Kyle Mandli | Building Predictive Models for Real-Time Storm Surge Impact
Though the mechanisms that cause them are vastly different, tsunamis and storm surge are both dangerous geophysical phenomena that can lead to destructive flooding and significant damage to coastal areas.
Assistant Professor of Applied Mathematics
—Photo by Eileen Barroso
Tsunamis, which can be caused by earthquakes, mudslides, or volcanic eruptions, send swells of ocean water, sometimes reaching heights of more than 100 feet, onto land. They can spread hundreds of feet inland, causing devastating property damage and loss of life in populated areas.
Storm surge is an abnormal rise in sea level, generated by a storm system. When the surge enters shallow water, it is driven ashore by storm winds. The flooding caused by this surge can erode beaches and dunes and damage infrastructure.
The severity of each is affected by underwater topography, namely the shallowness of the water in its path.
Kyle Mandli, assistant professor of applied mathematics, is working to develop better mathematical representations and computer models of these phenomena in order to predict their potential impact, help emergency managers make critical evacuation decisions, and evaluate the effectiveness of protection measures, such as levees. He examines the physics behind why these phenomena intensify in shallow water and applies advanced algorithms to quickly and accurately predict these events.
“My research specifically involves adapting current models so they can be easily analyzed in a depth-averaged context and implementing robust and efficient solvers (mathematical software)—including accelerator technology, such as graphics cards—to simulate these flows,” explains Mandli. “We must be careful when we add additional detail to our simulations, as adding too much will make it so that we cannot compute a tsunami or storm surge fast enough to make a reliable prediction.”
An example of how Mandli and his colleagues approach this research is their examination of a scenario in which a hurricane is approaching the coast and there is limited time to provide forecasts showing where storm surge flooding will occur and how bad it will be. While such forecasting is critical for emergency managers, there are a number of unknown factors that can make the task difficult.
“These may include the hurricane's trajectory and intensity, how much on-land rainfall is contributing to flooding, and the current location of sandbars,” explains Mandli. “In order to handle these uncertainties, we run multiple simulations that try to capture variations in the storm’s track, maximum wind speeds, and more.”
While he has always been fascinated by the ocean and its dynamics, Mandli says his real interest in this research stemmed from the fact that he started graduate school right around the time the 2004 Indian Ocean earthquake and resulting tsunami hit Sumatra, Indonesia.
“One of my fellow graduate students was already studying tsunamis and I became very interested in his work,” explains Mandli. “I wanted to improve the models we were using to predict vulnerabilities to storms and tsunamis and provide means for emergency personnel to make informed decisions. Currently, we’re also attempting to model storms based on future climate change and the effects elements such as sea level rise might have on storm surge and its resulting damage.”
Prior to joining Columbia, Mandli was a research associate in the computational hydraulics group at the Institute for Computational and Engineering Sciences at the University of Texas–Austin. He is a member of the Society for Industrial and Applied Mathematics (SIAM) and the American Geophysical Union (AGU).
BS, University of Wisconsin, 2004; MSc, University of Washington, 2005; PhD, University of Washington, 2011
—by Amy Biemiller