Investigating the seismic structure and visibility of dynamic plume models with seismic array methods

Mantle plumes may play a major role in the transport of heat and mass through the Earth, but establishing their existence and structure using seismology has proven challenging and controversial. Previous studies have mainly focused on imaging plumes using waveform modelling and inversion (i.e. tomography). In this study we investigate the potential visibility of mantle plumes using array methods, and in particular whether we can detect seismic scattering from the plumes. By combining geodynamic modelling with mineral physics data we compute 'seismic' plumes whose shape and structure correspond to dynamically plausible thermochemical plumes.We use these seismic models to perform a full-waveform simulation, sending seismic waves through the plumes, in order to generate synthetic seismograms. Using velocity spectral analysis and slowness-backazimuth plots, we are unable to detect scattering. However at longer dominant periods (25 s) we see several arrivals from outside the plane of the great circle path, that are consistent with an apparent bending of the wave front around the plume conduit. At shorter periods (15 s), these arrivals are less obvious and less strong, consistent with the expected changes in the waves' behaviour at higher frequencies. We also detect reflections off the iron-rich chemical pile which serves as the plume source in the D region, indicating that D reflections may not always be due to a phase transformation. We suggest that slowness-backazimuth analysis may be a useful tool to locate mantle plumes in real array data sets. However, it is important to analyse the data at different dominant periods since, depending on the width of the plume, there is probably an optimum frequency band at which the plume is most visible. Our results also show the importance of studying the incoming energy in all directions, so that any apparently out-of-plane arrivals can be correctly interpreted.