Hurricane Matthew could leave 7 million in the dark

JACKSONVILLE, Fla. – Hurricane Matthew could knock out power for approximately 7 million people in the United States in a wide swath stretching from Miami to the Carolinas.

That's according to power outage forecasts by researchers at the University of Michigan, Ohio State University and Texas A&M University. Current as of 2 a.m. Eastern Daylight Time Wednesday morning, the forecasts are likely to change quickly as the storm progresses, and will be updated frequently on the team's website.

"We'll be running new power outage forecasts every six hours," said Seth Guikema, U-M associate professor of industrial and operations engineering. "With a storm this size, a small wobble isn't going to change things very much. Even if it stays over the ocean, there are likely to be substantial power outage impacts."

The forecast shows the storm affecting Florida by Thursday night and the Carolinas by Saturday evening.

To make the forecasts, the researchers have developed a predictive model that begins with the National Hurricane Center's weather forecast and uses data like population density, tree cover and soil moisture levels to calculate the probability of a power outage in a given area.

The team—which also includes Steven Quiring at OSU and Brent McRoberts at Texas A&M—has been making these forecasts for a decade. Their model accurately predicted that superstorm Sandy would knock out power for nearly 10 million people in 2012.

One of the focal points of the model is how the weather is likely to affect trees.

"When strong winds hit an area, especially when the ground is wet, trees fall over and pull power lines down and break poles and supporting structures," Guikema said. "Flooding causes outages too, but mostly it's the interaction between the wind and trees, except for the coastal margin. Wetter soil makes it more likely that trees will fall over."

The power outage model considers the maximum three-second wind gust at each census tract, as well as how long they can expect winds in a particular area to stay above 45 mph. It also factors in data from the U.S. Department of Agriculture about the amount and type of tree cover, and soil moisture models from the University of Washington. When combined with census data, this can predict which areas will be affected by power outages.

The team is sharing detailed models with utility companies along the eastern seaboard to help them deploy repair crews and other resources before, during and after the storm. The models are part of a larger project that aims to predict the probability of power outages from a wide variety of events, including less severe but more frequent incidents like thunderstorms, heatwaves and blizzards. Ultimately, Guikema envisions a rolling model, updated daily, that utility companies could use to allocate resources on a day-to-day basis.