Full-waveform Joint Inversion of Teleseismic Cross-convolution, Receiver Function, and Ambient Noise Data for High-resolution Lithospheric Seismic Imaging
Dec 12, 2024·
Mijian Xu
,
Kai Wang
,
Tianshi Liu
,
Nanqiao Du
,
Ping Tong
,
Qingya Liu
·4 min read
Abstract
Full-waveform inversion (FWI) is a popular technique for imaging lithospheric structures due to its high resolution and robustness. This technique plays a crucial role in understanding various geodynamic processes, including subduction, continental collision, and intraplate volcanism. However, conventional FWI are often limited by the trade-off between the source and structural parameters, leading to uncertainties in the inverted velocity models. Here, we propose a novel objective function for teleseismic FWI based on single and double-difference of cross-convolution. This approach eliminates the need for estimating the source time function during inversion, thus reducing the associated uncertainties. Additionally, we combine teleseismic cross-convolution, receiver function, and ambient noise data to invert for lithospheric S-wave velocity structure. The complementary sensitivity kernels and frequency contents of these different datasets enhance the resolution of the inversion. We apply the joint inversion technique to investigate intraplate volcanism and subduction processes, yielding high-resolution lithospheric velocity structures.
Date
Dec 12, 2024 12:00 AM
Event
AGU Fall Meeting 2024
Location
Washington, D.C.
Related Publications
Xu, M., Wang, K., Chen, J., He, J., Liu, Q., Liu, Y., Huang, Z., and Tong, P. (2025). Multilevel mechanisms driving intraplate volcanism in central mongolia
revealed by adjoint waveform tomography of receiver function and ambient noise data. Earth and Planetary Science Letters, 650:119137.
Xu, M., Wang, K., Chen, J., Yu, D., and Tong, P. (2023). Receiver function adjoint tomography for three-dimensional high-resolution seismic array imaging:
Methodology and applications in southeastern tibet. Geophysical Research Letters, 50(19):e2023GL104077.
Synthetic Test: thickened crustal model and comparison with conventional 1D RF inversion
We first conducted a synthetic test to evaluate the capability of the RF-FWI to recover the 3-D structure with a thickened crust. The results show that the shape of the velocity anomaly is well recovered, particularly for S-wave velicity. We further compared the RF-FWI with the conventional 1-D receiver function inversion. The results demonstrate that the RF-FWI can provide more accurate result.
Synthetic Test: continental subduction
We further conducted a synthetic test to evaluate the capability of the RF-FWI to recover the 3-D structure with a continental subduction. The results show high-resolution low-velocity continental slab and high-velocity mantle wedge.
The evolution of the RF-FWI for the continental subduction model.
Application of RF-FWI to SE Tibet
In this case, we selected teleseismic receiver functions recorded by a linear seismic array in southeastern Tibet. We used the adjoint method to invert for 3-D Vp, Vs, and density structures. The checkerboard test demonstrates the capability of the receiver function FWI to recover the 3-D structure of the lithosphere.
(a) Tectonics in the SE Asia. Gray lines denote major tectonic boundaries. The red box shows the study region. (b) Topographic map of the Ailaoshan-Red River shear zone and adjacent regions. The 24 red triangles denote seismic stations deployed by Nanjing University (NJU). The 14 blue triangles denote stations deployed by the ChinArray project. The two brown triangles are stations from China National Seismic Network. Blue lines represent boundary faults. The black box shows the boundaries of the model. The abbreviations are ALSF: Ailaoshan Fault; ARSZ: Ailaoshan Red River shear zone; RRF: Red River Fault. (c) Earthquakes with agglomerative clustering. The earthquakes are grouped into 31 clusters indicated by different colors. (d) Locations of virtual events. Colors denote weights of events.
The 3-D checkerboard test for the receiver function adjoint tomography. The perturbations in Vp, Vs and density of the target model relative to the initial model are shown in horizontal slices (a) at depths of 10, 30, 50, and 70 km, and in the vertical section (e) along the AA’ profile (green line in panel (a)). Panels (b)(d) show the recovered perturbations of Vp, Vs and density, respectively. Panels (f)–(h) show the inverted perturbations of Vp, Vs, and density in the vertical section, respectively.
(a) Map views of the Vs model at depths of 14, 28, 45, and 65 km. (b) Cross-section of the Vs anomalies relative to M00 along profile AA’ (red line in panel (a)). Panels (c) and (d) demonstrate cross sections of the Vs model along AA’ revealed by the receiver function adjoint tomography (RFAT) and the 1-D joint inversion of P-wave receiver functions (PRFs) and surface wave dispersions, respectively. The respective green and magenta arrows in panels (b) and (c) denote the extending direction of the low-velocity zone. The dashed lines in panels (b) and (c) denote the Moho. (e) The observed PRFs (black curves), the synthetic PRFs computed in the model generated by RFAT (red curves), and the synthetic PRFs computed in the final model generated by the 1-D joint inversion (blue dashed curves) at station X1.53188. The pink area indicates the time window for measuring misfits around the Ps phase between the observed and synthetic PRFs. (f) The corresponding back-azimuth of each event. (g) The measured misfit for each event.
Joint FWI of Receiver Function and Ambient Noise
We further combined the receiver function and ambient noise data to invert for the lithospheric S-wave velocity structure. The joint inversion technique enhances the resolution of the lithospheric structure. We conducted a retrieval test to evaluate the resolution of the RF-FWI, ANAT, and joint FWI. The results show that the joint FWI effectively combines sensitivity of the RF-FWI and ANAT.
Retrieval tests. Perturbations of R00 relative to the final model M10 (a–d), the model $R_{EGF}$ derived from ambient noise adjoint tomography (e–h) (ANAT), the model $R_{PRF}$ obtained via receiver function adjoint tomography (i–l) (RFAT), and the model $R_{Joint}$ derived from joint adjoint waveform tomography inverting receiver functions and ambient noise data (m–p) (JointAT).
Joint FWI of Teleseismic Cross-convolution, and Ambient Noise
We further propose a novel objective function for teleseismic FWI based on single and double-difference of cross-convolution. This approach eliminates the need for estimating the source time function during inversion, thus reducing the associated uncertainties. We combine teleseismic cross-convolution, and ambient noise data to invert for lithospheric S-wave velocity structure. We apply the joint inversion technique to investigate Yakutat subduction processes in the southern Alaska, yielding high-resolution lithospheric velocity structures.
Checkerboard resolution test in the southern Alaska.
