UNIVERSITY PARK, Pa. — Researchers at Penn State are part of a multi-university team selected to receive $6.6 million in recommended funding from the National Oceanic and Atmospheric Administration (NOAA). The group will establish a new multi-university data assimilation consortium to improve weather forecasts using enhanced numerical weather prediction systems.
The Penn State team includes four researchers from the College of Earth and Minerals Sciences: Xingchao Chen, assistant professor of meteorology, Steven Greybush, associate professor of meteorology, David Stensrud, professor of meteorology and atmospheric science, and Yunji Zhang, assistant professor of meteorology. Other universities in the consortium, named the Consortium for Advanced Data Assimilation Research and Education, or CADRE, include the lead institution, the University of Oklahoma, as well as Colorado State University, Howard University, University of Maryland and the University of Utah. Additionally, the University of Illinois Urbana-Champaign, University at Albany, State University of New York and City College of New York will participate as non-funded collaborators.
“Nearly every week we have another extreme weather event in the United States that impacts our communities, causes injuries and loss of life, and leads to hundreds of billions of dollars in damage annually,” Stensrud said. “CADRE will partner closely with NOAA to improve computer model forecasts of weather extremes, including tornadoes, hurricanes, winter storms, heavy rain, heat waves, wildfires and storm surges. These computer model forecasts form the basis of the weather forecasts widely used by society.”
CADRE's focus is on making transformational advances to data assimilation — the science of combining observations with numerical models to obtain the best estimate of the state of a system as it evolves over time. It is a critical element in producing daily weather forecasts, Stensrud said.
“Data assimilation is a way of taking observations that are collected at different locations and from different sensors and putting them together in a way that makes sense physically using our best understanding,” Stensrud said. “We have observations from satellites, weather balloons, ground observations, aircraft and radar, as well as previous computer model forecasts. Data assimilation brings all this information together using a statistical framework and forms a physically consistent three-dimensional picture of the atmosphere using the variables that are needed to operate our weather forecast models.”
Although weather forecasting accuracy has vastly improved in the last several decades, thanks in part to improvements in data assimilation, the entire weather forecast process needs to be upgraded continually given new technological developments, such as artificial intelligence. Computer models, such as the Global Forecast System used by NOAA to produce weather forecasts, need constant updates, including the assimilation of new data, according to the CADRE team.
Next-generation data assimilation faces significant challenges associated with high-resolution, multiscale, coupled Earth system modeling and a large amount of diverse and complex observations. Serious gaps in the data assimilation workforce and the lack of sustained innovative data assimilation research support inhibit addressing these challenges, the team said. CADRE will help develop the next generation of scientists with expertise in data assimilation.
“CADRE will partner closely with NOAA to advance data assimilation education, leading to growth in the data assimilation workforce, and will conduct innovative research to address fundamental challenges in numerical weather prediction,” Stensrud said. “We want to help NOAA make optimal use of the data they currently have, leading to better weather forecasts that people can use to make the best decisions for themselves and their families when severe weather occurs.”
Expected outputs from the consortium will be used to improve computer weather forecast predictions using the Unified Forecast System (UFS), a community-based, coupled, comprehensive Earth modeling system. The consortium will also work closely with NOAA’s Earth Prediction Innovation Center to advance new data assimilation science into operations within the UFS.
Stensrud, Chen, Greybush, and Zhang are members of the Penn State Center for Advanced Data Assimilation and Predictability Techniques, referred to as ADAPT, which seeks to integrate and enhance the existing strength and expertise in cutting-edge data assimilation and predictability research across Penn State.