Improving Accuracy of Real-Time Storm Surge Inundation Predictions

Emergency managers rely on fast and accurate storm surge predictions from numerical models to make decisions and estimate damages during storm events. One of the challenges for such models is providing a high level of resolution along the coast without significantly increasing the computational time. Models with large domains, such as the ADvanced CIRCulation (ADCIRC) model used in this study, are accurate in predicting water levels and their variation in complex coastal regions, however their spatial resolution may limit their predictions of flooding at the scale of buildings, roadways, and critical infrastructure.

A new tool has been developed that uses Geographic Information System (GIS) scripts to enhance the resolution of maximum water level predictions at the boundary of predicted flooding using a high-resolution Digital Elevation Model (DEM). The water levels predicted by the lower resolution model are extrapolated outward to where the water would intersect with the higher resolution elevation dataset. The result is a highly-refined flooding boundary that represents inundation on scales smaller than the typical ADCIRC mesh resolution. This tool can process a 15-m DEM for all 32 coastal counties of the state of North Carolina in less than 15 minutes during a storm event.

Comparison of results using spatial building datasets showed that for a simulation of Hurricane Matthew, 2,353 buildings were predicted to be flooded in Carteret County, NC, prior to enhancing resolution and 3,298 post-enhancement, an increase of 40 percent. In Dare County, the increase was 22 percent. This dramatic increase in flooded buildings shows the importance of achieving high accuracy in floodplains, as a relatively small change in predicted flooding extent can have a substantial impact on the predicted number of flooded buildings. The validity of these results was tested via comparisons to results of an ADCIRC model with the same 15-m resolution as the DEM in Dare County. Dare County is a coastal region with widely-varying topography and land cover, and preliminary comparisons have shown that the GIS method is accurate in coastal regions with steeper slopes and less accurate in flatter, low-lying areas.

N Tull (2018). “Improving Accuracy of Real-Time Storm Surge Inundation Predictions,” North Carolina State University.

Thesis Defense: Nelson Tull

Nelson Tull. “Improving Accuracy of Real-Time Storm Surge Inundation Predictions.” Department of Civil, Construction and Environmental Engineering, North Carolina State University, Raleigh, North Carolina, 23 May 2018.

Poster: ASBPA Coastal Conference 2017

N Tull, JC Dietrich, TE Langan, H Mitasova, BO Blanton, JG Fleming, RA Luettich. “Improving Accuracy of Real-Time Storm Surge Inundation Predictions Using GRASS GIS.” ASBPA National Coastal Conference 2017, Fort Lauderdale, Florida, 25 October 2017.

Poster presented by Nelson Tull at the ASBPA National Coastal Conference 2017.

Continue reading →

News: Enhancing Storm Surge Visualization

2017/10/04 – Science Node
The aim of storm surge models: When a storm approaches, emergency managers want fast and accurate forecasts

Although our model provides water level predictions from the deep ocean all the way to the coastal floodplains, the system is limited by the model’s resolution. Topographic features at scales smaller than 500 feet, such as roadways or narrow stream channels, are often not included in the models because of the computer time needed to produce such high-resolution outputs. Because of this limitation, the extent of flooding can be underpredicted by the model.

2017/08/31 – National Consortium for Data Science
Data Fellows project aims to make storm surge predictions faster and more accurate

Continuing our North Carolina Sea Grant project and with new support from the National Consortium of Data Science, we are developing a method to improve prediction of the true flooding extent by combining the results of our model with more accurate elevation datasets.

To perform this prediction of the flooding extent, we use a Geographic Information System (GIS) called GRASS GIS that specializes in processing very large amounts of data. The project has two major objectives. The first is to process the modeled water levels and the elevation data set together, producing a map showing the extent of predicted flooding. When the modeled water levels are greater than the land elevation, flooding extends outward into neighboring, unflooded areas in the data set. By mapping the model results to the higher resolution data sets on elevation, we can create more accurate surge forecasts of overland flooding.

2017/08/08 – NC Sea Grant Coastwatch Currents
Fast, Accurate Forecasts of Coastal Flooding: Enhancing Visualization of Storm Surge Guidance to Support Emergency Managers

Storm surge models must be both fast and accurate to give coastal communities the guidance they need to prepare for and respond to a storm. Perhaps just as important is the need for these forecasts to be visualized in a way that is meaningful and useable by emergency managers.

ADCIRC forecasts are currently visualized using Kalpana, a Python script that converts the model output into formats compatible with commonly-used visualization applications such as ArcGIS and Google Earth. With support from the National Consortium of Data Science (NCDS) and in partnership with North Carolina Emergency Management (NCEM), our team has developed a new visualization method that makes use of enhanced topographic resolution along the flooding boundary. This results in modeled storm surge extending farther into estuaries and floodplains, increasing the accuracy of the forecast.