Under southern Japan and Ryukyu Island Arc, the Pacific plate and the Philippine Sea plate both subducted under the Eurasian plate, forming a complex double-plate subduction system. Statistical results of seismic activity reflect that there are significant differences in the influence of the two subducted plates on the mantle transition zone. It is currently unclear whether the stagnant slab beneath this region is a single Pacific plate or a mixed stagnant body consisting of the Pacific plate and the Philippine Sea plate. The properties, geometric characteristics, and morphological changes, as well as its interaction with the mantle material, collectively affect the tectonic evolution of the East China Sea Shelf Basin and the Chinese continental margin. The accumulation of the stagnant slab within the mantle transition zone and its interaction with the mantle material can cause variations in the depth of the 660 km discontinuity at the bottom of the mantle transition zone. Cold stagnant slab material induces depression of the 660 km discontinuity. By studying the depth of the 660 km discontinuity and its nearby velocity structure, we can explore the depth variation characteristics of the stagnant slab in the mantle transition zone beneath the southern Japan and Ryukyu Island Arc, obtain the lateral differences of the plate subduction process. This will deepen our understanding of the integrity of plate subduction in the Northwest Pacific Ocean and contribute to the study of the geodynamic processes of plate subduction. When the seismic wave passes through the 660 km discontinuity, the triplicated seismic phases will be generated. Utilizing the arrival times and amplitude characteristics of the triplicated waveforms recorded by a dense seismic network, we can effectively constrain the morphology of the discontinuity and the velocity structure near it.

In this study, we selected a deep earthquake that occurred in the Izu-Bonin recorded by the China National Seismic Network （CNSN）. The earthquake had a focal depth of 472 km, which avoided the influence of the triplicated waveforms associated with the 410 km discontinuity. The surface projections of the P-wave turning points are located in the southern Japan and Ryukyu Island Arc, and the characteristics of the triplicated waveforms can be used to constrain the depth of the 660 km discontinuity and its nearby velocity structure. Through identification and analysis of the morphological characteristics in P-wave triplicated waveforms, we divided the study area into 13 profiles, delineating horizontal differences in velocity structures based on distinct features of triplicated waveforms features. Building upon previous research findings and the characteristics of observed waveforms, we constructed a foundational velocity structure model. Employing a grid search methodology, we generated various parameter models. Theoretical seismic waveforms were subsequently computed using reflectivity method and compared against observed waveforms. The optimal velocity structure was determined via maximization of cross-correlation coefficients. Specifically, during the fitting process of P-wave triplicated waveforms, we refined waveform selection to minimize interference from non-triplicated waveforms, thereby augmenting fitting precision.

Our results reveal that: ① There is a high-velocity layer above the 660 km discontinuity with depth of 490−720 km and P-wave velocity anomaly of 1.0%−3.0%; ② The 660 km discontinuity depressed 20−60 km with velocity increment of 1.5%−3.0%, and the depression depth gradually decreases from south to north.

Drawing on insights gleaned from seismic tomography and tele-seismic data, we propose that the thickness of the high-velocity anomaly layer above the 660 km discontinuity significantly surpasses that of the subducting Pacific slab. This discrepancy likely arises from the halting of the downward motion of the subducted oceanic slab within the mantle transition zone by the 660 km discontinuity. This interruption triggers a lateral shift in motion, leading to the entrapment and subsequent buildup of the slab material within the mantle transition zone. Moreover, compared to previous studies based on seismic activity and tomography, our findings position the high-velocity anomaly layer at a depth deeper than typically reachable by the subduction of the Philippine Sea Plate, suggesting its composition primarily comprises material from the stagnant western Pacific slab rather than from the Philippine Sea Plate. Additionally, the depression of the 660 km discontinuity ranges from 20 to 80 km, exceeding depths attributed solely to thermal anomalies. We suggest that this depression may also involve phase transformations within the stagnant material brought by the subducted slab. The gradual reduction in depression depth from south to north likely reflects variations in subduction dynamics along the north-south direction.