The magma plumbing systems of volcanoes in subduction and divergent tectonic settings are relatively well known, whereas those of intraplate volcanoes remain elusive; robust geophysical information on the magma pathways and storage zones is lacking. We inverted magnetotelluric data to image the magma plumbing system of an intraplate monogenetic volcanic field located above the stagnant Pacific slab in northeast China. We identified a complex, vertically aligned, low-resistivity anomaly system extending from the asthenosphere to the surface consisting of reservoirs with finger- to lens-like geometries. We show that magma forms as CO2-rich melts in a 150-km-deep asthenospheric plume crossing the whole lithosphere as hydrated melt, inducing underplating at 50 km depth, evolving in crustal reservoirs, and erupting along dikes. Intraplate volcanoes are characterized by low degrees of melting and low magma supply rates. Their plumbing systems have a geometry not so different from that of volcanoes in subduction settings.

Volcanoes on Earth are concentrated along subduction zones, mid-ocean ridges, and continental rifts. Volcanism in intraplate settings is poorly understood and is interpreted as the result of mantle melting processes induced by the rising of hot plumes or by the release of fluids from stagnant slabs (Chen et al., 2017). The magma plumbing systems of intraplate monogenetic volcanoes above stagnant slabs are enigmatic and lack robust geophysical constraints. In particular, the full magma pathway from the asthenosphere to the surface as well as the size and geometry of the storage zones are poorly known (Smith and Németh, 2017). Northeast China has monogenetic volcanic fields such as the Jingpohu monogenetic volcanic field (JMVF) and a metasomatized mantle plume located above the ~500-km-deep stagnant Pacific slab (Figs. 1A1D; Chen et al., 2017; Li et al., 2020b). We obtained well-resolved images of the JMVF magma plumbing system from a depth of ~150 km to the surface by inverting magnetotelluric (MT) data, which are useful in identifying magma storage zones below volcanoes (Hill et al., 2009; Comeau et al., 2015). Data were collected from an array of MT stations centered on the JMVF (Fig. 1E). We recognized a complex resistivity pattern below the JMVF. The observed anomalies, integrated with rock-physics models, provide a well-constrained picture of the structure of the plumbing system of a monogenetic volcanic field from the asthenosphere to the surface. Our results at the JMVF provide new insights into the geometry, size, and depth of magma reservoirs for volcanoes in intraplate settings (Smith and Németh, 2017).

JMVF activity mainly developed in the Pleistocene and Holocene, with the last eruptions occurring in historical times (2470 and 3940 yr B.P.; Wei et al., 2003). The JMVF vents follow two main alignments: a large-scale northeast-southwest alignment and a local north-south alignment (Fig. 1F). The northeast-southwest alignment is controlled by the left-lateral, ENE-WSW– to northeast-southwest–striking Dunhua-Mishan fault (DMF) system, which affects the entire crust and the mantle to depths of 35–40 km (Xu et al., 2017). The Yilan-Yitong fault (YYF) system is located north of the JMVF and bounds the Songliao sedimentary basin. The JMVF eruptions produced ~2 km3 of basaltic, trachybasaltic and basanitic lava flows and scorias. The compositions of basalts record mixing between enriched mantle 1 (EM1) and depleted mantle (DM) sources and contamination by lower-crust material (Zhang et al., 2000; Yan and Zhao, 2008). The alkaline volcanism of northeast China, including the JMVF, involved carbonate melts (C melts) supplied from the mantle transition zone (MTZ) by the eclogites in the stagnant Pacific slab and peridotites associated with the overlying mantle wedge (Figs. 1B1D; Xu et al., 2020; Zhang et al., 2020).

Magnetotelluric data were collected using nine long-period MT (LMT) and 31 broadband MT (BBMT) stations (Fig. 1E) arranged around the JMVF. We deployed 25 BBMT instruments in a grid with an average spacing of ~33 km. In the central area, six BBMT stations were added to the network to better constrain the crust structure beneath the JMVF. The BBMT data were collected using a minimum acquisition time of 20 h and processed to obtain full impedance with periods from ~0.003 s to 2000 s, while >20 d LMT data had responses with periods from 10 s to ~20,000 s. Representative apparent resistivity and phase curves of off-diagonal and full-impedance tensors around JMVF are reported in Figures S1 and S2, respectively, in the Supplemental Material1. A three-dimensional, nonlinear, conjugate-gradient inversion technique was applied to convert the MT responses to resistivity images from data sets including off-diagonal and full-impedance tensors. Compared to the off-diagonal impedance inversion model, the full-impedance inversion model showed a more complicated resistivity pattern at crustal depth (Fig. S3), likely due to some distortion of diagonal impedances by noise (Fig. S2). As the two prominent anomalies in the mantle are key to understanding the deep structure below JMVF, we preferred the inversion model from off-diagonal impedances. Selection of this model also considered the facts that the fitting of the full-impedance tensor was not satisfactory, that of off-diagonal data was good (Fig. S1), and the diagonal component was often significantly smaller than the off-diagonal impedance (e.g., curve on site 47 in Fig. S2). Details on the data elaboration, inversion methods, evaluation of the quality of the inversion from off-diagonal impedance tensor data, and sensitivity analysis are provided in the Supplemental Material and Figures S4–S9.

Figure 2 shows the resistivity structure of the JMVF and surrounding area. The anomaly pattern is characterized by deep-rooted, vertically aligned, low-resistivity zones in the mantle and shallower anomalies at the Moho and in the crust. From the surface to ~20 km depth, the crust beneath the JMVF is generally highly resistive, with resistivity values up to ~104 Ω·m (anomaly R in Figs. 2D and 2E). These values are consistent with the occurrence of the Precambrian and Paleozoic plutonic rocks of the Zhangguangcai Range (Xu et al., 2017). Localized sheet-like low-resistivity anomalies also exist, which are mostly distributed along the DMF at depths of 10–15 km (Figs. 2D and 2E), reflecting the presence of intracrustal fluid-rich zones, the depths of which are possibly controlled by crustal discontinuities (Wannamaker et al., 2009). A low-resistivity zone (anomaly c in Figs. 2A and 2E) is observed between depths of 20 and 40 km. This ~50-km-wide anomaly is located to the north of the JMVF in the middle-lower crust. The average resistivity is ~2–3 Ω · m, but in the central part, it is characterized by values <1 Ω · m (Fig. S3). Below the JMVF, an up to 50–70-km-wide, low-resistivity anomaly, anomaly d (~20 Ω · m), extends down to the lithosphere-asthenosphere boundary at a depth of ~75 km (Figs. 2B, 2D, and 2E; Yan and Zhao, 2008). At depths of ~130 km, the mantle becomes less resistive than 25 Ω · m in a spindle-like zone slightly tilted upward to anomaly d. We defined this low-resistivity zone as anomaly e (Figs. 2C and 2D). It probably represents the top of an up to 400-km-deep, low-resistivity anomaly detected in the asthenosphere (Fig. 1C; Li et al., 2020b). In its central area, the resistivity becomes as low as 15 Ω · m. In this moderately resistive situation, the 20,000 s period of our LMT data is sufficient to guarantee such a depth of investigation. The resistivity of the anomalies detected must be interpreted with some caution because MT data are primarily sensitive to the integrated conductance of conductive bodies. However, the inversions with identical data coverage to real observations demonstrate that the detected anomalies at depths from 20 km to 150 km are resolved rather well (Fig. S6). The robustness of the single anomaly described above is demonstrated by the results of model perturbation analysis (Figs. S8 and S9).

Anomaly e in the asthenosphere may be ascribed to high-temperature carbonate melts, peridotite, and/or hydrated melts in the asthenosphere (Fig. 3). Producing this low-resistivity anomaly with only high temperatures requires temperatures as high as ~1800 K (Gardés et al., 2014), which are incompatible with the pressure and temperature conditions for peridotite melts (~1500 K) and exceed the mantle adiabat (~1600 K). Alternatively, we can use olivine containing ~200 ppm H2O to interpret the observed resistivity. However, such a high water content corresponds to ~700–1000 ppm H2O in peridotite, i.e., well above the mantle water storage capacity (Hirschmann et al., 2009); in addition, this range of H2O contents is not consistent with the relatively dry source(s) of the JMVF and northeast China basalts (Hsu and Chen, 1998). These difficulties lead us to associate anomaly e with C melts (Gaillard et al., 2008), where ~0.03 vol% of interconnected C melts may reproduce the observed resistivity values (Fig. 3). The reduction in S-wave velocity (Fig. 2D; Kang et al., 2016) in correspondence with anomaly e could reflect the occurrence of such melts. Results from the MAGLAB platform (Massuyeau et al., 2021) suggest a melt with ~300 ppm CO2 and 240 ppm H2O for anomaly e (Fig. S10). The widespread mantle-derived CO2 around the JMVF area (Liu et al., 2018), the experimental petrological data on the stability of C melts (Zhang et al., 2020), the presence of alkali olivine basalts (Zhang et al., 2000), and the available isotopic data (δ26Mg values of −0.6‰ to −0.30‰ in basalts) (Li et al., 2017) support our interpretation. Anomaly e thus represents a volume where C melts occur beneath intraplate volcanoes, a feature mainly observed in divergent geodynamic contexts (Key et al., 2013; Sifré et al., 2014). C melts from the asthenosphere ascend across their stability field at a depth of ~120 km (Key et al., 2013; Sifré et al., 2014). Consequently, the low resistivity of anomaly d, originating from a depth of ~100 km, should be ascribed not only to C melts (Key et al., 2013) but also to peridotite partial melting (Yan and Zhao, 2008) triggered by ~200 ppm H2O (Fig. 3; Hirschmann et al., 2009). The melting in anomaly d may be only ~1–2 vol% with a relatively low water content (1–2 wt%; Fig. S11; Hsu and Chen, 1998), and the low resistivity at the top of anomaly d suggests magma accumulation at the base of the crust, which could explain the ~2%–3% seismic velocity decrease (Fig. 2E; Fan and Chen, 2019). In the central sector, anomaly d shows a resistivity of 2–3 Ω·m, a value suggesting ~10 vol% of melt in its core (Fig. S11). The anomalies detected below the JMVF suggest, on average, low melting fractions (4–6 vol%) and water contents (<1.5 wt%). These values are significantly lower than the up to 12 vol% melting fraction and 2.5–9 wt% water content of the sources feeding the volcanoes in subduction settings (Comeau et al., 2015; Laumonier et al., 2017).

Therefore, anomaly d represents the mantle source of the JMVF basalts. In this scenario, the EM1 component of the JMVF basalts (Hsu and Chen, 1998) could be lithospheric, thermally eroded underplated material (Konter and Becker, 2012), implying active underplating processes. In the JMVF, ~2 km3 of basaltic lavas erupted from the source in ~8000 yr (Yan and Zhao, 2008), equal to an average magma production rate of 2.5 × 10−4 km3/yr, a value within the 10−5–10−3 km3/yr range estimated for alkaline volcanoes in intraplate settings and significantly lower than that of subduction volcanoes (0.4–0.6 km3/yr; Crisp, 1984). We suggest that the JMVF magmatic system is characterized by a high ratio between intrusions and erupted products. This observation and the low melt fraction and water content of the JMVF reservoirs may explain why the magma production rate of intraplate volcanoes is lower than that of volcanoes in subduction settings. The low-resistivity midcrust anomaly c is located just above anomaly d. We interpret anomaly c as the shallow magma reservoir of the JMVF. This reservoir could have been the source of the andesitic magma erupted in the Eocene (Qin et al., 2008). The 1–2 Ω·m bulk resistivity of anomaly c implies an ~10 vol% melt with 2.5 wt% water (Li et al., 2020a). The presence of this midcrustal hot melting zone is supported by (1) a low-velocity zone in the same region (Fig. 2E; Pang et al., 2016), (2) an ~100 kg/m3 density decrease at 20–30 km below the JMVF (Suo et al., 2015), and (3) the occurrence of seismicity confined to the upper 12 km of the crust (Fig. 2D). The hypocentral depths suggest a brittle-ductile transition shallower than the ~20 km depth expected for an intraplate setting with underlying Paleozoic basement (Xu et al., 2017). This could reflect the thermal anomaly associated with the low-resistivity anomaly c. Although the conduits for eruptions cannot be seen in our image due to the large instrument spacing, transcrustal magma transfer can occur, according to field data, along the dikes associated with major faults (Xu et al., 2017).

Our results reveal the multilevel structure of the JMVF plumbing system, which is characterized by (Fig. 4): (1) a deeper zone located in the asthenospheric mantle fed by a C-melt–rich plume, possibly ascending from the MTZ above the stagnant Pacific slab (Ma et al., 2019; Xu et al., 2020); (2) an intermediate storage zone extending from the upper mantle to the Moho, in which hydrated peridotite melts form and underplating processes develop; (3) a shallower, intracrustal reservoir; and (4) a surface, fault-guided dike system responsible for the alignment of monogenetic vents. With the exception of the asthenospheric mantle zone, this configuration is unexpectedly very similar to that of silicic stratovolcanoes typical of arc settings (Afanasyev et al., 2018).

Our study reveals the complete architecture of the magma plumbing system of an intraplate volcanic field above a stagnant slab. Our results suggest that the main differences between silicic, mainly explosive subduction volcanism and lower-energy, basaltic intraplate volcanism are related to the low water content and melt fraction of the mantle above stagnant slabs and not to the geometry of the plumbing system. Our data highlight the role of C melts and poorly hydrated silicate melts in the formation of magma reservoirs beneath intraplate volcanoes.

We thank G. Egbert for sharing his ModEM package codes. We are grateful to Denghai Bai, Xiangyun Hu, and V. Sepe for critical discussions. We are grateful for the computational resources from Jilin Kingti Geoexploration Tech, Ltd. (Changchun, China). This work was supported by the National Natural Science Foundation of China (grants 42074080 and 91858211) and Istituto Nazionale di Geofisica e Vulcanologia (Italy) funds to G. Ventura. We gratefully acknowledge editor Chris Clark for handling, and Hanchao Jian, Fabrice Gaillard, and an anonymous reviewer for manuscript reviews.

1Supplemental Material. Details on the 3-D magnetotelluric inversion, sensitivity test of resistivity anomalies and melt fraction estimation, and Figures S1–S11. Please visit https://doi.org/10.1130/GEOL.S.14903307 to access the supplemental material, and contact editing@geosociety.org with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.