It does not happen often that a supercomputer turns a hundred years of geological orthodoxy on its head, but that is just what occurred when scientists looked under the Pacific and discovered seismic structures that, in all fairness, should not exist. The finding, facilitated by the computing prowess of the Piz Daint supercomputer and a new generation of seismic imaging, has left Earth scientists elated and confused.
For many years, the conventional wisdom was that such positive seismic anomalies in the lower mantle areas where earthquake waves move more rapidly were trustworthy fingerprints of old, chilled tectonic plates that had been subducted into the mantle. These “slabs” were always positioned beneath known or reconstructed zones of subduction, consistent with the dance of plate tectonics. But the new research from ETH Zurich and Caltech, released in Scientific Reports, put a monkey wrench into this story. “Apparently, such zones in the Earth’s mantle are much more widespread than previously thought,” said Thomas Schouten, author of the study and doctoral student at ETH Zurich’s Geological Institute.
The key breakthrough came with the use of full-waveform inversion (FWI), a computationally demanding technology that processes complete seismograms instead of individual wave arrivals. It records the intricate ballet of reflected, refracted, and scattered seismic waves and provides a much richer and volumetric view of the interior of the Earth than classical travel-time tomography. “It’s like a doctor who has been examining blood circulation with ultrasound for decades and finds arteries exactly where he expects them. Then if you give him a new, better examination tool, he suddenly sees an artery in the buttock that doesn’t really belong there. That’s exactly how we feel about the new findings,” said ETH Zurich Professor Andreas Fichtner.
The most dramatic anomaly occurred under the western Pacific, between 900 and 1,200 kilometers in depth. By plate tectonic theory, this zone should be free of subducted plate fragments. The Pacific Plate is a huge, stable slab, and there is no evidence of subduction in its recent geological history. And yet, the FWI models detected an enormous, high-velocity structure material that flows like cold, dense rock, but with no apparent tectonic origin.
The consequences are deep. Traditional tomography has never been able to resolve in the interiors of oceans and continents, where earthquake sources and seismic stations are rare. FWI, leveraging the complete richness of seismic information and advances in GPU-boosted wave propagation solvers, now reveals a hidden heterogeneity in the mantle that was previously unseen. The REVEAL model, as an example, learned more than six million distinct waveforms, computing synthetic seismograms that generalize to out-of-sample data and thereby confirming the validity of these newly imaged anomalies.
And what might these mysterious structures be? The researchers advance two leading hypotheses. One is that they are ancient, silica-rich regions ancient relics from the formation of the mantle some four billion years ago, somehow imprinted amidst the ceaseless churn of mantle convection. The second is that they are iron-rich rock accumulations, formed as an evolutionary consequence of mantle dynamics on a geological timescale. Experiments at high pressure have shown that iron-rich ferromagnesian silicates can crystallize at the core–mantle boundary, producing phases up to 20% denser than normal mantle silicates. These dense, iron-enriched post-perovskite phases would reduce seismic velocities very significantly and explain both the high- and low-velocity zones found in the D″ layer.
The mineralogical richness of the lower mantle is complemented by advances in experimental and computational mineral physics that have recently been made. Research indicates that the composition of the lower mantle is not uniform; rather, it is likely to consist of domains with varying Mg/Si ratios and iron abundance, affecting seismic wave velocities that cannot be accounted for simply by temperature. The existence of Fe2+-rich bridgmanite and iron spin-state changes at high pressures contributes additional facets of complexity, influencing both elasticity and density.
The discovery of these anomalies also calls into question the long-standing practice of using seismic wave speed anomalies as direct proxies for past subduction and mantle temperature. Statistical analyses of FWI models show no significant correlation between these deep anomalies and reconstructed subduction zones, challenging the assumption that all positive anomalies are simply cold slabs. Instead, Earth’s lower mantle appears far more compositionally diverse and dynamically complex than previously imagined.
As the discipline advances, the emphasis will turn toward combining seismic imaging with mineral physics, geochemistry, and geodynamics in an effort to untangle the real nature of these enigmas. For the present, the Pacific’s deep mantle enigma remains a reminder: even the strongest models can merely tantalize at what is hidden deep beneath our feet.
