein et al., 1999). It is increasingly evident that the solution space for adaptation in low-lying coastal areas is shrinking due to the progression of climate change (Haasnoot et al., 2021). Although adaptation has the ability to substantially reduce the negative impacts of SLR (Hinkel et al., 2014), there have been few assessment studies that evaluated the effectiveness and cost of adaptation. Hard engineering protection measures are commonly used to reduce coastal flooding and to drain or store excess water from intense precipitation. Many coastal cities, particularly densely populated and high-resource archetypes, have planned and are planning to continue a protection-based strategy (IPCC, 2022). Tamura et al. (2019) and Kumano et al. (2021) evaluated protection, including both gray and green infrastructures, globally but did not make comparisons with retreat or relocation. Retreat is a strategy for reducing exposure and, ultimately, the risks facing coastal settlements by moving people, assets and activities out of coastal hazard zones (IPCC, 2019). Relocation (managed retreat) is generally considered to be a viable adaptation response for communities in areas impacted by, or at risk of, inundation or other hydrological changes (IPCC, 2022). Diaz (2016) developed an optimization model for least-cost adaptation decisions in response to SLR and storm surge in the Coastal Impact and Adaptation Model and concluded that protection is more expensive than retreat in most countries. However, the cost estimation methods used in her study and in other relevant studies are based on numerous parameterizations and are not necessarily derived from detailed empirical data (c.f., André et al., 2016; Hallegatte et al., 2013; Lincke & Hinkel, 2021). Most Japanese relocation studies have focused on relocation from tsunami disasters and floods in specific cities (c.f., Higashino & Murao, 2021; Takeda & Tsuda, 2015) but rarely have analyzed relocation as it relates to SLR. Imamura et al. (2022) estimated the costs of relocation as a response to SLR in Japan for the first time using empirical economic data. However, the range of relocation costs in Imamura et al. (2022) was rather broad due to the uncertainty regarding the relocation sites, which is inconvenient when comparing relocation costs with damage and protection costs. This study narrows the range of possible relocation sites by taking into account salient geographical characteristics.

The present study was primarily focused on a comparison of protection and relocation. It considered two counterfactual cases in which it was assumed that only protection or only relocation is applied in all the vulnerable coastal areas of Japan. The study did not consider SLR accommodation strategies, which would offer intermediate features and more diverse options. The estimates provided here effectively bound the range of possible adaptation costs and will arguably enhance the discussion of near-future countermeasures against SLR due to climate change.

2 Methodology

2.1 Assessing impacts of SLR

Figure 1 outlines the method used in this study to assess the inundation impacts of sea level change. The method employs a GCM, together with tide and topography data. Inundation damage in coastal zones is considered a significant consequence of SLR. Here, inundated areas and temporal changes in inundation were estimated using topographic data, astronomical high-tide data (that is, mean higher high-water level: MHHWL), and sea surface height data (global steric sea level). Potentially inundated areas without adaptation were identified by comparing sea surface height and land elevation. In the identification process, the areas of the grids of sea water were extended landward if the height of the neighboring landward grid was below the sea surface height. The reference year was set to 2010, which only includes tidal data. Vertical land movement, such as subsidence, was ignored.

Fig. 1
figure 1

Outline of the method

Full size image

Topography data were obtained from digital national land information as a tertiary mesh with approximately 1-km resolution (MLIT, 2011). Tidal data were obtained from the TPXO9.0 (Egbert & Erofeeva, 2002). High tides (MHHWL), which occur roughly twice monthly, were combined with the four major component tides (M2, principal lunar semi-diurnal tide; K1, luni-solar diurnal tide; S2, principal solar semi-diurnal tide; and O1, principal lunar diurnal tide).

Future sea surface height data were taken from the MIROC-ESM-CHEM model (Watanabe et al., 2011) of CMIP5 (Coupled Model Intercomparison Project Phase 5), which was demonstrated as reasonably reproducing ocean and marine ecosystems around Japan. Tsuchida et al. (2018) evaluated the uncertainty of global SLR impacts using eight GCMs under various RCPs and SSPs, and showed that, among them, results with MIROC-ESM-CHEM had the largest impacts. Hence, the impacts here are generally the highest among commonly used models. Left to future work is a sensitivity analysis using multiple climate models.

RCP refers to future changes in greenhouse gases and aerosols, and the amount of SLR varies with each of the pathways. This study employed RCP8.5 and RCP2.6, which can be considered the maximum and moderate scenarios for SLR. SSP affects the socio-economic components due to SLR. O’Neill et al. (2016) suggested that RCP8.5-SSP5 is one of the more probable “tier 1” combinations among the more than 20 scenarios of RCP and SSP, together with RCP2.6-SSP1, RCP4.5-SSP2, and RCP7.0-SSP3. Our analysis was performed under RCP8.5-SSP5 and RCP2.6-SSP1. SSP5 is regarded as a “fossil-fueled development” scenario, which may emit the highest greenhouse gases relevant to RCP8.5 while SSP1 is regarded as “sustainability” scenario, which may emit the lower greenhouse gases relevant to RCP2.6. Mean SLR around Japan of MIROC-ESM-CHEM increases by about 27 cm under RCP2.6 and about 56 cm under RCP8.5 from 2010 to 2100. The effects of permanent inundation by SLR due to climate change were considered, while the effects of temporary flooding, such as storm surges by typhoons and tsunamis due to earthquakes, were not. In this study, the GCM results did not indicate the catastrophic cases involving more than 1 m SLR in 2100 and more than 5 m SLR by 2300 that have been pointed out by IPCC (2021).

Socioeconomic scenarios were obtained from the tertiary mesh SSP population and land use projections of the National Institute for Environmental Studies, Japan (NIES, 2021a, 2021b; Yoshikawa et al., 2022), which began distribution in 2021. Among the five SSPs (SSP1–SSP5), SSP5 projects the highest Japanese population and GDP in 2100, while SSP3 projects the largest world population and SSP5 projects the largest world GDP. SSP1 projects the second-largest population and GDP in Japan. The Japanese population is projected to decrease to 79.4 million by 2100 in SSP5 and to 72.9 million in SSP1, as compared to the 2015 population of 125.6 million (NIES, 2021a). Population scenarios were updated in the 2nd edition from Oba et al. (2021). The Japanese GDP is projected to be around 530 trillion yen (2015 prices) by 2050 and 704 trillion yen by 2100 in SSP1, while it is projected to be around 632 trillion yen by 2050 and 1119 trillion yen by 2100 in SSP5 (NIES, 2021a).

2.2 Assessing economic damage

Economic damage was estimated according to the “Manual for economic evaluation of flood control investment (draft)” (flood control manual, MLIT, 2020). The flood control manual is used to assess the economic benefits and cost-effectiveness that results from the development of flood control facilities such as dikes and dams in Japan (MLIT, 2020). This method for estimating economic damage requires a substantial amount of data, including future projections of population and land use, such as agriculture, housing, other buildings, and infrastructure. The tertiary mesh SSP population and land use projections noted above were used.

There is some debate as to whether the manual for estimating temporal inundations such as floods can be applied to a permanent inundation such as SLR. Japan has protected coastal area, and, to date, there have been no empirical data for establishing the economic damages associated with SLR. If factors such as national land loss, the impact on Japan’s Exclusive Economic Zone (EEZ), and various socio-cultural impacts are considered, the damages can be enormous and extend well beyond the economic aspects. Nevertheless, the study employed this method and compared it with others as a basis for developing a better approach in the future.

Table 1 provides examples of damage estimation using MLIT (2020). Economic damage is generally calculated by multiplying the unit cost (e.g., per area or per capita), magnitude (e.g., the inundated area or affected people in the sector), and damage rates by the inundation depth. Table 2 shows the damage rates by inundation depth for each category. Relatively large damage rates from the flood control manual are set here since SLR is permanent rather than temporal inundation. Tamura & Yokoki (2021) preliminarily assessed the economic damage only under RCP8.5-SSP5 and the constant 2015 land use scenario. Kodama et al. (2022) and the present study analyze the uncertainty of future projections, employing land use scenarios and two combinations of RCP/SSP after the release of NIES (2021b). This study updates the unit cost from Kodama et al. (2022) based on the MLIT correction in August 2022 (MLIT, 2022). Unless otherwise specified, all monetary values are converted to 2015 Japanese yen and are not discounted, as in Hinkel et al. (2014). Our previous studies (e.g., Kodama et al., 2022; Oba et al., 2021) used 2005 prices. Deflators and unit costs vary according to SSP.

Table 1 Examples of damage estimation using MLIT (2020)
Full size table
Table 2 Damage rates by inundation depth
Full size table

2.3 Assessing cost of adaptation

2.3.1 Protection

For the purpose of evaluating SLR protection costs, all coastlines in the inundation area were assumed to be protected by dikes. It was assumed that coastlines with tides higher than the ground level in 2020 are already protected up to the tidal height and that land areas should be protected until 2100. The term “new construction” is used for dikes constructed where the tide level exceeds the ground level and the height of existing dikes due to SLR; the term “reconstruction” is used when a dike is constructed at the same height as an existing dike whose life has exceeded its 30-year expected lifetime. The height of the dikes is adjusted according to SLR through 2100.

Future adaptation costs were obtained by multiplying the length of the inundated coastline to be protected and the unit cost of protection using dikes. Tamura et al. (2019) examined the design standards of coastal defenses such as sea dikes and constructed a database of the unit costs of such projects based on international design standards as recorded in numerous reports. The database includes 455 dike cost values from 20 countries. Tsuchida et al. (2019) and Kumano et al. (2021) estimated the unit cost of a dike using the height of the dike and the GDP per capita. Unit costs also vary according to SSP.

Two types of unit cost of protection were used to examine the robustness of our estimation. First, the study slightly updated the above data for dikes in Japan, and regression analysis was conducted again. The unit costs of protection using dikes were then simulated by incorporating the required height of dikes and GDP per capita (Type 1). The unit cost of Type 1 dike is estimated as in Eq. (1):

$$c_{1} = 8412\;H^{3} + 304\;pGDP$$
(1)

where c1 is the unit cost of dike construction (US$/km), H is the height of the dike (m), and pGDP is the GDP per capita (US$). 1 US$ was converted to 121 yen in 2015 price.

Second, the study also applied the JSCE report (2018), which calculated the current unit cost of dikes built as preparation for serious natural disasters in Japan (Type 2). This unit cost of a Type 2 dike also includes the height of the dike as in Eq. (2):

$$c_{2} = 7.0 \cdot H$$
(2)

where c2 is the unit cost of dike construction (100 million yen/km). The future unit costs of protection using dikes were calculated by multiplying the growth rate of GDP per capita by these current unit costs. Here, adaptation costs are for the normal level of hazards (rank C) and do not include more serious countermeasures for huge earthquakes and tsunami (rank A or S).

2.3.2 Relocation

The method used in this study improves upon the cost estimation method for relocation in Imamura et al. (2022), which followed the conventional scheme prescribed in the “Project for promoting community relocation for disaster prevention” (MLIT, 2021) for serious natural disasters and used abundant existing empirical economic data on the cost of civil engineering works. The relocation cost consists of the costs of land acquisition, land development, infrastructure construction, the purchase of inundated residential and agricultural land, the subsidized costs of acquiring housing and land at the relocation site, and the subsidized costs of other expenses, such as moving costs, calculated on a per-household basis. The total relocation cost is estimated by multiplying the relocation cost per household by the number of inundation-affected households and the rate of future price increases. This total represents the cost of relocating the entire inundation-affected population at each time point, based on a land requirement of 660 m2 per household, which includes a house and the social infrastructure needed for the community. In order to simplify the estimation, it is assumed that all inundation-affected households will live in single-family homes on the relocation site. This assumption is clearly unrealistic for densely populated areas such as Tokyo. However, the scheme was designed to be implemented in rural areas or small communities, and no corresponding scheme exists for relocations in densely populated areas or nationwide. It should be noted by the reader that the relocation costs estimated here are based on the application of the current system.

The present study differs from Imamura et al. (2022) in that it has a narrower range of relocation costs. Imamura et al. (2022) assumed that land development costs would vary depending on the slope of the relocation site and stipulated that the upper limit of these costs would be in the case of relocation to land with a slope of 25–30 degrees. The present study collected data on the maximum slopes (c.f., MLIT, 2011) of land areas dedicated to various land uses (c.f., NIES, 2021b; Yoshikawa et al., 2022) in Japan and applied a 1 km square grid to establish the area and slope of the land available for relocation. The available land includes paddy fields, other cropland, forests, and wasteland. The amount of land area required for relocation was, as noted, calculated by multiplying the number of inundation-affected household by 660 m2, the area per household needed at the relocation site. Table 3 shows both the land required and the land available. The available area with a slope of no more than 30 degrees (“≤ 30 deg.”) accounts for most of the land in Japan, which has a total land area of approximately 378,000 km2. As shown in the table, the area available for relocation would be sufficient even if all relocation sites were constrained to land with a slope of three degrees or less (i.e., flat land). However, such a constraint would reduce the candidate sites for relocation to roughly one-tenth of the sites that could be used if no slope-related constraints were imposed. In order not to exclude too many patterns of relocation, the present study determined the upper limit of land development costs by setting the maximum slope allowable for relocation at 10–15 degrees [vs. 25–30 degrees in Imamura et al. (2022)], which would qualify nearly half the candidate sites that would be available if there were no slope constraints.

Table 3 Required land area for relocation and available land area according to slope levels in Japan (km2)
Full size table

Another rather minor change in the cost estimation setup that differentiates this study from that of Imamura et al. (2022) was the assumption that inundation-affected residents of Tokyo would be relocated to Saitama prefecture rather than to other areas of Tokyo prefecture. Table 4 shows the required land area and the available land area having a slope of no more than 15 degrees in the two prefectures. As indicated in Table 4, if the relocation sites were restricted to land with a slope no greater than 15 degrees, there would be insufficient relocation sites in Tokyo. Consequently, Saitama prefecture was considered to be a suitable location for relocation from Tokyo because it is adjacent to Tokyo and has sufficient area.

Table 4 Required land area for relocation and available land area in Tokyo and Saitama prefecture (km2)
Full size table

3 Results

3.1 Impacts and economic damage

Figure 2 shows the potentially inundated areas in Japan in 2100 (as compared with 2010 sea levels). Figure 3 represents the temporal change in the potentially inundated areas under RCP2.6 and under RCP8.5 in 8 regions compiled from 47 prefectures with coastlines. The potentially inundated area was determined to be approximately 2111–2127 km2 in 2050 and 2261–2598 km2 in 2100, under RCP2.6 and RCP8.5. The Kyushu and Okinawa region shows the largest future inundation, mainly derived from the coastal areas of Saga, Fukuoka, and Kumamoto, which surround Ariake bay. The Chubu region has the second-largest potentially inundated area, mainly due to Aichi and Mie, which surround Ise bay. The coastal areas of Ise bay and Saga in the Kyushu and Okinawa region have large potential inundation from tides due to their low elevation; inundation in the Tokyo area is expected to increase as the result of the predicted SLR.

Fig. 2
figure 2

Potentially inundated area (RCP8.5)

Full size image
Fig. 3
figure 3

Temporal change of potentially inundated areas in 8 regions (bar: RCP8.5; line: RCP2.6)

Full size image

The affected population was estimated to be 4.45–4.70 million in 2050 and 3.76–4.92 million in 2100, depending on the RCP/SSP (Fig. 4). As in Oba et al. (2021) and Kodama et al. (2022), three major bays, namely, Tokyo, Ise, and Osaka bays, will have large affected populations. Under RCP8.5-SSP5, the affected populations in Tokyo, Chiba, and Kanagawa are expected to increase through 2100, while the affected populations in most of the other prefectures are expected to increase until 2050 and then decrease through 2100. The affected population under RCP2.6-SSP1 will peak around 2030 and then decline. The decreases noted here reflect the overall shrinkage of the Japanese population. The percentage of total population is larger in RCP8.5-SSP5 than in RCP2.6-SSP1.

Fig. 4
figure 4

Affected population without adaptation (left: total population; right: % of total population in each SSP)

Full size image

Figure 5 shows the economic damages under RCP8.5-SSP5 and RCP2.6-SSP1. The damages were estimated to vary within 151–181 trillion yen in 2050 and within 243–455 trillion yen in 2100. The economic damages were derived mainly from public facilities and residential and commercial buildings under both RCP8.5-SSP5 and RCP2.6-SSP1. The other economic damages, such as those related to agriculture and golf, were extremely minor. Although agricultural land accounted for 20–30% of the total inundated area, the associated economic damages were much smaller than those for developed land. The economic damages determined by applying the “flood control manual” were 24–26 times as large as those calculated with the previous estimation method, as shown in Oba et al. (2021). The result closely matches that of Yanai et al. (2017), which compared two methods for determining the inundation impacts of Ise bay by applying them to the record typhoon that struck the area in 1951. The “flood control manual” method applied here took into account more, and more detailed, economic factors and produced a higher estimate than did the previously applied macro econometric method. The “flood control manual” method indicated a greater impact around the three major bays, mainly due to the damage done to buildings and the size of the affected population. Economic damage under RCP8.5-SSP5 was approximately 10% less than in Tamura & Yokoki (2021), which preliminarily used constant land use in 2015, reflecting the future decrease in residential and commercial buildings. In 2005 prices, the revised economic damage based on the MLIT (2022) correction was roughly 0.01% less than that reported in Kodama et al. (2022).

Fig. 5
figure 5

Economic damage without adaptation (left: RCP8.5-SSP5, right: RCP2.6-SSP1)

Full size image

It is noteworthy that all the impacts (inundated area, affected population, and economic damage) under RCP2.6 were smaller than those under RCP8.5, suggesting the importance of greenhouse gas mitigation efforts.

3.2 Cost of adaptations

3.2.1 Protection

Figure 6 shows the length of dikes and the distribution of the height of the dikes required for protection that would be reconstructed under RCP8.5 and RCP2.6. The length of dikes generally increases according to the progress of SLR. RCP8.5 requires approximately 440 km of dike from 2090 to 2100, with 42% of the dike over 2 m in height. On the other hand, the required dike length is approximately 384 km from 2090 to 2100 for RCP2.6, with 36% of the dike over 2 m in height.

Fig. 6
figure 6

Length of dikes and distribution of height of dikes required for protection to be reconstructed (left: RCP8.5, right: RCP2.6)

Full size image

Figure 7 presents protection costs against SLR, based on the length, height, and unit cost of dikes. Protection costs under RCP2.6-SSP1 were less than those under RCP8.5-SSP5. This is because lower SLR in RCP2.6 corresponds to lower required length and height of dikes and SSP1 generally costs less than SSP5. The total cost of protection under RCP2.6-SSP1 was estimated to be approximately 39.7 trillion yen for Type 1 and 53.4 trillion yen for Type 2 in 2100. Similarly, the total cost under RCP8.5-SSP5 was estimated to be 54.4 trillion yen for Type 1 and 84.5 trillion yen for Type 2 in 2100.

Fig. 7
figure 7

Protection cost (left: Type 1, right: Type 2)

Full size image

3.2.2 Relocation

Figure 8 compares the relocation costs in the present study with those in Imamura et al. (2022). The range of relocation costs in the present study was within the lower half of the range in Imamura et al. (2022). This is due to the difference between the development cost for land with a 25–30 degree slope and the development cost for land with a 10–15 degree slope.

Fig. 8
figure 8

Relocation cost

Full size image

Figure 9 gives the details of the cost of residential relocation under RCP8.5-SSP5 and RCP2.6-SSP1. The total cost of relocation under RCP2.6-SSP1 was estimated to be 116 [92–140] trillion yen in 2050 and 161 [121–202] trillion yen in 2100, whereas the total cost under RCP8.5-SSP5 was estimated to be 134 [105–163] trillion yen in 2050 and 286 [208–363] trillion yen in 2100. Land acquisition and development accounted for the highest cost, followed by the cost of purchasing agricultural and residential lands.

Fig. 9
figure 9

Details of relocation cost (left: RCP8.5-SSP5, right: RCP2.6-SSP1)

Full size image

Both protection and relocation costs under RCP2.6-SSP1 were less than those under RCP8.5-SSP5 after 2030. It was found that the costs of relocation could exceed those of protection in Japan under the current framework of relocation policies. Diaz (2016) showed that, globally, relocation costs are lower than protection costs but that relocation costs in Japan are slightly higher than the cost of protection. The results of the present study partially support Diaz (2016); however, the difference between relocation and protection costs in the present study is much larger than in Diaz (2016). As noted earlier, the conventional scheme described in the “Project for promoting community relocation for disaster prevention” (MLIT, 2021) is designed for application to small communities. The implementation of relocation measures in densely populated areas may well require various other costs in addition to those provided for under this scheme, and thus, the difference between the protection and relocation costs can be expected to be even wider.

4 Conclusion

This study evaluated SLR and its inundation effects in the coastal areas of Japan using the latest climate scenarios, tidal data, and SSPs for population and land use, with a resolution of 1 km. Economic assessment was incorporated by including SSP land use scenarios, also at a 1-km resolution. One unique property of this study was its use of Japanese SSP population and land use projections for this economic assessment. The uncertainty of future pathways RCP2.6-SSP1 and RCP8.5-SSP5 was carefully considered. The potentially inundated area of Japan was estimated to be 2111–2127 km2 in 2050 and 2261–2598 km2 in 2100, depending on the RCP. The affected population was estimated to be 4.45–4.70 million in 2050 and 3.76–4.92 million in 2100, depending on the RCP/SSP combination. The economic damage was estimated to be 151–181 trillion yen in 2050 and 243–455 trillion yen in 2100. All the impacts (inundated area, affected population, and economic damage) under RCP2.6-SSP1 were significantly smaller than those under RCP8.5-SSP5. It is evident that mitigation efforts can substantially reduce serious impacts in coastal areas.

Given that the solution space for adaptation in low-lying coastal areas is shrinking due to the progression of climate change, this study also examined various coastal adaptations using a consistent framework. The study evaluated the cost and effectiveness of protection and relocation adaptations using empirical data on the cost of constructing dikes and the cost of relocating houses and infrastructure. The study could consistently compare the costs of protection and relocation, in contrast to analyses on single adaptations such as Tamura et al. (2019). The total cost of protection under RCP2.6-SSP1 was estimated to be approximately 39.7–53.4 trillion yen in 2100, while the total cost under RCP8.5-SSP5 was estimated to be 54.4–84.5 trillion yen in 2100. The total cost of residential relocation was estimated to be approximately 92–163 trillion yen in 2050 and 121–363 trillion yen in 2100. In addition to the above impacts, both the protection and relocation costs under RCP2.6 were less than those under RCP8.5. It was found that the costs of relocation in Japan can exceed the costs of protection under the current framework of relocation policies. The costs of these two adaptations can also be decreased by mitigation efforts.

Finally, the study examined two counterfactual cases, where only protection or only relocation measures would be applied to all the coastal areas of Japan. Clearly, it would be more realistic to assess a portfolio with more diverse options that include accommodation. Protection can be mainly implemented in densely populated areas, while relocation can be mainly undertaken in vulnerable and suburban areas, in light of the socioeconomic conditions of these areas. Such options should be discussed among local municipalities using the results reported here as a basis for the discussion. Even given this limitation, however, the study identified the possible range of adaptation costs and outlined some important implications for balancing coastal adaptations.