A comparison of static versus dynamic models of sea-level rise impacts to atolls: Insights from the Northwestern Hawaiian Islands

Two inundation events in 2011 underscored the potential for elevated water levels to damage infrastructure and impact terrestrial ecosystems on the low-lying Northwestern Hawaiian Islands in the Papahānaumokuākea Marine National Monument. The goal of this study was to compare passive GIS-based "bathtub" inundation models to those that include dynamic water levels caused by wave-induced set-up and run-up for two end-member island morphologies: Midway, a classic atoll with islets on the shallow (2-8 m) atoll rim and a deep, central lagoon; and Laysan, which is characterized by a deep (20-30 m) atoll rim and an islet at the center of the atoll. Vulnerability to elevated water levels was assessed using US Army Corps of Engineers hindcast wind and wave data to drive coupled Delft3D wave, current, and water level models for the atolls. The resulting model data were then used to compute run-up elevations using a parametric run-up equation under both present conditions and future sea-level rise scenarios. In both geomorphologies, wave heights and wavelengths adjacent to the islet shorelines increased more than three times and four times, respectively, with increasing values of sea-level rise, as more deep-water wave energy could propagate over the atoll rim and larger wind-driven waves could develop on the atoll. Although these increases in water depth resulted in decreased set-up, the larger wave heights and longer wavelengths due to sea-level rise increased the resulting wave-induced run-up. Run-up values were spatially heterogeneous and dependent on the direction of incident wave direction, bathymetry, and islet configuration. Islet inundation was modeled to increase substantially when wave-driven effects were included, suggesting that inundation and impacts to infrastructure and terrestrial habitats will occur at lower values of predicted sea-level rise, and thus sooner in the 21st century, than suggested by passive GIS-based "bathtub" inundation models.