Machine learning for data-driven discovery in solid Earth geoscience.

Understanding the behavior of Earth through the diverse fields of the solid Earth geosciences is an increasingly important task. It is made challenging by the complex, interacting, and multiscale processes needed to understand Earth's behavior and by the inaccessibility of nearly all of Earth's subsurface to direct observation. Substantial increases in data availability and in the increasingly realistic character of computer simulations hold promise for accelerating progress, but developing a deeper understanding based on these capabilities is itself challenging. Machine learning will play a key role in this effort. We review the state of the field and make recommendations for how progress might be broadened and accelerated.

Compositional heterogeneity near the base of the mantle transition zone beneath Hawaii.

Global seismic discontinuities near 410 and 660 km depth in Earth's mantle are expressions of solid-state phase transitions. These transitions modulate thermal and material fluxes across the mantle and variations in their depth are often attributed to temperature anomalies. Here we use novel seismic array analysis of SS waves reflecting off the 410 and 660 below the Hawaiian hotspot. We find amplitude-distance trends in reflectivity that imply lateral variations in wavespeed and density contrasts across 660 for which thermodynamic modeling precludes a thermal origin. No such variations are found along the 410. The inferred 660 contrasts can be explained by mantle composition varying from average (pyrolitic) mantle beneath Hawaii to a mixture with more melt-depleted harzburgite southeast of the hotspot. Such compositional segregation was predicted, from petrological and numerical convection studies, to occur near hot deep mantle upwellings like the one often invoked to cause volcanic activity on Hawaii.