Three‐dimensional electromagnetic imaging of upwelling fluids in the Kyushu subduction zone, Japan

Abstract

A three-dimensional (3-D) lithospheric-scale electrical resistivity model, developed using network-magnetotelluric (network-MT) data, contains structures associated with arc magmatism beneath Kyushu Island in the Southwest Japan arc. Kyushu Island, where the Philippine Sea Plate (PSP) subducts beneath the Eurasian plate, can be divided into northern and southern volcanic regions separated by a nonvolcanic region. Many active Quaternary volcanoes occur along the volcanic front (VF) associated with the PSP in the two volcanic regions. Our 3-D electrical resistivity model shows three different shapes of upwelling fluid-like conductive anomalies, indicative of either slab-derived aqueous fluid and/or partial melt beneath the volcanic and nonvolcanic regions. A conductive anomaly in the northern volcanic region, located at some distance from the subducting PSP, extends from the surface to depths of <100 km, whereas another conductive anomaly in the southern volcanic region, located along the subducting PSP at >70 km depth, extends from the surface to depths of >100 km. In the nonvolcanic region, the upper region of a relatively conductive anomaly extends upward to a depth of~50 km along the subducting plate. The degrees of magmatism and the relative contribution of slab-derived fluids to the magmatism vary spatially in the one nonvolcanic and two volcanic regions.

3‐D electrical resistivity structure based on geomagnetic transfer functions exploring the features of arc magmatism beneath Kyushu, Southwest Japan Arc

“…In this study, geomagnetic TF data were introduced to improve the resolution at crustal depths of an electrical resistivity structure model obtained previously by using only Network‐MT data [ Hata et al ., ]. Observed three‐component geomagnetic variations are converted to TFs, expressed as Hz ( r , T ) = Ax ( r , T ) Hx ( r , T ) + By ( r , T ) Hy ( r , T ), where Hz ( r , T ) is the vertical geomagnetic field, Hx ( r , T ) and Hy ( r , T ) are the horizontal components of the geomagnetic field, r is the location of observation sites, and T is the period.…”

“…Further theoretical details of the Network‐MT method and analysis are described by Uyeshima []; details of the Network‐MT observations on Kyushu and the processing procedure for the Network‐MT RFs are described by Hata et al . [, ]. In our previous 3‐D inverse modeling, the RFs for 72 dipoles with high S/N ratios (Figure a) were determined for 14 periods in the range of 480–40,960 s, using the RRRMT data processing code [ Chave et al ., ; Chave and Thomson , ].…”

“…Little is known about the cause of the separation of the active volcanic regions in Kyushu and about the relationship between magmatism (and/or volcanism) and electrical resistivity structures in the mantle wedge beneath Kyushu. To explore these questions in our previous studies, we performed two‐dimensional (2‐D) and three‐dimensional (3‐D) inversion analyses to obtain electrical resistivity structures beneath all of Kyushu, using Network‐Magnetotelluric (Network‐MT) data obtained at 14 periods between 480 and 40,960 s; these data contain geoelectromagnetic information at lithospheric scales [ Hata et al ., , ]. Two major findings of previous studies are described below.…”

“…In this study, we generated a 3‐D electrical resistivity structure model by inverting available geomagnetic TF data, obtained on Kyushu and three small islands (as shown in Figure a), by using a 3‐D inversion code [ Siripunvaraporn and Egbert , ]. A previous 3‐D model [ Hata et al ., ], which was based on inverted Network‐MT data (black lines with two dots in Figure a) by using another 3‐D inversion code [ Siripunvaraporn et al ., ], is used as the starting model for the 3‐D inverse modeling in this study. Our new 3‐D model, for which the synthetic induction arrows are consistent with observed arrows, provides a higher‐resolution model than the previous model [ Hata et al ., ], especially at shallow depths.…”

“…A previous 3‐D model [ Hata et al ., ], which was based on inverted Network‐MT data (black lines with two dots in Figure a) by using another 3‐D inversion code [ Siripunvaraporn et al ., ], is used as the starting model for the 3‐D inverse modeling in this study. Our new 3‐D model, for which the synthetic induction arrows are consistent with observed arrows, provides a higher‐resolution model than the previous model [ Hata et al ., ], especially at shallow depths.…”