A Missing Link in Predicting Hurricane Damage

By Donatella Pasqualini

Aftermath of Hurricane Matthew, 2016 Credit: Paul Brennan Pixabay

There’s an old saying that generals always fight the last war. In communities hit by hurricanes, it could be said that governments often fight the last storm.

That situation is understandable. When a hurricane moves through a city, enormous amounts of data are collected about its path and any storm surge; other related flooding; or power outages. Each hurricane that passes through our coastal communities is followed by a series of “fixes” that draw from those past data to make our power plants and other key facilities more resilient. Yet what if such fixes aren’t really fixes at all?

This is an important question for local governments, utilities and the military, which stands to lose 128 military bases to eroding coastlines, according to a 2016 report by the Union of Concerned Scientists. Knowing which mitigation strategies to pursue is critical for protecting communities and valuable assets — but it’s not easy.

For example, after Hurricane Sandy devastated the Atlantic coast in 2012, resulting in some $70 billion worth of damage and power outages for about eight million customers across 21 states, the Delaware Bay area’s major utility spent billions of dollars to shore up electrical substations that were flooded by the storm’s water surge. According to a model we’ve developed at Los Alamos National Laboratory, however, those fixes might not help when the next big storm hits.

The reason for this problem is that existing storm-impact models are missing a critical piece of data: eroding coastlines. The Delaware estuary loses a full acre of coastline each day, which significantly impacts how quickly and where water will rise. But incorporating data about coastlines into models is extremely difficult because the variables that cause erosion are interrelated: the type of soil along a coastline and its salinity influence what kind of vegetation will grow there, which, in turn, impacts how rising flood waters will move — which impacts erosion, which impacts the soil and salinity, and so on. This interdependence makes the problems nonlinear. And nonlinear problems are very difficult to solve computationally.

Using supercomputers at Los Alamos, we’ve developed a model that processes data about the physical characteristics of coastlines — including soil, vegetation and sensitivity to erosion — and how those characteristics would interact with a hurricane and its storm surge.

Although many studies have been conducted in laboratory settings to understand changing coastlines, the limitations of lab research mean that the resulting models are only relevant for a very small geographical area. To overcome this issue, Los Alamos has developed a model that describes the physics of changing coastlines. Because it’s based on physics and not a single geographical area, the model can be applied on a regional scale and include any coastal city.

Eroding coastlines are only one piece of the puzzle. To get the full picture, we integrate this model with several others — for ocean, vegetation, land and electrical power — along with simulationsto show how extreme weather propagates through the networks — to create a physics-based, fully automated framework that can be employed the next time a hurricane is barreling toward a coast. By coupling details of the storm’s rate of travel with all the other data, the framework would approximate its impact on critical infrastructure, such as electrical substations and water treatment facilities.

But the framework can do more than assess the impact of an existing storm. It can also predict how future hurricanes will affect a community. By combining the data of 10,000 possible storms and predicted sea levels, it can forecast virtually every outcome — including how many people will be without power, which substations are more likely to flood and whether salt water might contaminate the water table and reduce the supply of drinking water.

Furthermore, the tool helps local governments and utilities determine which critical infrastructure they should prioritize when investing in protective measures. For example, if a utility knows that a power substation serving a large hospital is at risk of flooding, it can raise that substation or redesign the network to be able to reroute power to the hospital another way.

While these computations currently run on supercomputers, we will move the framework to the cloud in the coming year, so decision makers will be able to input the variables for their coastal areas and get the results they need to protect their critical infrastructure. By providing access to the model’s results, our goal is to get more data and feedback to ultimately improve the accuracy of forecasts.

Our hope is that all of this will result in better preparedness for the next big storm. So instead of looking back and fighting the last war, we’ll be looking ahead — and ready for whatever comes next.

Donatella Pasqualini is a physicist in the Information Systems and Modeling group at Los Alamos National Laboratory. This work is funded by Los Alamos’s Laboratory Directed Research and Development program.

This story first appeared in Scientific American