Planning for a hurricane is a complicated process involving many stakeholders and varying degrees of uncertainty. Accurate predictions of storm surge and wave heights are vital to decision-making before, during and after the storm. Creating these predictions through modeling software can be expensive and time-consuming. When dealing with hurricanes, time is critical for emergency managers and other officials.
Helping decision-makers to save valuable prediction time is CRC Principal Investigator Dr. Casey Dietrich of North Carolina State University (NCSU). His project, “Improving the Efficiency of Wave and Surge Models via Adaptive Mesh Resolution,” involves collaboration with co-PI Dr. Clint Dawson at the University of Texas at Austin. Their project focuses on speeding up a widely used prediction tool, ADCIRC. His work with North Carolina Emergency Management during Hurricane Matthew in 2016, and his contributions to developing future disaster resilience specialists, have helped make significant contributions to disaster preparation and recovery.
Speeding up predictions
Dr. Dietrich is investigating new ways to optimize coastal ocean models as part of his CRC project to improve ADCIRC, a storm surge prediction tool used by state and federal emergency management officials, among others. The results of his research will make the tool faster, more efficient, and more accurate when predicting flooding from storms.
“We are working with how ADCIRC is employed on a parallel computer, and trying to better use the computing resources that are available so that we can make those predictions much faster,” Dr. Dietrich said.
Parallel computing is the division of a larger simulation into smaller pieces, which are then solved faster by sharing the workload over many CPUs.
To improve the model’s run time and predictions, Dr. Dietrich’s team is applying an adaptive mesh resolution – in which the model’s representation of the coastal ocean environment will change as an approaching storm evolves – to ADCIRC. This change from the current, static mesh will reduce the computing power needed to run the model, freeing up resources to allow ADCIRC to run faster and provide more accurate predictions.
The current goal is for ADCIRC forecasts to be simulated within about 60 minutes per advisory. By streamlining ADCIRC’s computing process, Dr. Dietrich estimates his work will shave 15 to 30 minutes from the previous prediction time, giving emergency managers valuable decision-making time during storms. This will allow communities to respond more quickly to changing weather conditions when deciding where to deploy resources and aid and whether to evacuate.
Speeding up the ADCIRC tool will not only save time when making critical decisions during storms, but money as well. Since models like ADCIRC can be expensive to operate, the project will reduce the costs associated with running it over time. Improving the model also has long-term benefits for planners, who can use the predictions between storms to set flood insurance rates and design more resiliently for the future.
“We think every ADCIRC user will benefit from this project,” Dr. Dietrich said, “A faster ADCIRC will be better not just in forecasting, but also for long-term engineering, design and planning.”
During the 2016 hurricane season, Dr. Dietrich was among several CRC researchers assisting emergency managers and hurricane forecasters with tracking Hurricane Matthew. Dr. Dietrich visited the North Carolina Emergency Management (NCEM) State Emergency Operations Center to observe how emergency managers there use ADCIRC during immediate storm response. He said the professionals described ADCIRC as their “eyes on the coast,” and were excited about the prediction data the model provided. More about how CRC researchers aided in storm impact predictions and immediate response can be found here.
In addition to Dr. Dietrich’s work with North Carolina emergency managers during the storm, he has also worked to understand how Matthew affected water levels along the entire southeast U.S. coastline, from Florida through North Carolina. The researchers are using data from hundreds of observations during the storm to validate water levels predicted by the ADCIRC model, turning the disaster into an opportunity to improve the model’s accuracy for future storm predictions. They found that storm surge, when combined with tides, led to uneven increases in the water levels into estuaries along the Georgia and South Carolina coasts, and that the flooding would have been very different if the storm had occurred earlier or moved faster. More about how CRC researchers have aided Matthew recovery efforts can be found here.
Dr. Dietrich also helped train future disaster resilience specialists in 2017 through his leadership in the second annual SUMmer Research Experience (SUMREX) Program. This program invites students from CRC Education and Workforce Development (E&WD) partner universities to participate in CRC research projects in an immersive summer research internship lasting up to 10 weeks. Over the course of the internship, these students engage in meaningful, hands-on research experiences that can lead to future graduate studies and careers in disaster resilience fields.
Part of this research experience included a one-day exchange activity where students from a Johnson C. Smith University (JCSU) education project met with Dr. Dietrich. Nine of JCSU’s students visited the Department of Civil, Construction, and Environmental Engineering at NCSU, where they learned about computing-intensive coastal resilience research. Dr. Dietrich led the JCSU students through presentations by faculty members and discussions with his own graduate students, where they learned more about using computational tools to address various research problems in civil and environmental engineering.
“We had the opportunity to go around to each student’s work area and hear their stories on what they all created,” said Imyer Majors, then a computer engineering major at JCSU. “I love the honesty they gave on the difficulties they were faced with in certain areas of their projects, and how they were able to think of different ways to solve them.”