When a hurricane approaches land, forecasters predict its effects on the coastal ocean, such as how high the water will rise above the normal tides (in a process called storm surge) and which regions are likely to be flooded. These predictions require many computer simulations to account for uncertainties in the storm’s size, track, and intensity. To be fast, these simulations use simplified representations of the coastline and the ocean physics. Simulations with fine-scale representations have been shown to be more accurate, but they are far too slow on current supercomputers to use when time is limited to achieve reliable predictions. This trade-off has limited the accuracy of real-time simulations and increases the uncertainty for decision-makers and coastal residents.
This project will develop, test, and implement ways to embed fine-scale information into coarse-scale storm surge models using high-resolution elevation maps to correct mass balances, bottom friction, and other quantities. The resulting models will keep most of the high-resolution accuracy while having speeds comparable to the simpler coarse models which will lead to more accurate pre-storm simulations, improving decision-making for policy-makers, emergency management personnel, and coastal residents. The work performed in this project will not only enable increased accuracy in ensemble surge forecasts, but will also decrease computational costs for a given accuracy in higher resolution studies.
It will enable entirely new types of studies including decadal-level simulations using reanalysis products or climate model outputs. This approach also opens the way for dynamical global surge/tide simulations, which do not presently exist. Results will add little to costs, while significantly increasing accuracy. The project team will ensure adoption of these results by implementing findings into two widely-used storm surge models, by working in concert with a governmental-academic-industry advisory committee, and by disseminating results through existing model code repositories. Three graduate students and three undergraduate students per year will be trained. An immersive fluid mechanics theater will be developed both for undergraduate teaching, and as part of outreach programs for local schools.
Parameterizations for unresolved processes in numerical models are standard in fields as far ranging as turbulence and porous media transport, but are sorely lacking in coastal flooding applications. As in those fields, rigorous development of up-scaled models holds the potential for a transformative leap in the way surge models are used to forecast coastal inundation. By building a framework on a sound physical foundation, and by incorporating and adapting ideas from other fields, the project team will develop novel sub-grid methods that will be physically consistent, robust, and thus flexible for widespread use. Using established theoretical methodologies coupled with existing high-resolution data and new numerical simulations, the project team will develop scale-dependent closure corrections to mass and momentum balance equations. Sub-grid closures will span a hierarchy of three approaches with increasing complexity, ranging from hand-calculable simple closures to high-order multiscale numerical corrections.
This will allow for a user-chosen compromise between speed, accuracy, and data availability. By rigorously addressing this closure hierarchy, this project will develop a much stronger physical understanding of how very specific flow and land features impact hydrodynamics at different scales. Specifically, this research will lead to new insights on how coastal flooding is controlled by unresolved flows through marshes, natural channels, and man-made canals, and how best to model these unresolved scales. This will assist not only in forecast operations, but also in understanding and designing protective infrastructure.
AB Kennedy, D Bolster, D Wirasaet, JC Dietrich. “PREEVENTS Track 2: Collaborative Research: Subgrid-Scale Corrections to Increase the Accuracy and Efficiency of Storm Surge Models.” National Science Foundation, Directorate for Geosciences, Prediction of and Resilience against Extreme Events (PREEVENTS), 2017/09/01 to 2021/08/31, $1,252,526 (Dietrich: $320,001).
