The discovery by UCI scientists of a new way to predict winter rainfall in the southwest U.S. inspired them to start a new project to combine data science and climate research. Amir AghaKouchak / UCI

Investigators from UCI’s Henry Samueli School of Engineering, Donald Bren School of Information & Computer Science, and School of Physical Sciences will work together in a new initiative, funded by the National Science Foundation, to perfect the use of data science in climate studies. The Transdisciplinary Research in Principles of Data Science plus Climate project includes collaborators from the University of Wisconsin-Madison and the University of Chicago.

The TRIPODS+X research team will jointly pursue the creation of new methodologies in machine learning and network estimation to develop a better understanding of the Earth’s climate system and its regional hydrologic impacts. “Classical tools fail to account for complex dependence structures and higher-order interactions among features, and for nonstationary dynamics,” said UCI co-investigator Efi Foufoula-Georgiou, Distinguished Professor of civil & environmental engineering. “Because of this, important aspects of climate systems are poorly understood, limiting our ability to improve the accuracy of regional forecasts. We believe that data science has a lot to offer in exploring climate data and model outputs to understand and attribute climate modes of variability and change to develop better predictive models.”

Foufoula-Georgiou is joined by UCI colleagues Padhraic Smyth, Chancellor’s Professor, computer science, and director of UCI’s Data Science Initiative; and James Randerson, the Ralph J. Cicerone chair and Chancellor’s Professor of Earth system science. The inspiration for this collaboration came from a recent project conducted by Foufoula-Georgiou, Randerson and other UCI scientists in which a new interhemispheric teleconnection was discovered, allowing climatologists to predict winter rainfall in the Southeastern United States by monitoring sea surface temperatures near New Zealand during the summer. The discovery was made through exhaustive analysis of historic climate data. Foufoula-Georgiou said she thinks there are many other globe-spanning connections yet to be found, and that novel data science tools that have become available in recent years will provide a significant boost to their efforts. “It is vitally important that we fully understand the factors that determine regional climate variability,” she said. “The implications are clear for the economy, security and environmental sustainability for many regions around the world.”