Hydrothermal characteristics of the Mienhua submarine volcano in the southernmost Okinawa trough

The Mienhua submarine volcano (MHV) is located in the southernmost Okinawa Trough and exhibits vigorous hydrothermal activity. This paper presents a detailed volcanic morphological analysis of the MHV, which has not been fully explored or discussed in terms of its hydrothermal characteristics and their relationship to hydrothermal activities. The MHV is situated at a water depth of 1370 m and has a width of 2.2 km and relief height of 220 m. The MHV is an asymmetric volcano composed of two summits in the center, rugged mounds in the west, and flat regions in the east. Two hydrothermal vents, Devil Chimney and Witch Mound were discovered through acoustic flares and a high backscatter intensity in the eastern flank. Based on three repeated mappings, no noticeable depth changes were observed at the acoustic flare zone compared to the area of the rugged mounds and two summits. In addition, a sediment core collected in a high backscatter intensity patch displayed low magnetic susceptibility, which could be influenced by the high methane levels in hydrothermal fluid that flows through sediments. An acoustically transparent zone adjoining active flares was observed in the sub-bottom profiles in the southeastern flank of the MHV, suggesting that the morphological and volcanic features are mainly concentrated in the southeast. Based on the seabed characteristics and the distribution of sills and reflectors in the multichannel seismic profiles, we inferred that the MHV is an off-axis vent that has experienced at least two stages of morphological development.


Introduction
The Okinawa Trough is a back-arc rift basin that includes the Ryukyu arc-trench system ( Fig. 1a) (Kimura et al. 1988;Lee et al. 1980;Letouzey and Kimura 1986;Sibuet et al. 1998Sibuet et al. , 1987Uyeda 1987). Hydrothermal activities form abundant massive sulfide deposits on the seafloor have been reported in the northern, middle, and southern Okinawa Trough (e.g., Chen et al. 2019;Chou et al. 2019;Glasby and Notsu 2003;Kimura et al. 1988;Nakamura et al. 2015;Uyeda 1987;Wen et al. 2016). Both temporal and spatial distribution of silldriven systems affect the lifespan of hydrothermal activities if a host sill is an off-axis system or the host sill becomes cooling (Teske et al. 2019). Thus, the distribution of hydrothermal activity is a key to revealing the way of magma activity.
The Mienhua submarine volcano (MHV), located in the southern Okinawa Trough (SOT), displayed vigorous hydrothermal activities (Fig. 1b). Chemosynthetic faunal communities were observed on a mound with a hydrothermal plume on the southern side of the MHV foothills (Chou et al. 2019;Kuo et al. 2019). Significant methane concentrations and slightly high 3 He/ 4 He ratios in bottom seawater were found around the crater cone of the MHV (Su 2017). The existence of hydrothermal activities may be accompanied by rapid changes in topography. For instance, the North Big 1 3 10 Page 2 of 15 Chimney mound in the middle Okinawa Trough presented a growing chimney with a height of 15 m within 25 months (Kawagucci et al. 2013;Nozaki et al. 2016). However, mapping data were previously unavailable for the MHV; as a result, the geophysical and geological characteristics associated with its hydrothermal activity are not well known.
Using a multibeam echosounder system, bathymetry, water column, and backscatter intensity, data can be collected simultaneously to study hydrothermal vents (Chadwick et al. 2018;Jones et al. 2010;Minami and Ohara 2020). Morphological changes in volcanic and hydrothermal areas have been reported using repeated bathymetric surveys. Furthermore, significant depth differences, acoustic water columns, and high backscatter intensity anomalies have been successfully used to identify hydrothermal features (e.g., Colbo et al. 2014;Nakamura et al. 2015). Additionally, repeated multibeam echosounder surveys have the potential to detect detailed variations in the morphology and characteristics of seabeds and, thus, can provide a better understanding of hydrothermal systems.
Here, we describe the results of triple repeated bathymetric mapping (Fig. 1c), backscatter intensity, and water column data of the MHV and then discuss the depth changes of the volcano and its hydrothermal features. The sub-bottom characteristics are identified by sub-bottom profiling and multichannel seismic data, and remotely operated vehicle (ROV) observations and core samples were analyzed to verify the spatial distribution of geophysical data. We aim to (1) describe the geophysical and geological characteristics of the MHV, (2) understand where hydrothermal activities occur, and (3) propose a conceptual model of the evolution of the MHV.

Geological background
The Okinawa Trough, which is part of the East China Sea, is an incipient intracontinental back-arc basin formed by extension within the Ryukyu trench-arc system (Fig. 1a) (Kimura et al. 1988;Lee et al. 1980;Letouzey and Kimura 1986;Sibuet et al. 1998Sibuet et al. , 1987Uyeda 1987). Sibuet et al. (1998) indicated that the subduction slab in the SOT is steeper than that in the middle Okinawa Trough. Zhang et al. (2019) reported that volcanic rocks from the SOT contain a more significant  Minami and Ohara (2020). Red triangles show subaerial active volcanoes from Wang et al. (2004). NTVZ Northern Taiwan Volcanic Zone. The box shows the location of b. b Regional bathymetry map (~ 100-m grid). The depth contour interval is 250 m. The box outlines the area shown in Figs. 1c, 2, 3a, and 4d. Thick black lines mark the location of multichannel seismic profiles shown in Fig. 6. Pink dots mark the location of normal faults observed on the seismic profiles. Gray dash lines mark the location of the multichannel seismic survey lines in the study area. The yellow stars indicate six main mineral potential sites. MHSC Mien-Hun submarine canyon, YK4-1 Yonaguni Knoll IV-1, FDV-1 and FDV-2 Fir Dragon Volcano 1 and 2, GLM Geolin Mound, PFZ Penglai Fault Zone. c Triple ship-borne multibeam survey tracks and sub-bottom profiling survey lines in the MHV proportion of subduction components than those from the middle Okinawa Trough. Different tectonic activities along the Okinawa Trough make the seafloor morphology broader and shallower in the north and narrower and deeper in the south (Sibuet et al. 1998). The initial opening occurred during the Miocene in the northern and middle parts of the Okinawa Trough, and the southern part of the Trough opened during the Pleistocene (Hsu et al. 2001;Park et al. 1998). This indicates that the emplacement and tectonic setting of the East China Sea shelf basin were controlled by the Okinawa Trough, which may be suited even westwards before the Miocene (Huang 1992;Lin et al. 2005).
The six hydrothermal fields in the back-arc basin of the SOT include the MHV, Yonaguni Knoll IV-1 (YK4-1), Geolin Mounds (GLM), Fire Dragon Volcano 1 and 2 (FDV-1 and FDV-2), and Penglai Fault Zone (PFZ)   (Fig. 1b). In addition to the MHV, two remarkable hydrothermal fields in the SOT are the YK4-1 and GLM (Fig. 1b). In the YK4-1 field, diverse microbial communities with black smokers originating from vital hydrothermal activities and liquid CO 2 have been detected (Nunoura and Takai 2009). YK4-1 is located in a symmetric rift system called the Yaeyama Rift. An on-axis intrusion is linearly continuous along the rift system and has led to the formation of hydrothermal sites (Arai et al. 2017;Minami and Ohara 2020). Gas plumes and temperature anomalies have also been reported in the GLM (Chou et al. 2019;Hsu et al. 2019). The GLM is located approximately 16 km from the MHV (Fig. 1b). Seismic characteristics indicate that E-W directional faults diminish toward the spreading center, where a blanking zone near the GLM is observed (Chen et al. 2020a). This finding indicates that the axial spreading center of the SOT is close to the GLM (Chen et al. 2020a;Hsu et al. 2019;Lin et al. 2019b).
Several volcanic groups have been observed at the southern end of the Okinawa Trough (Chang et al. 2017;Tsai et al. 2017). These volcanic fields from northern Taiwan to the East China Sea are known as the Northern Taiwan Volcanic Zone (NTVZ) (Fig. 1a), which is currently in its post-collision stage (Chang et al. 2017;Hsiao et al. 1998). Extensional magmatism has caused magmatic activity since the Quaternary (Wang et al. 2004). As a result, many onshore and offshore islets have been reported from south to north, including the Tsaolingshan, Kuanyinshan, Keelung Volcanic Group, Tatun Volcanic Group, Mianhua Islet, Pengjia Islet, Kobisho, and Sekibisho (Wang et al. 2004).

Bathymetry, backscatter intensity and acoustic water column
Three multibeam echosounder surveys were conducted on the MHV by the R/V New Ocean Researcher 2 (NOR2) in July 2019, R/V New Ocean Researcher 3 (NOR3) in August 2019, and R/V New Ocean Researcher 1 (NOR1) in February 2020. The three surveys covered two periods of 50 and 240 days to collect bathymetry (Figs. 2a and 3), backscatter intensity (Fig. 2b), and water column data (Fig. 4a, b) to investigate the seabed characteristics and seafloor changes of the MHV. The data-acquisition parameters for each survey are listed in Table 1. Mappings with 100% coverage, 50% overlap, and ~ 2000 m of swath width between survey lines were conducted within the three repeated surveys (Fig. 1c). Navigation positioning data were provided by the Kongsberg Seapath 380-5, a differential global positioning system (DGPS). At least four sound velocity profiles were obtained every day during the survey.
The bathymetry data were processed with CARIS software by accounting for sound velocity variations, tides, and standard quality control. The data quality of each survey met the International Hydrographic Organization (IHO) S-44 standards for 1A order (IHO 2020). The maximum total vertical uncertainty (TVU) and total horizontal uncertainty (THU) were 9.9 m and 12.9 m, respectively. A backscatter intensity map was created using SIPS Backscatter in CARIS software. Acoustic features referred to as "flares" in the water column were visually identified for each ping by using the swath editor tool in CARIS. Finally, the bathymetry data and backscatter images were gridded to 10 m.

Multichannel seismic data and sub-bottom profiles
Multichannel seismic (Fig. 1b) and sub-bottom profiling data ( Fig. 1c) were acquired by R/V Ocean Researcher 1 (OR1) during cruises OR1-1146 in 2016 and OR1-1205 in 2018. The source of the multichannel seismic system was based on three air gun arrays within 505 cubic inches of high-pressure air, and the shot spacing was approximately 25 m. Seismic data were recorded using a 72-channel streamer in 2016 and a 120-channel streamer in 2018. The receiver spacing of the streamers was 12.5 m to provide a common depth point spacing within 6.25 m. The seismic data were processed using trace editing, geometry building, bandpass filtering, amplitude compensation, predictive deconvolution, spiking noise removal, velocity analysis, normal move-out correction, trace balance, stacking, water velocity F-K time migration, and water column mute tools, for the multichannel seismic data.
The sub-bottom profiling data were collected using a Bathy-2010 chirp sonar with an 8 cm vertical resolution for up to 50 m penetration. Both the multichannel seismic and sub-bottom profiling data were extracted using the SEG-Y data format. Then, we input the processed seismic data and sub-bottom profiles into IHS Kingdom software for further display and interpretation.

ROV observations and sampling
The 3000 m-class ROV (Forum Subsea Technologies, Triton XLX 56) used in this study was 3 m long, 1.8 m wide, 2.2 m high, and weighed 5.2 tons. Two ROV dives were conducted by R/V Legend (LGD) to collect water samples, geothermal information, and full high-definition videos of the hydrothermal vent during the LGD-1902 and LGD-1908 cruises in 2019. During each dive, continuous video observations were collected within approximately 3 h, and the dives were approximately 7 km in length and performed along several gentle slopes of the MHV. The ROV observations can provide optical images as ground truths for the analysis of multibeam echosounder surveys. An METS methane sensor (S/N G56-E270) was mounted at the bottom center of the ROV to provide in situ measurements. The detected concentration of methane in this instrument presented a range between 0.05 and 10 μmol/L. We used a temperature probe and a self-made Teflon-lined pressure-retaining water collector operated by two ROV manipulator arms to measure the temperature and obtain the water samples. The temperature resolution of the sensors in these instruments reached 0.1 mK. The volume of the custom-made water container was 500 mL.

Sediment samples and analyses
Two gravity cores near the MHV, OR1-1231-MHV-1-GC (Core-G1) and OR1-1202-VB2-GC2 (Core-G2), were acquired for geochemical and sedimentological analyses (Fig. 2). Core-G1 recovered 92 cm of sediment from a water depth of 1408 m during cruise OR1-1231 in 2019. Core-G2 recovered 96 cm of sediment from a water depth of 1424 m during cruise OR1-1202 in 2018. The distance between the two core sites was 772 m. These two sites were designed to obtain sedimentary information and hydrothermal characteristics near the MHV and to understand the hydrothermal activities in the SOT. Pore water samples were collected using Rhizon samplers when the cores were retrieved on the deck (Dickens et al. 2007). The Rhizon samplers were inserted horizontally into pre-drilled holes 10 cm apart in the core liner, and the holes were sealed after sampling. Magnesium ion (Mg 2+ ) concentration in the pore water samples was determined using an Ion Chromatograph (882 Compact IC plus, Metrohm, Swiss). A multi-sensor core-logger (Geotek, NSCL-S) was used to measure the sediment properties of the whole core after it reached thermal equilibrium. A series of physical property measurements were conducted, including the gamma density, resistivity, and p-wave velocity. Magnetic susceptibility measurements were also applied at an interval of 0.5 cm along the cores. After the above scanning, the cores were split into working and archive halves.
Core surface photographs were taken, and core descriptions were obtained for the detailed deposit conditions within the archived halves. Discrete samples of core sediments, such as trace elements in the sediments, were taken from the working halves for further analysis.

Seafloor morphology
The characteristics of the MHV were illustrated using a multibeam echosounder bathymetric map (Fig. 2a) and a slope map derived from the same data (Fig. 4d). The MHV is located downstream of the Mienhua submarine canyon (Figs. 1b, 2). Two summits of the MHV are situated in the west and east, spaced approximately 400 m apart, and separated by a long depression from the two summits. The MHV is located at a water depth of 1370 m, with a height of 220 m, a basal diameter of 2200 m, an average slope angle of 8.9°, and a volume of 0.226 km 3 . A rough topography with several gully-like features is scattered across the western, northern, and southern flanks of the surveyed area. However, on the eastern flank of the surveyed area, a flat region was observed with a slope of < 5° (Fig. 4d). Thus, the MHV is an asymmetric volcano (Fig. 2).
Repeated mapping surveys revealed depth differences in small regions, although dramatic depth changes were not observed over 240 days (Fig. 3a). The depth change was mostly ± 10 m (Fig. 3a). Depth changes less than 10 m can be ignored because of the uncertainty (TVU and THU) of the data; thus, small differences in depth were discounted in the flat basin. For example, the eastern flank of the MHV showed gentle variations, with a mean of ± 5 m over 240 days (Table 2). However, depth changes greater than 10 m were located on steep slopes (Fig. 5). The maximum positive depth difference was 15 m and located near the two summits ( Fig. 3b) and the northern slope, at water depths between 1300 and 1400 m (Fig. 3c). In addition, the seafloor morphology in the two areas showed dynamic changes at 50, 190, and 240 days (Fig. 3). Decreasing depths were measured over 50 days. In contrast, increasing depth was observed over 190 days.

Backscatter intensity
The backscatter intensity map (Fig. 2b) shows widely distributed high backscattering patches from − 3 to − 16 dB in and around the volcano, especially in areas with gully-like features (Fig. 2b, dark gray). The flat smooth seafloor on the eastern flank of the MHV also exhibited a high backscatter intensity. Chou et al. (2019) reported hydrothermal activity with increasing water temperature anomalies in the eastern flank of the MHV, and the locations are shown as red circles in Fig. 2b. The distribution of the red circles is consistent with the high backscatter intensity patches.

Flare distribution identified by the water column
In the surveyed area (Fig. 2), 22 seepage sites were found in the eastern MHV (Fig. 4d). The backscatter intensities of the bottom in the seep sites were between − 5.8 and − 22.5 dB. A compilation of representative seabed characteristics beneath all flare groups is presented in Table 2. These gas bubble accumulations, with a length of up to several hundred meters, can be detected by the multibeam echosounder as gas flares or gas plumes in the water column (Fig. 4a, b). The distributions of the 20 detected flares after 240 days were the same except for F1-11 and F2-1 (Fig. 4d). F1-11 was observed off the volcano in the south, while F2-1 was located close to the other flare groups but further north.

Sub-bottom and multichannel seismic profiles
The sub-bottom profiling data across the volcanic cones were analyzed to reveal the sub-bottom characteristics of the MHV and their relationship with gas emissions. Three echo types were observed in the study area (Fig. 4c): (1) acoustically chaotic masses, (2) acoustically transparent masses, and (3) acoustically stratified sediments. The seabed's rough topography and flare distribution present almost acoustically chaotic masses. In the area near the flare-distributed zone, the flat seafloor shows acoustically transparent masses in shallow strata. Clear sub-parallel reflectors, such as acoustically stratified sediments, were observed at a distance from the location of the flares (Fig. 4d).
The multichannel seismic profiles show the volcanic features beneath the MHV and nearby areas. The strata present inhomogeneous amplitude reflectors from northwest to south (Fig. 6). The amplitude of the sub-parallel reflectors on the northwest side of the MHV is lower than that of the stratified reflectors on the south side of the MHV. Two vertical blanking zones were beneath the GLM and MHV (black arrows in Fig. 6b). Another vertical blanking zone is on the southeastern side of the MHV (blue arrows in Fig. 6b). Gently up-dipping reflectors (green triangles in Fig. 6b) and horizontal blanking zones (yellow blocks in Fig. 6b) were widely distributed across the sub-parallel reflectors. Several high-amplitude anomalies with saucer-shaped features were observed (red lines in Fig. 6b). These saucer-shaped features with high-amplitude anomalies have been suggested to represent volcanic sills in other igneous complexes (Cukur et al. 2010;Rocchi et al. 2007). In addition, dozens of faults have developed from the NTVZ to the SOT (Fig. 6). Moreover, compared with the southeastern flanks of the MHV, these faults do not emerge on the seafloor in the northwestern flanks. The chaotic reflectors are discontinuous near the MHV (Fig. 6). In addition, a bright amplitude reflector is located on the northwest side of the MHV (Fig. 6c), and the high-amplitude anomaly reflector extending to a normal fault is shallower than the anomalies in the NTVZ. Based on the seismic amplitude and fault distribution, the boundaries of different volcanic features were located near the MHV (Fig. 6).

ROV observations
Two hydrothermal vents (Fig. 7), Witch Mound and Devil Chimney, in the MHV were named according to their appearance in the ROV optical images (Table 3). The Witch Mound, which is approximately 2 m in height and presents imbricated cobble deposits (Fig. 7a), is covered by white tubeworms, mussels, and other biomes on the surface. Several venting spots were discovered in the ROV image due to the optic fluctuation caused by the upgoing hot water. The temperature of the venting fluid mixed with seawater is 2.6 °C higher than that of the surrounding seawater. The methane concentration around the Witch Mound reached 2.4 µmol/L. The Devil Chimney, which is approximately 4 m in height, has two chimney vents, and its appearance in the ROV image is similar to a two-horned demon. The left horn is 30 cm high, and the right horn is 20 cm above the head structure of the chimney (Fig. 7b). Gas bubbles were observed above the two chimney vents. The temperature of the venting fluid mixed with seawater near the left chimney vent was 8.4 °C higher than that of the surrounding seawater. The methane concentration around the Devil Chimney reached 0.8 µmol/L. Furthermore, the water sample collected from the vent had a sulfide smell, with a pH of 6.00. Compared to those of other hydrothermal sites in the northern and middle Okinawa Trough (pH = 4.65-5.70; Kawagucci 2015) and in the sediment-starved back-arc and mid-ocean ridge systems (pH = 2-5; Kawagucci 2015), the Devil Chimney in the SOT presents similar acidity to other hydrothermal sites in the northern and middle Okinawa Trough but much higher pH values than in the sediment-poor back-arc and mid-ocean ridge systems. The results showed that the decomposition of sedimentary organic matter buffered the pH due to the abundance of sediments in the SOT.

Sediment description and properties
Sediment cores acquired from high-and low-backscatter intensity seabeds exhibit different properties and characteristics. Core-G1, which was collected from a high backscatter seabed (Table 3), presents dark-grey silt within two light-grey fine sand layers of mixed silts and three fine sand layers (Fig. 8a). A pungent sulfide smell was observed when Core-G1 was retrieved on the deck. Sheet debris and black speckles were scattered in Core-G1. The bulk density result for Core-G1 can be split into two sections at a core length of approximately 40 cm. The density from the seafloor to 40 cm increases slightly from 1.39 to 1.62 g/cm 3 , whereas the density from 41 to 92 cm increases dramatically from 1.52 to 1.87 g/cm 3 . Core-G2 was collected from a low backscatter seabed (Table 3). It presents dark-grey homogeneous silt within two thin, fine sand layers (Fig. 8b), and its bulk density increased slightly downward from 1.41 to 1.76 g/cm 3 .
The magnetic susceptibility of the sediments and Mg 2+ concentration in the pore water samples showed variable characteristics between Core-G1 and Core-G2. Core-G1 had low magnetic susceptibility values below 8.5 × 10 −5 SI, whereas Core-G2 had high magnetic susceptibility values between 1.45 × 10 −4 and 2.96 × 10 −4 SI (Fig. 8). In Core-G1, the concentration of Mg 2+ in the pore water samples decreased with increasing depth (Fig. 8a). The average concentration was 42.14 mmol/L, and the lowest concentration was 29.80 mmol/L. Core-G2, however, showed a stable pattern at different depths (Fig. 8b), and the average and lowest

Seabed hydrothermal activities and related characteristics
Variations in backscatter intensity and magnetic susceptibility are used to analyze seabed hydrothermal characteristics. Backscatter is the measure of sound scattered back through acoustic reflection and scattering. Harder bottom type reflects more sound than softer bottom type; a rough bottom scatters more acoustical energy than a smooth one (Lurton et al. 2015). Thus, high backscatter indicates relatively hard material deposits or surfaces with high roughness and has been verified in many other volcanic areas (e.g., Innangi et al. 2016;Jones et al. 2010;Minami and Ohara 2020). In Fig. 2b, the flat seafloor on the eastern flank of the MHV shows a high backscatter intensity, and these high backscatter values did not appear to be caused by the rough topography, which suggests the presence of hard material deposits. In addition, the distribution of acoustically chaotic masses in the sub-bottom profiling images was beneath the high-intensity patches (Figs. 2b,4d). Acoustic signals appear to have difficulty penetrating the seabed with hard and coarse sediments. Furthermore, the distribution of high-intensity patches in Fig. 2b is consistent with the locations of hydrothermal activities previously determined by video surveys (Chou et al. 2019). The Witch Mound and Devil Chimney (Fig. 7) are also located at high backscatter intensities (Table 3). Based on the above observations, we infer that the hydrothermal seabed could form a cap covered by hard sediments, such as imbricated cobble and pyroclastic deposits. Thus, patches with high backscatter intensity could be an indicator of hydrothermal seabeds. Low magnetic susceptibility resulting from hydrothermal fluids occurs in MHV. The hydrothermal systems in the SOT develop at the sediment-rich continental margin with huge sediments sourced from the surrounding continents and islands (Huh et al. 2006;Nakamura et al. 2015;Sibuet et al. 1987). In general, the main iron-bearing mineral in marine sediments is magnetite, which has high magnetic susceptibility. Core-G2, the reference sample with a low backscatter intensity (Table 3), presents high magnetic susceptibility (Fig. 8b). However, the low magnetic susceptibility of Core-G1 (Fig. 8a), which was located near the flares (Fig. 4d), indicated a change in iron-bearing minerals in the sediments. The reaction of hydrothermal fluids with magnetite in sediments may have changed the mineral composition of Core-G1. The Mg 2+ concentration in the pore water of Core-G1 (29.8 mmol/L) was significantly lower than that of seawater (~ 54.1 mmol/L) and showed a significant downward trend with increasing depth (Fig. 8a). In addition, the hydrothermal fluids are characterized by Mg 2+ close to 0 mmol/L (Sakai et al. 1990), indicating the presence of distinct mixed hydrothermal fluids. The fluctuating density of Core-G1 also suggests that the sediments may have been disturbed by the Fig. 7 Optical images, seafloor temperatures and methane concentration measurements acquired by ROV dives at two hydrothermal vents: a Witch Mound and b Devil Chimney. Temperature and methane concentration anomalies are observed in the water column data around the two hydrothermal sites. See the location in Fig. 2a Table 3  fluid migration. As the main gas components of hydrothermal fluids in the Okinawa Trough are CO 2 , CH 4 , and H 2 S (Kawagucci et al. 2011(Kawagucci et al. , 2013Nakamura et al. 2015;Su 2019), we infer that the anaerobic methane oxidation reaction will lead to the change of magnetic susceptibility (März et al. 2008). Once a high upward methane flux occurs within a vent zone, it induces the reduction of iron-bearing minerals to pyrite (Novosel et al. 2005), followed by a decrease in magnetization, which reduces magnetic susceptibility.
In addition, methane flux may form authigenic iron sulfide nodules in sediments, which are described as black speckles in methane-enriched regions (Horng 2018). Similarly, black speckles and sheet debris were also observed in Core-G1 (Fig. 8a). Based on the above observations, the sediments of the MHV hydrothermal field are rich in methane fluids, leading to low magnetic susceptibility. This explanation also supports the hypothesized relationship between hydrothermal activity and magnetic anomalies reported by Huang et al. (2019).

Fluid migration and morphological change in the MHV hydrothermal field
Hydrothermal activities were mostly located east of the MHV hydrothermal field. First, most flares were distributed in a similar position on the eastern flank of the MHV over 240 days (Fig. 4d). Furthermore, most flares have a high backscatter intensity of approximately − 13 dB ( Table 2), indicating that gas emissions occur in the hard seabed. This is consistent with the distribution of acoustically chaotic masses (Fig. 4d). In fact, it is unexpected that the seafloor above the chaotic echo type can exhibit flares. The mechanism of gas emission involves fluid accumulation beneath the seafloor and the venting out of gases due to pressure variations. Once the fluids accumulate into high permeability layers, the accumulated fluids in the sub-bottom profiling data are visualized as an acoustically 'transparent' layer or 'wipe-out' zones (Papatheodorou et al. 1993). However, a hard seabed was observed, which may have resulted in less penetrations and unclear images, even if fluids have accumulated. Thus, although imbricated cobble and pyroclastic deposits cover the seafloor of the chaotic zone, we infer that the high porosity of pyroclastic material may play an important role in lateral fluid migration. A similar case was reported by León et al. (2014). Consequently, flares can be emitted from the seabed through local seafloor highs or structural fractures on the hydrothermal seabed. Fluid seepage in a hydrothermal field may lead to depth differences, such as chimney growth (Kawagucci et al. 2013;Nozaki et al. 2016). Once a large volume of fluid is present and vented out, such as during a volcanic eruption, the volcano can experience dramatic changes in morphology. Simultaneously, an eruption event can induce tremors and even earthquakes (e.g., Lin et al. 2005;Novosel et al. 2005). According to earthquake records from the Central Weather Bureau of Taiwan, one earthquake 22 km away from MHV, with a magnitude of 4.02, occurred within the survey period of 240 days. Figure 3 shows no significant differences in depth between the three bathymetry surveys. Thus, we ruled out the existence of large igneous activity during the three surveys.
Compared to dramatic hydrothermal activities, low-tomoderate hydrothermal activities may be revealed by highfrequency multibeam echosounder surveys. Small morphological features can be observed by ultra-high resolution autonomous underwater vehicle (AUV) mapping (e.g., Chen et al. 2020b;Minami and Ohara 2020). Because of resolution limitations, the morphological changes caused by small hydrothermal activities were ignored. However, Fig. 3b, c show dynamic changes in the two summits and the northern slope over 50 and 190 days, respectively. The decreasing and increasing seabeds exhibited a vertical reciprocating process. The vertical reciprocating seabeds are mostly located on steep slopes (Fig. 5), which probably indicates that the mapping was uncertain or the material for growth was reduced. Tremors have been widely observed in the survey area (Lin et al. 2019a(Lin et al. , 2007, which indicates abundant hydrothermal activity. The slope degree and shaking may cause some hills to collapse, which otherwise seems to erode over a period of 50 days (Fig. 3). Then, continuous small hydrothermal activities cause the seabed to increase, and the material was deposited over a period of 190 days (Fig. 3). These observations indicate that low-to-moderate hydrothermal activity may be displayed by high-frequency multibeam echosounder surveys. In contrast, the depth difference in the flare zone was less than 5 m, which cannot be identified in high-frequency multibeam echosounder surveys (Table 2), despite the flatness of the region. The flat zones were more active than the northern slope (Fig. 3c). These flares indicate that high-flux gas bubbles can destroy the initially grown material. Thus, although the seabed on the eastern flank of the MHV was active because of the flare, we did not capture the dramatic morphological changes.

Hydrothermal evolution in the MHV
The MHV is far from the axial spreading center of the SOT (Figs. 1, 6); thus, the MHV is likely to be an off-axis volcano. In a back-arc rifting setting, most volcanoes are distributed along a spreading center (e.g., Ishibashi and Urabe 1995). At the axial spreading center, heat is driven by both shallow sills and deep magma, which results in tectonic spreading and maintains a persistently high heat flow over timescales much longer than the life cycle of the off-axis sill (e.g., Arai et al. 2017;Lizarralde et al. 2010;Teske et al. 2019). However, the MHV is far from both hydrothermal fields, namely, the GLM and YK4-1 (Arai et al. 2017;Chen et al. 2020a;Hsu et al. 2019;Lin et al. 2019b;Minami and Ohara 2020). We noted that the amplitude of the reflectors and the emerging features of the faults are distinct between the north and south flanks of the MHV. The saucer-shaped sill in Fig. 6c is shallow and similar to the high-amplitude anomalies shown in the spreading center of the SOT. The evolution of the MHV was influenced by the SOT. In addition, we infer that the MHV developed via igneous activities that occurred over at least two intrusion phases. First, the asymmetric volcano shows that the backscatter intensity, volcanism morphology, and slope differ between the western and eastern flanks (Figs. 2, 4d). Gully-like features that deposit hard sediments with high backscatter intensity are observed in the western MHV (Fig. 2b), and the flat region in the eastern MHV also presents high backscatter intensity (Fig. 3). If these high backscatter intensity patches are hydrothermal seabeds (Fig. 2b), then the two extrusion phases can explain the morphological differences. Second, most of the flares were distributed in the eastern MHV. The distribution of the transparent zone in Fig. 4 is not at the center of the volcano. The black and blue arrows in Fig. 6b show blanking zones beneath the central and southeastern parts of the MHV, respectively, thus implying two heat vents. These distributions revealed an asymmetric pattern instead of a radiated distribution. Thus, the asymmetric evolution may explain why the flat region is more active than the rough topography.
Based on the observations and data analyses, we propose a simplified conceptual model to explain the impact of asymmetric heat on hydrothermal evolution (Fig. 9). In the initial stage (Fig. 9a), the volcano had a rough topography formed by an off-axis sill. For a long time, cooling occurred until the other sill supported the southeastern side of the volcano Fig. 9 Conceptual models present the positions of the hydrothermal seabed and transition zone in an off-axis vent (not drawn to scale). a Initial stage: a sill provided heat to form the hydrothermal seabed. b After a long time: cooling occurred after a long time. c A new heat source in the present: the other sill added heat to increase the distribution of the hydrothermal seabed and transition zone. Thus, the MHV becomes an asymmetric volcano Page 13 of 15 10 (Fig. 9b, c). The sills varied in temperature and viscosity, which induced variations in morphology during volcanic extrusion. At the same time, heat caused the hydrothermal seabed to transition from immature to mature, which also led to the development of a transition zone in the lateral distribution area (yellow seabed of Fig. 9c), such as in the transparent zone shown in Fig. 4. Thus, when hydrothermal circulation is an eastward shift, flares mainly emit off the center and are distributed on the eastern side (blue flares of Fig. 9c). This interpretation explains why the MHV has asymmetric hydrothermal activity. Similar to other off-axis volcanoes (e.g., Alexander and Macdonald 1996;Bosworth 1987;Leroy et al. 2010), the off-axis MHV may provide a potential opportunity to study sills and mantle plumes from the spreading center along the SOT.

Conclusions
(1) Devil Chimney and Witch Mound with high backscatter intensities were discovered at the MHV, and these two hydrothermal vents were confirmed to have hydrothermal characteristics using temperature and methane concentration probes.
(2) Low magnetic susceptibility was observed at a high backscatter intensity patch. We infer that the low magnetic susceptibility resulted from sediment-associated hydrothermal activity. (3) The MHV is an asymmetric volcano, and dramatic differences in morphology are observed, with rugged mounds in the west and flat regions in the east. Most flares were detected on the eastern side of the MHV region. The echo type of the flare zone belongs to the acoustically chaotic zone, which is located close to the acoustically transparent zone. Although no flares were detected in the acoustically transparent zone, we suggest that this transparent echo type is a transition zone from the seabed with active hydrothermal activity to the sediments. Thus, the off-center transition zone can reveal the distribution of the seabed with hydrothermal activity. (4) The MHV is an off-axis vent located approximately 16 km from the spreading center. The amplitude of the multichannel seismic reflectors and distribution of the sills and blanking zones indicates that the MHV is influenced by the SOT. In addition, hydrothermal activity is mainly located on the eastern side of the volcano.
If the life cycle of the sill is considered, the MHV likely presented morphological development in at least two stages. (5) Depth differences between the collapsing and hydrothermal activities were observed, although the MHV was scanned using ship-borne bathymetry. The multi-beam echosounder system with high-frequency surveys provides an effective method for detecting a volcanic region, excluding the active flare zone. However, AUV ultra-high-resolution mapping, long-term surveys (more than ten years), sampling, and rock analysis should be carefully examined to improve our understanding of volcano evolution.