Small Island Developing States under threat by rising seas even in a 1.5 °C warming world

The study addresses the vulnerability of Small Island Developing States (SIDS) to sea-level rise and coastal extremes, seeking to quantify current and future coastal flood risk, losses, and damages under different greenhouse gas emission trajectories. SIDS encompass diverse geographies and socioeconomic contexts but collectively face high exposure due to concentrated people and assets in coastal zones and constraints on adaptation. Prior IPCC assessments recognized their vulnerability, yet there is a lack of comprehensive, quantitative, and comparable estimates of flood risk, loss, and damage across all SIDS under mitigation pathways including a 1.5 °C warming target. The purpose is to provide a global, consistent assessment of coastal flood hazard, exposure, and direct economic damages for SIDS throughout the 21st century and beyond, isolating climatic effects and then integrating socioeconomic dynamics, to inform mitigation benefits and adaptation needs.

The paper notes that while there is growing attention on climate risks in SIDS, most studies focus on single states or regional subsets, using differing datasets and methods that hinder comparability. Global assessments including SIDS often lack detail or provide broad overviews without SIDS-specific focus. The literature identifies SIDS as vulnerable since the IPCC First Assessment Report, with multiple drivers including tropical cyclones, swell and wind-sea waves, microtidal regimes, and threatened coral ecosystems. Previous work highlights the role of nearshore bathymetry/topography, natural defenses (reefs, mangroves), and uncertainties in future extreme water-level changes. There is evidence of high historical losses (e.g., Hurricanes Dorian and Maria), shoreline changes, and the importance and limits of nature-based solutions. Studies also underscore the need for improved DEMs, representation of nonlinear tide–surge–wave interactions, and integrating socioeconomic pathways and adaptation in risk analyses.

The assessment employs the LISCOAST (Large-scale Integrated Sea-level and Coastal Assessment Tool) framework to quantify coastal hazard, exposure, vulnerability, and risk for all SIDS under five SSP-based climate scenarios spanning 1.5 °C (SSP1-1.9) to very high emissions (SSP5-8.5). Hazard modelling: Extreme sea levels (ESLs) combine mean sea level (MSL), tides, storm surges, and wave run-up. A global coupled reanalysis using SCHISM (2D barotropic) and WWM-V spectral wave model on an unstructured grid (~50 km offshore to ~2 km nearshore) is forced by ERA5 winds and pressures for 1980–2020, initially without tides to isolate weather-driven components. Nonlinear tide–wave–surge interactions are corrected using a 10-year reanalysis with tides and copula-based adjustments. Quantile mapping bias correction (satellite altimetry) is applied to water-level anomaly and significant wave height. Tropical cyclone (TC) contributions are enhanced via Delft3D-FM simulations forced by IBTRACS best-track data, superseding ERA5-based values when higher. Wave transformation uses spectral peak propagation along a global 1-km coastal transect dataset, applying Snell’s law and computing breaking wave height and run-up (Stockdon formula) to derive time series of run-up. ESL attenuation by coral reefs and mangroves is incorporated via reduction coefficients from published datasets. Tides are added using FES2014, considering spring–neap variability and high-tide water level. Future ESLs to 2100 integrate AR6 CMIP6-based probabilistic relative SLR (including steric, dynamic, cryospheric, land-water storage, GIA) with modeled changes in tidal elevations and present meteorological tides, acknowledging high uncertainty in future storm surge and wave changes. All components are represented as PDFs and combined via Monte Carlo to derive ESL return values for multiple return periods (1 to 5,000 years) through non-stationary extreme value analysis. Inundation modelling: Permanent inundation (land below future high-tide water level) is identified via a bathtub approach. Episodic flooding is simulated with Lisflood-ACC (inertial 2D solver) at 30 m resolution, using ESLs as forcing, land-use-derived roughness, and computational domains extended up to 200 km inland to capture hydrologically connected areas. Elevation data: GLO-30 DEM with post-processing using global LiDAR observations (DeltaDEM) to reduce vertical bias and correct for buildings/vegetation, improving upon SRTM-based datasets. Coastal protection: Present protection standards are estimated per 1-km coastal segment using proxies—GDP per capita classes, presence of ports/airports, urban/artificial surfaces, and population density within the 1-in-500-year flood area—with rules mapping socioeconomic indicators and critical infrastructure to assumed protection levels (minimum 1-in-1-year up to 1-in-50-year events depending on indicators and income class). Reverse calibration with PCRAFI, local studies, and expert judgement is applied where needed. Exposure and vulnerability: Present exposure uses WorldPop 2020; land cover from ESA WorldCover 2020 (10 m), corrected using 30 m HBASE built-up extent. Vulnerability is represented via depth-damage functions per land-use class; asset values scaled by 5-arcmin GDP per capita distribution. Future exposure/economy: SSP-consistent projections of population density and urbanization (1 km) and country-level GDP (IIASA/OECD) are downscaled to adjust asset values; urban population is used as proxy for urbanization change. For SIDS missing in SSP datasets, k-means clustering on geographic and socioeconomic variables imputes relative changes from similar countries. Risk metrics: For each segment and time step (1981–2100), PDFs of flooded area (FA), population exposed (PE), and direct damages (D) are derived across return periods and integrated to compute Expected Annual Flooded Area (EAFA), Expected Annual People Exposed (EAPE), and Expected Annual Damage (EAD). Adaptation sensitivity: Curves of 2100 EAD versus increased protection standard are generated to estimate additional protection heights/standards needed to keep 2100 EAD at present-day levels, with indicative unit costs per coastline km drawn from previous studies.

Present-day (baseline): - EAFA across all SIDS: 3,568 km² (5th–95th: 1,460–10,368), about 0.31% (0.13%–0.94%) of total SIDS land area. Regional shares: Caribbean ~50%, Pacific 42%, East Atlantic 5%, Indian Ocean 1%. Highest fraction of land flooded: Indian Ocean (0.71%), East Atlantic (0.61%), Caribbean (0.32%), Pacific (0.28%). - EAPE: 131,315 people (49,834–415,472), or 0.18% (0.07%–0.58%) of SIDS population. Largest contributors: Haiti (~22%), Papua New Guinea and Cuba (~11% each). Highest national shares: Vanuatu (1.96%), Belize (1.59%), with several others >1%. - EAD: US$1.64 billion (0.69–5.15) in 2020 dollars; 0.13% (0.06%–0.41%) of SIDS’ GDP. Losses concentrated in Caribbean (49%) and Pacific (46.5%); highest EAD/GDP in East Atlantic (median 1% (0.5%–5.9%)). National present-day EAD/GDP notable for Belize (3.17%), Bahamas (1.75%), Vanuatu (1.47%), Turks and Caicos (1.4%), Papua New Guinea (1.24%). Mid-century and end-century hazard (no socioeconomic change): - By 2050, EAFA more than triples, to 14,224 (9,809–19,601) km² under 1.5 °C and 15,620 (11,175–21,151) km² under very high emissions. - By 2100, EAFA reaches 23,890 (13,810–38,059) km² under SSP2-4.5 (≈1.1–3.2% of SIDS area). Under 1.5 °C: 19,213 (1,460–31,697) km²; under SSP5-8.5: 29,515 (18,330–46,463) km². - Permanently inundated land (below high-tide): median 7,653 km² (1.5 °C) to 16,274 km² (SSP5-8.5) by 2100; by 2150, 11,530–28,083 km² (1.0%–2.3% of area) for SSP1-1.9 to SSP5-8.5. Mitigation benefits (no socioeconomic change): - Reducing from very high (SSP5-8.5) to moderate (SSP2-4.5) emissions cuts end-of-century EAD by 23% overall (rising to 27.5% by 2150). Regional median reductions: Indian Ocean 24%, Caribbean 23%, East Atlantic 19%, Pacific 19%. - Further mitigation to 1.5 °C (SSP1-1.9) yields an additional 24.5% EAD reduction by 2100 (30.2% by 2150). Countries with >55% total mitigation benefits (SSP5-8.5 vs SSP1-1.9) include Barbados, Cayman Islands, Dominica, Saint Lucia, Northern Mariana Islands, Puerto Rico, Singapore, Suriname, Trinidad and Tobago. Lower benefits for steep volcanic islands (e.g., American Samoa, Tonga, French Polynesia). Low-lying Tuvalu and Maldives remain severely impacted under all scenarios, with smaller inter-scenario differences as key assets are already exposed. Regional/national highlights: - East Atlantic: Guinea-Bissau dominates regional exposure (90%) and damages (>70%). EAD as %GDP ranges from 5.4% (1.5 °C) to 8.6% (very high emissions). Bahrain: 2.4%–4.3% of GDP; Maldives: 9.2%–12.5% of GDP by 2100 (climate effects only). Longer term: - By late 22nd century (SSP1-2.6), SLR 0.26–1.92 m; EAFA 7,515–24,711 km² (0.6%–2.0% of area). By late 23rd century, SLR can exceed 3.1 m (and up to >16 m under SSP5-8.5 likely range), implying ≥6% of SIDS area annually flooded for high-end emissions, with severe impacts for Tuvalu, Marshall Islands, Cayman Islands, Bahamas, Maldives. Socioeconomic dynamics included: - Population projections: Under SSP1-2.6, EAPE is 28% lower than with climate change alone (population declines to 54 million by 2100); under SSP5-8.5, EAPE is 35% lower. Under SSP3-7.0, increased coastal population raises EAPE by 57% by 2100; Indian and Pacific SIDS see the largest increases in coastal populations. - Economic growth amplifies damages: 2100 EAD (US$ billions) including socioeconomic change: SSP1-1.9: 128.3 (4.4–225.3); SSP1-2.6: 140.9 (59.7–240.1); SSP2-4.5: 133.5 (66.6–224.3); SSP3-7.0: 81.48 (42.2–137.9); SSP5-8.5: 337.2 (198.8–526.4). Relative to 2100 GDP: 3.5% (SSP1-1.9), 3.8% (SSP1-2.6), 3.2% (SSP2-4.5), 6.6% (SSP5-8.5); these are 82%, 81%, 27%, and 104% higher, respectively, than climate-only effects due to coastal concentration of assets. SSP3-7.0 is the exception, showing lower EAD/GDP than climate-only due to economic trajectories.

The study provides the first comprehensive, consistent global assessment of coastal flood risk for all SIDS, quantifying present and future flooded areas, population exposure, and direct economic damages across emissions pathways, and explicitly comparing climate-only impacts with those including socioeconomic change. The findings confirm that even with stringent mitigation to 1.5 °C, SIDS face a more-than order-of-magnitude increase in coastal flood risk by 2100, with several countries experiencing losses that represent substantial shares of GDP, likely necessitating difficult adaptation measures and potential managed retreat in low-lying zones. The analysis demonstrates significant benefits of mitigation, especially in low-lying islands and certain regions, but also highlights residual risks that require substantial adaptation investments. Results emphasize the critical role of natural defenses (reefs, mangroves) and the need to preserve and manage them, improvements in coastal elevation/bathymetry datasets, and the importance of socioeconomic development patterns (urbanization and asset concentration in coastal areas) in shaping risk. The framework can guide identification of hotspots and prioritization of adaptation and loss-and-damage support, while underlining that local-scale decisions need more detailed, place-based studies.

This work advances coastal risk assessment for SIDS by integrating probabilistic sea-level projections, tide–surge–wave interactions, ecosystem attenuation, and 2D hydraulic inundation with exposure and vulnerability to estimate present and future EAFA, EAPE, and EAD under five SSP scenarios. It quantifies large and rising risks, substantial mitigation benefits (up to and beyond mid-century), and significant residual damages even under 1.5 °C, underscoring urgent needs for adaptation and targeted financial support for loss and damage. To maintain damages at today’s levels by 2100, many SIDS would need upgraded protection capable of withstanding >1 m increases in extreme sea levels under high emissions. Future research should: improve topo-bathymetric data and high-resolution DEMs; develop dynamic models of reef and mangrove evolution and their protective capacities under warming and acidification; better resolve future changes in storm surges and waves; refine coastal protection inventories and standards; integrate adaptive capacity, governance, and human behavior dynamics; and robustly evaluate costs, benefits, and feasibility of nature-based and engineered adaptations, including relocation strategies.

As a large-scale assessment, results are constrained by epistemic, computational, and data limitations and should not be used directly for local adaptation decisions. Key limitations include: reliance on global DEMs despite improvements (remaining vertical errors and representation of coastal micro-topography and channels); uncertainty in future storm surge and wave projections (future meteorological components largely held constant relative to SLR); proxy-based estimation of current coastal protection standards with reverse calibration where needed; assumption of static vulnerability and no improvement or degradation in natural and artificial defenses; assumption that reefs and mangroves maintain their protective capacity despite potential bleaching, degradation, and other stressors; and uncertainty from non-linear tide–surge–wave interactions and from socioeconomic projections. These render estimates conservative in some respects (e.g., if ecosystems degrade, impacts would likely be higher).