UC professor develops ways to improve flood forecasts
In February, the Ohio River reached levels of over 60 feet. Brown, murky water flooded into the city, drowning homes, playgrounds, parks and streets. Rocks and mud tumbled down hills. Sewers overflowed into streets. Paul Brown Stadium, home of the Cincinnati Bengals, stood alone like a castle, a moat of coffee-with-cream water converging around its perimeter.
To put 60 feet into context, the flood stage for the Ohio River is 52 feet, and the normal pool of the river is about 25 feet. At 35 feet over its average level, the Ohio River becomes a destructive force.
February, however, was not simply an anomaly. In 1937, an overflowing Ohio River showed just what it’s capable of. In January of that year, the river rose to more than 80 feet, taking 385 lives and displacing more than one million people. At the flood’s peak, the Ohio River covered one-fifth of the city of Cincinnati.
Since that fateful month in 1937, the Ohio River’s crest has filled over the flood stage 55 times, hitting that mark eight times in the last decade, according to the National Weather Service.
Though Congress signed the Flood Act of 1938, authorizing projects to build dams, locks and other flood-protection works, efforts to control nature have always had their limitations. At the end of the day, Mother Nature has the last word.
So, when floods happen, how do we better prepare for them? One solution lies in the way we predict floods.
University of Cincinnati assistant professor Patrick Ray, Ph.D., seeks to mitigate the uncertainty that comes with flood forecasting. He recently received a $40,000 grant for his project, “Visualization of Uncertainty of Forecast Component Parts.” The grant was administered by the National Oceanic and Atmospheric Administration and the University Corporation for Atmospheric Research.
“This project is about helping people to target their resources – their time, energy and money – to reducing the uncertainties in flood forecasting,” says Ray.
The National Weather Service employs hundreds of operational hydrologists to predict 5-day flooding forecasts based on the hundreds of data sets spread across each river basin. The hydrologists read and analyze this data and then send out public service announcements that notify citizens of potential flooding threats. The problem, explains Ray, is that this system is not that accurate.
“It’s an intuitive process,” Ray says. “Operational hydrologists have these computer models with variables they can adjust, but much of it is trial-and-error.”
The Ohio River Forecast Center uses computer models that use variables like precipitation, soil moisture and the hydrologic system to simulate water conditions in the region. Much of these simulations, however, are still riddled with uncertainty. Ray is proposing to make this data more accessible to forecasters through statistical analysis tools and data visualization techniques, empowering forecasters to reduce forecast errors.
“We’re coming in with statistical analysis to run simulations really fast for them ahead of time and help coach them,” Ray says.
“We’re not going to replace them,” he adds. “We still need the human intuition for successful flood forecasting.”
A flood haven
Months like February of this year are not uncommon in the Ohio River Basin. Flooding here is driven by the geography, weather patterns and sheer size of the region, Ray says. While the Ohio River is associated with historic river towns like Pittsburgh, Cincinnati and Louisville, it influences 14 states and includes a population of more than 25 million people.
The Ohio River Basin looks a bit like an upside down ice cream cone – the cone’s bottom nearly reaches Lake Erie, almost touching Buffalo, New York. The rest of the “cone” covers a good chunk of Western Pennsylvania, nearly all of West Virginia, minus the eastern panhandle, and the lower three-fourths of Ohio. The rest of the basin, the “ice cream” part, covers mostly all of Kentucky, Indiana, Tennessee and a slice of eastern Illinois and sneaks into the southern states of Mississippi, Alabama, Georgia and both Carolinas.
All the rain that falls on any of these areas eventually drains toward the Ohio River. It’s no surprise then why the Ohio River is so prone to flooding.
“We’re talking about a huge river with huge amounts of water,” says Ray. “And the big flood times are typically in the late winter and early spring.”
Ray explains that increased rainfall in the spring, compounded by the “rain-on-snow” events, where rain slides off snow and ice much faster than soil, creates ideal conditions for flooding. As winter turns to spring, that same snow melts and ends up in the river.
The earth’s groundwater is also more saturated in the spring. “In those few wet months, the sponge of the earth is sustaining the water in this river,” says Ray. “The Ohio River is so big, and there’s so much rain in the winter time, that we fill up the earth with water and it pours into here, causing flooding.”
In fact, in the wet season, the water flow of the Ohio River is ten times faster than the dry season of late summer.
With additional snowmelt, increased rainfall and nowhere for all that water to go, you get what happened in February: the largest Ohio River flooding in 20 years.
With several million people living right along the river, this becomes a problem.
Crunching numbers
On top of property loss and damages – over $10 billion in damages in 2016 alone – floods on the Ohio River cost lives. That’s why it’s so important to predict accurately when flooding will occur.
When forecasting floods, Ray is most concerned with three variables in five-day flood forecasts: precipitation, antecedent soil moisture and the flood-routing hydrograph. Ray can adjust all three of these variables in a model simulation to determine river levels and flood possibilities.
Precipitation refers directly to the weather forecast. As we all know, weather is unpredictable and forecasting weather is rarely an exact science. Ray is interested in how to better assess how to adjust flood forecasting when the precipitation prediction is off.
The antecedent soil condition refers to the groundwater that feeds the river. Think of it like a sponge: How much water is the earth going to absorb, and how much water is going to feed into the river?
Finally, the hydrograph is probably the most complicated variable Ray will address. “The hydrograph essentially means how fast the water travels to the river from where it fell on the ground.” For example, if it’s raining in Columbus, knowing that it will take two weeks as opposed to three weeks to get to the Ohio River will affect forecasters’ predictions for flooding.
By developing statistical tools that account for all these variables, Ray can reduce the uncertainties in flood forecasting in real-time, leading to more accurate flood predictions.
A picture is worth a thousand words
Ultimately, Ray hopes to present this complex information in a way that can help people make smart decisions. He plans to do this through data visualization techniques ranging from graphs to 3D plots to diagrams.
“The end goal is to help these operational hydrologists working at the Ohio River Forecast Center to visualize the data they are working with,” Ray says.
Ray talks about the art of data display and is even working with students from UC’s College of Design, Architecture, Art and Planning to find the best ways to present this data. Instead of just handing over a bunch of Excel spreadsheets to the forecast center, Ray will present this information in a digestible medium like a graph that can lead to a better understanding of the data, influence decision-making and help develop strategies for managing forecast uncertainty.
In the long run, Ray hopes this work will not only save lives and property but also contribute to larger scale projects related urban planning and infrastructure development and restoration. By determining the gaps in data of flood forecasting, Ray’s project could lead to recommendations for various flood control measures like the installation of soil moisture meters.
Though flooding will never go away, Ray’s work can improve the ways we determine the timing of floods and the magnitude of their destruction. With additional analytical tools and data visualization techniques, flood forecasters can close the gap on uncertainties in their predictions, keeping our heads above water.
Featured image at top: The Ohio River flows between Cincinnati, Ohio, and Newport, Kentucky. The Ohio River Basin spans across 14 states and includes over 25 million people. Photo/Carol M. Highsmith Archive/Library of Congress, Prints and Photographs Division.
Find your path
Interested in encouraging others as you find your path to academic success? If so, apply to UC as an undergrad or graduate inspirational Bearcat. As part of UC's Next Lives Here strategic direction, UC's students makes positive and real-world impact.
Related Stories
UC 1819 Innovation Hub workshop spreads AI awareness
June 28, 2024
Microsoft and Disrupt Now joined forces to run a June 24-27 seminar at UC’s 1819 Innovation Hub teaching Cincinnati residents how to responsibly use AI.
UC opens Blood Cancer Healing Center
June 28, 2024
Media outlets including WLWT, Local 12, Spectrum News, the Cincinnati Enquirer and Cleveland.com highlighted the opening of the University of Cincinnati Cancer Center's Blood Cancer Healing Center.
New project aims to better support teen mothers in Adams County
June 27, 2024
WCPO highlighted a partnership between the University of Cincinnati and the Adams County Health Department that is aiming to provide better support for teen mothers in the county.