High-resolution (1.5 m) mapping from the autonomous underwater vehicle (AUV) D. Allan B. of West Mata Volcano in the northern Lau Basin is used to identify the processes that construct and modify the volcano. The surface consists largely of volcaniclastic debris that forms smooth slopes to the NW and SE, with smaller lava flows forming gently sloping plateaus concentrated along the ENE and WSW rift zones, and more elongate flows radiating from the summit. Two active volcanic vents, Prometheus and Hades, are located ∼50 and ∼150 m WSW of the 1159 m summit, respectively, and are slightly NW of the ridgeline so the most abundant clastic deposits are emplaced on the NW flank. This eruptive activity and the location of vents appears to have been persistent for more than a decade, based on comparison of ship-based bathymetric surveys in 1996 and 2008–2010, which show positive depth changes up to 96 m on the summit and north flank of the volcano. The widespread distribution of clastic deposits downslope from the rift zones, as well as from the current vents, suggests that pyroclastic activity occurs at least as deep as 2200 m. The similar morphology of additional nearby volcanoes suggests that they too have abundant pyroclastic deposits.
Submarine eruptions are clearly common events but to date only NW-Rota-1 in the Marianas Islands [Embley et al., 2006; Chadwick et al., 2008a; Deardorff et al., 2011] and West Mata volcano in the northern Lau Basin [Resing et al., 2011] have been directly observed so that volcanic landforms and deposits can be directly linked to processes observed in situ. The activity at West Mata was first detected as an event plume in the water column in November 2008 [Baker et al., 2009]. The eruption site was located in May 2009 by the remotely operated vehicle (ROV) Jason II [Resing and Embley, 2009; Resing et al., 2011] during NOAA-NSF-MBARI cruise TN234 on the R/V Thompson, and was found to be still active in March 2010 during a subsequent NOAA cruise [Baker et al., 2010; Dziak et al., 2010]. The MBARI Mapping autonomous underwater vehicle (AUV) D. Allan B. collected swath bathymetric data during the 2009 cruise to produce 1.5 m lateral resolution bathymetry of the summit, the WSW and ENE rift systems, and about half the lower flanks of the volcano. This high-resolution AUV bathymetry reveals the fine-scale morphology of the volcano's eruptive vents and rift zones. On a coarser scale, comparisons of repeated surveys of the volcano between 1996 and 2010 with ship-mounted multibeam sonars show depth changes on the north flank that we interpret are due to deposits from eruptive activity near the summit. Here we describe and interpret these data to understand the geologic setting of the active eruptive vents, the long-term accumulation rate of eruptive deposits, and the structure of West Mata volcano. West Mata is one of several morphologically similar volcanoes in the northern Lau Basin so defining the eruptive processes that formed one of them will provide insight into the formation of the others as well.
2. Geologic Setting and Background
West Mata volcano is located in the northernmost Lau Basin about midway between the Tonga volcanic arc and the Northeast Lau Spreading Center, southwest of Samoa (Figure 1). Surface ship multibeam bathymetry collected during the 2008 and 2009 cruises showed West Mata to be a nearly conical structure with smooth slopes and rift zones extending WSW and ENE from the summit (Figure 2a), and mapped other similarly constructed and oriented seamounts in the area [Embley et al., 2009]. The volcano rises steeply from seafloor depths as great as 2900 m near the tip of the WSW rift zone to a summit at 1159 m and has a volume of ∼26.6 km3. Two active volcanic vents named Hades and Prometheus were discovered at ∼1200 and ∼1175 m depth, respectively, near the summit [Resing and Embley, 2009; Resing et al., 2011]. The vents were actively erupting lava of boninite composition in May 2009 [Michael et al., 2009; Rubin et al., 2009; Resing et al., 2011] with vigorous ejection of lava clasts (lava fountains of Head and Wilson ) alternating with strombolian bubble-burst activity and effusion of small pillow flows at the Hades vent and fire fountains from the Prometheus vent [Clague et al., 2009b; Resing et al., 2011]. The summit region, including the two active vents [Caress et al., 2009], and the two major rift zones were the primary targets of the high-resolution AUV mapping effort and the locations of five successful ROV JASON II dives during R/V Thompson cruise TN234 in May 2009.
3. Data Acquisition and Processing
The MBARI Mapping AUV D. Allan B. is an MBARI designed autonomous vehicle [Caress et al., 2008b; Paduan et al., 2009a] that surveys 50–75 m above the bottom at 1.5 m/s (3 knots), with an endurance of about 18 h. It is outfitted with a Reson 7125 200 kHz multibeam sonar, an Edgetech FS-AU sonar package including 110 and 410 kHz chirp side scan and 2–16 kHz chirp subbottom profiler, a SeaBird SBE-49 fastCat CTD, a Paroscientific pressure sensor, a three component magnetometer, and for some missions, including these in the Lau Basin, a nephelometer and an Eh sensor. The 7125 generates 256 1° × 1° beams across a 135° swath, producing soundings with 1.5 m wide beam footprints at the 75 m altitude used for these surveys. The repeatable vertical precision of the bathymetry is 10 cm following tidal correction. The real-time navigation is provided by a Kearfott SeaDevil inertial navigation system (INS) that integrates a 24 cm ring laser gyro, accelerometers, and a 300 kHz Doppler Velocimeter Log (DVL) through a Kalman filter. The INS is initialized by GPS at the surface and aided during descent by USBL fixes transmitted to the vehicle using an acoustic modem. The INS achieves a position accuracy of 0.05% distance traveled provided DVL bottom lock is maintained.
The AUV multibeam data were processed using the MB-System open source software package [Caress and Chayes, 1996, 2010]. The MB-System bathymetry processing workflow begins with the calculation of the bathymetry from raw multibeam travel times and angles using the INS attitude data and a water sound speed calculated from on-board conductivity, temperature and pressure-depth (CTD) values. The bathymetry are then interactively edited in both a single-file waterfall view (program MBedit) and a multifile, survey view (program MBeditviz). The latter tool allows editing in a 3D “cloud of soundings” context where the soundings displayed are associated with interactively selected areas of the survey. Following bathymetry editing, the navigation is adjusted to match features in overlapping swathes using the MBnavadjust tool, resulting in a navigation model with relative accuracy equivalent to the lateral bathymetric resolution. In the case of the 75 m altitude West Mata surveys, the lateral resolution is 1.5 m. The AUV bathymetry are also referenced to GPS-navigated, hull mounted multibeam bathymetry collected during the cruise from R/V Thompson, resulting in an absolute position accuracy on the order of 10 m. The maps presented in this paper derive from 2-m resolution grids with a 5-m interpolation generated using a footprint-weighted-mean algorithm of program MBgrid in which the grid cell depth values are calculated as weighted averages of the overlying soundings. Each sounding weighting is represented by a 2D Gaussian function stretched over the sounding's footprint on the seafloor (determined by beam widths, beam angles, and sonar altitude) and partitioned among the grid cells overlaid by that footprint. The footprint-weighted-mean algorithm proceeds in two steps. The first, low-resolution step provides a seafloor slope model that is then used to tilt each footprint according to the local slope during the high-resolution step, allowing proper representation of steep slopes in maximum-resolution grids. Specific features in the high-resolution map data can be co-located with the same features observed during USBL or LBL navigated ROV dives.
The AUV D. Allan B. completed two mapping surveys on West Mata Volcano during cruise TN234. The first survey did not collect subbottom data due to the expected thin sediment cover and the side scan data was compromised due to an incorrect instrument setting. The second survey collected subbottom data that did not detect any sediment cover, but did record acoustic noise from the erupting vents, as will be described in a later section. Due to the steepness of the slopes on West Mata, and despite our planned mission altitude of 75 m (instead of 50 m), uphill survey lines repeatedly aborted when the slopes exceeded the maximum 30 m/minute climb rate of the vehicle while it advanced at 3 knots. When this occurred, the vehicle is programmed to stop forward thrust and float upwards until reaching a safe altitude to avoid colliding with the bottom. These disruptions of the survey created additional artifacts in the data that limit the resolution.
Much of our interpretation here depends on combined mapping and dive observations made elsewhere, since dive observations are quite limited at West Mata. Clague and Paduan  summarize the identification of many submarine volcanic features as mapped at different resolutions, including the 1–1.5 m resolution achieved by the MBARI mapping AUV.
Ship-based multibeam sonar bathymetry was collected over West Mata during expeditions in 1996, 2008, 2009, and 2010 (Table 1). Comparison of these surveys to identify significant depth changes generally followed the methods described in the study by Chadwick et al. [2008b]. The data from each survey were gridded at a cell size of 20 m using MB-System software, with spline interpolation to the adjacent 5 grid cells to fill any small data gaps. Comparison of the surveys in areas away from the volcano suggests that there is a systematic depth bias of +10 m in the 1996 survey, and so we have subtracted that amount from the 1996 data (making the seafloor depths shallower). After this correction, the grids from each year's survey were subtracted from one another to look for significant depth changes.
Table 1. Ship-Based Multibeam Sonar Surveys at West Mata Volcano That Were Compared
R/V Kilo Moana
The high-resolution AUV bathymetric data (Figures 2b and 3) show the same gross features seen in the surface ship multibeam data (Figure 2a) with prominent rift zones extending away from the summit, which is the apex of a conical structure with a nearly circular base. Even at 1.5 m resolution, the northwest and southeast flanks of the volcano are smooth. The rift zones consist of stair-step lava plateaus that are shingled on each other. The ENE rift zone curves from ENE to E as it deepens; the WSW rift zone has a west-oriented limb at its terminus. The summit is a narrow, irregular ridge aligned with the two rift zones. There are no summit pits or obvious caldera structure, although remnant inward-facing walls, perhaps fragments of a former caldera, were mapped (Figures 4a and 5). Cone-shaped volcanic features are rare and restricted to the lowermost portions of the two rift zones.
4.1. Summit Region
The characteristics of the summit are illustrated with an AUV bathymetric map of the summit (Figure 4a), a slope map derived from the same data (Figure 4b), and an oblique perspective view from the N (Figure 5). The locations of the Prometheus and Hades vents are indicated, as are lava and clastic sediment samples collected by ROV JASON II. These vents and their ongoing eruptions will be described elsewhere, but are situated in embayments along the NW side of the summit ridge and upper WSW rift zone, respectively. They are roughly 90 m apart and separated by a rough textured lava flow from the summit ridge that terminates as a smooth lobe. Downslope of the vents to the NW, the bottom is not characterized by obvious lava flow lobes, but rather by a smooth steep slope (Figure 5) with subtle, thin “sheet-like” fans superposed on one another. The slope (Figure 4b) decreases with increasing depth, especially on the NW flank, so that near the summit it averages 34–35° whereas lower on the flanks it averages 27–28°.
Other embayments in the summit ridge, on both the NW and SE sides, have smooth slopes below. Flow lobes radiate from the summit ridge to the N, NW, S and SE; all were emplaced during earlier eruptions or earlier phases of this eruption. Within 300 m of the summit, the rift zones consist of gently sloping, broad lava plateaus that are crudely wedge-shaped with their apex pointed upslope. These nearly flat-lying plateaus differ from those farther downrift in having upturned outer margins, but all have outer slopes generally steeper than 60° and as steep as 80°.
4.2. ENE Rift Zone
Three portions of the ENE rift are illustrated in Figures 6–8, with progressively increasing depth. Starting on the upper part of the rift zone, two linear features characterized by aligned pits are shown in Figure 6. The pits are as large as 40 m across, 20 m deep and occur in strings as long as 400 m. These probably indicate the locations of eruptive fissures cross through several successively deeper lava plateaus that have sharp scalloped margins both on the downrift and lateral sides. Numerous lobate structures cascade down the otherwise smooth northwest and southeast flanks of the rift zone. Figure 7 shows a different style of eruptive fissure consisting of linear toughs as long as 200–250 m between low ridges. The trough to adjacent ridges has relief up to 10 m. The shingled plateau-like structures, up to 300 m long and 200 m wide with steep margins as tall as 45 m, are less apparent in this middle part of the ENE rift zone and the plateau margins are not as strongly scalloped, although they are still steep. The distal part of the ENE rift, characterized by near-circular cones, is shown in Figure 8. To the east of the cones, the nearly flat seafloor has subtle channels and shallow circular depressions.
4.3. WSW Rift Zone
The upper WSW rift zone (Figure 9) is characterized by shingled, near-circular, slightly domed plateaus with steep margins (see inset in Figure 9). Deeper along the rift, the plateaus become broader and flatter and have scalloped margins. The northwest and southeast flanks are mainly smooth, with some hummocky mounds. The lowermost rift zone (Figure 10) consists mainly of coalesced cones, including a few with flat tops (see inset of Figure 10). Many of these cones have steep sides and smooth lower slopes and are widely distributed near the rift zone. Two parallel hummocky ridges (Figure 11), separated by a zone of smooth clastic debris, fit together roughly like jigsaw pieces, but are offset from one another by 140 m (lower half) and 270 m (upper half).
4.4. North and South Flanks
The north flank, best seen in the bathymetry (Figure 2b) and the slope map (Figure 4b), is almost entirely smooth from near the summit to at least halfway down to the surrounding seafloor. The south flank is also mainly smooth, with a narrow linear zone of rougher morphology extending S from the summit to the base of the volcano. This rough zone resembles the two rift zones in some ways, but we interpret that it is not one (see discussion in Results section). The upper part (1380–1820 m depth, Figure 12) shows sheet-like structures that are somewhat different from those along the rift zones in that they are smoother in texture, broader, and more fluid-shaped in outline. The lobes on the lower slope (2090–2590 m depth) more closely resemble the plateaus along the ENE and WSW rift zones, but their surfaces are not as flat or as nearly horizontal as the other rift plateaus. In some cases, they resemble the low cones on the deepest parts of the other rifts. A narrow smooth zone separates the upper and lower portions of this rough zone and other smooth slopes fill in between lobes between 1860 and 2290 m depth (Figure 2b). A second smaller array of lobes extends to the SE from the summit (Figure 2b).
4.5. Acoustic Noise Over the Summit
The AUV's subbottom chirp system collected data during the second mission, which ran up the entire WSW rift, over the summit, and down the upper half of the ENE rift zone. The upper portion of the profile (Figure 13) was affected by acoustic noise but only near the summit. As the AUV traveled over the summit, noise in the 3–10 kHz range was recorded as dark bands on the subbottom profile (Figures 13b and 13c); the locations of these bands are shown on Figures 4a and 13a. There are two narrow, lower amplitude bands located near Hades vent (points 2 and 3) and a broader higher-amplitude band located close to Prometheus vent (between points 4 and 5). In all three cases the apparent noise has a constant amplitude over each entire sweep, and so does not represent water column reverberations, multiples, or scattering from particles or bubbles in the water column. Instead, these bands image ambient sound present at a 75 m altitude only at three locations along the vehicle track.
4.6. Depth Differences Between Ship-Based Bathymetric Surveys
Significant depth changes are observed between the 1996 and the 2008–2010 bathymetric surveys (Table 1). However, the quality and density of soundings in the 1996 multibeam survey are significantly lower than those in the more recent surveys, which results in comparisons that are noisier and more difficult to interpret than the comparisons made exclusively with data from recent generation sonars. The noise level in these kinds of comparisons is generally about ±10 m [Chadwick et al. 2008b], but in this case it is closer to ±30 m. During gridding of the sonar data, we have minimized interpolation between individual soundings, because false depth differences can be created where data in one survey are sparse and are interpolated through gaps or at the edges of swath coverage. This is essentially an issue of resolution; denser coverage in the newer surveys can make fine-scale features appear to be “new” that were not previously well-resolved by the older sparser soundings. Therefore, our confidence in interpreting where depth changes are real is highest where (1) the sonar coverage is good in both surveys, (2) the differences appear over broad areas and are not coincident with fine-scale features appearing in the later surveys, (3) the magnitude of the depth change is well above the background “noise” level, and (4) the shape and location of the depth changes make geologic sense.
The largest depth differences between 1996 and 2010 are at the summit and on the north slope of the volcano, downslope of the eruptive vents discovered in 2009 (Figure 14), and these are the only differences that we interpret to be real because they meet the criteria above (Figure 14, and outlined in Figures 2b and 4a). The apparent depth changes along the WSW rift and the south flank are not reliable because they are in areas that were not well-insonified in the 1996 survey, although their locations could make sense geologically. The areas of apparent negative depth change on the NW and NE flanks (blue areas in Figure 14a) coincide with areas of low data density in the 1996 survey, and are also deemed to be artifacts. When the areas interpreted to be unreliable depth differences are excluded from the map (Figure 14b) the remaining higher confidence depth differences extend at least 3 km north from the summit to depths >2600 m.
We calculate the areas and volumes of depth change in two ways: (1) including depth differences above +10 m (the typical background noise level found in other such comparisons) and (2) including only those differences above a more conservative +30 m (perhaps more appropriate for this comparison due to the relatively noisy 1996 survey). Depth differences greater than +30 m in thickness have an area of 2.2 × 106 m2 and a volume of 8.0 × 107 m3. If we include the lower-confidence area outlined by +10 m depth differences, the area covered more than doubles to 4.5 × 106 m2 and the volume increases to 1.4 × 108 m3. The maximum depth change between the surveys is 88 m near the summit, and 96 m on the north flank. We interpret these positive depth changes as due to the deposition of pyroclastic ejecta and fragmental lavas that were erupted from the summit vents and remobilized downslope during this 14-year time interval. Another important result is that no significant depth differences (above the typical ±10-m noise level) are evident between the 2008, 2009, and 2010 surveys, suggesting that the eruptive flux during the past several years has been small in comparison to the preceding decade.
The smooth bottom that characterizes so much of the steep slopes of West Mata consists of clastic volcanic debris, as observed and sampled by the ROV JASON II. The clastic debris fans make the gross shape of the volcano appear more conical than if there were only effusive lava flows like the plateaus along the rift zones or the cones at the base of the rift zones. The conical form of the volcano is defined in the low-resolution ship-based bathymetry as well (Figure 2a). Other such conical seamounts of the NE Lau Basin [Embley et al., 2009] may be composed dominantly of clastic volcanic debris as well. The upper 34–35° slopes consist of talus draped by sediment (Figure 15a) and chutes of sand-sized clastic debris (Figure 15b). The lower 27–28° slopes are close to the critical angle of repose for water-saturated sand and are interpreted to be almost entirely composed of volcanic sand. The widespread distribution of such clastic deposits is consistent with the dominant eruptive style observed at the Prometheus vent (location in Figure 4a), where low lava fountains produce mainly pyroclastic fragments [Clague et al., 2009b]. Hades vent exhibited a range of eruptive behaviors ranging from low lava fountains, strombolian bubble bursts, to effusion of pillow lavas. Many of the pillow flows observed are elongate tubular pillows that form on steep slopes (Figure 15c). Many of the pillow flows fragmented as they flowed down the steep slopes and produced pillow breccia that mixed with other primary clastic debris, similar to the lithologies shown in Figure 15a, and interpreted to be responsible for the talus that steepens the upper slopes.
Smooth slopes that we interpret as fine clastic debris formed during pyroclastic eruptions along the rift zones, and coarser talus shed from the lava flows, plateaus, and cones, can be traced upslope perpendicular to contours to the rift zones at depths as great as 2350 m, suggesting that explosive pyroclastic activity on West Mata is common at least this deep, and much deeper than most theoretical models suggest [Head and Wilson, 2003; White et al., 2003] without extraordinary initial volatile contents or accumulation of volatiles. Previous studies suggest that strombolian bubble-burst basalt eruptions occur along the mid-ocean ridge system for volatile-poor mid-ocean ridge basalt at least as deep as 1600 m deep on Axial Seamount on the Juan de Fuca Ridge [Helo et al., 2011], 1750 m on the mid-Atlantic Ridge near the Azores platform [Fouquet et al., 1998; Eissen et al., 2003], 3800 m on the Gorda Ridge [Clague et al., 2003b, 2009a], and 4000–4116 m deep on the Gakkel Ridge [Sohn et al., 2008]. Deep water strombolian activity of more volatile rich lavas has also been observed at 550–560 m depth on NW Rota-1 in the Marianas arc [Chadwick et al., 2008a; Deardorff et al., 2011] for basaltic-andesitic lava, at inferred at least as deep as 590 m depth offshore Oahu [Clague et al., 2006], 1300 m at Loihi Seamount [Clague et al., 2003a; Schipper et al., 2010], and 4300 m for volatile-rich strongly alkalic lavas in the North Arch volcanic field [Davis and Clague, 2006]. The distribution of clastic debris on West Mata suggests that boninite eruptions can also be pyroclastic much deeper than the activity observed at the active vents near the summit at 1175–1200 m depth.
The lava plateaus are bounded by steep slopes consisting of pillow talus or breccia (Figure 15d), occasionally draped by pillow flows (Figure 15e). The nearly flat-topped plateaus are commonly covered or partly covered by thick deposits of mainly sand-sized pyroclastic volcanic fragments (Figure 15f). These volcanic sands are most likely fall deposits that settled from pyroclast-rich plumes from nearby eruptions. The glass chemistry of the clasts in the deposits shown in Figure 15f includes several compositions, none that match the composition of the ongoing eruption [Clague et al., 2009b] located only 300–400 m away. We infer the clastics were deposited during several earlier eruptions and that the clastic deposits are not widely dispersed by being lofted high in the water column and settling to the surface, as proposed for mid-ocean ridge eruptions with dominant effusive and minor pyroclastic activity [Clague et al., 2009a]. The compositions, dispersal, and deposition of the pyroclastic debris at West Mata will be discussed in a subsequent paper.
We interpret the steep talus ramps surrounding the plateaus to have formed as part of the active volcanic processes that built the plateaus, rather than during a subsequent period of erosion. At the Gorda Ridge, steep pillow mounds erupted in 1996 [Chadwick et al., 1998] form similar talus or pillow breccia at the base of steep slopes. Clague et al.  argued that the talus formed during the eruption as pillow lava flows cascaded over near-vertical margins of the mounds and fragmented. Similar talus deposits draped by pillow lava occur at Cage Seamount at the CoAxial segment of the Juan de Fuca Ridge (Embley et al. , and observed during ROV Doc Ricketts dive D77 in 2009) and are also thought to have formed as part of the eruptive sequence that built the flat-topped seamount. The same mechanism may form much of the steep talus observed below the plateau margins, a view consistent with the presence of pillow lavas draping the talus (Figure 15e).
5.2. Eruptive Fissures, Lava Flows, and Cones
The linear troughs between low ridges (Figure 7) resemble subaerial spatter ramparts along eruptive fissures, and are either similar structures, presumably constructed of fragmental (non-agglutinated) spatter or small pillow ridges built on parallel fissures. Pillow ridges aligned along eruptive fissures are common along the mid-ocean ridge system and have been described for the 1986 North Cleft eruption [Chadwick and Embley, 1994], the 1993 and 1982–1991 CoAxial eruptions [Embley et al., 2000] on the Juan de Fuca Ridge, at the 1996 northern Gorda Ridge eruption [Chadwick et al., 1998], and along the axial volcanic ridge in the mid-Atlantic [Searle et al., 2010] and East Pacific Rise [White et al., 2000; Bohnenstiehl et al., 2008]. The prevalence of smooth clastic deposits downslope from these features, however, supports the idea that they are tephra or spatter ramparts.
The series of aligned collapses (Figure 6) could be either “skylights” along a lava tube, as described along Kilauea's submarine Puna Ridge [Smith et al., 2002] or drainbacks from a lava pond perched above an eruptive fissure. We prefer the latter interpretation because of their orientation parallel to the rift.
The large-scale morphology of West Mata, the adjacent East Mata, and five other nearby volcanoes are all dominated by two well-developed WSW and ENE-oriented rift zones [Embley et al., 2009]. The dual rift zones observed at these volcanoes probably reflect edifice construction within an extensional stress regime that focuses dike intrusion along orientations perpendicular to the dominant extensional stress. The rift zones of West and East Mata all exhibit curvature to the right away from the summit. This consistent downslope rift zone curvature could reflect temporal changes in the stress field or interaction between nearby extensional fault zones.
The dual rifts on West Mata and other nearby volcanoes suggest that the prominent lobate structures that extend southward from the summit to the base at West Mata (Figure 12) are not a third rift. They likely reflect the path taken by some of the more voluminous flows erupted at the summit. These lobate flows cascaded downslopes as steep as 35°, making them quite different from long tube-fed flows on land [e.g., Clague et al., 1999] or underwater [Clague et al., 2002] that advance downslopes of only a few degrees. On steeper slopes on land, flows tend to form channelized àà flows [e.g., Kauahikaua et al., 2002], and we have observed similar channelized sheet flows within the caldera at Axial Volcano on the Juan de Fuca Ridge [Clague et al., 2007]. The lack of channels or “skylights” observed in the high-resolution bathymetry for any of these lobes, coupled with the steepness of the slope, support the idea that these lobes are not tube-fed lava flows. The lobes lower on the slopes are more similar to pillow mounds or the lava plateaus along the rift zones but seem unlikely to be the distal ends of lava flows. They may be pillow mounds that formed early in the history of the volcano that have escaped burial by subsequent clastic debris, although there is no obvious reason this area should have remained unburied. A plausible origin of the lobes on the upper slopes, despite their thickness and steep outer slopes, is that they are flows of coarse clastic debris produced near the vents during extended effusive activity near the summit, and that cascaded down the slope as debris avalanches. Alternatively, this entire rough zone could consist of thick lava flows erupted near the summit. This region warrants further exploration to determine its origin.
The plateaus constructed along the rift zones (Figures 6, 9, and 11) resemble structures observed along the submarine rift of Kilauea Volcano [Smith et al., 2002] and in the Galapagos [Geist et al., 2006] and may have formed in a manner analogous to the formation of circular flat-topped cones that are proposed to form from moderate steady effusion of lava that creates a perched lava lake surrounded by levees [Clague et al., 2000]. On steep slopes, these circular constructs may be modified by preferential levee rupture on the steep downslope sides. Their margins are clearly modified by collapses, as will be discussed below in section 5.4.
The flows beyond the distal end of the ENE rift zone (Figure 8) resemble sheet flows with subtle relief indicative of drainouts and collapses [Ballard et al., 1979; Fundis et al., 2010] and presumably formed during voluminous eruptions at the base of the rift zone. We have mapped similar flow morphologies with the AUV at Axial Volcano, and at the CoAxial and North Cleft segments of the Juan de Fuca Ridge [Clague and Paduan, 2009]. Similar flows with subtle relief occur on nearly flat terrain inside the caldera at Axial Volcano. Along the ridge axis, these high-effusion rate flows form intricate deep channels lined by numerous lava pillars [Francheteau et al., 1979; Gregg and Chadwick, 1996; Soule et al., 2005] and the advancing flow fronts are pillow lavas with sometimes collapsed lobate flows adjacent to the channels.
5.3. Possible Caldera Remnants?
Near the summit at 1320 m depth on the NNE and WSW rifts, two plateaus have slightly upturned downslope margins and very steep outer slopes. These features could be formed by eruptive centers that shifted location through time, but would require that plateaus erupted farther downrift grew taller than those closer to the summit and that two such plateaus fortuitously formed at the same depth, one on each rift. As an alternative, we propose that these features, labeled on Figures 4a and 5, are the possible rim remnants of a prior caldera. Such a caldera would have been about 700 m across and would be almost entirely filled by flows and debris from eruptions postdating the caldera. If these features are remnants of a former caldera, then West Mata has had a longer, more complex history than its rather simple overall conical shape would imply. It would also suggest that periods of much higher magma flux led to development of a magma reservoir within the volcanic edifice, implying much more rapid throughput of lava and higher eruption rates at times in the past. That only small remnants remain on the two rift zones would also imply efficient burial or destruction of the rim of this caldera on the NW and SE flanks. Perhaps the lava plateaus down-rift of these possible caldera remnants formed during such a period of higher magma flux. Other morphologically similar volcanoes show no evidence of summit calderas in the lower-resolution bathymetry. The former existence of a caldera remains speculative, pending further dive observations.
Many of the plateaus constructed along the rift zones have scalloped margins that closely resemble slump or landslide headwalls [e.g., Moore, 1977; Hampton and Lee, 1996]. These lava plateaus bear a striking resemblance to lava benches formed at Kilauea where lava enters the sea [Mattox and Mangan, 1997], and like the Kilauea benches, developed on top of unconsolidated volcanic sediment. During construction, the Kilauea lava benches commonly slide away, such that the coastline retreats and the process repeats. Much of the talus and slide chutes may have resulted from this process and probably formed during eruptions that constructed the plateaus.
Slumping may also have occurred on a much larger scale. On the south side of the WSW rift, a ridge that parallels the rift zone may represent a large portion of the rift that slumped to the south (Figure 11). There are alternate interpretations, such as a wider-than-normal rift zone here, or a relocation of the rift zone, as is known to occur along submarine Hawaiian rifts [e.g., Eakins and Robinson, 2006]. If our preferred interpretation is correct, however, then the inferred headwall is nearly 1 km long and the blocks fit roughly together in jigsaw fashion. Fitting the pieces together requires two separate slumps with the northeastern one sliding about 270 m at 185° and the southwestern one sliding about 140 m at about 170°. Both blocks remained largely intact. Such displaced, but largely intact, slump blocks are not known to us from other locations. Such large-scale dismemberment of the volcano probably reflects the unconsolidated character of much of the slopes, even those now draped by lava flows, and indicates that much of the volume of West Mata is composed of unstable clastic deposits.
West Mata, or similar submarine volcanoes made mostly of clastic volcanic debris, may require long-term extensive hydrothermal circulation to lithify the mainly glassy clastic debris and to stabilize the slopes and allow for emergence as new islands.
5.5. Extent of Eruptive Activity in May 2009
The AUV drove 75 m above the entire rift system and over the summit, but the acoustic noise recorded on the subbottom chirp system was restricted to the summit, and only for a track length of ∼180 m (Figure 13). The noise was recorded at much higher frequencies than at NW Rota-1, although this mainly reflects different sampling rates and recording systems [Chadwick et al., 2008a]. The noise recorded by the AUV chirp system was restricted near the Hades and Prometheus vents, and was almost certainly noise generated by the eruptions at these vents. Other embayments in the summit ridge that have clastic aprons below them may mark other vents that were previously active. Our rock and sediment sampling confirm that lavas and pyroclastic fragments along the upper rift zones, but outside the immediate summit region, have different compositions than the 2009 eruption. The comparison of bathymetric maps from 1996 and 2008–2010 suggest more widespread changes than the distribution of lava and clastic compositions and flow ages estimated from visual observations indicate (see Figure 4a). There are several possible explanations for this difference. The range of compositions of flows and pyroclasts within the +30 m shallower area could simply reflect changing lava compositions during perhaps as much as 14 years of activity. Visually “old” lava flows could be caused by rapid alteration of flows due to extensive hydrothermal discharge, as observed in the area. It is also possible that the 1996 survey did not accurately capture the bathymetry of the volcano and some of the margins of the calculated depth differences are erroneous. The resolution of the AUV survey presented here will allow more accurate volume changes to be calculated if the eruption continues and a repeat AUV survey is conducted.
5.6. Possible Long-Term Rate of Eruptive Activity
The comparison of ship-based multibeam surveys between 1996 and 2008–2010 provides a longer-term view of eruptive activity at the volcano. The location of the depth changes at West Mata also shows that vents near the summit have been the main eruption sites since 1996. The conservative volume of depth change at the summit and on the north flank, calculated from the area with > +30 m change, amounted to 8.0 × 107 m3 over this 14-year period, or an average eruption rate of 5.7 × 106 m3/yr if the volcano was active during this entire period. If the eruption began near 2008, then the annual eruption rate could be 2–5 times higher, depending on when activity began. For comparison, this lower long-term rate is about half the accumulation rate of pyroclastic material determined at NW Rota-1 volcano in the Mariana arc between 2003 and 2006 [Walker et al., 2008], and about a quarter that at Monowai volcano in the Kermadec arc between 1998 and 2007 [Chadwick et al. 2008b], both of which are also known to be persistently active. This is somewhat counter-intuitive since the activity observed at W Mata by ROV in May 2009 appeared to be more energetic and voluminous that that previously seen at NW Rota-1, but the eruption rate at the latter is also known to vary over time by several orders of magnitude [Chadwick et al., 2009; Dziak et al., 2009], and is a reminder that short-term visual observations are not necessarily representative of long-term rates. For comparison with other volcanic settings, the eruption rate at W Mata is also about half the magma supply rate estimated from inflation measurements at Axial Seamount, a hot spot influenced volcano on the Juan de Fuca Ridge [Chadwick et al., 2006; Nooner and Chadwick, 2009], but <8% of the long-term magma supply rate at Kilauea volcano, Hawaii [Dzurisin et al., 1984; Dvorak and Dzurisin, 1993].
Mapping of the rift zones, summit, and parts of the flanks of West Mata volcano with a sonar-equipped AUV show that the flanks of the volcano primarily consist of clastic debris and that the WSW and ENE rift zones consist of shingled lava plateaus. Additional short lava flows and sheet-like debris avalanche deposits emanating from the summit region cascade down the flanks. This shows that eruptive vents are restricted to the summit and rift zones, extend down to depths of at least 2200 m. The predominance of clastic debris downslope from the summit suggests that eruptions there tend to be mainly pyroclastic whereas rift eruptions tend to be more effusive. Small landslides have modified the margins of nearly all the lava plateaus on the rift zones and may have transported large blocks several hundred meters downslope. The volcanic activity observed in May 2009 was restricted to vents located within 150 m of the summit, and this seems to have been the case for at least the last decade. Analysis of ship-based multibeam sonar surveys between 1996 and 2008–2010 shows depth changes up to 96 m on the summit and north slope of the volcano, which we interpret as long-term accumulation of pyroclastic debris downslope of the summit eruptive vents.
R/V Thompson cruise TN234, with Joe Resing and Bob Embley as chief scientists, was funded by the National Science Foundation Marine Geology and Geophysics program (OCE-0929411, -0929881, -0930025, and -0934660), NOAA-Ocean Exploration and Research, and the Pacific Marine Environmental Laboratory. Operational and personnel costs for the Mapping AUV were supported through a grant to MBARI from the David and Lucile Packard Foundation and shipping and travel costs were supported by NSF grant OCE-0934278 for the AUV technical team. Discussions at sea and post-cruise with Ken Rubin, Bob Embley, and Joe Resing improved our understanding of West Mata. This is PMEL contribution 3723.