During major storm events such as hurricanes, emergency managers rely on fast and accurate forecasting models to make important decisions concerning public safety. These models can be computationally costly and cannot quickly make predictions at the highest geospatial resolution. However, model output can be post-processed to mimic high-resolution results with minimal additional computational cost. This research proposes methods for improvement in the accuracy of downscaling (enhancing the resolution of) a real-time storm surge forecasting model. Such improvements to downscaling methods include 1) expansion in its spatial applicability, 2) adding physics using water surface slopes, and 3) adding physics using friction losses across the ground surface.
This research builds upon a process that uses maximum water elevation output from the Advanced Circulation (ADCIRC) model and downscales these results to a finer resolution by extrapolating the water levels to small-scale topography. This downscaling process is referred to as the static method. The method was originally designed for use in North Carolina (NC), where results from an ADCIRC model designed specifically for NC were downscaled to a set of NC topographical data. By joining the static method with an ADCIRC output visualization tool, the downscaling process is now able to run faster with the same level of accuracy and can run on any ADCIRC model with downscaling data from any geographical region or given resolution. This process is used to provide extra guidance to emergency managers and decision makers during hurricanes.
The downscaling process is also improved by adding physics using the slopes method and the head loss method. The slopes method incorporates the slopes of the water levels produced by ADCIRC, rather than only the value of the water level. By interpolating ADCIRC output water elevation points into a smooth surface, slopes of this surface can be used to influence the elevations of downscaled water levels. The head loss method adds friction loss due to variations in the ground surface based on land cover types and friction associated with each type. As water travels over any surface, head loss, or a loss in energy, occurs at different rates depending on the surface roughness. This rudimentary hydrologic principle is applied to increase the accuracy of the downscaling process at minimal cost. The downscaling methods are applied for results from an ADCIRC simulation used in real-time forecasting, and then compared with results from an ADCIRC simulation with 10 times more resolution in Carteret County, NC. The static method tends to over-estimate the flood extents, and the slopes method is similar. However, the head loss method generates a downscaled flooding extent that is a close match to the predictions from the higher-resolution, full-physics model.
By improving the accuracy of downscaling methods at minimal computational cost and expanding the applicability of these downscaling methods, these methods can be used by emergency managers to provide a better estimation of flooding extents while simulating storm events.
CA Rucker (2020). “Improving the Accuracy of a Real-Time ADCIRC Storm Surge Downscaling Model,” North Carolina State University.
Hurricanes or typhoons are multi-hazard events that usually result in strong winds, storm surge, waves, and debris flow. A community-level multi-hazard hurricane risk analysis approach is proposed herein to account for the combined impacts of hazards driven by hurricanes including surge, wave, and wind. A tightly coupled ADCIRC and SWAN model is used to account for the surge and wave hazard. Community-level exposure analysis is conducted using a portfolio of building archetypes associated with each hazard. A building-level hurricane vulnerability model is developed using fragility functions to account for content, building envelope, and structural damage. These fragility functions calculate the exceedance probability of predefined damage states associated with each hazard. Then, a building damage state is calculated based on the maximum probability of being in each damage state corresponding to each hazard. The proposed hurricane risk model is then applied to Waveland, Mississippi, a community that was severely impacted by Hurricane Katrina in 2005. The main contribution of this research is modeling the community-level hurricane vulnerability in terms of damage to the building envelope and interior contents driven by surge, wave, and wind using fragility functions to provide a comprehensive model for resilience-informed decision-making.
As coastal circulation models have evolved to predict storm-induced flooding, they must include progressively more overland regions that are normally dry, to where now it is possible for more than half of the domain to be needed in none or only some of the computations. While this evolution has improved real-time forecasting and long-term mitigation of coastal flooding, it poses a problem for parallelization in an HPC environment, especially for static paradigms in which the workload is balanced only at the start of the simulation. In this study, a dynamic rebalancing of computational work is developed for a finite-element-based, shallow-water, ocean circulation model of extensive overland flooding. The implementation has a low overhead cost, and we demonstrate a realistic hurricane-forced coastal flooding simulation can achieve peak speed-ups near 45% over the static case, thus operating now at 80−90% efficiency.
During coastal storms, forecasters and researchers use numerical models to predict the magnitude and extent of coastal flooding. These models must represent the large regions that may be affected by a storm, and thus, they can be computationally costly and may not use the highest geospatial resolution. However, predicted flood extents can be downscaled (by increasing resolution) as a post-processing step. Existing downscaling methods use either a static extrapolation of the flooding as a flat surface, or rely on subsequent simulations with nested, full-physics models at higher resolution. This research explores a middle way, in which the downscaling includes simplified physics to improve accuracy. Using results from a state-of-the-art model, we downscale its flood predictions with three methods: (1) static, in which the water surface elevations are extrapolated horizontally until they intersect the ground surface; (2) slopes, in which the gradient of the water surface is used; and (3) head loss, which accounts for energy losses due to land cover characteristics. The downscaling methods are then evaluated for forecasts and hindcasts of Hurricane Florence (2018), which caused widespread flooding in North Carolina. The static and slopes methods tend to over-estimate the flood extents. However, the head loss method generates a downscaled flooding extent that is a close match to the predictions from a higher-resolution, full-physics model. These results are encouraging for the use of these downscaling methods to support decision-making during coastal storms.
Predicting building damage as a function of hurricane hazards, building attributes, and the interaction between hazard and building attributes is a key to understanding how significant interaction reflects variation hazard intensity effect on damage based on building attribute levels. This paper develops multihazard hurricane fragility models for wood structure homes considering interaction between hazard and building attributes. Fragility models are developed for ordered categorical damage states (DS) and binary collapse / no collapse. Exterior physical damage and building attributes from rapid assessment in coastal Mississippi following Hurricane Katrina (2005), high-resolution numerical hindcast hazard intensities from the Simulating WAves Nearshore and ADvanced CIRCulation (SWAN+ADCIRC) models, and base flood elevation values are used as model input. Leave-one-out cross-validation (LOOCV) is used to evaluate model prediction accuracy. Eleven and forty-nine combinations of global damage response variables and main explanatory variables, respectively, were investigated and evaluated. Of these models, one DS and one collapse model met the rejection criteria. These models were refitted considering interaction terms. Maximum 3-s gust wind speed and maximum significant wave height were found to be factors that significantly affect damage. The interaction between maximum significant wave height and number of stories was the significant interaction term for the DS and collapse models. For every 0.3 m (0.98 ft) increase in maximum significant wave height, the estimated odds of being in a higher rather than in a lower damage state for DS model were found to be 1.95 times greater for one- rather than for two-story buildings. For every 0.3 m (0.98 ft) increase in maximum significant wave height, the estimated odds of collapse were found to be 2.23 times greater for one- rather than for two-story buildings. Model prediction accuracy was 84% and 91% for DS and collapse models, respectively. This paper does not consider the full hazard intensity experienced in Hurricane Katrina; rather, it focuses on single-family homes in a defined study area subjected to wind, wave, and storm surge hazards. Thus, the findings of this paper are not applicable for events with hazards that exceed those experienced in the study area, from which the models were derived.
This research describes a method to improve the accuracy and efficiency of coastal flooding predictions. First, an existing model is used to explore the effect of storm forward speed and timing on tides and storm surge during Hurricane Matthew (2016). It is hypothesized that the spatial variability of Matthew’s effects on total water levels is due to the surge interacting nonlinearly with tides. If the storm occurred a few hours earlier or later, then the largest surges would have been shifted to other regions of the U.S. southeast coast. A change in forward speed of the storm also should alter its associated flooding due to differences in the duration over which the storm impacts the coastal waters. If the storm had moved faster, then the peak water levels would have increased along the coast, but the overall volume of inundation would have decreased. Then this research explores ways to increase the model’s accuracy and efficiency. To better represent Matthew’s effects, a mesh with detailed coverage of the coastal regions from Florida to North Carolina was developed by combining regional meshes originally developed for floodplain mapping. Compared to predictions using the earlier model, the new mesh allows for simulations of inundation that better match to observations especially inland.
Then, to best utilize this new mesh, a multi-resolution approach is implemented to use meshes of varying resolution when and where it is required. It is hypothesized that by `switching’ from coarse- to fine-resolution meshes, with the resolution in the fine mesh concentrated only at specific coastal regions influenced by the storm at that point in time, both accuracy and computational gains can be achieved. As the storm approaches the coastline and the landfall location becomes more certain, the simulation will switch to a fine-resolution mesh that describes the coastal features in that region. Application of the approach during Hurricanes Matthew and Florence revealed the predictions to improve in both accuracy and efficiency, as compared to that from single simulations on coarse- and fine-resolution meshes, respectively.
Barrier islands are susceptible to erosion, overwash, and breaching during intense storms. However, these processes are not represented typically in large-domain models for storm surge and coastal inundation. In this study, we explore the requirements for bridging the gap between dune-scale morphodynamic and region-scale flooding models. A high-resolution XBeach model is developed to represent the morphodynamics during Hurricane Isabel (2003) in the North Carolina (NC) Outer Banks. The model domain is extended to more than 30 km of Hatteras Island and is thus larger than in previous studies. The predicted dune erosion is in good agreement with post-storm observed topography, and an ‘‘excellent’’ Skill Score of 0.59 is obtained on this large domain. Sensitivity studies show the morphodynamic model accuracy is decreased as the mesh spacing is coarsened in the cross-shore direction, but the results are less sensitive to the alongshore resolution. A new metric to assess model skill, Water Overpassing Area (WOA), is introduced to account for the available flow pathway over the dune crest. Together, these findings allow for upscaled parameterizations of erosion in larger-domain models. The updated topography, obtained from XBeach prediction, is applied in a region-scale flooding model, thus allowing for enhanced flooding predictions in communities along the Outer Banks. It is found that, even using a fixed topography in region-scale model, the flooding predictions are improved significantly when post-storm topography from XBeach is implemented. These findings can be generalized to similar barrier island systems, which are common along the U.S. Gulf and Atlantic coasts.
Missing building attributes are problematic for development of data-based fragility models. Relative to other disciplines, the application of imputation techniques is limited in the field of engineering. Current imputation techniques to replace missing building attributes lack evaluations of imputation model performance, which ensure accuracy and validity of the imputed data. This paper presents two imputation approaches, along with imputation diagnostic and comparison approaches, for binary building attribute data with missing observations. Predictive mean matching (PMM) and multiple imputation (MI) are used to impute foundation type and number of stories attributes. The diagnostic approach, based on the logistic regression goodness-of-fit test, is used to evaluate the imputation model fit. The comparison approach, based on the percentage of correctly imputed observations, is used to evaluate the imputation model performance. A data set of single-family homes damaged by the 2005 Hurricane Katrina is used to demonstrate implementation of the methodology. Based on the comparison approach, PMM models showed 9% and 2% greater accuracy than MI models in imputing foundation type and number of stories, respectively.