Dr Tianjue Li (left) and Associate Professor Ping Tong (right) have used advanced algorithms to observe underground magma flows with unprecedented range and accuracy.
Magma, or molten rock, exists deep within the Earth and occasionally bursts onto the surface in spectacular volcanic eruptions. Inside the Earth’s solid crust, magma flows through huge networks of underground caverns called magmatic systems. These are important for geothermal power generation, since geothermal plants often benefit from being situated above them.
But before magma flows can be exploited for geothermal energy, we must find them, and this can be difficult. A case in point is the Geysers-Clear Lake system in California, USA, which drives the world’s largest complex of geothermal power plants, providing about 9 percent of green power production in the state. Despite its importance, the Geysers-Clear Lake system is still not understood in its entirety. Using seismic imaging, scientists have previously found only a shallow system of magma caverns, down to around ten kilometers below the surface.
Now, applied mathematicians at Nanyang Technological University (NTU), Singapore, have developed a computational technique that dramatically extends our ability to detect underground magma. Writing in the Proceedings of the National Academy of Sciences in March 2024, Associate Professor Ping Tong and his co-authors report that their new seismic imaging method has discovered complex magma flows as much as 23 kilometers below the surface of Geysers-Clear Lake, far deeper than previously observed.
Feeling The Earth Move
Much of the information we have about the Earth’s crust is gleaned from observations of seismic waves. These powerful vibrations, usually caused by earthquakes, travel faster in solid rock than in magma, and also bend and reflect as they move through the crust. Using these features, scientists can convert seismic wave measurements into information about the structure of the crust. However, existing “seismic imaging” techniques relying on local earthquake data can only visualize the upper half of the crust. This makes them unreliable for studying magmatic systems, which usually originate from the lower half of the crust.
Seismic imaging is what mathematicians call an “inverse problem”. If the crust structure were known, it would be relatively easy to predict what the seismic wave data would be. The opposite task – using seismic wave data to determine crust structure – turns out to be mathematically and computationally demanding.
To surpass established seismic imaging techniques, Prof. Tong and two colleagues used a series of computational algorithms to analyze historical seismic data collected over the past 20 years.
A sample of the seismic data that Prof. Tong and his team used in their study. Figure excerpted from Li, Wu, and Tong, PNAS (2024).
“Seismic waves come in different varieties. Traditional seismic imaging relies on Primary Waves, which arrive directly from seismic sources such as earthquakes,” explains Prof. Tong, who is a faculty member at both the School of Physical and Mathematical Sciences and the Asian School of the Environment in NTU. “Other seismic waves, called PmP Waves, are reflected off the boundary between the crust, the solid outer layer of the Earth, and the mantle below.”
Previously, PmP Waves have seldom been used for seismic imaging as they are harder to detect. PmP Wave signals are also harder to interpret, since the waves take a more circuitous route from the seismic source (e.g., an earthquake) to the detector.
Using an assortment of advanced algorithms, Prof. Tong and his team managed to pick out PmP Wave signals from historical seismic data, and then convert them into a wealth of untapped geological information.
Big, Hot, and Deep
With the Primary Wave and PmP Wave data, the team performed extensive mathematical modeling to solve the seismic inverse problem. They were then able to generate the first detailed map for the system of magma caverns beneath Geysers-Clear Lake.
A geologic map of the Geysers-Clear Lake area created by Prof. Tong and his group, revealing a previously-unobserved deep magma reservoir. Figure excerpted from Li, Wu, and Tong, PNAS (2024).
Their results unveil an extensive network of magma chambers running through the entire crust, deeper and more complicated than geologists had previously observed. For example, in addition to a known shallow magma reservoir primarily composed of silica, Prof. Tong and his colleagues discovered a deep magma reservoir composed of magnesium and iron, extending down to the boundary between the crust and the mantle.
“This deep magma reservoir is interesting, as it’s the actual heat source for the Geysers-Clear Lake geothermal plants,” says Dr Tianjue Li, a former PhD student in Prof. Tong’s group and the first author on the new paper, who is now a postdoctoral researcher in the same group. “Researchers have suspected it is there, since the sheer amount of geothermal heat in the area meant there had to be a deeper magma source. But before our study, no one had any direct evidence for it.”
One immediate benefit of this discovery is that it provides scientists with accurate information that can be used to predict how long the Geysers-Clear Lake geothermal field can continue generating green energy.
The study has also shed light on the geological history of the Geysers-Clear Lake area. “From the detailed crust data, we have been able to work out how the underground magma ascended over the past two million years, including the formation of the shallow reservoir after an eruption 10,000 years ago,” explains Dr Li, the postdoctoral researcher.
Looking forward, Prof. Tong intends to develop more sophisticated seismic imaging methods and machine learning-based tools for PmP data identification and picking. He aims to apply his seismic imaging techniques to other geologically important areas of the world. With better information about how magma flows are distributed deep within the crust, scientists and engineers will be able to situate new geothermal plants, as well as optimise the operations of existing plants.