Deformation of the Tonga Slab: Evidence for Interaction with a Small-scale Secondary Plume in the Transition Zone

Abstract:

Magali I. Billen and John Z. Bikoba

The Tonga Slab is the most seismically active subduction zone providing a uniquely detailed picture of the internal deformation of the slab, with apparent warping and folding, from the surface through the transition zone. Here, we investigate the dynamical origin of a irregular feature in the seismicity within the transition zone located at 21-28^oS, using 3D visualization and analysis of the seismicity and compression/tension (P/T) axis from the moment tensor solutions to characterize the geometry of, and the orientation of forces acting on, the slab. This irregular feature can be described as narrow region of upward deflection of the slab, with a gap in seismicity beyond (down-dip of) the deflected region, and flanked by two narrow V-shaped gaps in seismicity suggestive of tearing of the slab. The P/T axis show a dominate down-dip orientation of the P axis above the deflection point, which rotate to a nearly vertical orientation within the central region of the deflected slab. The adjacent attached regions (down-dip of the two flanking slab gaps) also have rotated and more heterogeneous P/T axis orientations. In contrast, the adjacent section of the slab to the north of 21^oS has continuous seismicity throughout the transition zone, with a roughly uniform planar shape, and generally down-dip orientation of the P axis. We explore three possible hypothesis for the observed deformation including: 1) deflection due to a buoyant metastable olivine wedge, 2) a buckling feature in the slab as previously proposed by Myhill (GJI., 2013), and interaction with a small-scale, secondary plume upwelling below the slab. If the newly-observed gaps in seismicity indicate physical gaps or significant thinning of the slab, then these observations are not consistent with the buckling hypothesis. The lack of significant along-strike variation in slab age or subduction rate also suggests that a localized region of metastable olivine is unlikely. Therefore, we test the third hypothesis using a simple 3D geodynamical model of a planar dipping slab overlying a localized buoyant upwelling (radius < 150 km). We present comparisons of the observations to the model predictions for the subsequent deformation of the slab and orientations of principal stress axis within the slab.