Research

Active projects

Samoa-Lau Ocean Observing Network (SaLOON)

We deployed 29 ocean-bottom seismographs (OBSs) and collected rock samples in the Tonga-Samoa region to investigate the interactions between the Tonga subduction zone and the Samoan mantle plume. Visit our project website: https://sites.google.com/msu.edu/saloon

Applications of machine learning on OBS data

Applications of machine learning in seismology have greatly improved our capability of detecting earthquakes in large seismic data archives. Most of these efforts have been focused on continental shallow earthquakes, but here we introduce an integrated deep-learning-based workflow to detect deep earthquakes recorded by a temporary array of ocean-bottom seismographs (OBSs) and land-based stations in the Tonga subduction zone. We develop a new phase picker, PhaseNet-TF, to detect and pick P- and S-wave arrivals in the time–frequency domain. The frequency-domain information is critical for analysing OBS data, particularly the horizontal components, because they are contaminated by signals of ocean-bottom currents and other noise sources in certain frequency bands. PhaseNet-TF shows a much better performance in picking S waves at OBSs and land stations compared to its predecessor PhaseNet. The predicted phases are associated using an improved Gaussian Mixture Model Associator GaMMA-1D and then relocated with a double-difference package teletomoDD. We further enhance the model performance with a semi-supervised learning approach by iteratively refining labelled data and retraining PhaseNet-TF. This approach effectively suppresses false picks and significantly improves the detection of small earthquakes. The new catalogue of Tonga deep earthquakes contains more than 10 times more events compared to the reference catalogue that was analysed manually. This deep-learning-enhanced catalogue reveals Tonga seismicity in unprecedented detail, and better defines the lateral extent of the double-seismic zone at intermediate depths and the location of four large deep-focus earthquakes relative to background seismicity. It also offers new potential for deciphering deep earthquake mechanisms, refining tomographic models, and understanding of subduction processes.

PhaseNet-TF is available at https://github.com/swei-seismo/PhaseNet-TF

Reference:

Xi, Z.#, S. S. Wei, W. Zhu, G. Beroza, Y. Jie#, N. Saloor# (2024), Deep Learning for Deep Earthquakes: Insights from OBS Observations of the Tonga Subduction Zone, Geophys. J. Int., 238(2), 1073-1088, doi: 10.1093/gji/ggae200. (Open Access).

Deep water and material cycle

The Pacific Plate picks up water near the trench, when the slab cracks, and seawater reacts with the rock to form hydrous minerals. Most of the hydrous minerals become unstableand break down 70 to 160 km beneath the surface, releasing a large amount of water that triggers intermediate-depth earthquakes. Some portion of the hydrous minerals, however, may survive to greater depths. How deep does water go? Is there any water in the mantle transition zone? What is the Earth's water budget? How does this change our understanding of the Earth's evolution? I am using various seismic techniques to investigate the water storage in the mantle.

Tonga-Lau subduction system

According to the theory of plate tectonics, oceanic lithosphere, the rigid outermost shell of the Earth, is generated along mid-ocean ridges, and finally subducts into the interior of the Earth along oceanictrenches. The Pacific plate subducts beneath the Australian plate along the Tonga Trench, and induces the volcanic islands of Tonga. On the other hand, a series of back-arc spreading centers, i.e. new born mid-ocean ridges, exist west of the trench in the Lau basin, an oceanic basin between Tonga and Fiji. Therefore, the Tonga-Lau-Fiji region exhibits both the birthplace and the grave of oceanic lithosphere within an area of a few hundreds miles. By studying the mantle structure in this region, we can better understand how the lithosphere is born, how it dies, and how these two processes interact with each other.

I am investigating the seismic velocity and attenuation structures of the mantle wedge by analyzing data from 52 broadband ocean bottom seismographs (OBSs) and 17 island-based seismic stations deployed in this region from 2009 to 2010. (Project ELSC supported by Ridge 2000 Program under the National Science Foundation).

One of our studies uses Rayleigh wave tomography to infer the distribution of partial melt below the Lau Basin, revealing an unexpected relationship between the amount of in-situ melt and the water content of the magma. This indicates that the water carried by the down-going Pacific plate enhances melt extraction, and plays a crucial role in melting mantle and producing new seafloor at the nearby Lau basin.

Intermediate-depth earthquakes, occurring at depths of 70-300 km, are observed at most subduction zones though their mechanism remains controversial, as the pressure (P) and temperature (T) are too high to allow brittle failure. More intriguingly, double seismic zones (DSZ), in which intermediate-depth earthquakes occur along two layers parallel to the dip of the subducting slab and separated by 20-40 km, have been observed in several subduction zones. We relocate intermediate-depth earthquakes in Tonga using data from local and global seismic stations to provide an unprecedented image of the Tonga DSZ. The maximum depth and seismicity patterns of the DSZ suggest a temperature activated mechanism of triggering earthquakes, which further implies the slab dehydration processes.

Press release:

A more general introduction to this research can be found in WashU newsroom: "To speed up magma, add water".
A podcast about this research interviewed by Hold That Thought from Arts & Sciences at WashU: "Discovery in the Lau Basin"
A more general introduction to this research can be found in WashU newsroom: "Release of water shakes Pacific Plate at depth".

Reference:

Wei, S. S., et al. (2015), Seismic Evidence of Effects of Water on Melt Transport in the Lau Back-arc Mantle, Nature, 518(7539), 395-398, doi: 10.1038/nature14113.
Wei, S. S., et al. (2016), Upper mantle structure of the Tonga-Lau-Fiji region from Rayleigh wave tomography, Geochem. Geophys. Geosyst., 17(11), 4705-4724, doi: 10.1002/2016GC006656.
Wei, S. S. and D. A. Wiens (2018), P-wave attenuation structure of the Lau back-arc basin and implications for mantle wedge processes, Earth Planet. Sci. Lett., 502, 187-199, doi: 10.1016/j.epsl.2018.09.005.
Wei, S. S. and D. A. Wiens (2020), High bulk and shear attenuation due to partial melt in the Tonga-Lau back-arc mantle, J. Geophys. Res., 125(1), e2019JB017527, doi: 10.1029/2019JB017527.

Alaska Peninsula

The Alaska Peninsula, with its magnificent diversity of seismic and magmatic activities, was identified by the community as a top priority site to address fundamental questions of subduction processes. Particularly, the along-strike variations and segmentation of seismic activity are critical for understanding the hydration state of the subducting slab, plate coupling, and earthquake hazard assessment. The newly recovered Alaska Amphibious Community Seismic Experiment (AACSE) and the adjacent stations of EarthScope Transportable Array and USGS Alaska Earthquake Center provide unprecedented high-quality seismic data to constrain this diverse subduction system.

Reference:

Wei, S. S., P. Ruprecht, S. L. Gable*, E. Huggins, N. Ruppert, L. Gao, and H. Zhang (2021), Along-strike variations in intermediate-depth seismicity and arc magmatism along the Alaska Peninsula, Earth Planet. Sci. Lett., 563, 116878, doi: 10.1016/j.epsl.2021.116878.
Wang, F.*, S. S. Wei, C. Drooff, J. L. Elliott, J. T. Freymueller, N. A. Ruppert, and H. Zhang (2024), Fluids control along-strike variations in the Alaska megathrust slip, Earth Planet. Sci. Lett., 633, 118655, doi:10.1016/j.epsl.2024.118655.

Hawaiian mantle plume head in the lower mantle

The Hawaiian-Emperor seamount chain that includes the Hawaiian volcanoes is created by the Hawaiian mantle plume. Although the mantle plume hypothesis predicts an oceanic plateau produced by massive decompression melting during the initiation stage of the Hawaiian hotspot, the fate of this plateau is unclear. We discovered a megameter-scale portion of thickened oceanic crust in the uppermost lower mantle west of the Sea of Okhotsk by stacking seismic waveforms of SS precursors. We propose that this thick crust represents a major part of the oceanic plateau that was created by the Hawaiian plume head about 100 Ma ago and subducted 20–30 Ma ago. Our discovery provides temporal and spatial clues of the early history of the Hawaiian plume for future plate reconstructions.

Reference: Wei, S. S., P. M. Shearer, C. Lithgow-Bertelloni, L. Stixrude, and D. Tian (2020), Oceanic plateau of the Hawaiian mantle plume head subducted to the uppermost lower mantle, Science, 370(6519), 983-987, doi: 10.1126/science.abd0312.

Press release:

IRIS Science Highlights (a wonderful article for the general public): The mystery of the missing plume head
A general introduction to this research can be found at MSUToday: MSU researchers discover 'missing' piece of Hawaii's formation

Mantle transition zone discontinuities and SS precursors

The mantle transition zone (MTZ), bounded by the 410- and 660-km seismic discontinuities, plays an important role in Earth’s evolution and mantle convection. The seismic discontinuity at 410 km depth is usually attributed to an isochemical phase transformation from olivine to wadsleyite. In addition to this globally observed feature, a low-velocity layer immediately above the 410-km discontinuity (LVL-410 hereinafter) has been observed regionally in many places. We analyze waveforms of SS precursors recorded by global stations to investigate lateral heterogeneities of upper-mantle discontinuities on a global scale. Our new results show a sporadic LVL-410 worldwide, including East Asia, western North America, eastern South America, the Pacific Ocean, and possibly the Indian Ocean. The best data coverage is for the Pacific Ocean, where the LVL-410 covers 33–50% of the resolved region. We interpret the LVL-410 as partial melting due to dehydration of ascending mantle across the 410-km discontinuity, which is predicted by the transition zone water filter hypothesis. The strong lateral heterogeneity of the LVL-410 in our observations suggests partial melting with varying intensities across the Pacific, and further provides indirect evidence of a hydrous mantle transition zone with laterally varying water content.

Reference:

Wei, S. S. and P. M. Shearer (2017), A sporadic low-velocity layer atop the 410-km discontinuity beneath the Pacific Ocean, J. Geophys. Res, doi: 10.1002/2017JB014100.
Tian, D.*, M. Lv, S. S. Wei, S. M. Dorfman, and P. M. Shearer (2020), Global variations of the 520- and 560-km discontinuities, Earth Planet. Sci. Lett., 552, doi: 10.1016/j.epsl.2020.116600.
Hao, S., S. S. Wei, and P. M. Shearer (2024), Substantial global radial variations of basalt content near the 660-km discontinuity, AGU Advances, 5(6), e2024AV001409, doi: 10.1029/2024AV001409. (Open Access)

A presentation about our discoveries in the mid mantle:

Past projects

Seismicity in the northern Tibetan Plateau (INDEPTH IV)

A total of 400 regional earthquakes were located in northern Tibetan Plateau from data recorded by INDEPTH IV and PKU Eastern Kunlun arrays from May 2007 to June 2009. The distribution of these earthquakes is compatible with a continuously deforming Tibetan lithosphere. Most earthquakes occur at a depth range of 0‐15 km, but no event is deeper than 30 km. This observation strongly supports the existence of a hot and weak lower crust beneath the northern Tibet. The crustal seismogenic zone appears slightly thicker beneath the northern Tibet than in the southern plateau, possibly reflecting a difference in the rheological (dry vs. wet) structure of the crust. The absence of lower crustal and uppermost mantle earthquakes in northern Tibet is consistent with a localized asthenospheric upwelling under the Qiangtang and Songpan‐Ganze terranes. Finally, the lack of mantle earthquakes should be fully addressed in any models of subduction in northern Tibet.

Project INDEPTH (International Deep Profiling of Tibet and the Himalaya) is a multidisciplinary geophysical and geological investigation of the Himalayas and Tibet. The passive seismic array of phase IV consisted 98 PASSCAL instruments deployed from May 2007 to June 2009, covering northern Tibetan Plateau.

Reference: Wei, S., et al. (2010), Regional earthquakes in northern Tibetan Plateau: Implications for lithospheric strength in Tibet, Geophys. Res. Lett., 37(19), L19307, doi:10.1029/2010gl044800.

Hainan mantle plume

The Hainan mantle plume has been proposed to explain the largest igneous province in Lei-Qiong region of southern China. We analyzed the data from regional broadband seismic stations using the receiver function method to study the seismic constrains of this plume. The shear wave velocity structures indicate high velocity anomalies in the upper crust and low velocity anomalies in the lower crust beneath the Leizhou Peninsula. This is the consequence of the eruption/intrusion of the mafic rocks in the upper crust and the partial melting in the lower crust. The migration images for the shallow structure of Lei-Qiong region show significant depression of the Moho discontinuity beneath the Leizhou Peninsula, where the Cenozoic basalt mostly outcropped, which is about 15 km deeper than that beneath the adjacent South China block and Hainan Island. This is consistent with the existence of the mantle plume as the up-welling mantle materials thickened the crust of Leizhou Peninsula. In addition, the migration imaging for the upper mantle shows a thinner transition zone (with the top boundary at 425 km depth and the bottom boundary at 650 km depth) beneath the Lei-Qiong region, which indicates the temperature is higher by ~200˚C than the surrounding mantle.

Reference: Wei, S. S. and Y. J. Chen (2016), Seismic evidence of the Hainan mantle plume by receiver function analysis in southern China, Geophys. Res. Lett., 43(17), 8978-8985, doi: 10.1002/2016GL069513.

Page updated

Report abuse