A Scientist's Life: Alice Gabriel

Alice Gabriel is a seismologist at Scripps Institution of Oceanography at UC San Diego. She received her bachelor’s and master’s degrees from Technical University Dresden in Germany in 2004 and 2008, respectively. She received her PhD from ETH Zurich in Switzerland in 2013 and was a visiting student at Caltech and in Saudi Arabia. After postdoctoral work and being an assistant and then associate professor in Munich, Germany, she joined Scripps Oceanography’s Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics (IGPP) in 2022.

explorations now (en): What do you do for a living?

Alice Gabriel (AG): I study earthquakes. I'm specifically interested in the physics of earthquakes, and I use very large supercomputers to connect all kinds of observations, all kinds of data with equations to better understand earthquakes.

We have seen recently in the big Turkey earthquakes this year and the Ridgecrest earthquake in 2019 here in California that earthquakes can jump from structure to structure, from fault to fault, and produce much larger shaking than we might think. Understanding why and how geologic faults move is one of the grand challenges in earth science. We have increasingly dense observations. We have satellites in space that can image what happens during earthquakes. We have seismic arrays all over the globe. We have submarine seismometers. But we don't know much about the physics of how rocks break at the depths where earthquakes happen.

I really try to find out how earthquakes start, how they move and how they stop. And each of these questions is interesting in itself.

en: What are some of the main questions those in your field are trying to answer?

AG: One of the really big questions is what causes earthquakes? Why can't we predict them? How do they break through the crust and how do they stop? And this is all related to the fact that we don't understand how rocks break. We don't understand the energy budget of earthquakes. These are really fundamental questions on some basic physics of how our earth works.

We have forces loading faults over decades or over hundreds of years sometimes and then suddenly you have an earthquake occurring on the timescale of seconds. On some large faults, these earthquakes occur rarely, so it's really difficult to prepare for them. It’s not just a big bang and it’s over. There’s a lot of interesting complexity in how and what happens exactly at the source of this process. That is what we're after. We're using ideas from laboratory experiments that break small samples of rocks. We’re using seismological data. We use geodesy, so space satellites or other GNSS stations. You can also use tsunami data. Using supercomputing, we can use all of these data together that are typically isolated. Different fields and different scientists are building all these datasets. We can build models that connect all of them to better understand how rocks break and how they produce earthquakes.

I'm really excited about the opportunities we have now. We have increasingly dense observations. We have, as we all know, increasing power of computers that allows us to model earthquakes. However, at the same time, it is a difficult situation to be in after a large earthquake happens because we see all the devastation. As a seismologist, I often feel like I've failed because I wasn't able to predict that earthquake, but I'm convinced that the research I'm doing will help to better prepare for next events and to be better prepared for the devastation that it might cause.

en: What are the tools you use in your research?

AG: I'm a mostly theoretical scientist, but I do use tools. The tools I use are coming from applied math, from computer science. I'm working with colleagues that develop smart algorithms that come up with good ways to solve equations. I'm also using really modern hardware. So tech really in the sense of computers, such as large supercomputers, which are some of the biggest machines worldwide. I also use cloud distributed systems, which are supercomputers that are not really assembled in one place, but they're distributed over many servers in the cloud.

One of the most recent projects I'm working on is studying the two large earthquakes that happened in Turkey earlier this year. I do that in collaboration with a lot of colleagues at IGPP. We have been setting up physics-based models to understand each of these earthquakes and also how they talk to each other. One of these models requires 150,000 CPU hours. So that means one computer core, like a chip like you have in your phone. Just one core would need to calculate for 150,000 hours to come to the solution that we're after. For one model, however, we do that, of course, on large machines. So for example, you can scale this up to 8,000 nodes and each node has 48 cores and then you get the result in a few hours.

en: What got you into seismology?

AG: I'm a physicist by training and I started to work on material science. I was actually looking into how to produce better chips for computers, but at some point I wanted to do something that's more impactful for society. Then I did an exchange year in New Zealand where I learned a lot about geology. I learned about fossils and tectonics and also about geophysics. So the physics of how our earth is shaping up and how it does what it does. I got really interested and then I moved my PhD to seismology.

en: Why did you want to come to Scripps Oceanography?

AG: I came here recently and the reason is there's fantastic students and fantastic postdocs I’m able to work with, and a fantastic set of colleagues, a large geophysics group that brings a lot of expertise in geodesy and seismology. I also think I fit right in because IGPP has a longstanding tradition in theoretical seismology.

At the same time, they really value methods and technology. This is what I'm doing in my research. I am a theoretical scientist, but I'm interested in instrumentation in the sense of computing. I'm interested in technology in the sense of large computers using modern hardware and software.