Can lower mantle plumes be located using seismic tomography and transition zone topography?

Hotspot volcanism is well explained by fluid dynamical models of plumes rising from great depth in the Earth's mantle. Seismic imaging of these suggested plumes has been difficult and somewhat controversial, as imaged low velocity regions in tomographic models may be due to choices in the inversion techniques. Observations of transition zone topography may be a more robust indication of the presence of hotter-than-normal mantle, but in general the topography of the 670-km phase boundary appears flat, suggesting in part that hotspots have an upper mantle origin. We present new high-resolution finite element models of plumes rising from the core-mantle-boundary using compressible convection with phase changes. Plumes that have a buoyancy flux similar to that of the Hawaiian or Iceland hotspots show a broad high velocity conduit in the lower mantle with significant thinning and time-dependent behavior in the upper mantle. Seismic wave propagation modeling using SHAXI demonstrates that the presence of the low velocity plume causes a distinct later arrival of seismic waves that travel through the upper mantle, but wave front healing renders the strong velocity contrasts in the lower mantle invisible at the Earth's surface. The predicted 400-km phase boundary topography occurs over short wavelengths and is consistent with the observations at a significant number of hotspots. The broad plume anomaly in the uppermost lower mantle causes a slight and very broad uplift of the 670-km topography. This is consistent with the absence of significant topography of 670 as seen by the limited aperture seismic studies. Our modeling suggests that the absence of 670 km topography and difficulties of imaging plumes in tomographic models are inherently due to plume morphology and wave front healing.