Global survey shows planners use widely varying sea-level rise projections for coastal adaptation

The study addresses how sea-level rise (SLR) projections are being selected and applied by coastal practitioners in planning and adaptation decisions worldwide. With hundreds of millions living in coastal zones and substantial anticipated adaptation costs, the choice of SLR scenarios affects societal, ecological, and economic outcomes, and involves risks of over- or under-investment. Sea-level science has advanced substantially through successive IPCC assessments, expanding from median estimates to include high-end scenarios and uncertainty drivers such as ice-sheet sensitivity and tipping points. However, growing uncertainty complicates decision analysis and requires better translation to practice. Despite decades of guidance and a growing body of adaptation planning, there has been limited assessment of the actual SLR scenarios used by practitioners and how these inform the science-policy interface. This study disseminated a global, multilingual survey to coastal managers to document time horizons, projection structures and values, sources of science, and application contexts, aiming to reveal patterns of use and standardization in SLR projections.

The paper situates its inquiry within extensive sea-level and adaptation literature, including IPCC assessments (AR5, SROCC, AR6) and research on high-end SLR, uncertainty, and decision frameworks. Prior work emphasizes evolving understanding of ice-sheet contributions and the need to incorporate widening uncertainties, including high-end scenarios, into risk management. Guidance since the 1990s (IPCC coastal guidance, national frameworks) and regional efforts have promoted adaptation, while climate services literature highlights the usability gap and the importance of co-production and boundary organizations. Few studies have systematically assessed how European or global coastal managers select and use SLR projections; existing works point to heterogeneous approaches and the need for better integration of multiple scenarios and robust decision methods.

Design: A global, confidential questionnaire survey targeted coastal managers involved in SLR planning across all habitable continents. The survey (Qualtrics) ran from November 2020 to August 2021 in English and eight professionally translated languages (Arabic, Chinese, French, Hebrew, Japanese, Korean, Portuguese, Spanish), verified by native speakers. Sampling and recruitment: Snowball sampling was used due to challenges defining a complete sampling frame. Regional leads engaged personal networks, diverse seeding within regions, trust-building contacts, persistent follow-ups, two sampling waves, and multiple response modes (online, PDF, phone). Additional targeting of cities conducting SLR planning leveraged prior publications and adaptation portals. Despite efforts, gaps and uneven representation remained. Sample: 253 managers from 49 countries responded: Africa (10), Asia (39), Europe (31), North America (126), Australia/Oceania (44), South America (3). Jurisdictional levels: local (65%), sub-national (24%), national (9.6%), with populations ranging from small towns to large cities and states/countries, totaling over 1 billion people represented. Questionnaire structure: For users of SLR projections (22 questions, four sections): (1) inclusion of future sea levels in planning; (2) details of local policies, development dates, regulatory strength, 2050/2100 projections, and consideration beyond 2100; (3) scientific basis and processes included; (4) application of projections, decision criteria, planning approaches, and update frequency. Non-users answered five questions on coastal planning and hazard engagement. Analysis: Given non-random sampling, analyses were descriptive and qualitative. The team summarized geographic patterns of use, projection data structures (A: single; B: low/high; C: low/intermediate/high; D: low/intermediate/high/high-end), and projection values, handling anomalies (e.g., replacing nonsensical zeros with NA and relabeling three structures to reflect data). Spatial assessments were conducted by continent and region; internal consistency across structures and external consistency with IPCC AR5 and SROCC ranges were evaluated.

• Sample and use groups: 253 respondents from 49 countries; 181 (72%) in Group 1 with formal plans including SLR projections; 67 (26%) Group 2 trying to use projections without formal policy; 5 (2%) Group 3 not using SLR projections.
• Continental patterns (Group 1 share): Europe 87% (N=31), Australia/Oceania 84% (N=44), North America 77% (N=126); Africa 50% (N=10); Asia 36% (N=39); South America 33% (N=3). Regional contrasts include high use in North/West Europe (95%) and lower in Southern Europe (50%), and a US vs Caribbean dichotomy (US 80% vs Caribbean Islands 20%).
• Country examples: New Zealand 90% (N=10) and the United Kingdom 100% (N=8) of respondents use SLR projections; Japan reported 80% (N=5) not using SLR in planning at survey time (policy has since changed to include SLR). Western Africa respondents reported no use in planning.
• Projection data structures (among 143 using formal structures): Structure A (single estimate) 76 respondents (53.1%); Structure B (low/high) 20 (14.0%); Structure C (low/intermediate/high) 28 (19.6%); Structure D (low/intermediate/high/high-end) 19 (13.3%). Structure A is the majority on every continent (e.g., Oceania 78.6%, Asia 72.7%, Africa 66.7%). Forty additional respondents used bespoke local structures.
• High-end scenario usage (Structure D): Observed in the United States (17 locations), Northern/Western Europe, New Zealand/Australia, and Northern Africa; not observed elsewhere in the sample. Canada vs USA contrast: Canada largely uses single estimates (A: 84%; C: 16%), while the USA exhibits diverse structures (A 24%, B 19%, C 28%, D 29%).
• 2100 projection values (N=135 across structures; rounded): • Structure A (N=71): median 0.90 m; min 0 m (8 locations); max 2.03 m (Hayward, California, USA). • Structure B (N=19): median low 0.61 m; median high 1.40 m. • Structure C (N=26): median low 0.42 m; median intermediate 0.71 m; median high 1.21 m. • Structure D (N=19): median low 0.53 m; median intermediate 1.19 m; median high 1.91 m; median high-end 3.05 m.
• Comparison to IPCC AR5 and SROCC: Many planning values for 2100 fall below or above the IPCC likely ranges; 119 responses exceeded the AR5/SROCC RCP8.5 global value of 0.98 m, reflecting regional guidance, relative SLR, or inclusion of higher-end possibilities with low IPCC confidence at the time.
• Inference: Countries with robust national guidance and longer adaptation histories (e.g., UK, New Zealand) show greater assimilation of SLR projections. There is no global standard; many practitioners rely on single values despite best-practice guidance favoring multiple scenarios and risk-based methods.

The first global survey of coastal practitioners reveals broad incorporation of SLR projections into planning, yet substantial heterogeneity and a prevalent reliance on single estimates. While single values are necessary for later-stage engineering design, current best practice recommends multiple scenarios aligned with different climate futures and risk-based, robust adaptation approaches. High-end scenarios can be valuable for stress-testing and bounding long-term options but may be misapplied, potentially prompting unnecessarily costly or disruptive measures. Variability in approaches likely reflects differences in decision contexts (risk tolerance, asset lifetimes, planning horizons), sources of projections (national/state guidance, co-produced science, authoritative guidance), and regional conditions (e.g., relative SLR from subsidence). Many reported values diverge from IPCC global ranges due to regionalization, relative SLR adjustments, or inclusion of high-end estimates. The unequal geographic representation and non-random sampling limit inference about global drivers (e.g., GDP, education), though patterns suggest that strong national guidance and adaptation histories support better integration. Improving climate services, translation of uncertainty, and peer learning can enhance the usability and appropriate application of SLR science.

This study provides the first global dataset and assessment of how coastal practitioners select and use SLR projections for adaptation planning. It documents widespread use but highly inconsistent practices and a frequent reliance on single-value projections, with limited global standardization. The findings underscore the need for clearer guidance on multiple-scenario use, appropriate application of high-end scenarios, and improved science-policy translation through climate services and co-production. Future research should expand sample diversity, link specific decision contexts and use cases to projection structures and values, examine how national guidance filters into local decisions, and analyze planning documents to clarify interpretation and application of SLR guidance.

• Sampling: Snowball sampling is non-random and subject to bias; responses are unevenly distributed geographically and dominated by North America, limiting generalizability.
• Representation: Under-representation of regions such as the Caribbean, Latin America, Africa, and Southeast Asia; potential cultural and resource-related nonresponse.
• Measurement: Respondents may have interpreted requested values differently (e.g., inclusion of storm surge, regional sea-level signals, vertical land motion) versus mean sea-level change; some anomalies required data cleaning (e.g., zero values set to NA).
• Linkage to decisions: The survey design did not allow robust matching of specific use cases or decision contexts (risk tolerance, planning horizons) to the projection structures/values reported.
• Source attribution: Limited ability to trace specific guidance documents or scientific sources underlying reported projections; regionalization effects not systematically assessed.
• Temporal changes: Policies can change after survey completion (e.g., Japan), so findings reflect the survey period (2020–2021).