Seismic tomography: What does it tell us about Hawaii?
California Institute of Technology
Mantle plumes are hypothetical hot, narrow mantle upwellings originating at a thermal boundary layer that are often invoked to explain volcanism with unusual geophysical and geochemical characteristics. In models of seismic velocity one should look for mantle plumes as narrow, continuous near-vertical structures with anomalously low wave velocity.
In model S20RTS, we observe in the upper mantle transition zone low velocity structures beneath 8 of the 37 hotspots from the compilation of Sleep , including Hawaii. It is indeed possible to connect the low velocity anomaly in the transition zone beneath Hawaii to a broad region of anomalously low velocity in D" beneath the central Pacific. It is not obvious though, that this structure is the seismic signature of a mantle plume because it is strongly distorted and it is, due the inherent 1000-km scale lateral model resolution, much wider than expected for plumes.
While global models provide a global perspective of mantle
heterogeneity, it is my opinion that the detection of mantle plumes requires
1000-km aperture dense arrays that provide 100-km scale resolution below
the 670-km discontinuity. The spread of hot plume material below the 670-km
phase transition and its narrow tail in the transition zone are expected
plume attributes that cannot be seen in global seismic models at this
time. The averaging kernels at around 700 km depth are, in particular
for oceanic regions at least 400 km in depth. In order to decide whether
a feature is too small to be resolved, or simply does not exist, its size
must be estimated and compared with the model resolution in the volume
There was disucssion of recent results announced by a group at Princeton University, which purport to show narrow, vertical columms of low-wave-speed material extending throughout much of the mantle beneath Hawaii. The general feeling of the seismologists present was that it was unclear whether anything new had been found, since the data set used was known to exhibit these features in prior inversions. Insufficient information was as yet available regarding the method and resolution test performed, for the result to be objectively assessed.