Momentum

Sea ice was thought to block sunlight and limit the growth of microscopic marine plants living under the ice, but in 2012 scientists discovered a massive bloom of phytoplankton beneath three-foot thick ice in the Chukchi Sea, north of Alaska. The unexpected bloom, which extended 62 miles under the ice pack, was fueled by sunlight penetrating the ice through pools of melted water on its surface.

The National Science Foundation awarded two UD researchers and a colleague at the University of Maryland a three-year, $201,655 grant to model critical changes in Arctic sea ice to detect and predict the growth and movement of large algae blooms in the Arctic Ocean, related to that region's warming climate. The interdisciplinary project combines artificial intelligence, theoretical physics and biology.

"In the past, the algae beneath the Arctic ice didn't bloom or produce too much energy because there was a nice balance," said Ivan Sudakow, assistant professor of physics and the project's principal investigator. "But once Arctic melting begins on this really rapid scale and the algae is absorbing solar energy through these melt ponds, it increases this process of the ice melting, leading our climate system to a tipping point."

Sudakow and co-principal investigator Vijayan Asari, director of the Vision Lab, are developing machine learning tools to analyze Arctic sea ice data. Sudakow will use the data to build new models to describe plankton dynamics and test the hypothesis about melt pond transformation triggering under-ice algae blooms.

The NSF-funded project builds on past work by Sudakow, who developed a model to better predict the effects of climate change on Arctic sea ice. The discovery, which was published in New Journal of Physics, also gained attention from Scientific American, Physics World, Eos and WIRED.

Invented in 1920, the Ising model shows how natural systems can behave in related ways, such as showing phase transitions between solid, liquid and gaseous states of matter. Sudakow, who specializes in mathematical modeling of physical and living systems, will work with an undergraduate student researcher to develop new Ising-based models to simulate changes in plankton dynamics.

"Standard mathematical methods like differential equations are not really helpful here, because they require a lot of parameters, a lot of data, and need to be solved numerically," Sudakow said. "We are going to use statistical mechanics, the methods of this branch of physics, to actually find a simple way to model algae bloom as a critical phenomenon."

The lack of expert annotated data about under-ice phytoplankton is an issue, said Asari, professor of electrical and computer engineering. Using a semi-automated algorithm, Vision Lab researchers are creating synthetic data by analyzing aerial images from satellites and aircraft and marking the location of melt ponds. They then re-run the program with the marked data to precisely detect and predict these areas of interest, based on their current movement characteristics.

"You will be able to detect and identify the ponds and their growth, and predict the dynamics of phytoplankton growth and movement," Asari said. "This will provide a lot of information for agencies and authorities to make the appropriate decisions to stop the formation and thickening of the algae."

Sudakow said the project's goal is to understand the interaction between phytoplankton, sea ice, melt ponds and the global climate system to determine possible tipping points in the climate system related to the phytoplankton bloom.

"Some small changes in the connection between the ecosystem and the climate system are actually leading to big changes in the state of the climate system," Sudakow said.