Fine-scale structure of the mid-mantle characterised by global stacks of PP precursors

Abstract

Subduction zones are likely a major source of compositional heterogeneities in the mantle, which may preserve a record of the subduction history and mantle convection processes. The fine-scale structure associated with mantle heterogeneities can be studied using the scattered seismic wavefield that arrives as coda to or as energy preceding many body wave arrivals. In this study we analyse precursors to PP by creating stacks recorded at globally distributed stations. We create stacks aligned on the PP arrival in 5° distance bins (with range 70–120°) from 600 earthquakes recorded at 193 stations stacking a total of 7320 seismic records. As the energy trailing the direct P arrival, the P coda, interferes with the PP precursors, we suppress the P coda by subtracting a best fitting exponential curve to this energy. The resultant stacks show that PP precursors related to scattering from heterogeneities in the mantle are present for all distances. Lateral variations are explored by producing two regional stacks across the Atlantic and Pacific hemispheres, but we find only negligible differences in the precursory signature between these two regions. The similarity of these two regions suggests that well mixed subducted material can survive at upper and mid-mantle depth. To describe the scattered wavefield in the mantle, we compare the global stacks to synthetic seismograms generated using a Monte Carlo phonon scattering technique. We propose a best-fitting layered heterogeneity model, BRT2017, characterised by a three layer mantle with a background heterogeneity strength (ϵ=0.8%) and a depth-interval of increased heterogeneity strength (ϵ=1%) between 1000 km and 1800 km. The scalelength of heterogeneity is found to be 8 km throughout the mantle. Since mantle heterogeneity of 8 km scale may be linked to subducted oceanic crust, the detection of increased heterogeneity at mid-mantle depths could be associated with stalled slabs due to increases in viscosity, supporting recent observations of mantle viscosity increases due to the iron spin transition at depths of ∼1000 km.