Downhole data and cores collected during International Ocean Discovery Program (IODP) Expedition 376 at Brothers volcano, Kermadec arc, provide unprecedented, in situ views of volcanic facies and fluid pathways in an actively forming volcanogenic massive sulfide (VMS) ore deposit. Brothers volcano is a submarine caldera with extensive sea floor hydrothermal alteration. Downhole data were collected in two holes: Hole U1530A at the NW Caldera and Hole U1528D at the Upper Cone.

Textural analysis of microresistivity images in Hole U1530A provides a continuous image facies record that greatly improves findings based upon sporadic and partial (18%) core recovery. Between 90 and 214 meters below sea floor (mbsf), the heterogeneous image facies with local pattern variations is consistent with the volcaniclastic facies interpreted from cores. Between 232 and 445 mbsf, a volcanic facies was not recognizable in cores because of overprinting alteration, apart from five intervals of coherent lava flows that were less altered. Based on the fairly constant petrophysical data, Vp-porosity relationship, and presence of five to six coherent image facies intervals on the microresistivity image, we propose that the apparent volcaniclastic textures observed on cores and microresistivity images beneath 232 mbsf are dominantly lava flows. The change from volcaniclastic to dominant lava flow facies occurs over a transition zone (214–232 mbsf) where all petrophysical properties gradually change. In Hole U1528D, cores and petrophysical data show a similar transition from deep coherent lava flows to shallower, largely volcaniclastic sequences at ~270 mbsf.

Down to 232 mbsf in Hole U1530A and 360 mbsf in Hole U1528D, the overall first-order downward decrease in porosity is interpreted to be caused by compaction and increased alteration intensity. Volcanic facies and fractures exert a second-order local control on petrophysical properties. Beneath 232 mbsf in Hole U1530A, the prolonged hydrothermal activity is inferred to have diminished local petrophysical property variations within the proposed lava flow-dominated rock package. High downhole fluid temperatures in Hole U1528D contrast with the moderate temperatures in Hole U1530A. Permeable zones show a mix of structural (inferred fault in Hole U1530A) and lithological controls in both holes. Some low-permeability layers and/or lithological interfaces possibly focus fluids laterally in higher-permeability layers, which may act as a trap for metal-rich fluids to form stratabound massive sulfides and deposits. Matrix is likely too low in permeability to conduct fluids but provides perfect conditions for the storage of super saline brines. In Hole U1530A, located near active vents at the sea floor, the relatively low fluid temperature and the alteration overprint of moderate temperature demonstrate the high spatial and temporal variations at Brothers volcano.

The implications of the new stratigraphy and controls on permeability proposed here for Brothers volcano include a better understanding of the following: 1) submarine volcanic eruption sequences, 2) permeability in active submarine volcanoes, and 3) the formation of volcanogenic massive sulfide deposits on (and near) the sea floor.

Active submarine arc caldera volcanoes are a source of natural hazards (e.g., Jutzeler et al., 2014) but also have potential for renewable geothermal energy generation (e.g., Shnell et al., 2019) and can be hosts to volcanogenic massive sulfide (VMS) deposits that are significant sources of metals (largely Cu, Zn, Pb ± Au; e.g., Large et al., 2001; de Ronde et al., 2005). Sulfides can accumulate in discordant, pipe-like geometries, e.g., when associated with faults or stratabound in permeable layers capped by an impermeable layer (Tornos et al., 2015).

Primary lithological facies, fracturing, and associated permeability are all key to understanding the formation of VMS deposits. Large, tabular VMS deposits can be formed subseafloor by replacement (Franklin et al., 2005), a process that occurs more easily in reactive and/or porous rocks such as volcaniclastic facies (Tornos et al., 2015). Subseafloor replacement of existing lithologies has been documented during previous drilling of active submarine hydrothermal systems (Trans-Atlantic Geotraverse [TAG], Mid-Atlantic Ridge [Humphris et al., 1995]; Bent Hill massive sulfide [BHMS] deposit in Middle Valley, northern Juan de Fuca spreading centre [Zierenberg et al., 1998]). In these hydrothermal systems, hydrothermal fluids infilled primary pore space below an impermeable cap, or directly above such a cap after the cap had been breached by hydrothermal fluids (Doyle and Allen, 2003; Galley et al., 2007).

Drill cores are important when evaluating permeability. International Ocean Discovery Program (IODP) Expedition 376 drilled a series of boreholes at Brothers volcano in 2018 (de Ronde et al., 2019a, b). Brothers volcano is an active submarine caldera hosting hydrothermal systems and is one of 34 large volcanic complexes along the Kermadec arc (de Ronde et al., 2001, 2003; Fig. 1). Overall core recovery was 17.9% (222.37 m of core recovered from a 1244-m total drilled interval). Pieces of core totaling generally ≤2 m (up to ~3.7 m) were recovered every 4.8 m of drilled depth (de Ronde et al., 2019b). Although of immense value, the core record is scattered throughout the drilled interval. Intense hydrothermal alteration precluded interpretation of volcanic textures in cores in the bottom half of Hole U1530A (de Ronde et al., 2019d).

Continuous downhole measurements of petrophysical properties and temperature at Brothers volcano offer an opportunity to revisit the core-based stratigraphy and identify current fluid pathways. Downhole measurements were acquired in two holes (Fig. 1): Hole U1528D at the Upper Cone site (de Ronde et al., 2019c) and Hole U1530A at the NW Caldera site (de Ronde et al., 2019d). Petrophysical properties measured on both cores and downhole measurements link the two scales and environments of observations. Textural analyses of cores were made at submillimeter to decimeter scale. Core-based petrophysical property measurements have centimeter resolution and were not analyzed under their subsurface conditions (i.e., pressure, temperature, pore fluid chemistry). By contrast, downhole measurements are centimeter to decimeter resolution and measured in situ over hundreds of meters.

Downhole images have significantly helped to identify volcanic textures and refine core-based stratigraphy in oceanic crust (e.g., Brewer et al., 1999; Tominaga, 2013, and references therein), but are rare for offshore volcanoes (Tominaga et al., 2015) and hydrothermally altered silicic rocks actively venting hydrothermal fluids (Binns et al., 2002; Bartetzko et al., 2003).

Petrophysical properties vary as a function of both original microstructure/porosity and that resulting from alteration (Bartetzko et al., 2005; Durán et al., 2019). Heap et al. (2019) show that acid-sulphate alteration can increase or decrease porosity in andesites, while Kennedy et al. (2020) and Kanakiya et al. (2021) show that the effect of this type of alteration in a conduit is lithology dependent. Other types of hydrothermal alteration minerals, such as clay or chlorite, can have variable effects on porosity (Julia et al., 2014; Wyering et al., 2014). Clays can decrease wave speeds and resistivity (Han et al., 2011), particularly as they absorb water in their microstructure (e.g., smectite). Compaction during burial also reduces porosity with increasing depth (Bischoff et al., 2019). In the oceanic crust (Pezard, 1990; Bartetzko et al., 2001) and at Pual Ridge in the Eastern Manus Basin (PACMANUS), Papua New Guinea (Bartetzko et al., 2003), coherent lavas were shown to have higher resistivity than volcaniclastic rocks. Volcanic facies and petrophysical properties in turn affect hydraulic properties (Heap et al., 2017). Permeable rocks include relatively coarser-grained volcaniclastic facies or a fractured lava carapace, whereas low-permeability rocks include relatively finer-grained volcaniclastic facies (e.g., Large et al., 2001; Garden et al., 2017).

Here, we present petrophysical and textural properties in relation to permeable zones within Brothers volcano, using downhole and core-based measurements from the NW Caldera and the Upper Cone acquired during IODP Expedition 376. Textural analysis of the microresistivity images acquired in Hole U1530A (this study) refines the core-based stratigraphy presented in de Ronde et al. (2019d) and provides indications of fracture and vein characteristics. By combining textural interpretation of a microresistivity image and petrophysical properties, we propose a new interpretation of facies where alteration masked volcanic textures in cores and intervals of low recovery. Permeable zones inferred from temperature logs reveal controls on modern permeability in this active submarine hydrothermal system. By relating the current permeable zones to the volcanic facies and structures directly in the subsurface, we aim to constrain controls on fluid pathways and, hence, on possible future mineralization.

Brothers volcano is a caldera. The 3- to 3.5-km-diameter caldera floor lying at ~1,800 m below sea level is surrounded by 290- to 530-m-high walls. It contains an elongate NE-SW–oriented, 1.5- to 2-km-wide and 350-m-high postcollapse cone (Upper Cone) and a smaller satellite cone (Lower Cone; de Ronde et al., 2005; Embley et al., 2012; Fig. 1).

Brothers volcano is host to two active, but very distinct, hydrothermal systems (de Ronde et al., 2005; 2011): 1) a modified seawater-derived system at the NW Caldera, Upper Caldera, and West Caldera sites that vents high-temperature (<320°C), moderately acidic (pH = 3.2) hydrothermal fluids at the sea floor and 2) a magmatically-influenced hydrothermal system at the Upper and Lower Cones where vent fluids expelled at the sea floor have lower temperatures (<120°C) and very low pH (down to 1.9). Numerical models of hydrothermal fluid flow show that high-permeability faults on the inner slopes of the caldera and porous rocks under the Upper Cone are required to explain today’s sea floor venting (Gruen et al., 2012).

During Expedition 376, up to five holes were drilled at each of five different sites in water depths ranging from 1,230 to 1,730 m (Fig. 1; de Ronde et al., 2019b). All data presented here are in meters below sea floor (mbsf). Hole U1530A was drilled from immediately above an exposed stockwork zone in the central part of the NW Caldera vent field (20 m lateral distance from actively venting Cu-Zn-Au–rich sulfide chimneys) down to 453 mbsf. Holes at Site U1528 were drilled from within a ~25-m-diameter summit crater of the Upper Cone, in the upflow zone of this magmatic fluid-dominated hydrothermal system. Hole U1528D reached 298 mbsf (Fig. 1). The other three sites include Site U1527 on the caldera rim on the western margin of the NW Caldera, Site U1529 on the caldera floor adjacent to the western caldera walls, and Site U1531 between the Lower and Upper Cones. No downhole measurements were acquired in holes at these three sites.

All of the 222.4 m of core recovered from Brothers volcano is of dacitic composition. Stratigraphy and alteration defined from the cores is presented in detail in de Ronde et al. (2019b-e) and summarized for Holes U1530A and U1528D in Figs. 2 and 3, respectively. Petrophysical properties of rare unaltered and slightly altered cores across the sites are variable (de Ronde et al., 2019b). This paper focuses on hydrothermally altered sections of Holes U1528D and U1530A, particularly where downhole measurements were acquired (60–334 and 67–447 mbsf, respectively).

Hydrothermal alteration is pervasive and intense throughout the drilled intervals. Different alteration assemblages, called “alteration types” (de Ronde et al., 2019a, b), occur at all sites and are briefly summarized here. At Hole U1528D, Upper Cone, cores have three different, commonly alternating acid-sulfate (magmatic) alteration assemblages of high temperature (230°–350°C), all indicative of magmatic volatiles having interacted with heated seawater. At Hole U1530A, NW Caldera, a low-temperature mineral assemblage is found down to 30 mbsf. Between 30 and 225 mbsf, a medium-temperature (≤250°C) assemblage is ascribed to hydrothermal fluids of modified seawater composition. Between 225 and 453 mbsf (bottom of the hole), a higher-temperature assemblage (230°–350°C) indicates acid-sulfate fluids. This high-temperature assemblage is intercalated with, and locally overprinted by, a lower-temperature (<250°C), fracture-controlled alteration assemblage consistent with the incursion of heated seawater. The similar acid-sulfate fluids at the magmatically-influenced Hole U1528D (Upper Cone) and at >225 mbsf in Hole U1530A (NW Caldera), recognized from alteration mineral assemblages, show that the deep parts of this site have previously undergone the same processes that are currently occurring at the Upper Cone. Based on Expedition 376 drilling results, de Ronde et al. (2019a) proposed that hydrothermal activity occurred pre- and postcaldera collapse.

In the intervals covered by downhole measurements, facies include lava and volcaniclastic deposits grouped in igneous units (de Ronde et al., 2019b), briefly summarized here. In Hole U1528D, volcaniclastic facies form Igneous Unit 2a (61.30–150.53 mbsf) and Igneous Unit 2c (162.50–269.03 mbsf), and lavas form Igneous Unit 2b (152.90–160.17 mbsf) and Igneous Unit 3 (272.90–355.07 mbsf). In Hole U1530A, Igneous Unit 3 (59.62–64.40 mbsf) is a largely coherent plagioclase-phyric lava with pseudomorphs after glomerocrystic plagioclase. Igneous Unit 4 (64.40–218.21 mbsf) is composed of volcaniclastic facies—poorly sorted monomict and polymict lapillistone. There are also subordinate horizons of lapilli-tuff and tuff with lapilli. Igneous Unit 5 (220.70–448.68 mbsf) consists of volcanic rocks of unknown facies with five discrete horizons of less-altered plagioclase-phyric lava containing pseudomorphs after glomerocrystic plagioclase.

Boundaries between these igneous units defined from cores coincide with the tops and bottoms of a recovered core (or core piece) due to partial core recovery. Therefore, these corebased boundaries may be offset by several meters compared to the real depth below sea floor. Igneous units and alteration types do not correlate across holes (de Ronde et al., 2019b).

Drilling parameters

Drilling parameters, such as rate of penetration and torque, relate to rock strength (Hamada et al., 2018). Drilling parameter logs were produced by selecting the time intervals of each core cutting from the continuous drilling parameters recorded at 1-second time intervals. In addition to drilling parameters per se, H2S odor emanating from specific cores is used as an additional, low-confidence indicator of fluids (in this case, a gas) moving through rock. Indeed, H2S odor occurred only at specific depths and was commonly associated with occurrence of native sulfur in veins in the core, also suggesting fluid pathways (de Ronde et al., 2019b).

Petrophysical measurements on cores

Porosity, P-wave velocity (Vp), bulk, and grain densities were measured on 8-cm3 core samples at room pressure and temperature at each site (de Ronde et al., 2019e). Grain density (also known as matrix density) is the density of the rock components without the effect of porosity. In the context of Brothers volcano, a high grain density is consistent with large proportions of pyrite (density of pyrite is ~5 g/cm3). Vp measurements were conducted on samples saturated with seawater. All measurements followed standard IODP practices (de Ronde et al., 2019e).

Acquisition and processing of wireline downhole measurement profiles

Principles, configurations, and operations of wireline logging tools are described in de Ronde et al. (2019e) and at https://iodp.tamu.edu/tools/logging/index.html. Principles and operations of wireline downhole measurements for Expedition 376 were similar to other IODP expeditions (e.g., see Bartetzko et al. [2003] for ODP Leg 193 at PACMANUS) and are briefly summarized below.

During wireline logging operations, a combination of logging tools (tool string) was lowered into the Hole Using a cable. Measurements were performed while pulling the tools upward. In Hole U1528D, the high-temperature tool string included gamma-density, photoelectric factor, caliper (hole diameter), spectral gamma ray, and temperature-recording instruments (Fig. 1b; App. 1, Table A1). In Hole U1530A, which had a lower temperature, the “Triple Combo” tool string was run first and included electrical resistivity, gamma-density, photoelectric factor, caliper, neutron-porosity, spectral gamma ray, and temperature instruments. The second “FMS-Sonic” tool string measured high-resolution microresistivity images with the Formation MicroScanner (FMS) and included caliper, sonic velocity, spectral gamma ray, and temperature-recording instruments.

An enlarged borehole may cause a lack of contact between the tool and borehole wall and thus can affect the quality of the measurements, particularly for the density profile and microresistivity images (e.g., Bartetzko et al., 2003; Massiot et al., 2015). The diameter of Hole U1530A exceeds the maximum aperture of the caliper tools (43.9 cm) in many places, down to 350 mbsf (Fig. 2). Hole U1528D is locally enlarged (Fig. 3). Although these dimensions affect logging data and are carefully considered in any interpretation, core sample and downhole measurements are generally in good agreement, which yields good confidence in the reliability of downhole measurements (de Ronde et al., 2019c, d; Figs. 2, 3). Direct depth matching of the core and downhole measurements was not conducted because of partial core recovery (29 and 17% at Holes U1528D and U1530A, respectively) and fragmentation (i.e., pieces rarely >20 cm long). The IODP standard is to push all core pieces to the top of each drilled section (commonly 4.8-m-long intervals). There can therefore be variations of ±10 m between cores and logs. However, core observations were used to calibrate logging results, with considerations of depth matching limitation.

Data acquired within the drill pipe were not considered except for natural gamma, which shows variations even though the signal is very attenuated. Porosity and density profile data with high correction curves were excluded (App. 1, Table A1), as they mainly correspond to local variations in borehole diameter. In Hole U1528D, the density profile data were further affected by the high temperatures below 287 mbsf and, therefore, were also excluded.

Spectral gamma ray data provide the relative proportions of potassium (K), thorium (Th), and uranium (U) by measuring natural gamma radiation. Total natural gamma radiation measured on whole cores was systematically underestimated due to core fragmentation (de Ronde et al., 2019b). In both holes, the natural gamma ray values are largely influenced by the K, which has been shown to be highly mobile using shipboard inductively coupled plasma-atomic emission spectroscopy (ICP-AES) measurements associated with varied alteration mineralogy (de Ronde et al., 2019c, d).

The sonic log measures Vp. To improve data quality, the sonic log in Hole U1530A was reprocessed with a smaller window than that used in the automatic processing. Vp measured downhole shows similar variations along the hole to measurements on the cores. However, the downhole Vp profile is consistently ~0.5 km/s slower than core measurement. As this discrepancy occurs along the entire logged interval, including below 400 mbsf where the hole’s enlargement is small, hole size is probably not the main contributor to the difference (Fig. 2). Undersaturation of core samples (Mielke et al., 2017), or lack of confining pressure (Ludwig et al., 1998) applied to the core samples during their measurement, would underestimate Vp and therefore cannot explain the discrepancy. We infer that this slower Vp in downhole measurements compared to core measurements is caused by (1) the frequency scale (dispersion) difference between applied wave speed on cores (~500 kHz) and downhole measurement (~10 kHz; Adam and Otheim, 2013) and (2) the presence of macrofractures in the wall rock formation that were not present in the core samples. Thus, we consider the variations of Vp measured by both core and log as being reliable, with a possible shift of ±0.5 km/s.

Electrical resistivity in volcanic rocks is particularly sensitive to fractures and porosity. Fractures or other void spaces filled with conductive materials such as seawater, clay minerals, oxide, or sulfide minerals can cause electrical resistivity to decrease, provided that the fractures or void spaces are connected (Pezard, 1990; Bartetzko et al., 2003). In Hole U1530A, electrical resistivity profiles were measured at six depths into the formation, up to 1.3 m. The “resistivity true” (RT) parameter, derived from these six measurements and variably affected by drilling fluid incursion into the rock, is the most representative of the intact formation resistivity and, therefore, is used in this study.

Acquisition, processing, and interpretation of microresistivity images in Hole U1530A

Electrical microresistivity images show contrasts in resistivity at the borehole wall by pressing four pads with electrodes against the borehole wall, resulting in high-resolution images (i.e., a vertical resolution of 5 mm). In Hole U1530A, two logging passes were used in tandem between 90 and 445.5 mbsf. In places, they cover different sections of the borehole, which increases azimuthal coverage. Vertical adjustment of the two passes would have resulted in numerous stretches and shrinking of a few tens of centimeters and was thus not performed. Microresistivity images were reprocessed using the Recall™ software. Static normalization is used to compare rock types along the hole, and dynamic normalization is used for resolving variations within a few meters (e.g., Rider, 1996). Picking of sinusoids representing planar features (i.e., fractures or layers) was done with consideration of both passes simultaneously to avoid “double counting.” Borehole enlargement resulted in poor contact of the sensors with the borehole wall at some depths and a deterioration of image quality in these areas. Combined with the partial azimuthal coverage and bias against sampling steeply dipping features (e.g., Martel, 1999), planar features are overall undersampled. We assigned either a “high” or “low” confidence level for the image facies and fractures.

Following the approach of genetic interpretation of textures in volcanic deposits by McPhie et al. (1993), we use a multistep method to describe and interpret volcanic facies from downhole measurements. Facies described from downhole images are called “image facies.” Core descriptions and resulting interpreted genetic facies from cores used for comparison with the image facies are entirely from de Ronde et al. (2019d).

Firstly, two image facies are distinguished from the microresistivity image, “coherent” and “apparent volcaniclastic,” similar to volcanic facies defined from core samples (Table 1; Fig. 4). Intervals that show high resistivity and homogeneous background, often containing networks consisting of short, low-resistivity cracks that cannot be fitted by a sinusoid across the image, are labelled “coherent image facies” (Tables 12; Fig. 4K-L). This crack network pattern is here named “honeycomb.” Honeycomb pattern is similar to polygonal cooling joints in lavas, although a similar pattern has been described by others as stockwork veining (de Ronde et al., 2019b). Conversely, intervals that show patches having a contrast to a homogeneous to granular background of varied resistivity are labelled “apparent volcaniclastic image facies” (Tables 12; Fig. 4A-B, F-G). Patches are of high resistivity, low resistivity, or low resistivity with a high-resistivity halo. Boundaries between intervals of different image facies are described as either sharp or gradational (reported in App. 2).

Image facies identification criteria used in this study are similar to those used to interpret microresistivity images from other locations, such as 1) the hydrothermally altered silicic volcanic rocks of the submarine active hydrothermal vent field hosted in hydrothermally altered dacitic rocks at PACMANUS (Bartetzko et al., 2003, 2006; Arnold et al., 2007), 2) the hydrothermally altered rhyolite lavas and (volcaniclastic) ignimbrites at the Ngatamariki and Wairakei Geothermal Fields, New Zealand (Halwa et al., 2013; Milicich et al., 2018), 3) the Tamu Massif submarine supervolcano, Shatsky Rise (Tominaga et al., 2015), and 4) oceanic crust basalt (e.g., Brewer et al., 1998; Tominaga and Umino, 2010).

Secondly, a genetic origin is proposed for the image facies where volcanic textures observed in nearby cores corroborate the image facies. “Coherent image facies” are interpreted as lavas formed from cooling and solidification of molten lava or magma (McPhie et al., 1993). “Apparent volcaniclastic image facies” are interpreted as volcaniclastic deposits if nearby drill cores support this interpretation. In this case, resistivity patches are likely clasts. Volcaniclastic deposits are composed of volcanic particles of any shape and size (McPhie et al., 1993).

Thirdly, where alteration masks volcanic textures in cores (i.e., in most of the 222.7- to 448.68-mbsf interval), “apparent volcaniclastic image facies” are interpreted as apparent volcaniclastic facies. The genetic origin is proposed using the other petrophysical measurements (downhole and on core samples) and drilling parameters. Apparent volcaniclastic facies result from patchy or domanial alteration or fracture- and joint-controlled alteration (McPhie et al., 1993). In many cases, the apparent textures superficially resemble welded ignimbrite or coarse lithic breccia (McPhie et al., 1993). The pseudoclastic texture can thus appear to have “patches” on the microresistivity image. Such pseudoclastic textures are very common in altered coherent lavas and intrusions (McPhie et al., 1993).

Microresistivity images also provide observations of fracture and vein occurrence and orientation. The distinction between open fractures filled with seawater and veins fully (or partially) infilled by clay or pyrite is equivocal, as they would all appear to be of low resistivity (Pezard, 1990; Lévy et al., 2018; Nono et al., 2020). Conversely, planar features of high resistivity are veins infilled with resistive minerals such as silica. Planar features of low resistivity with a high-resistivity halo are most likely to be open and permeable fractures, as this pattern can result from a strong electric contrast between seawater filling the fracture and the resistive wall rock (Massiot et al., 2017). However, this halo pattern can also be caused by a sealed clay vein with a resistive mineral halo, or the halo could represent a buildup of excess current along the fracture due to materials with unusually high resistivity within the fracture (e.g., Lofts and Bourke, 1999). Thus, in this study, we use the generic term “fracture” to represent open fractures, partially open fractures, and veins (closed fractures). In addition to distinct fractures observed around the borehole, we report intervals of honeycomb fracturing and of densely fractured zones likely made from multiple distinct fractures that cannot be resolved individually (Table 2).

Acquisition and analysis of the temperature profiles

Downhole temperature profiles measure the temperature of the fluid within the borehole. These temperature profiles reflect the formation temperature but also depend on the amount and duration of cold seawater circulation in the hole during drilling. At Brothers volcano, multiple temperature logs were acquired with several tools and at different stages of drilling and logging, and after the holes heated up (de Ronde et al., 2019b). In this work, we use the high-resolution wireline profiles measured with the MTEM tool in Holes U1528D and U1530A within 15 hours of stopping cold water circulation (App. 1, Table A1). We also report heat-up measurements where temperature was measured at the bottom of the hole just at the end of drilling with the ETBS tool (de Ronde et al., 2019b-e). During the ~15-minute-long heat-up measurements, drilling seawater circulation was stopped.

The location of the temperature sensor at the top of the ~30-m-long tool string affected fluid circulation differently while moving the tool down and up the hole, resulting in depth mismatch. The depth of temperature profile measured as the tool went down the hole was adjusted by matching the profiles with those of the uphole measurement, the latter used as a reference for this study (App. 1, Table A1).

Temperature gradients were derived from each temperature profile (three in Hole U1528D and seven in Hole U1530A). For each hole, deviations from the overall trends that occur at similar depths on successive temperature and temperature gradient profiles are considered indicative of permeable zones (Grant and Bixley, 2011).

Image facies and fractures using the microresistivity images for Hole U1530A (90–445.5 mbsf)

Image facies: Five, possibly six, coherent image facies intervals are clearly identified in the microresistivity image, all occurring below 235 mbsf. Drilled thicknesses of coherent image facies range from 1 to >12 m (the latter at the bottom of the microresistivity image) (Figs. 2, 4; Table 3). Two coherent image facies intervals are situated one above the other. The 360- to 366-mbsf interval shows a more homogeneous background and fractured zone than the distinct honeycomb fracture pattern between 366 and 367 mbsf. However, it is unclear whether these two coherent facies intervals represent distinct units or relate to a single unit with a different fracture pattern. Between 236.5 and 248 mbsf, the microresistivity image appears homogeneous but without the honeycomb pattern (Fig. 5), so it could be either a nonfractured, coherent image facies interval or an apparent volcaniclastic image facies interval having patches smaller than 5 mm (below image resolution).

The coherent image facies interval identified in the deepest part of the microresistivity image (433–445.5 mbsf) corresponds to a lava flow identified in cores (de Ronde et al., 2019d; Fig. 2) between 429.1 and 429.77 m (i.e., core 89R) and between 438.7 and 448.38 m (i.e., cores 91–93R). The coherent image facies interval at 299 to 317 mbsf corresponds to a lava flow in cores 63 through 65R (i.e., 304.36–314.47 mbsf). Due to depth-matching discrepancies, the 299- to 317-mbsf interval of coherent image facies may correspond to the 289.98- to 291.05- and 304.38- to 414.47-mbsf intervals of lava flows identified in cores. Conversely, the 400.57- to 405.47-mbsf lava flow interval identified in cores may correspond to the 408- to 410.5-mbsf interval of undefined image facies.

Apart from these intervals that show the very distinct coherent image facies, the 90- to 445.5-mbsf imaged interval shows apparent volcaniclastic image facies. The numerous patches of both higher and lower resistivity (with respect to the background), yielding the heterogeneous appearance, relate to contrasting amounts of minerals such as quartz (resistive) versus clay, pyrite, and pore water (conductive) in resistivity patches compared to the background image (Fig. 4A-B, F-G). The resistivity patches have varied size, frequency, and resistivity content (high, low, or low with a high-resistivity halo). There is no systematic variation of patch characteristics with depth (Fig. 2). There is an overall lesser occurrence of resistivity patches between 160 and 282 mbsf than in the rest of the imaged interval. Based on core descriptions, the resistivity patches are either (1) clasts in volcaniclastic deposits down to ~220 mbsf or (2) a mottled alteration texture imparted on either volcaniclastic deposits or lava flows beneath 220 mbsf (de Ronde et al., 2019a). Thus, between 90 and 223 mbsf, the apparent volcaniclastic image facies is interpreted to be volcaniclastic deposits with variable clast patch sizes, shapes, and frequency. The genetic origin of apparent volcaniclastic image facies between 223 and 445 mbsf is evaluated in the “Discussion” section in the context of other petrophysical properties.

Five thin intervals (i.e., 252–253.5, 367–368, 386–388, 408–410.5, and 431–433 mbsf) have insufficient image quality to designate an image facies. These 1- to 2.5-m-thick intervals are referred to as “undefined.” Undefined intervals immediately adjacent to coherent image facies interpreted as lava flows (e.g., 431.2–433 mbsf) may be rubble zones of lava flows or the fractured chilled margins of intrusions. The 252- to 253.5-mbsf undefined interval, directly beneath an inferred fault (see following section), is likely caused by intense fracturing, decreasing the contact between the tool pads and the wall rock and, therefore, decreasing image quality.

Characteristics of fractured zones and distinct fractures: Nearly 30 intervals are consistent with the presence of multiple crosscutting fractures of low resistivity, here termed “fractured zones.” They commonly appear blurry in the microresistivity image due to deterioration of the borehole wall (Figs. 2, 5).

Identification of distinct fractures was made difficult by only partial azimuthal coverage and poor tool pad contact with the borehole wall in places. In addition, some flattened, low-resistivity patches may appear to align on a sinusoid on the unrolled image but could also be randomly distributed. As a result, only 53 distinct fractures were identified throughout the imaged interval with a high degree of confidence, i.e., where the sinusoids could clearly be traced across at least three tool pads. High-confidence fractures traced on all pads are either of low resistivity (n = 43) or of low resistivity with a high-resistivity halo (n = 10). An additional 97 fractures were identified with a lower degree of confidence that have either high or low resistivity. Distinct fracture characteristics are reported in Appendix 2. This is in contrast to the 475 distinct veins and fractures identified in cores in the 90- to 448-mbsf interval, which had ~17% core recovery (de Ronde et al., 2019d).

The distinct fractures have a random strike and moderate dip magnitude (i.e., 70% have a dip magnitude between 20° and 50°). The limited number of steeply dipping fractures identified on the microresistivity image is likely an artefact of the enlarged borehole and the gap between the microresistivity image tool pads, in addition to the undersampling of fractures subparallel to the borehole (e.g., Martel, 1999). In cores, the average dip magnitude of fractures is similarly moderate (54°), with most of the dips (72%) being greater than 45° (de Ronde et al., 2019d).

Apparent fracture apertures at the borehole wall measured on the microresistivity image are <2 cm, except at 252 mbsf (3 cm) and 99 mbsf (4 cm). The fracture at 252 mbsf is very clear, dipping 40° towards the north (striking east-west; Fig. 5). Apparent apertures of >2 cm in places along this fracture are likely caused by spalling from the borehole wall. This fracture is directly underlain by numerous, distinct fractures, and fracture frequency decreases with depth to 258 mbsf. Most halo fractures identified in the microresistivity image are located in this interval (i.e., eight out of the 10 are between 234 and 274 mbsf; the remaining two are at 364 and 385 mbsf within coherent image facies intervals). These characteristics are consistent with the fracture at 252 mbsf being a fault, underlain by its damage zone.

Core-based petrophysical properties across Brothers volcano

Petrophysical properties of altered core samples show an overall decrease of porosity and an increase of both bulk density and Vp with depth (Fig. 6). Vp is more variable within this trend, particularly in Hole U1528D, where Vp repeatedly increases and decreases downhole (e.g., 2.3 to 3.5–4.2 km/s over tens of meters; Figs. 3, 6). Local variations away from the general trends with depth are described in more detail below.

Porosity is inversely correlated with bulk density. Linear regression slopes for Sites U1527, U1528 (data from Holes U1528C and U1528D), and U1530 are similar (Fig. 6B). Hole U1530A has a relatively high linear regression intercept that results from the higher grain density in this hole (2.9 ± 0.1 g/cm3) than in Hole U1527C and holes at Site U1528 (2.7 ± 0.1 g/cm3), consistent with the higher pyrite content (both disseminated and in veins) observed in the cores (Fig. 6A; Table 3; de Ronde et al., 2019b). Porosity is also inversely correlated with Vp. Porosity shows more scatter and a shallower slope of the regressions for holes at Site U1528 compared to Hole U1530A (Fig. 6c). This scatter is not strictly due to the number of measurements per site, considering that the combined holes at Site U1528 provided the highest number of samples.

In Hole U1528D, lava flows and volcaniclastic deposits (as defined from core observations; de Ronde et al., 2019c) do not have distinct petrophysical property signatures, having similar ranges of porosity, density, and Vp values (Fig. 7). In Hole U1530A, the intervals of coherent image facies (lava flows) display a lower porosity and higher bulk density, Vp, and resistivity than the volcaniclastic image facies (volcaniclastic deposits). However, all these lavas are located deeper than the volcaniclastic deposits (Table 3), and the volcaniclastic deposits show the trend of porosity decreases with depth. Thus, this difference in petrophysical property between lava and volcaniclastic deposits may be a combined effect of facies, alteration, and depth.

Microstructures likely influence Vp in addition to porosity. Measurements of Vp and Vs under effective pressure of two samples from Hole U1530A, at 122.3 and 242.6 mbsf (Adam and Massiot, unpub. data), suggest that they have different microstructure characteristics. Indeed, although the sample from 122.3 mbsf has higher porosity than the sample from 242.6 mbsf, the lower pressure dependence of Vp and Vs suggests that the sample from 122.3 mbsf has stiffer pores than the sample from 242.6 mbsf, the latter having more compliant pores such as microfractures (Han, 2016). This different pressure dependence that indicates stiff, low-compliance pores confirms that the sample from 122.3 mbsf is of volcaniclastic facies (as identified in cores and microresistivity images). By contrast, pressure dependence suggests that the sample from 242.6 mbsf is consistent with a lava with microfractures (undefined facies in cores and apparent volcaniclastic image facies).

Downhole measurement-based petrophysical units

Downhole measurements show similar overall variations to discrete measurements on core samples. As downhole measurements are continuous, they reveal more details of variations with depth and improve delineation of lithological boundaries.

Within the overall trends noted with depth on discrete core samples, intervals of distinct petrophysical properties are defined within Holes U1528D and U1530A when considering the combination of continuous downhole measurements and drilling parameters (Figs. 2, 7; Table 3). Here, these classifications are called “petrophysical units” (PU) as opposed to “igneous units” (IU), defined from textural characteristics in cores (see de Ronde et al., 2019b). PU descriptors are detailed in Appendix 1, Tables A2 and A3. Key aspects are described below.

U1530A petrophysical units: In Hole U1530A, six petrophysical units labeled PU-A through PU-F have been identified (Fig. 2; Table 3; App. 1, Table A2). In the absence of downhole measurements at the top of the hole, PU-A and PU-B are defined from core measurements and are the same as igneous units IU1 and IU2. PU-A and PU-B are separated by a clear peak in uranium, even though the signal has been attenuated by the drill pipe. PU-C (59.62–70 mbsf) is similar to the lava flow of IU3 (59.62–64.4 mbsf), but slightly thicker. PU-C is characterized by low porosity, high bulk density, and high Vp.

PU-D (70–86 mbsf) is very distinct across all the petrophysical data. It has very high porosity, low bulk density, high grain density, low Vp, and very low resistivity (median of 0.7 Ω·m, slightly higher than seawater resistivity of 0.2 Ω·m). The core samples were visibly rich in clay, confirmed by X-ray diffraction (XRD; abundant illite and moderate smectite). The high grain density reflects the abundance of pyrite noted in the core descriptions and an increase in Fe2O and S (de Ronde et al., 2019d), yet with a porosity so high that the bulk density remains low.

PU-E (86–214 mbsf) and PU-F (232–445 mbsf) intervals are similar to IU4 (64.4–218.21 mbsf) and IU5 (222.7–448.68 mbsf) but with different upper and lower boundaries. The top of PU-E (86 mbsf) was noted as a formation change during drilling by a higher torque. PU-E shows stepwise variations in most petrophysical properties, whereas PU-F has fairly constant petrophysical properties with depth. PU-F has higher density, Vp, resistivity, and torque, and lower porosity, than PU-E. Rather than a sharp boundary, PU-E and PU-F are separated by an 18-m-thick transition zone (214–232 mbsf), wherein there are steady increases of resistivity, Vp, and torque as well as decreases in porosity and drilling rate of penetration. PU-E and PU-F are also discriminated by different Vp:porosity relationships (Fig. 8); that is, PU-E shows a wide variability of porosity for a narrow range of Vp, opposite to that for PU-F. A single exponential relationship fits the entire core data reasonably well, but not the downhole data, so the distinction between PU-E and PU-F is preferred. These Vp:porosity relationships are consistent with PU-E being volcaniclastic deposits with stiff, low-compliance pores and PU-F being dominated by lava flows with microfractures (Wilkens et al., 1991). PU-E and PU-F are further separated into subcategories that have more subtle variations of petrophysical properties (Fig. 2; App. 1, Table A2).

U1528 petrophysical units: Petrophysical units in Hole U1528D are similar to the igneous units, partly because of fewer available downhole measurements. However, boundaries are defined at slightly different depths and subcategories (Fig. 3; Table 3; App. 1, Table A3). PU-C (140–163 mbsf), defined by high porosity and caliper and low Vp, is thicker than Igneous Unit IU2b (lava flow, 159.90–160.17 mbsf).

Temperature profiles and permeable zones in Holes U1530A and U1528D

Variations between successive temperature measurements show that the temperature was still equilibrating after a cooling period induced by drilling. This includes heat-up measurements at the bottom of Holes U1528D and U1530A just after drilling, where temperature increases by ~0.4°C/min over ~15min (Fig. 9; de Ronde et al., 2019c, d). Therefore, fluids moving between the formation and the holes during the temperature measurements are a mix of seawater and formation fluids.

The shapes of the temperature profiles in Holes U1528D and U1530A are different (de Ronde et al., 2019a). That is, Hole U1528D shows isothermal sections separated by discrete zones of temperature increase (or decrease). This profile indicates upward, nearly adiabatic flow of warm formation fluids that enter the hole at distinct depths, then flow up the hole (Fig. 9A). Fluid discharge out of the re-entry cones was observed at the sea floor, on submarine cameras, during and after drilling. By contrast, the temperature in Hole U1530A gently increases down the hole (Fig. 9B), even though the hole was spudded only ~20 m from active vents discharging fluids up to 320°C.

Hole U1528D is hotter than Hole U1530A. Successive temperature profiles in Hole U1528D (max. temperature of 247°C, 6.5 h after cessation of cold drilling fluid circulation) show that temperatures were still increasing (i.e., they were still equilibrating; Fig. 9A). This is consistent with the geochemical signatures of three borehole fluid samples indicative of input of magmatic fluids to unaltered seawater at temperatures of at least 350°C (fluids collected at ~279 and 313 mbsf at various times after stopping drilling fluid circulations and then 23 days later at 160 mbsf; de Ronde et al., 2019d). In Hole U1530A, the bottom temperature increased from 40° to 94°C in the eight hours between stopping cold drilling fluid circulation and the first wireline run. In the four subsequent hours (between the two wireline log runs), bottom hole temperature decreased to 40°C (de Ronde et al., 2019d). This decrease of temperature in the hole is potentially caused by infiltration of seawater in the open hole. This is consistent with the seawater-like composition of a borehole fluid sample collected at 435 mbsf after the other downhole measurements (de Ronde et al., 2019d).

In Hole U1528D, temperature and temperature gradient profiles show isothermal sections separated by four zones of temperature increase, and one of temperature decrease (Fig. 9A). Two of these zones (145–160 and ~295 mbsf, both with temperature increases) are interpreted to be structurally controlled permeable zones. Indeed, they coincide with (1) local borehole diameter (caliper) increases, (2) abundant native sulfur observed in veins, (3) a high fracture density in the cores (with both variable dip magnitude and high vein density), and (4) an H2S odor emanating from the cores (Fig. 3). The other temperature anomalies at 90 to 105 mbsf (small incremental temperature increases), ~230 mbsf (small temperature decrease), and 250 to 275 mbsf (marked temperature decrease) contain fewer veins and open fractures in the cores and are thus considered to have permeability controlled by the matrix and/or lithological contacts rather than structures. The decrease in temperature at 250 to 275 mbsf may be caused by delayed heating after cold drilling fluids pervaded the formation.

In Hole U1530A, two zones (190–210 and 255–285 mbsf) have a small deviation from a concave-upward temperature profile that also coincides with stepwise increases in the temperature gradient, indicating permeable zones (Fig. 9B). A few other possible narrow zones show similar deviations from the general temperature profile but were not as easily replicated on successive temperature runs and also do not coincide with stepwise increases in temperature gradient. The permeable zone at 255 to 285 mbsf is interpreted to be structurally controlled because (1) most of the halo fractures are identified in the microresistivity image located in this interval, (2) fractures and veins of varied dip magnitude were observed in the cores, and (3) a noticeable H2S odor emanated from the cores (Figs. 2, 5). This zone sits directly beneath the possible fault described at 252 mbsf (Fig. 5 and below). By contrast, the temperature gradient change at 190 to 210 mbsf is unlikely to be structurally controlled, as it lacks H2S odor and any varied vein/fracture orientation in the cores. It is possible that the fractured zone identified in the microresistivity image at 205 mbsf contributes to local permeability, although many other fractured zones do not coincide with temperature gradient change, and hence this parameter is not considered characteristic. Instead, permeability at 190 to 210 mbsf is more likely to be controlled by the matrix and/or lithological contacts.

Major controls on petrophysical properties: depth, facies, fractures, and the ambiguous role of alteration

Increasing depth below sea floor is a first-order control on decreasing porosity and increasing bulk density and Vp down to 232 mbsf in Hole U1530A and to 360 mbsf (bottom of hole) in Hole U1528D (Fig. 6). We interpret these first-order trends to be the effects of compaction modulated by increased duration of alteration and filling of pores with secondary mineral phases. For example, in Hole U1530A, the overall gradual decrease of porosity down to 232 mbsf occurs across varied facies (i.e., lava flows [PU-C] and volcaniclastic deposits [PU-E]) and three types of alteration mineral assemblages. Compaction in volcanic edifices has been demonstrated experimentally (e.g., Heap et al., 2015) and has been estimated to be between 20 and 30% in a buried volcano (Bischoff et al., 2019). The effects of alteration have been documented at the TAG submarine hydrothermal mound, where the general increase of Vp with depth (up to 106.80 mbsf) has been interpreted to be increased cementation by quartz and pyrite (Ludwig et al., 1998), and a similar interpretation can be made for Holes U1530A and U1528D down to 232 and 360 mbsf, respectively.

Conversely, beneath 232 mbsf in Hole U1530A, porosity shows only a minor decreasing trend with depth. Porosity is lower beneath 232 mbsf (PU-F, 24 ± 10%) than in shallower formations (34 ± 4% in 85–214 mbsf). In this deep interval that has likely been subject to pre- and post-caldera collapse hydrothermal conditions (de Ronde et al., 2019a), compaction and hydrothermal alteration may not have as much local effect on petrophysical properties compared to shallower and younger sections. At Brothers volcano, extensive hydrothermal alteration has been occurring for more than 15,000 years (Ditchburn and de Ronde, 2017), whereas dating of mineralization (and, by proxy, the alteration) shows it can happen rapidly (e.g., de Ronde et al., 2005; 2011).

Within the first-order depth-controlled trend, there are second-order local variations of porosity, density, and Vp reflecting the local effects of fracturing and facies within each hole. For example, in Hole U1528D, the lava flow at 140 to 163 mbsf (PU-C) shows an increase followed by a decrease in porosity, although the range of values is similar to those units directly above and below (Fig. 3). In Hole U1530A, the volcaniclastic deposit interval between 159.5 and 184.5 mbsf, which has few but large resistivity patches on the microresistivity image, has locally higher porosity (50%) than the surrounding volcaniclastic facies (40%) that have more abundant resistivity patches (Fig. 2). Such local variability of textures and petrophysical properties has been observed at PACMANUS (Bartetzko et al., 2003) and is typical of volcanic environments (Heap et al., 2017).

Within the 232- to 448-mbsf interval in Hole U1530A, Vp varies by up to 1 km/s for a particular porosity. As this variability is not associated with systematic mineralogical variations, increasing fracture density and porosity likely cause the second-order Vp and resistivity variations. Indeed, fractures are compliant but occupy small volumes within the rocks while at the same time providing connectivity for the fluids. Increased compliance in a rock due to microfractures slows wave speeds (Nur and Simmons, 1969), while connected pores and fracture networks diminish the electrical resistivity of rocks (Brace and Orange, 1968).

In Hole U1530A, petrophysical properties stand out in the volcaniclastic facies interval of PU-D (70–85 mbsf). In this interval, discrete core measurements show the highest porosity (59 ± 8%), highest grain density (3.2 ± 0.2 g/cm3), and lowest Vp (2.2 ± 0.5 km/s) within the entire hole. Downhole porosity values are even higher (76 ± 11%), and the formation resistivity is the lowest of the hole (0.74 ± 0.13 Ω·m). Together, these petrophysical properties indicate a zone rich in pyrite (a high-density and conductive mineral) and water in the matrix, in fractures, and trapped in clays. This interpretation is consistent with cores having clay-rich clasts surrounded by a silica- and pyrite-rich matrix, and with the presence of illite and smectite in XRD (de Ronde et al., 2019d). The fracture density measured on the cores in this interval from 70 to 85 mbsf is an average of five to 10 fractures per 10 cm, with several network veins, whereas the interval from 60 to 70 mbsf contains an average of one fracture per 10 cm (de Ronde et al., 2019d). The lava flow at 60 to 70 mbsf (PU-C), located directly above this clay-rich, fractured 70- to 85-mbsf interval (PU-D), may have acted as a barrier to vertical flow, focusing the fluids in the top of PU-D, intensifying alteration to clay and leading to precipitation of pyrite.

There is no indication of stockwork zones (see de Ronde et al., 2019B) or massive sulfide deposits in Holes U1528D or U1530A in the intervals that have downhole data (i.e., below 61 and 69 mbsf, respectively). Stockwork zones at PACMANUS were determined to have high photoelectric factor (>6 barns/electron), increased density (1.8–2.5 g/cm3), and high and variable gamma ray values (Bartetzko et al., 2006). In Hole U1528D, the photoelectric factor exceeds 4 barns/electron (up to 6.7 barns/electron) only at the bottom of the logged interval. Heavy minerals (including those containing U) that can generate this high photoelectric factor commonly accumulate at the bottom of the hole after falling down from the borehole walls or from unrecovered core sections, so this high value is not diagnostic of the formation in place. The highest photoelectric factor value in U1530A is at 76 to 85 mbsf (~4 barns/electron) at the bottom of PU-D and is consistent with the high grain density indicating pyrite, although the low bulk density (1.9 g/cm3) is not consistent with a stockwork zone. This lack of any indication of a stockwork zone in the petrophysical properties below ~60 mbsf might be consistent with the lack of observed multicentimeter-wide veins in cores.

Controls on electrical resistivity in Hole U1530A

Facies, fractures, and, potentially, alteration mineral assemblages are the main factors controlling electrical conductivity of the hydrothermally altered formations for both electrical images and formation resistivity in Hole U1530A, as documented in other volcanic rocks (e.g., mid-ocean ridge basalts [Pezard, 1990]; hydrothermally altered dacites [Bartetzko et al., 2003, 2006]).

The resistivity profile in Hole U1530A is distinctly different in different facies, identified as 1) clay-rich volcaniclastic deposits (PU-D), 2) volcaniclastic deposits (PU-E), and 3) coherent lava flows (within PU-F; Fig. 2). Coherent lava flows with honeycomb fracturing identified in the microresistivity images have the highest formation resistivities (Table 3; Fig. 7). At PACMANUS, this high resistivity in coherent lava flows was interpreted to correspond to a lower proportion of interconnected pore space (i.e., filled with water and/or conductive alteration minerals, such as phyllosilicates) when compared to volcaniclastic facies. The same characteristics of fluid pathways through interconnected pore space in lava flows are likely at Brothers. Volcanic textures (such as clast size and frequency in volcaniclastic deposits) and fractures are likely to cause the second-order controls on formation resistivity, themselves related to facies and, sometimes, local porosity changes.

In Hole U1530A, the effects of hydrothermal alteration mineral assemblage on resistivity are ambiguous, given the available data. Chlorite, smectite, and, to some extent, illite are conductive (i.e., low-resistivity) minerals (e.g., Pezard 1990; Lévy et al., 2018). These minerals are identified throughout this hole as either major or minor minerals based on XRD data, core observations, and petrography (de Ronde et al., 2019d). Variations in relative amounts of these clay minerals are not associated with variations in resistivity (Fig. 2). Similarly, the K content (measured downHole Using the natural gamma ray profile and in core samples using ICP-AES [de Ronde et al., 2019d]), which reflects the illite content, is consistently high (2–2.5%) between 80 and 189 mbsf, whereas resistivity varies between 1 and 3.5 Ω·m (Fig. 2). In addition, there are zones of high resistivity (>5 Ω·m) in each interval of Alteration Types II, IV, and V (Fig. 2). The effect of pyrite content on resistivity is not readily quantifiable, as pyrite is disseminated in variable amounts in the matrix and veins within the cores. Thus, determination of the local effects of hydrothermal alteration on resistivity at Brothers volcano awaits quantitative estimation of chlorite, smectite, illite, and pyrite, and further laboratory measurements of resistivity on cores (e.g., Lévy et al., 2019).

Textures in Hole U1530A from microresistivity images

Microresistivity images acquired from Hole U1530A over a 346-m-long interval highlight the variability in texture and fracturing. In the 96- to 223-mbsf interval, consisting of volcaniclastic deposits, microresistivity images show that clasts vary in size, frequency, and mineral content, reflected by low- and high-resistivity patches. Lava flows characterized by a coherent image facies are clearly distinguished on the microresistivity image. The variability in texture over short distances is also evident at the sea floor at the NW Caldera (Fig. 10).

Below 223 mbsf, volcanic textures in most cores are too altered to determine the facies (i.e., lava flow or volcaniclastic), apart from some distinct, less-altered intervals of plagioclase-phyric lava flows. Similarly, distinct coherent (lava flow) intervals are recognized in the microresistivity image, but the majority of the interval has an apparent volcaniclastic image facies that cannot be assigned to a genetic origin based on the microresistivity image alone. This interval may consist of (1) a succession of lava flows, (2) monomictic volcaniclastic facies made of clasts of the same composition as the lava flows, (3) a volcaniclastic deposit, or (4) a combination of the first three cases. To propose a genetic origin beneath the 223-mbsf (PUF) interval, we rely on the following independent petrophysical data:

  1. All petrophysical and drilling properties markedly change between 214 and 232 mbsf. Beneath 232 mbsf, torque, density, Vp, and resistivity are higher; porosity is lower.

  2. Between 232 and 448 mbsf, petrophysical properties of discrete core samples show similar properties whether they are in cores identified as less-altered plagioclase-phyric lavas or generally as altered volcanic rocks (Fig. 7).

  3. The Vp-porosity relationship follows a linear trend consistent with the 232- to 448-mbsf interval being lava flows, and distinct from the volcaniclastic deposits above 223 mbsf (Fig. 8).

  4. The fairly constant high intensity of magnetization and magnetic susceptibility at depths >235 mbsf, in contrast to <235 mbsf, is consistent with a package dominated by lava flows rather than an interlayering of lava flows within volcaniclastic facies (Caratori Tontini et al., in press).

  5. The lower pressure dependency of Vp and Vs for the sample from 242.6 mbsf compared to the sample from 122.4 mbsf suggests that the sample from 242.6 mbsf is a lava with microfractures (Adam and Massiot, unpub. data).

  6. The resistivity profile for Hole U1530A is similar to PACMANUS Hole U1189C, i.e., low resistivity with volcaniclastic-dominated facies in the upper part and higher resistivity with lava-dominated lithologies in the lower part with a 10-m-thick transition zone (Arnold et al., 2007).

Hydrothermal alteration is not fundamentally different between 232 and 445 mbsf and the volcaniclastic deposits between 180 and 223 mbsf. Firstly, alteration intensity is ranked as “intense” between 0.095 and 448.68 mbsf, and alteration intensity hinders detailed textural analysis in cores between 179.5 and 222.7 mbsf (de Ronde et al., 2019d). Secondly, quartz is a major component of the alteration mineral assemblage beneath 35.5 mbsf (de Ronde et al., 2019d). Despite silicification from hydrothermal alteration likely acting to homogenize petrophysical properties, the ubiquity of quartz throughout the drilled interval implies that silicification is not the only factor acting on the great change in all petrophysical properties between <223 and >232 mbsf. Thirdly, changes in other alteration mineral occurrence are not associated with variations in petrophysical properties (Fig. 2). These minerals include the following: illite, chlorite, and anhydrite (Alteration Type II); pyrite, illite, and smectite (Alteration Type IV); and diaspore, pyrophyllite, smectite, and rutile (Alteration Type V). For example, Alteration Type IV occurs in intervals of both volcaniclastic deposits (189.10–223.81 mbsf) and apparent volcaniclastic image facies (357.1–385.95 and 409.9–434.89 mbsf), but petrophysical properties are different (higher porosity and lower density, resistivity, and Vp) at 189.10 to 223.81 mbsf than in the deeper intervals.

Based on the combination of independent measurements that are all clearly different above 214 and below 232 mbsf, comparison with results from PACMANUS, and lack of definite evidence that hydrothermal alteration would cause these large petrophysical property changes, we propose that the 232- to 448-mbsf interval is dominated by lava flows. These lavas may be coherent with honeycomb fractures or may have a mottled alteration texture, appearing as pseudoclastic. Similar pseudoclastic textures have been described from hydrothermally altered ancient volcanic successions associated with massive sulfide deposits (Allen, 1988, 1992; McPhie and Allen, 1992; Paulick and McPhie, 1999; Doyle and McPhie, 2001, Paulick et al., 2004).

The 232- to 448-mbsf interval is separated from the volcaniclastic deposits at 85 to 214 mbsf (PU-E) by gradational changes of all the measured petrophysical parameters and torque (Figs. 2, 8). The possible fault at 252 mbsf and its underlying damage zone are located 20 m beneath the top of PU-F, so the boundary between PU-E and PU-F is not structural. Given the consistent dacitic composition of all igneous units at Brothers volcano and the lack of reworking (e.g., brecciation posteruption) shown by paleomagnetism (Caratori Tontini et al., in press), the gradational, great changes in all petrophysical and drilling data between PU-E and PU-F suggest a gradual change in rock texture.

Local variations in texture, fractures, and/or alteration in the 232- to 448-mbsf interval likely cause the small, second-order variations in petrophysical properties within the interval, proposed to consist dominantly of lava flows. The lower resistivity and Vp and higher porosity in the 280- to 300- and 318- to 360-mbsf intervals may correspond to increased fracture intensity or different prealteration porosity, or they may be a result of mineral dissolution in comparison to the nearby lava flows showing a coherent image facies. Although alteration mineralogy could also influence petrophysical properties within the 232- to 448-mbsf interval, we believe it plays a secondary role as smectite is not dominant in volume and its abundance (defined from XRD; de Ronde et al., 2019d) is not correlated to petrophysical properties. However, the available data cannot completely exclude the possibility of these intervals (or part thereof) being of volcaniclastic origin interlayered between lava flows (as documented in Hole U1189C at PACMANUS; Arnold et al., 2007) or monomictic clastic facies.

Mostly thin fractures and a possible fault in Hole U1530A, using microresistivity images

Numerous veins and fractures were observed in the cores, although very few were >2 cm wide. In the microresistivity images, fracture density is undersampled because of the limited azimuthal coverage and locally degraded image quality. This undersampling is partially mitigated by the interpretation of fractured zones rather than distinct fractures. The presence of numerous, wide veins at PACMANUS (Binns et al., 2002) and at the TAG (Humphris et al., 1995) submarine hydrothermal fields shows that partial recovery rate is not the only reason for undersampling of wide veins at Brothers volcano (total core recovery of 18% at Brothers volcano vs. 11% at PACMANUS and 10% at TAG). We therefore interpret the limited number of fractures >2 cm wide in recovered cores at Brothers volcano to indicate a rock mass hosting numerous fractures and veins, but these are mostly thin. Fractured zones identified in the microresistivity images may contain >2-cm-wide veins, but deteriorated image quality prevents us from making this assessment.

In Hole U1530A, the fracture at 252 mbsf is inferred to be a fault underlain by its damage zone that is permeable based on the microresistivity image (fractures with thick apparent apertures, majority of halo fractures), core observations (wide range in fracture and vein dip magnitudes), and permeability indicators (temperature gradient increase, H2S odor; Fig. 5). A prominent fracture striking north-south is observed directly beneath the main plane striking east-west, suggesting a varied fracture orientation, typical of fault zones. A very similar arrangement was documented in a microresistivity image at the Wairakei Geothermal Field, New Zealand, where a high fracture density, including numerous halo fractures, was noted in the hanging wall of a permeable normal fault (Massiot et al., 2017; McNamara et al., 2019). Similar to the Wairakei example, no displacement was observed on either side of the possible fault plane on the microresistivity image at 252 mbsf for Hole U1530A. Thus, in the absence of corroborative evidence such as stratigraphic offsets in nearby wells (as at Wairakei), interpretation of a fault in Hole U1530A remains tentative. An east-west strike for this possible fault is contrary to prominent structures on the sea floor in the NW Caldera area (which generally strike northeast-southwest and northwest-southeast) and to the orientation of the local caldera wall that slopes to the southeast (Embley et al., 2012). The orientation of the possible fault suggests it may be part of (1) the damage zone of a more prominent caldera ring fault, such as the E-W–striking fault just to the north of Site U1527 and visible on the bathymetry (Fig. 1), (2) a rotational slump, or scalloped faulting as observed on the bathymetric map (Fig. 1; Embley et al., 2012), or (3) pathways to some chimney structures in the area (de Ronde et al., 2019b).

Comparison of facies between Holes U1528D (Upper Cone) and U1530A (NW Caldera)

Holes U1528D and U1530A are characterized by dominantly volcaniclastic facies with a few thin plagioclase-phyric lava flows down to 266 and ~232 mbsf, respectively. Beneath, there are lava flows in Hole U1528D (266–355 mbsf), as well as five definite coherent lava flow intervals and a proposed dominantly lava flow facies (232–448 mbsf) in Hole U1530A. Therefore, based on downhole measurements, lava flows represent 31 and 12% of the drilled interval in Holes U1528D and U1530A (112 and 56 m drilled thickness, respectively), with an additional proposed 37% of dominantly lava flows in Hole U1530A (167 m drilled thickness).

The major difference between the two main Brothers volcano holes is that rocks in Hole U1528D are recent (because they are part of a post-caldera collapse cone) and relatively less pervasively altered than those in Hole U1530A. Indeed, primary volcanic textures are observable in cores throughout Hole U1528D (de Ronde et al., 2019c). We infer that this difference is reflected in the petrophysical properties, particularly Vp. In Hole U1528D, Vp cycles through successive highs and lows, consistent with a local variability of textures in volcaniclastic deposits (clast size and arrangement) and mostly coherent lava flows as defined from core observations. Conversely, in Hole U1530A, Vp shows stepwise increases down the hole and less local variability.

Petrophysical properties of PU-F (232–448 mbsf), Hole U1530A, may be the most representative of the deeper parts of the caldera that could be used for future interpretation of any geophysical surveys. Most of the measured Vp in altered core and downhole measurements exceeds the 2.5 km/s average Vp used for the time-to-depth conversion of the seismic reflection section (de Ronde et al., 2019b). Therefore, the original caldera floor depth interpreted from seismic reflection is likely 300 to 700 m deeper than the depth presented by de Ronde et al. (2019b). This estimate, however, requires further pressure-dependent measurements of Vp, consideration of faults and fractures (Tsuji and Iturrino, 2008), and consideration of low-temperature alteration overprints.

Multi-scale permeability and potential for mineralization

Permeability at Brothers volcano is driven by processes operating at several scales, discussed below from matrix to regional scales (Fig. 11). Present-day fluid pathways are indicative of conditions in which mineralization could eventually accumulate. Estimations of flow rates and formation permeability require a dedicated study, based on the hole acting as a cylindrical heat exchanger with the formation (see Becker et al., 1983; Fisher et al., 1997; Winslow et al., 2013). Here, we propose conceptual interpretations of controls on permeability based on facies and fractures interpreted in cores and microresistivity images that will support future modeling studies.

Given the similarities in host rock, alteration, and porosity profiles with depth, we assume that measured matrix permeabilities and the simplified analytical calculations at PACMANUS (Christiansen and Iturrino, 2004) are generally applicable to Brothers volcano (Farough et al., 2019; Parker and Farough, 2019). That is, matrix permeabilities of 10–18 to 10–15 m2 alone cannot sustain the observed hydrothermal venting, and faults and fractures strongly influence fluid flow and mineralization processes in the volcano. This inference is consistent with numerical models of Brothers volcano active sea floor discharge requiring the presence of high-permeability faults to match the measured outflows at the sea floor (Gruen et al., 2012). Such faults were not intersected during Expedition 376; however, there are numerous near-vertical scarps present along the caldera wall without a well-defined, continuous ring fault (Embley et al., 2012). The low matrix permeability is especially likely in the deeper parts of the caldera due to compaction. For any one porosity, the pore shape (which varies as a function of lithology and compaction) determines whether the rock matrix is effectively permeable (Violay et al., 2010). The limited number of fractures and veins >2 cm wide and the lack of permeable zones (as defined from temperature profiles) in the lava flows with honeycomb fracturing suggest that this pervasive fracture network provides only limited permeability. However, matrix permeability combined with pervasive fracture permeability allows alteration to become pervasive over time, enabling the storage of fluids and heat (Fig. 11A). Those conditions are perfect for the storage of the super saline brines found in fluid inclusions in the NW Caldera borehole and sea floor vents (de Ronde et al., 2019a; Deihl et al., 2020; Lee et al., in press) and predicted from the modeling of Gruen et al. (2014).

At the borehole scale, half of the permeable zones have evidence for structural controls in both Hole U1528D (145–160 and ~295 mbsf) and Hole U1530A (255–285 mbsf). It is unclear whether these fractures are primary cooling fractures or subsequent fractures. Indeed, in Hole U1528D, while the permeable zone at 145 to 160 mbsf corresponds to a clear interval of fractured lava flow (PU-C, 140–163 mbsf), only a thin interval of the deeper lava flows (PU-E, 266–355 mbsf) shows a temperature anomaly (i.e., at ~295 mbsf). In Hole U1530A, the possible fault’s damage zone at 252 to 275 mbsf likely provides a connected permeable fracture network (Fig. 11B). Conversely, the multiple fractures (<1 cm thick) and fractured zones (fractures possibly 1–10 cm thick) identified in the microresistivity image and in cores in Hole U1530A, away from the permeable zones, may be sealed today by alteration minerals (i.e., they are veins rather than open fractures), or they are insufficiently connected to the main flow pathways. Indeed, the vast majority of brittle structures observed in the core were completely filled veins, with the ratio of veins to fractures being almost 5 to 1 (de Ronde et al., 2019d).

Variations in lithology and/or alteration appear to form local barriers to fluid flow and, indeed, facilitate focusing of flow directly above, beneath, or along these permeability interfaces (Fig. 11C). In Hole U1528D, the permeable zone at 250 to 275 mbsf coincides with the interface between volcaniclastic deposits (PU-D) and lava flows (PU-E). The volcaniclastic deposits may have acted as a cap layer. In Hole U1530A, the noticeable H2S odor emanating from cores at 240 to 250 mbsf, above the permeable inferred fault, may have resulted from the weakening of the already fractured rock ahead of the drill bit, forming cracks (Davatzes and Hickman, 2010) large enough for gases but not necessarily liquids to move upward (Wiersberg et al., 2020). Focused flow directly beneath these cap layers, or directly above if the cap gets breached by hydrothermal fluids, may enhance fluid-rock interactions, similar to mineralization accumulation by replacement (McPhie and Allen, 1992; Doyle and Allen, 2003; Tornos et al., 2015).

Hydrothermal alteration reflects past circulation of fluids and also, in places, the role of change in facies on focusing fluid flow. In Hole U1530A, the lava flow at 60 to 70 mbsf (PU-C), showing very few veins in cores (one vein per 10 cm on average) and located directly above a clay-rich interval, may have acted as a barrier to vertical flow, focusing and harboring fluids in the underlying volcaniclastic deposits. The focusing of fluids is further supported by the presence of several network veins in cores (5–10 veins per cm on average), mostly filled with pyrite, in the 70- to 85-mbsf interval.

In some cases, current permeable zones coincide with changes in hydrothermal alteration that reflect past permeability. In Hole U1528D, the tops of permeable zones showing the least evidence for current structural permeability (i.e., ~100 and 235 mbsf) are located at alteration mineral assemblage and geochemical boundaries indicative of different fluid temperatures and acidity (de Ronde et al., 2019a). Similarly, the tops of both permeable zones in Hole U1530A coincide with drops in K interpreted to be indicative of extensive dissolution of plagioclase and other igneous minerals at high temperature. The 190- to 210-mbsf permeable zone in Hole U1530A coincides with the first appearance of the high-temperature alteration mineral pyrophyllite (Alteration Type IV; 230°–350°C). This mineral is indicative of earlier, high-temperature acid-sulfate fluids (Reyes, 1990; de Ronde et al., 2019a), which may have been capped by the overlying volcaniclastic deposits having only rare clasts on the microresistivity image. Therefore, some permeable zones seem to be long-lived.

Holes U1528D and U1530A provide endmembers of permeability styles (i.e., upflow of hot fluid versus downflow of cold fluid). At Hole U1528D, high-temperature fluids (>247°C) traveling up the hole and low-pH fluids sampled in the hole are consistent with the Upper Cone hosting active white smokers. At Hole U1530A, cores showing overprinting of higher-temperature mineral assemblages (up to 250°C) by relatively low-temperature minerals (<100°C; de Ronde et al., 2019a) are consistent with the relatively low downhole temperature profiles. This low temperature is surprising, however, given that Hole U1530A is situated <100 m away from black smokers discharging fluids up to 302°C (Berkenbosch et al., 2012), structurally sitting above an area of prominent, metal-rich stockwork (de Ronde et al., 2019b).

The proximity today of this relatively “cold” U1530A hole to active high-temperature black smokers requires one of the following: (1) a vertical permeability barrier such as a sealed fault or dike, (2) closely-spaced permeable faults providing downflow of cold water and upflow of hot fluids, (3) a focused, permeable fault supplying fluids to the black smokers in a low-permeability host rock that is incurring cold-water recharge, and/or (4) very narrow, local, convection cells bounded by low-permeability barriers to vertical flow (Fig. 11D). Option (3) is consistent with observations of exhumed calderas in which small-displacement faults (<150-m) have <5-m fluid transport corridors (e.g., Garden et al., 2017). All of the above options would result in narrow (and shallow) convection cells in the NW Caldera. Such narrow convection cells are consistent with heat flow measurements and the thermal model proposed by Caratori Tontini et al. (2019) and are also consistent with different fluid compositions measured at different vent sites within the field (de Ronde et al., 2011, 2019b; Kleint et al., 2019; Stucker et al., in press). Finally, distinct microbial communities are associated with these local upflow zones, which ultimately may be constrained by fluid flow pathways (Reysenbach et al., 2020).

Permeability also varies through time, as demonstrated by the intertwined and overprinted alteration mineral assemblages, including variable fluid inclusion populations (de Ronde et al., 2019a; Diehl et al., 2020; Lee et al., in press). Alteration overprint shows that current circulation of relatively cold fluids in the vicinity of Hole U1530A is likely recent.

On a caldera scale, the permeability architecture of Brothers volcano is likely 3-dimensional (Fig. 11E). IODP Expedition 376 successfully recovered cores and in situ measurements and fluids along a northwest-southeast profile. Arc-perpendicular extension across NE-SW–striking rift zones, tectonic ridges, and possible dikes identified from bathymetry (Figs. 1 and 11e; Embley et al., 2012; de Ronde et al., 2019b) potentially have high permeability, especially where they intersect the caldera walls. Evaluating this potential 3-D permeability structure awaits further sea floor and drilling measurements.

This study has highlighted the spatial and temporal complexity of volcanic facies and fluid flow in a submarine arc volcano, Brothers volcano. For the first time, downhole measurements reveal controls on fluid flow in this active submarine hydrothermal system. Continuous downhole measurements contribute to unravelling the stratigraphy whose delineation was limited by sporadic (generally <2 m every 4.8 m) core recovery. Based on the combination of microresistivity image facies and independent petrophysical properties, and guided by available core observations, we refine the stratigraphic boundaries and units defined from cores. In addition, intense hydrothermal alteration that mostly masks the original volcanic textures in the bottom part of Hole U1530A prevents the determination of volcanic facies. We propose that the 232- to 448-mbsf interval on Hole U1530A is dominated by lavas being either coherent with honeycomb fractures or having a mottled alteration texture, appearing as pseudoclastic. Core measurements remain essential to calibrate the interpretation of these downhole measurements performed in challenging conditions, and to complement the scale of investigation (i.e., mm to cm in cores, 0.1–100 m in downhole measurements).

Petrophysical properties at Brothers volcano, down to 232 mbsf in Hole U1530A and to 360 mbsf in Hole U1528D, are primarily controlled by the depth below sea floor, interpreted as a combined effect of compaction and alteration. Second-order controls include volcanic facies and fracturing. In Hole U1528D of the Upper Cone, the more recent lava flows have more locally varied petrophysical properties than in Hole U1530A of the NW Caldera.

Permeability at Brothers volcano is largely controlled by faults and fractures, but also by locally contrasting rock properties that focus fluids directly beneath, above, or along low-permeability interfaces. The low temperature and the alteration overprint encountered in Hole U1530A, located <100 m from actively discharging black smokers on the sea floor, are consistent with the narrow, shallow convection cells proposed for the NW Caldera by Caratori Tontini (2019). It remains unclear whether the convection cells are controlled by the fault system or lithological barriers. This study provided constraints on the hydrological regimes and host rocks hosting the hydrothermal systems of the NW Caldera and the Upper Cone at Brothers, and therefore the propensity for mineralizing fluids to accumulate and form an ore deposit on or near the sea floor. This work will add to models for the formation of sea floor VMS deposits associated with these volcanoes.

See Appendix 1 for intervals of downhole measurement and tool string composition, as well as for depth intervals and main descriptors for petrophysical units and subunits in Holes U1530A and U1528D. See Appendix 2 for the following:

  • -Hole U1530A downhole measurements

  • -Hole U1530A background resistivity

  • -Hole U1530A image facies

  • -Hole U1530A fracture zones

  • -Hole U1530A confidence in interpreting facies

  • -Hole U1530A distinct fracture characteristics

  • -Hole U1528D downhole measurements.

See Appendix 3 with extended versions of Figures 2 and 3.

This research used samples and data provided by the International Ocean Discovery Program (IODP). Logging data are available from http://mlp.ldeo.columbia.edu/logdb/scientific_ocean_drilling/. We thank the IODP technical staff (particularly Z. Mateo, E. Claassen, and D. Ferrell) and the D/V JOIDES Resolution crew for their invaluable support and perseverance during a challenging expedition. Co-chief Susan E. Humphris and IODP staff scientist Tobias W. Höfig provided great insights and logistics throughout the project. Thanks to staff at Lamont-Doherty, who processed data rapidly and facilitated onboard discussion, and to the shipboard science party, particularly A. Reyes and A. Farough. This research is supported by the New Zealand government via the Te Riu a Maui/Understanding Zealandia research program at GNS Science. The authors also thank M. Lawrence, A. Griffin, J. Burnell, R. Funnel, W. Kissling, R. Sutherland, and Y. Hamada for fruitful discussions. Constructive reviews by J. McPhie and A. Fisher significantly improved the manuscript.

Cécile Massiot is a geothermal scientist at GNS Science, New Zealand. She undertook her geoscience graduate degree at the École Nationale Supérieure de Géologie (ENSG; Nancy, France). After working on geothermal systems at Iceland Geosurvey (ÍSOR) and GNS science, she conducted her Ph.D. work in fracture system characterization in volcanic and metamorphic rocks at Victoria University of Wellington (New Zealand). Doctor Massiot specialises in characterizing fracture systems and fluid pathways, using borehole images and other data, to improve resource management and understanding of volcanic and tectonic processes. She conducted borehole data interpretations for the International Continental Scientific Drilling Program (ICDP) Iceland Deep Drilling Program and Deep Fault Drilling Program as well as for IODP Expedition 376.

Gold Open Access: This paper is published under the terms of the CC-BY-NC 3.0 license.

Supplementary data