Media contact
Article by Julie Leibach, Senior Science Writer, Nicholas School of the Environment
More than 99% of all freshwater ice is locked up in ice sheets cloaking Antarctica and Greenland.
Johnny Ryan, an assistant professor of ice and climate sciences at the Nicholas School of the Environment, uses a combination of satellites, drones and other tools to study how physical processes affect ice sheet dynamics in Greenland. Understanding how ice melts, moves and accumulates is critical to predicting future ice sheet changes as the climate warms.
In this Q&A, Ryan describes how the Greenland ice sheet is changing, the questions he鈥檚 chasing, and how artificial intelligence can help answer them.
Why is IT important to study the Greenland ice sheet?
It鈥檚 a very large ice mass containing a lot of frozen water, and in the current climate, it鈥檚 in a state of net mass loss, meaning that more ice is melting from the ice sheet than snowfall is accumulating. If you鈥檙e interested in sea level rise across the world, and you want to understand why sea level rise is happening, a large fraction of that is caused by the Greenland ice sheet melting.
It鈥檚 important for scientists to track how Greenland ice sheet mass is changing over time so that we can attribute sea level rise to different sources. By tracking changes in ice sheet mass, we can also develop models that can forecast how further climate change will impact the Greenland ice sheet and how fast sea levels might rise in the future.
What does it look like when you鈥檙e on the ice sheet?
It鈥檚 kind of difficult to gauge the actual scale of things since there are few features, like trees. The horizon is just a white line that looks like it goes on forever since the slope of the ice sheet is quite gentle. But when you fly over with a helicopter, or even better, walk on the actual surface, you realize that it鈥檚 not this white homogenous mass. Dust and black carbon can accumulate on the surface of the ice sheet, and so you can sometimes see patches of darker areas and brighter areas. There are also many large meltwater streams. In the summer, torrents of water flow through these streams. It鈥檚 a pretty powerful experience when you consider that this water used to be ice.
What are the big questions that your lab is trying to answer about Greenland?
We primarily focus on what we call the surface mass balance, which is basically the balance between processes that add mass to the ice sheet versus those that remove mass from the ice sheet. We call these accumulation and ablation processes, respectively, and we鈥檙e trying to understand what controls the rate at which they happen.
So, for example, if the ice sheet is darker, it鈥檚 going to absorb more solar radiation, and that鈥檚 going to cause more melt. But there鈥檚 also a lot of questions about what happens to this melt once it has been generated. Not all of that melt is going to make its way into the ocean. Some of it could be stored in air spaces, in the snow and the ice, or maybe even below the ice sheet. We鈥檙e trying to understand what controls melt, but then also how much of that melt is being transferred into the ocean.
You鈥檝e started incorporating machine learning, a type of artificial intelligence, into your research. How does this tool help?
We typically use observations from satellite instruments in space or sometimes from drones, but those observations can be limited. They might not cover the entire area that we want, or we may only get these observations every few days or few weeks, so we want to fill in the gaps both spatially and temporally. That鈥檚 one area where machine learning can come in. Models can learn from the data that we do have, and then interpolate in places where we don鈥檛 have as many observations.
The other way we鈥檙e using machine learning is to produce models for predicting ice sheet meltwater runoff. The traditional approach would be to build a model that uses climate data and physics to figure out how much energy is absorbed into the surface and how much meltwater runs off. But since we have observations of meltwater runoff, we can use a machine learning approach to understand the relationship between the climate data and runoff, and construct our own model of how these two things relate to each other.
What are some recent findings you鈥檝e made about how the Greenland ice sheet is changing?
We published a that showed that the synchronous drainage of several large supraglacial [surface] lakes is facilitated by the development of meltwater streams. We have also found evidence that meltwater causes more melt, because meltwater makes the surface darker, causing it to absorb more solar radiation than nearby surfaces that are not covered with water. We鈥檝e done some very to show that this could be quite an important process. That鈥檚 something that we鈥檒l be trying to study with our field work this August. We鈥檙e going to the Greenland ice sheet to measure how fast the ice melts in streams versus ice that鈥檚 not covered with water.
Are students joining you on this work?
We鈥檙e taking two Ph.D. students and three undergraduate students from Duke. They spend a lot of time behind computers looking at the Greenland ice sheet in satellite images, coding and processing data. I think it鈥檚 very important for them to get boots on the ground and be there to see the thing that they spend so much time studying.
This interview has been edited for brevity and clarity.