Climate scientist Kate Evans works at Oak Ridge National Laboratory (ORNL) on a variety of projects from using supercomputers to study the movements of ice sheets to developing a model to explore the impacts of storms on ocean currents. We recently had the opportunity to learn about her work advancing climate simulations and modeling and why an Indiana storm sparked her interest in Earth sciences.
Question: At Oak Ridge, you are part of the Computational Earth Sciences group. Why did you choose this field?
Kate Evans: I have always liked Earth science of all kinds, particularly big storms and events. When I was about 10 years old, a big storm system came through my town in Indiana. I was standing at our big French doors watching huge hunks of hail sail by and my mother was pleading with me to get in the basement.
I think it’s fascinating when the Earth uses its energy to produce something powerful, whether it’s a thunderstorm, volcanic eruption or the Northern lights. As far as the computational part, my college professors at Haverford usually included a computational component in science lab even though it was quite novel. It made me comfortable being around computers in the lab. In graduate school, it became clear that atmospheric scientists have always needed to work with big sets of data, whether they are observational or model-generated. Computing is a large part of almost all types of Earth science research.
Q: What (or who) inspired you to become a scientist?
KE: I have always been a stubborn and curious person, which I think are traits that lead to a love of the scientific method. Other events, such as a gift subscription to World magazine from my grandparents and two particularly memorable teachers I had in school really had an impact. For example, I had a calculus teacher that we called “Rigorous Rita.” She would jump up and down at the board with excitement when we got a proof going in class.
Q: Do you have advice for students interested in becoming scientists?
KE: The best thing I learned when becoming a scientist was not to be afraid of problems I couldn’t solve. Once I learned to take a deep breath and apply the logic and laws I had learned in class, facing a difficult test or homework set became a lot easier.
Q: What projects are you working on right now?
KE: I am on several projects that have disparate short term deliverables, but are linked in their long term goals. One project, funded by the Department of Energy’s Office of Biological and Environmental Research, involves the configuration and analysis of the Community Earth System Model (CESM) to run climate length simulations with higher spatial resolution than has previously been possible. The CESM is a global scale climate model that couples the atmosphere, land surface, ocean, sea ice and land ice together. With higher resolution, we can resolve tropical cyclones and mid-latitude storms in the atmosphere and investigate their interaction with other Earth system components. This will allow climate scientists to test many climate variability hypotheses such as the effect of long term shifting ocean currents on short term weather events, like tropical cyclones. Another project involves the development of new ways to move the climate simulations forward in time. As climate models are refined in space, the time scales of interest are also shortened and this has implications for simulation speed and accuracy.
Q: You are also the lead on the Scalable, Efficient and Accurate Community Ice Sheet Model Project. What is the goal of this initiative?
KE: We have been running atmosphere climate and weather simulations for decades, and that has improved knowledge of climate variability immeasurably. But the greatest short term impact on humans due to global warming is really sea level rise, and the land ice modeling community is less mature. Observations of land ice variability are starting to ramp up, but data is expensive and difficult to gather and trends are difficult to see over short time periods. Climate scientists would like to use simulations of continental ice sheets over Greenland and Antarctica to tell us the response of land ice to global temperature changes in the atmosphere and ocean.
The SEACISM project is tasked to take Glimmer-CISM, which is the existing serial land ice model included in the 2010 release of the CESM, and expand its capability to run much finer scale and complex problems. On fine grids, greater detail of ice sheet behavior can be resolved accurately, but the trade-off is greater computational expense. To address this, our project is applying new tools and algorithms developed through the Department’s Office of Advanced Scientific Computing Research programs for efficient and robust simulation of Glimmer-CISM so that predictive simulations of ice sheets can begin as soon as possible. For example, we are using the current solution method in Glimmer-CISM as a pre-conditioner to create more efficient and robust simulations of ice sheets.
Q: As part of the project, your team was granted 6 million hours on Oak Ridge’s Jaguar supercomputer and Argonne National Lab’s Intrepid. As you know, Jaguar is capable of 2.3 quadrillion calculations per second and Intrepid can complete up to 557 trillion calculations per second. Can you tell us about your experiences using these supercomputers? Why are these machines important?
KE: Almost all of us on the project have already been using these systems for climate simulation, so we have been able to use some of that experience for the development of Glimmer-CISM. We have designed our new version of Glimmer-CISM to scale to large processor counts and run efficiently in the near term. For example, unlike the atmosphere, land ice equations more naturally lend themselves to a hierarchy of spatial grids to achieve solutions quickly, and we plan to take advantage of this benefit.
As for importance, first imagine your brain as a supercomputer. Then imagine the number of locations on Earth where it would be useful to have data to assess the global climate over a long period of time. You have a sense of the huge data and computation involved. Imagine everything from the deep water currents near the poles that are affected by changes in heat and salt from ice sheets above and the root systems of large trees in the Amazon that drive carbon fluxes all across the tropics to the upper reaches of the global atmosphere where chemical reactions affect radiation from the sun and the shape and temperature of the ice on the coast of Greenland. All of those features and more are needed for understanding the climate, and you can get a sense of the scale of computation necessary to create a fully integrated global Earth system model.
Q: Just a few more questions - what can you never start a day at the lab without?
KE: Going to the building next door to get a coffee and bagel with colleagues.
Q: What is your favorite tool in the lab?
KE: Jaguar, but I can still only think properly if I have a pencil and paper in my hand.
Q: What do you enjoy doing in your free time?
KE: Spending time with my family or going for a long run.