Biological Science’s Cayelan Carey to expand modeling capabilities for daily water quantity and quality forecasts
As climate change brings on events affecting drinking water conditions in lakes and reservoirs — such as droughts, storms, and algal blooms — the Fulbright Future Scholar aims to give water managers the tools to forecast their effects as regularly as we predict the weather.
For years, water managers have used historical baselines to predict changes in water quality and quantity in lakes and reservoirs. If a big storm or drought were to hit in weeks to come, they could look back at the ebbs and flows in water conditions that came with past storms and droughts.
But climate change has thrown a wrench in that once reliable, longtime strategy, said Cayelan Carey, a freshwater scientist and professor of biological sciences in the Virginia Tech College of Science.
According to Carey, the once-reliable historical guidelines for conditions such as water temperature and oxygen are becoming less dependable. “There’s a much larger envelope of what the possible conditions are now versus what they were in the past,” she said.
Carey hopes to offer water managers a data-driven alternative. As a Fulbright Future Scholar, Carey traveled to Perth, Australia, earlier this year to work in collaboration with Matthew Hipsey of the University of Western Australia on a water quality and quantity forecasting system developed by Carey and Quinn Thomas, an associate professor in the Virginia Tech College of Natural Resources and Environment and a Data Science Faculty Fellow in the College of Science.
The system generates daily freshwater forecasts using real-time data from environmental sensors, and it represents a fusion of Hipsey’s modeling work and the forecasting software Carey and Thomas developed at Virginia Tech. The researchers put it to the test for the first time in South Australia’s Lake Alexandrina.
At Virginia Tech, Carey and Thomas have in recent years led the development of software called Forecasting Lake and Reservoir Ecosystems (FLARE), which produces daily water quality and quantity forecasts for the next two weeks for a drinking water reservoir in Roanoke. Core to the code Thomas used to build the system is Hipsey’s General Lake Model-Aquatic Eco-Dynamics (GLM-AED). It’s an open-source lake and reservoir model used in education and research by freshwater scientists around the world, and a rare find for free use, Carey said.
“His team is really at the forefront of developing open tools that anyone can use,” said Carey, who is a Roger Moore and Mojdeh Khatam-Moore Faculty Fellow in the College of Science and a member of the Global Change Center, part of the Fralin Life Sciences Institute. “The benefit there is that there’s then a community of people that are able to share code and work on the model together and improve it. That’s why we’ve made GLM-AED the core model in our forecasting framework, because we can actually access and edit all the code underneath it. And there are very, very few other open-model resources that are at the same caliber of a model as the GLM-AED model.”
In Hipsey’s model and his team’s willingness to work on it with others, Carey saw an opportunity for collaboration: Her team could bring the FLARE system’s forecasting capabilities to the GLM-AED model, bringing them closer to enabling forecasting for lakes and reservoirs worldwide.
Through virtual collaboration and then during Carey’s Fulbright term in Perth, Carey, Thomas, and Hipsey worked on the original model’s code to adapt it to daily forecasting as well as to build upon its capacity to simulate the effects of human activities on drinking water quality.
Previously, the GLM-AED model ran long-term simulations of historical conditions. Now, the hybrid GLM-FLARE system updates every day for 35 days into the future with new data collected and streamed wirelessly from sensors in the field to update the model, which runs every night for the next morning’s forecast. While together in Perth, the team also improved GLM-AED’s modeling of lake conditions such as freshwater ice, dissolved inorganic carbon, and pH.
Looking ahead, the team aims to automate the new GLM-FLARE system’s water quality forecasts, which will be delivered to water managers overseeing water bodies surrounding Lake Alexandrina. To Carey, Australia — where some regions are experiencing huge floods while others are facing drought — is the perfect testing ground for the new system.
“I think the water challenges in Australia are not unique, but they are really intense,” Carey said. “If we are able to make a tool that can work for this sector and meet these needs here, then I feel like we have stress tested our forecasting framework so that it could be generalizable.”
For Carey, the international collaboration she was able to carry out with Hipsey, supported by her Fulbright Future Scholarship, has been critical to making forecasting more broadly available. To encourage it further, the two published an open-source book that describes the model for researchers and students.
“Dr. Hipsey brought us this amazing lake model and code base,” Carey said. “And the open sharing part of it has really catalyzed all of this research. If we didn’t have that model and his willingness to collaborate, we couldn’t use it to make forecasts of tomorrow’s water.”