Identifying slab fragments in the lower mantle by comparing seismic and plate reconstruction models

We combine results from seismic tomography and plate motion history to investigate structure and evolution of slabs of subducted lithosphere in the lower mantle beneath western Pacific. We selected earthquakes and stations along a wide corridor from the Aleutians to Australia. We measured P and S differential travel times by waveform multi-channel phase cross-correlation (MCPCC). MCPCC enables to measure differential travel times without selecting a reference trace. We use phase cross-correlation rather than standard cross-correlation as it is better suited to measure differential travel times on the first onset even if the first onset is not the most energetic peak. Direct P and S wave measurements were complemented by PP-P, PcP-P, SS-S and ScS-S differential travel time measurements. Both P and S datasets were then inverted simultaneously to obtain high resolution P and S wave velocity models. Synthetic test were performed to assess the models quality and results are only interpreted in the well resolved areas. Paleomagnetic data were used to reconstruct the plate motion of western Pacific from 130 My ago to present. 130 My ago, the Panthalassa Ocean was made of 4 plates. The Pacific plate was initially separated by ridges to the Izanagi plate to the North-East, the Farallon plate to the East and the Phoenix plate to the South. The Izanagi plate subducted beneath Eurasia until 60 My ago and disappeared in the mantle when the ridge between Izanagi and Pacific plates subducted. In the South West Pacific, the Phoenix plate subducted beneath the Gondwana until 120My ago. Between 120-45 My ago a left-lateral transform fault took place between the Lord Howe Rise plate and the remaining portions of the Phoenix plate. During this period, a rift opened between the Australian plate and Lord Howe Rise plate. From 45 My to now, subduction took place between Pacific and Lord Howe Rise plate. Meanwhile, Australia and India plate have moved toward the North since 70 My and the collision between India and Asia occurred about 30 My ago. This northward motion induced the East-West opening of the West Philippine Basin and its rotation clock-wise between 50 and 40 My ago. This basin was then called the Philippine plate. The Philippine plate moved toward north from 30 My ago 10 My following the northward motion of Australia. 20 My ago, Philippine also rotated clockwise due to both the opening of the Shikoku and Parece Vela Basins and to the split of the Izu-Bonin Arc into two parts. The Philippine plate has been surrounded by subductions, the Ryuku and Philippine trenches to the west and the Izu-Bonin and Mariana trenches to the east since 10 My. Errors on the plate reconstruction models are due to the hypothesis of fixed hotspots and to the difficulty of determining precisely oceanic plate boundaries and plate relative motions. Therefore, we compare these models with tomographic images to obtain independent constraints on plate motion history. We identify the different slab fragments in the lower mantle and we are able to unravel subduction history over the last 120 My.