Mid-Century Projections of Waves and Water Levels Highlight the Dominant Role of Sea-Level Rise in the Salish Sea
Abstract. Using high-resolution, ensemble-based modeling of the CMIP6 High Resolution Model Intercomparison Project, this study evaluates potential changes to extreme coastal water levels and waves across the Salish Sea, the largest estuary on the North American west coast. Results of a well-validated (water level error ~10 cm) regional hydrodynamic and wave model indicate that extreme water levels (e.g., 30-year return period) are projected to increase across the region, while lower-magnitude events (e.g., annual) are projected to slightly decrease. Wave height projections are spatially heterogeneous but generally decrease in southern basins and increase in the far north basins. Seasonal decomposition reveals a modest decrease in winter water levels and an increase in summer water levels. Non-linear interactions between sea level rise (SLR) and hydrodynamic response were assessed across seven SLR scenarios (0–3 m), revealing localized amplification effects up to 10 cm, though these remain minor relative to the SLR signal. Overall, uncertainty across the CMIP6 ensemble is high in this region, rendering most modeled changes statistically non-significant and emphasizing the need for studies to consider an ensemble when looking at climate projections. The dominance of tidal forcing in extreme water levels, coupled with high inter-model variability, indicates that near-term climate-induced changes to oceanic drivers are unlikely to significantly alter coastal hazards in the Salish Sea by 2050. These findings support the use of present-day conditions as a proxy for near-future planning and emphasize SLR as the primary contributor to changing future coastal risk in the Salish Sea.
This manuscript presents an ensemble-based assessment of projected changes in waves and extreme water levels in the Salish Sea using high-resolution CMIP6 HighResMIP forcing, regional hydrodynamic and wave models, and sea-level rise scenarios. The study is scientifically sound, methodologically rigorous, and addresses a topic of high relevance to coastal hazard assessment and adaptation planning. A strength is the explicit treatment of climate-model uncertainty through an ensemble approach, which leads to the important conclusion that mid-century changes in oceanic forcing are generally small relative to inter-model variability, while sea-level rise remains the dominant driver of future coastal hazards. The manuscript is generally well written, the figures are informative, and the conclusions are supported by the analyses. The resulting dataset will likely be valuable for a broad range of future coastal hazard studies. I support publication after revision.
Major Comments
- Model resolution. Given the highly complex topography and wind climatology of the Salish Sea, I am not fully convinced that atmospheric forcing with resolutions of 25-100 km can adequately represent the mesoscale wind systems that drive much of the local wave climate. While the use of HighResMIP represents an improvement over standard CMIP products and is understandable from a computational standpoint, the manuscript should discuss more explicitly the limitations associated with unresolved gap winds, topographic steering, and local fetch-dependent processes. This issue is likely more important for projected wave changes than for water-level projections, and may contribute to the large inter-model spread observed throughout the analysis
- Ensemble spread. The large spread across the HighResMIP ensemble is a key element of the manuscript's conclusions. However, I am not convinced that this spread can be interpreted solely as uncertainty in future climate projections. The ensemble includes models with substantially different atmospheric resolutions (25-100 km), which may differ greatly in their ability to represent the complex wind field of the Salish Sea. As a result, the ensemble spread likely combines climate uncertainty with structural and resolution-dependent modelling uncertainty. I suggest that the authors acknowledge this limitation and discuss how it may affect the interpretation of statistical significance and ensemble-derived uncertainty estimates
- Statistical significance and ensemble size. A central conclusion is that most projected changes are not statistically significant. However, significance testing is based on seven CMIP6 models. While the authors acknowledge this limitation, the manuscript would benefit from a more detailed discussion of the statistical implications of such a sample size. Because the principal policy-relevant conclusion depends on the absence of statistically significant change, additional justification would strengthen the manuscript.
- Physical mechanisms underlying projected changes. The manuscript clearly documents projected changes but provides less discussion of the underlying physical drivers. Additional process-based interpretation would improve the scientific impact of the paper.
- Justification for SSP5-8.5-only analysis. The rationale for using SSP5-8.5 is plausible, particularly given the focus on 2050. However, several readers may question the choice of a single emissions pathway. I encourage the authors to strengthen the discussion by citing literature demonstrating that differences in oceanic forcing among SSPs remain relatively small through mid-century and therefore do not materially affect the study conclusions.
- Interactions. The methodology used to quantify non-linear SLR effects should be described more explicitly. The paper states that the bulk SLR signal was removed and only the nonlinear contribution is plotted, but the exact calculation is not fully clear. Please provide the mathematical formulation used to derive the nonlinear component, including the reference simulation against which SLR-induced changes were computed. This clarification would improve reproducibility and interpretation of the figure. Moreover, I encourage the authors to be slightly more cautious when characterizing these effects as "minor" or "relatively insignificant." While a 10 cm amplification is small compared to 3 m of SLR at the basin scale, it may still be relevant for threshold-based applications such as flood-frequency mapping, infrastructure design, and ecological transition zone.