Seismic heterogeneity in the deep mantle ranges in scale from 1,000’s of km as represented by the Large Low-Velocity Provinces (LLVPs) and imaged in seismic tomography down to 10’s of km as imaged through the scattered seismic wavefield. The origin of the small-scale heterogeneities is still debated yet enigmatic features with scale lengths from 10’s to 100’s of km, and known as ultralow-velocity zones (ULVZs) likely contributes to a large portion of the observed heterogeneity. ULVZs have been known to exist for nearly three decades, although a precise definition of what constitutes a ULVZ is still lacking.  These features are primarily characterized by their low seismic velocities, reported as being decreased by as much as 45% for S-waves and 25% for P-waves with respect to standard 1-D reference Earth models. ULVZs are typically best modeled as thin features on the order to 10 to 30 km in thickness, although some regions may exhibit greater thicknesses. ULVZ origins remain uncertain, as they have been discovered in a wide variety of settings, from directly underlying hot spot volcanoes and possibly representing the roots of whole mantle plumes, to accumulating near the boundaries of the Large Low-Velocity Provinces, to lying on the boundaries of regions of past subduction. Early interpretations focused on a partially molten origin as this could readily explain the ultra-reduced S-wave velocities. However, recent mineral physics efforts have demonstrated that compositional anomalies such as the Fe-enrichment of the mineral ferropericlase could also produce the required velocity decreases. Furthermore, the development of ULVZ material has also been proposed through many different mechanisms such as by melting of Mid-Ocean Ridge Basalt in downgoing slabs, through core-mantle reactions, or possibly through remnants of Earth’s early differentiation. Nonetheless, all features with low seismic velocities are collectively referred to as ULVZs, and their makeup, compositional or partially molten, and origin could be due to a combination of all of these scenarios.

In order to understand what these features physically represent, what they are, how they may have originated in the Earth, and what role they play in Earth evolution and dynamics we need to determine where these features exist and what their physical parameters are. Here we report on our efforts to globally map their locations and determine their physical properties using a combination of multiple seismic phases sensitive to the deepest layers of the Earth, 2- and 3-D seismic waveform modeling with the AxiSEM3D approach, and Bayesian inference. We present new maps of probable ULVZ existence covering more than 50% of the CMB by surface area. We show that as much as 20% of the CMB area may contain ULVZs and that several mega-sized ULVZs with length scales on the order of 1,000 km may exist.  ULVZs are found in all regions of the deep mantle from within LLVPs to regions of inferred past subduction. Furthermore, we show new details of the ULVZ beneath the Coral Sea and the Samoan mega-ULVZ, demonstrating it to be roughly 2× as large as presently thought and that it may contain a significant compositional component.