The continuously growing amount of seismic data collected worldwide is outpacing our abilities for analysis, since to date, such datasets have been analyzed in a human-expertintensive, supervised fashion. Moreover, analyses that are conducted can be strongly biased by the standard models employed by seismologists. In response to both of these challenges, we develop a new unsupervised machine learning framework for detecting and clustering seismic signals in continuous seismic records. Our approach combines a deep scattering network and a Gaussian mixture model to cluster seismic signal segments and detect novel structures. To illustrate the power of the framework, we analyze seismic data acquired during the June 2017 Nuugaatsiaq, Greenland landslide. We demonstrate the blind detection and recovery of the repeating precursory seismicity that was recorded before the main landslide rupture, which suggests that our approach could lead to more informative forecasting of the seismic activity in seismogenic areas.

Ambient seismic noise correlations are widely used for high-resolution surface-wave imaging of Earth's lithosphere. Similar observations of the seismic body waves that propagate through the interior of Earth would provide a window into the deep Earth. We report the observation of the mantle transition zone through noise correlations of P waves as they are reflected by the discontinuities associated with the top [410 kliometers (km)] and the bottom (660 km) of this zone. Our data demonstrate that high-resolution mapping of the mantle transition zone is possible without using earthquake sources.