Abrupt weather extremes, changing climate and frequent natural hazards such as floods and droughts create challenges for our nation’s aging reservoir systems. Tiantian Yang, Ph.D., an assistant professor in the Gallogly College of Engineering at the University of Oklahoma, has received a prestigious Faculty Early Career Development (CAREER) Award from the National Science Foundation to help mitigate these problems.
During the five-year project, Yang will seek to develop an integrated solution that addresses the variability and uncertainty of precipitation and develop a novel artificial intelligence and data mining tool to aid reservoir operators’ decision-making.
“We use machine learning and artificial intelligence tools to detect patterns of precipitation and improve subseasonal-to-seasonal forecasts so that we can better inform the future operation of reservoirs,” Yang said.
Over the past four decades, significant scientific advancements have been made in forecasting, linear programming, algorithm optimization and simulation model development to guide reservoir operations. However, with infrastructure that is sometimes 50-80 years old and ever-changing climate variability, including global precipitation, reservoirs must now behave differently.
To avoid reservoir failures, dam operators need two essential items: accurate and reliable hydrological forecasts with extended lead times, and powerful and adaptive tools to assist real-time, flexible decision-making about how much water to release and when.
“Dam operators need tools to prevent problems like those seen in recent years,” Yang said. “For example, in 2020, two reservoirs in Michigan overtopped because of problematic forecasts. With better forecasting tools, the operators could have lowered the water level and created a buffer before the heavy rainfall arrived.”
Yang’s research will leverage state-of-the-art deep learning models to discover and correct the errors associated with current precipitation forecasts from multiple forecasting models in the North American Multi-Model Ensemble dataset.
His work will include large-scale validation experiments on more than 671 watersheds and reservoir simulations on more than 316 dams across the United States. The artificial intelligence and data mining models will then be comprehensively tested in collaboration with the U.S. Bureau of Reclamation and the U.S. Army Corps of Engineers.
“We will collaborate with several reservoir agencies at the state and federal level to test our decision-making support tools, which intensively use our improved forecast data and integrated physical and artificial intelligence models,” Yang said. “Ultimately, this will help water managers better respond to and mitigate the impacts of potential weather extremes and climate uncertainties.”
In addition to the research component of this project, Yang plans to host a Water Festival exhibit at the National Weather Museum and Science Center, allowing visitors to witness the importance of hydrology, meteorology, water resources management and the impacts of extreme weather and climate.
An educational component of the project will provide student training using advanced active learning technologies. “This training, which includes computer programming in the Python language, complements the intensive research connecting big data, artificial intelligence, climate change and infrastructure operation in the interdisciplinary fields of environmental science, hydrology, water resources, meteorology and hydro-climatology sciences,” Yang said.