Volume 11, Issue 8
Free Access

Paving the seafloor: Volcanic emplacement processes during the 2005–2006 eruptions at the fast spreading East Pacific Rise, 9°50′N

A. T. Fundis

A. T. Fundis

Department of Geological Sciences, University of Florida, 241 Williamson Hall,, Gainesville, Florida, 32611 USA

Search for more papers by this author
S. A. Soule

S. A. Soule

Geology and Geophysics Department, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, 02453 USA

Search for more papers by this author
D. J. Fornari

D. J. Fornari

Geology and Geophysics Department, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, 02453 USA

Search for more papers by this author
M. R. Perfit

M. R. Perfit

Department of Geological Sciences, University of Florida, 241 Williamson Hall,, Gainesville, Florida, 32611 USA

Search for more papers by this author
First published: 31 August 2010
Citations: 44

Abstract

[1] The 2005–2006 eruptions near 9°50′N at the East Pacific Rise (EPR) marked the first observed repeat eruption at a mid-ocean ridge and provided a unique opportunity to deduce the emplacement dynamics of submarine lava flows. Since these new flows were documented in April 2006, a total of 40 deep-towed imaging surveys have been conducted with the Woods Hole Oceanographic Institution's (WHOI) TowCam system. More than 60,000 digital color images and high-resolution bathymetric profiles of the 2005–2006 flows from the TowCam surveys were analyzed for lava flow morphology and for the presence of kipukas, collapse features, faults and fissures. We use these data to quantify the spatial distributions of lava flow surface morphologies and to investigate how they relate to the physical characteristics of the ridge crest, such as seafloor slope, and inferred dynamics of flow emplacement. We conclude that lava effusion rate was the dominant factor controlling the observed morphological variations in the 2005–2006 flows. We also show that effusion rates were higher than in previously studied eruptions at this site and varied systematically along the length of the eruptive fissure. This is the first well-documented study in which variations in seafloor lava morphology can be directly related to a well documented ridge-crest eruption where effusion rate varied significantly.

1. Introduction

[2] The products of submarine volcanic eruptions at mid-ocean ridges (MORs) cover >60% of the planet's crust, yet, our understanding of volcanic processes in these settings has been limited by our inability to directly observe active eruptions. During the course of the past few decades of high-resolution seafloor mapping, only a few seafloor volcanic eruptions have been well documented [Embley et al., 1990, 1995; Chadwick et al., 1991, 1995; Haymon et al., 1993; Rubin et al., 1994; Gregg et al., 1996; Sinton et al., 2002], and in most cases there are gaps in our knowledge of both the size and continuity of individual lava flows that comprise the eruptive episodes. Volcanic deposition at MOR axes greatly impacts the permeability structure of the upper oceanic crust, and thus hydrothermal systems. It is therefore important to understand the mechanisms controlling lava emplacement and deposition at MORs. This study examines the flow surface morphology and intraflow features of the erupted products of submarine eruptions that occurred over ∼6 months in mid-2005 to early 2006 [Rubin et al., 2006, 2008; Tolstoy et al., 2006] in order to place constraints on the mechanisms of lava emplacement and deposition at a fast spreading MOR.

[3] In 2006, an eruption centered near the magmatically robust 9°50′N region of the EPR was discovered and the flow area fully mapped [Tolstoy et al., 2006; Soule et al., 2007] (Figure 1). Microseismicity data, Po-disequilibrium dating, and continuous time series temperature records collected at nearby hydrothermal vents indicate that the eruption occurred as a series of pulses between mid-2005 to January 2006 [Tolstoy et al., 2006; Cowen et al., 2007; Soule et al., 2007; Rubin et al., 2006, 2008; Fornari et al., 2010]. This discovery and subsequent investigations of the geological, hydrothermal and biological impacts of the eruptions provide the first documented evidence for a repeat eruption at a MOR site, in agreement with predictions of decadal recurrence intervals at this fast spreading ridge [Haymon et al., 1993; Perfit and Chadwick, 1998].

Details are in the caption following the image
Location and bathymetry of the East Pacific Rise 9°50′N [White et al., 2006]. Area covered by the 2005–2006 eruptions is outlined in black (derived from camera tow and side scan imagery data) and the four distinct regions of the flow are defined [Soule et al., 2007]. Hydrothermal vents are marked by red dots.

[4] The parameters that control submarine lava surface morphology differ from those that control subaerial flows. In submarine environments, lava temperature and crystal content remain relatively constant due to the rapid formation of a thick,insulating crust [Gregg et al., 1998; Fornari et al., 2004] and thus, do not change the rheology of the lava and the subsequent manner in which the lava surface deforms [e.g., Soule et al., 2004]. Experiments using wax as an analog to lava have shown that surface morphology is dependent on the timescales of cooling and advection [Fink and Griffiths, 1992; Gregg and Fink, 2000; Sakimoto and Gregg, 2001]. Due to the significant thermal gradient between lava and seawater, cooling rates of submarine lavas with differing surface morphologies should be relatively similar. Thus, prior to this study, the dominant parameter believed to control surface morphology was the timescale of advection, which is a proxy for the local volume flux influenced by eruption rate and pre-existing topography [e.g., Gregg and Fink, 2000; Sakimoto and Gregg, 2001].

[5] Here we use seafloor photography and high-resolution bathymetric profiles (1-m vertical precision) from deep-towed camera surveys and Alvin dive imagery to produce a geologic map of the volcanic features associated with this most recent eruption at the EPR near 9°50′N. Our analysis of the distribution of lava morphologies emplaced during 2005–2006 shows that lava effusion rate was the dominant factor controlling the observed morphological variations in the 2005–2006 flows. We also show that the effusion rates were higher than in previous eruptions (i.e., 1991–1992 [Haymon et al., 1993; Rubin et al., 1994; Gregg et al., 1996]) at this site and were variable along the length of the eruptive fissure.

2. Background

[6] The northern EPR between the Clipperton (∼10°N) and Siqueiros (∼8°30′N) transform faults is a fast spreading (5.5 cm/yr [Carbotte and Macdonald, 1992]) MOR and one of the best studied segments of the global ridge system [e.g., Lonsdale, 1977; Macdonald et al., 1984, 1992; Langmuir et al., 1986; Perfit et al., 1994; Fornari et al., 1998, 2004; Perfit and Chadwick, 1998; Carbotte et al., 2009; White et al., 2002, 2006; Soule et al., 2007; Escartín et al., 2007; Goss et al., 2010]. Over the past two decades, two documented eruptions have occurred along this segment of the EPR near the 9°50′N region, one in 1991–1992 [Haymon et al., 1993; Rubin et al., 1994] and the most recent one in 2005–2006 [Tolstoy et al., 2006; Rubin et al., 2006; Cowen et al., 2007; Soule et al., 2007]. Using data from deep-sea camera surveys and Alvin dive traverses, Soule et al. [2007] estimated that the 2005–2006 eruption was 4–5 times more voluminous than the rough estimate for the 1991–1992 eruption [Gregg et al., 1996]. The 2005–2006 eruptions were fed by eruptive fissures located within the axial summit trough (AST), which is the site of most of the primary eruptive fissures [Fornari et al., 1998, 2004] and known high- and low-temperature hydrothermal vents [e.g., Haymon et al., 1991, 1993; Von Damm et al., 1995; Shank et al., 1998]. A fissure system located ∼700 m east of the EPR axis in the 9°53′–9°55′N region was also active during the 2005–2006 flows (Figure 1). As suggested by Soule et al. [2007], the mapped flow boundary indicates that multiple flow lobes were produced along the ∼18 km eruptive fissure system between 9°46′N and 9°56′N, possibly reflecting discrete eruptive events [Rubin et al., 2006, 2008] during the 2005–2006 eruption. These lobes coincide with pre-existing, ∼10–50 m wide east-west trending lava channels that have been interpreted by Soule et al. [2005] to have served as distribution pathways for axially erupted lavas. The largest of these lobes, between 9°50′N and 9°52′N, coincides with the highest density of high-temperature hydrothermal vents along the ridge [Fornari et al., 2004; Soule et al., 2007], and for the purpose of this study it is referred to as the 9°51′N lobe. Lava deposition associated with the 2005–2006 flows occurred in four distinct regions: 1) the area referred to in this study as the 9°51′N lobe, considered to be the main flow unit, 2) flows produced by smaller axial eruptions north of 9°53′N through a discontinuous series of fissures and an axially sourced flow that was directed north by a ridge-parallel fault, 3) off axis flows erupted from fissures ∼700 m east of the axis near 9°54.5′N, and 4) the area between 9°49′N and 9°46′N where small volumes of axially erupted flows, found largely in the AST have been identified [Goss et al., 2010] (Figure 1).

[7] In the past, our understanding of the size of eruptions and the relationship between flow morphologies and eruptive processes along the MOR system have been fundamentally limited by our inability to unequivocally distinguish individual eruptive units. The key difference between this study and studies of seafloor characterization of other MOR sites where individual eruptions have been documented [e.g., Embley et al., 1990; Chadwick et al., 1991; Sinton et al., 2002] is that we were able to conduct surveys shortly after the eruptions occurred to confidently distinguish (by abrupt changes in sediment cover and glassiness of lava surfaces) and comprehensively survey the area affected by the eruptions. We surveyed the 2005–2006 flows with digital seafloor imaging on multiple cruises so that we could positively establish the boundaries of the new flows and collect detailed bathymetric data and lava samples.

[8] We use results from previous studies, including those presented by Kurras et al. [2000] and White et al. [2006] to help define the pre-existing lava morphology and topography in the study area, and to compare and contrast them to seafloor volcanic features created during the 2005–2006 eruption. White et al. [2002] presented results from a regional Argo I survey where they estimated the morphology between 9°08′N and 9°56′N was comprised of approximately 25% pillow lava, 50% lobate lava and 25% sheet lava. Kurras et al. [2000] used imagery from Alvin and deep-towed camera surveys to conduct a similar analysis over a smaller area, between 9°49′N and 9°52′N, that falls within the extent of the 2005–2006 eruptions. They reported that the crest of the EPR in this region was predominately covered by lobate (66%) and sheet flows (20%), consistent with the idea that fast spreading ridges typically produce high-effusion rate eruptions [Perfit and Chadwick, 1998]. Additionally, they reported that the 9°50′N area had experienced three types of volcanic emplacement processes including: (1) axial summit eruptions, (2) off-axis transport of axially erupted lava through channelized surface flows, and (3) off-axis eruptions. Engels et al. [2003] investigated the distribution and formation of lava collapse features observed on the EPR axis in the 9°37′N area and reported that collapse almost always occurred within lobate flows. Moreover, collapse pits <2 m in diameter were evenly distributed out to ∼300 m from the AST, while collapse features >2 m in diameter were concentrated within 100 m of the axis.

3. Methods

3.1. Deep-Towed Camera Surveys

[9] A total of 40 deep-towed imaging surveys were conducted in the study area during 2006–2007 using the Woods Hole Oceanographic Institution's TowCam system [Fornari, 2003]. Approximately 60,000 digital color images were collected. Bathymetric profiles (1-m vertical precision) were compiled using a 100 kHz altimeter and depth acquired using the TowCam's SeaBird SBE25 CTD. The TowCam was equipped with a color digital camera (3.5 megapixel) for 30 of the tows and a fiber optic camera (4 megapixel) was used during two cruises (AT15–6 and AT15–27) for 10 tows. TowCam's speed over the bottom was typically ≤0.5 knots. Towing altitude was normally 5–7 m above the seafloor, producing images that each cover an area of ∼4.5 m × 6 m of seafloor when the color digital camera was in use and ∼3.5 m × 3.5 m when the fiber optic system was utilized. Images were acquired at 10–15 s intervals, producing 0–20% of overlap between each successive image. During 18 surveys, the TowCam was navigated using a bottom-moored long baseline (LBL) transponder network [Hunt et al., 1974; Milne, 1983; Soule et al., 2007] that located the images to within 5–8 m [Haymon et al., 1991; Fornari et al., 1998]. During 22 of the surveys that were conducted outside the LBL network area, the TowCam was navigated using layback calculations that located the images to within ∼100 m [Soule et al., 2007] (Figure 2). TowCam image data were supplemented by down-looking video imagery collected during post-eruption Alvin dives. Dives 4204 and 4296 traversed areas south of the off-axis flow near 9°53′N and dive 4205 traversed the southern extent of the eruption near 9°46′N.

Details are in the caption following the image
Tracklines showing area of the 2005–2006 eruptions surveyed with TowCam. Blue lines represent navigational precision of ∼5 m, red lines represent ∼100-m precision and the green lines represent >100-m precision (see text for discussion).

3.2. Image Analysis

[10] Each TowCam image was classified for relative age and lava flow morphology. Boundaries of the 2005–2006 flows were easily identified by the lack of sediment on the flow surfaces and the presence of fresh glassy lava surfaces; these criteria provided the basis for relative age estimates (2005–2006 versus older flows). Lava flow surfaces in each of the images were classified into one of four morphological categories: pillow flow, lobate flow, sheet flow, and hackly flow, consistent with the descriptions used by Ballard and van Andel [1977], Ballard et al. [1982] and Kurras et al. [2000] (Figures 3a3d). In instances where multiple morphologies were present in an image, the dominant morphology by area was recorded. Three morphologic types of sheet flow (lineated, ropy and hackly) were observed and all are indicative of high flow rates at the time of emplacement [Gregg and Fink, 1995; Gregg et al., 1996; Chadwick et al., 1999]. Hackly flows were distinguished from other sheet flows because initial observations of the 2005–2006 flow morphologies from the TowCam observations and Alvin dives indicated that this eruption produced a higher percentage of hackly flows than previously existed in the region, especially within the 9°51′N lobe. The discrimination of hackly flows from lineated and ropy sheet flows is used here to help refine the emplacement mechanisms involved in erupting the 2005–2006 flows.

Details are in the caption following the image
Images showing examples of classification scheme used for images of the 2005–2006 flows. The four morphology types: (a) pillow, (b) lobate, (c) sheet, and (d) hackly. The three collapse types: (e) lobate blisters, (f) skylight collapse, and (g) lava pond collapse. (h) An example of a kipuka, an area of exposed flow older than the 2005–2006 flows and completely surrounded by the 2005–2006 flows. Horizontal scale across the bottom of each photograph is ∼3–5 m.

[11] Images were also classified for the presence and abundance of collapse pits, kipukas (defined in this study as an area of exposed flow older than the 2005–2006 flow and completely surrounded by the 2005–2006 flow), faults, and fissures (Figures 3e3h). Collapse pits were classified into three size categories using the terminology and size classes described by Engels et al. [2003]: lobate blisters (<2 m diameter), skylights (∼2–10 m diameter), and lava pond collapse (>10 m diameter). Kipukas were only recorded if they were identifiable in <3 image frames (ranging from <1 m to ∼6 m in diameter). Potentially larger kipukas that are not easily distinguished from true gaps in lava deposition were not included in our study. Faults and fissures were rare within the imaging surveys.

3.3. Analytical Methods

[12] Of ∼60,000 images collected, about half (27,205 images) imaged the 2005–2006 flows. Due to areas of extensive collapse in the new flows, it was not possible to classify flow morphologies in ∼10% of these images (2,104 images). Assuming each TowCam image covered ∼23 m2 of seafloor on average, the camera surveys covered ∼0.58 km2 (25,101 images). Images where morphologic and tectonic structures were classified represent less than 4% of the flow surface area. The surveys, however, evenly cover the extent of the 2005–2006 flows along and across axis (Figure 2), and we thus consider our analyses to be representative of the entire 2005–2006 eruption.

[13] Observations from the towed camera surveys were entered into a database and subsequently linked to time-stamped navigation and near-bottom profiles generated from the CTD depth and altitude data. These georeferenced observations were then compiled and used to plot distributions of the various lava flow morphologies and volcanic structures observed in the photographs (Figures 4 and 5).

Details are in the caption following the image
Maps showing the distribution of the four main types of morphology of the 2005–2006 eruptions surveyed with the TowCam. Each point represents an analyzed image and the presence of that specific morphology type.
Details are in the caption following the image
Maps showing the distribution of the three size classes of collapse and kipukas that formed in the 2005–2006 EPR flows. Lobate blister collapse is collapse <2 m in diameter, skylight collapse is ∼2–10 m in diameter and lava pond collapse is >10 m in diameter. Kipukas range from ∼1–6 m in diameter.

4. Results

4.1. Abundance of Lava Morphologies

[14] Lobate flows are the dominant lava morphology type observed in the 2005–2006 flows, accounting for 52% (13,106 images) of the survey area. Sheet flows comprise 31% (7807 images) of the observed morphologies, hackly flows account for 15% (3636 images) and pillow flows account for only 2% (552 images) of the survey area. Relative to the other three morphologic types, lobate flows are ubiquitous over the surveyed area but are the dominant morphology in the northern and southern regions of the flow away from of the 9°51′N lobe (Figure 6). Sheet flows are the dominant flow type flooring the AST and within flow channels produced by the 2005–2006 eruption outside of the AST. Over 60% of pillow flows in the surveyed area are located within 100 m of the mapped flow boundary; 87% of pillows in the imagery are located within 200 m of the flow boundary (Figure 7). Relative to the rest of the flow, pillows are more abundant at the northern and southernmost distal ends of the 2005–2006 flows (Figure 6).

Details are in the caption following the image
Observations of flow morphology and presence of collapse binned by 500 m of latitude. The percentage of the 2005–2006 flows surface area covered by the surveys was calculated by assuming each photograph covered an area of ∼23 m2.
Details are in the caption following the image
(a) Map showing the location of 100 m bins relative to the 2005–2006 flows boundary [Soule et al., 2007]. (b) >60% of pillow flows are within 100 m of the mapped flow boundary and >80% occur within 200 m, indicating that pillows formed from slow effusion rate and decreased lava supply at the distal flow margins. (c) Kipukas are also correlated with proximity to the flow boundary.

4.2. Morphological Transitions

[15] By examining the transitions between the different surface morphologies with respect to the AST, we can further constrain the genetic relationship between the various surface flow morphologies. There are two main types of morphological transitions seen within the 2005–2006 flows: 1) gradational due to changing flow conditions, and 2) superposition of separate flows. The gradational transitions are observed between sheet and lobate flows, sheet and hackly flows, lobate and pillow flows, and lobate and hackly flows (Figures 8a8c). Transitions between sheet and lobate lavas vary with respect to the length scales over which they occur. Typically, they occur over a few meters as flat sheet flows develop a more lobed surface. In other cases, where lobate flow appears to have transitioned into sheet flow, the transitions are more abrupt. The sheet flow in these cases appears to have flowed out from underneath the cooled lobate surface and is characterized by lineations that likely formed from being raked against the partially cooled upper surface of the lobate lava [Chadwick et al., 1999].

Details are in the caption following the image
Examples of transitions between different morphologies of the 2005–2006 EPR flows. (a) A sheet/lobate transitional morphology abruptly transitioning into a lineated sheet flow (∼6 m × 4.5 m). (b) A sheet flow abruptly transitioning into hackly flow (∼4.8 × 3.6 m). (c) Pillow flow developing toward a flow front (∼4.8 m × 3.6 m). (d) A potential example of superposition where a lobate flow appears to have flowed over a lineated sheet flow (∼7 m × 5.3 m).

[16] Transitions between sheet and hackly flows also vary with respect to the length scales over which they occur. In many cases, sheet flows transitioned to a jumbled flow where the sheet surface is characterized by small surface folds as the partially cooled and more brittle surface deforms due to continued flow of the underlying molten lava. These jumbled flows commonly transitioned into hackly flows as the folded surface of the flow continues to deform as it cools, ultimately breaking the glassy surface crust. The gradation from sheet to hackly flows in these cases typically occur over distances of 1–10 m. Sudden transitions between sheet and hackly flows, with no intermediate jumbled surface morphology, were also observed, and may reflect superposition of lavas erupted at different times.

[17] Transitions between lobate and pillow lavas occur over lengths <1 m, tend to occur toward the flow front termini, and are characterized by one to a few pillows branching out of well-defined lobate flows. Lobate flows near these transitions are more decorated and have distinct lobes that are minimally coalesced. The final type of gradational transition we observed is between lobate and hackly flows. These are similar to the abrupt transitions between sheet and hackly flows, where the flow surface abruptly changes character to a morphology representative of a higher rate of shear. These transitions tend to occur at the margins of lava channels, similar to results described by Soule et al. [2005].

[18] There are only 5 images from the TowCam surveys where potential superposition contacts were observed within the 2005–2006 flows. In these images, the morphologies do not appear to be transitioning from one type to another; rather it appears that a flow of one morphological type partially covered a flow of another flow morphology. For example, Figure 8d shows a lobate flow that appears to have flowed over a lineated sheet flow, both of which have been identified as 2005–2006 flows using the criteria we established. The spatially limited extent of such contacts suggest that it is either difficult to identify intraflow contacts or that such contacts do not commonly exist within the 2005–2006 eruptions.

[19] We used TowCam survey segments that were run perpendicular to the AST and that surveyed known lava channels to interpret the morphological evolution of a given lava flow as it moved away from an eruptive source, and to test whether the variations in distribution we see between the various morphologic transitions occurred along individual flow pathways with increasing distance from the eruptive fissure. Although individual flows are occasionally characterized by lobate flows at the flow front rather than pillow lavas, the general trend is that lava morphologies associated with higher flow rates (i.e., sheet, hackly and flat-lying lobate flows) occurred near the eruptive source and transition into morphologies that are typically associated with lower flow rates toward the termini of the flow. TowCam survey AT15–6:CT06 (see Figure S2), conducted perpendicular to the axis at ∼9°51.5′N, provides an example of this transition. The images from this camera tow show that the lava emplaced closest to the axis is a mixture of sheet and hackly flows which have no clear contacts, but rather which transition from one to the other over sub-meter length scales. Further from the axis, the sheet and hackly flows transition into dominantly lobate flows, and eventually into pillow flows at the termini of the flow front. Similar relations and transitions are seen throughout the surveys (see auxiliary material for examples).

4.3. Distribution of Volcanic Structures

4.3.1. Volcanic Collapse Pits

[20] Collapse pits are common features of submarine lava flows at fast and superfast spreading MORs. They form from flow inflation and post-eruption drainback of lava into eruptive fissures or down flow through lava conduits as supply from the vent diminishes [Fornari et al., 1998], or from a pressure gradient created along the upper crust of the flow due to the cooling of entrapped seawater-derived vapor [Engels et al., 2003; Perfit et al., 2003; Soule et al., 2006]. As described in previous studies, these features are depressions where 2–10 cm thick surface crusts have collapsed, revealing cavities beneath [e.g., Ballard et al., 1979; Sinton et al., 2002; Engels et al., 2003; Perfit et al., 2003; Fornari et al., 2004]. In this study, collapse pits cover 19.2% (4817 images) of the total surveyed area.

[21] The smallest collapse pits, lobate blisters, account for 17.8% of the overall collapse. Larger collapse pits, skylights, represent 27% of the overall collapse. Engels et al. [2003] found that lobate blisters were relatively evenly distributed within ∼300 m from the AST, with only a slightly higher frequency near the ridge axis. They also reported that the abundances of skylight and lava pond collapse peak dramatically within ∼100 m from the AST margin. The results we present here are similar since the distribution of both lobate blisters and skylight collapse generally decrease with increasing distance from the AST. However, in this study, the distribution of lobate blisters in addition to skylight collapse peaks within ∼100 m of the AST (Figure 9). While there is significant difference in the distribution of collapse across the AST, there is only a slight difference in the distribution of collapse pits along the AST. In the southern extent of the main flow unit between 9°49′ and 9°48′N, there is an increase in the percent collapse observed relative to the total number of observations (Figure 6).

Details are in the caption following the image
Map showing the location of 100 m bins relative to the AST. The distributions of collapse <2 m in diameter (lobate blisters) and collapse ∼2–10 m in diameter (skylight collapse) are similar relative to distance from the AST. Collapse features are primarily concentrated within 100 m of the AST.

[22] Lava pond collapse imaged in this study covers 9.8% (2659 images) of the survey area. The primary mechanism responsible for creating this aerially extensive collapse is post-eruption drainback of lava into primary eruptive fissures [Fornari et al., 1998] at the site of eruption. Therefore, as expected, lava pond collapse imaged in this study is almost completely concentrated within the AST where erupted lava is constrained and accumulates within the walls of the trough (Figure 5).

[23] The great majority of collapse features (∼90%) occur in lobate flows and the rest in sheet flows, which is consistent with results reported by Engels et al. [2003]. It is apparent that these networks of interconnected collapse areas play a role in the transport of lava during eruptions [e.g., Haymon et al., 1993; Schouten et al., 2002; Fornari et al., 1998, 2004]. We see evidence that subsurface pathways were exploited by 2005–2006 flows, with 2005–2006 lava imaged at the base of some collapse pits located in older flows (Figure 10a). We also observe isolated fragments of older lava crust lying on the surface of the 2005–2006 flows, that was presumably pushed up by lava flowing within these subsurface networks (Figure 10b).

Details are in the caption following the image
Evidence that subsurface pathways were utilized by the 2005–2006 EPR flows. (a) 2005–2006 pillow lava imaged at the base of collapse in older lobate terrain (∼6 m × 4.5 m). (b) Older lava crust broken and lying on the surface of a 2005–2006 flow that was presumably pushed up by lava flowing within subsurface networks (∼4.8 m × 3.6 m).

[24] It has been shown that vapor formed by lava-seawater interactions [Perfit et al., 2003; Chadwick, 2003; Engels et al., 2003; Soule et al., 2006] is important in the development of collapse features. We find features such as lava drips on the undersides of lava crusts throughout the 2005–2006 flows, supporting the involvement of seawater vapor within the active flow. Vapor pressure may also play a role in expelling older lava crusts onto the surface of the new flows.

4.3.2. Kipukas

[25] Kipukas <6 m in diameter were seen in 4.3% (1170 images) of the survey images. These areas of older lavas exposed through the new flows indicate that the flow is ∼1–2 m thick [Soule et al., 2007]; kipukas are associated with lobate forms in 88% of the observations, pillow forms in 9%, sheet flows in 3%, and hackly flows in <1% of the observations. These small kipukas are typically individual lobes of sheet flow or lobate flows that divert and wrap around areas of older flow. Kipukas commonly occur where there is no clear topographic barrier, or where there is an increase in seafloor roughness often when pillows or inflated lobate lavas characterize the pre-eruption surface. Although, kipukas are found throughout the survey area, they are mostly concentrated in the northern and southern regions of the flow area, following similar trends to the distribution of lobate and pillow flows. Additionally, an increase in the number of kipukas is correlated with proximity to flow boundaries (Figure 7).

4.3.3. Faults and Fissures

[26] Only one fault north of 9°52′N, against which the 2005–2006 flows dammed [Soule et al., 2007], and 14 fissures were seen in the survey images. The imaged fissures are associated with the AST or with the off-axis fissures north of 9°52′N. Escartín et al. [2007] noted that volcanism at MORs, and the fast spreading EPR specifically, efficiently buries faults within 2 km of the axis. Thus, the lack of faulting we see within the 2005–2006 flows, which is predominately within 2 km of the axis, is consistent with what we would expect for recently repaved seafloor, and consistent with studies of the same area prior to the eruption [Fornari et al., 2004; Escartín et al., 2007].

5. Discussion

[27] In this study, we use lava surface morphology as an indicator of relative effusion and flow rates for the 2005–2006 flows. Previous studies have suggested a variety of intrinsic and extrinsic parameters, in addition to effusion rate, that may influence the development of distinct flow morphologies including: lava rheology (i.e., composition, temperature, and crystallinity) and the pre-existing slope and terrain [e.g., Bonatti and Harrison, 1988; Rowland and Walker, 1990; Griffiths and Fink, 1992; Gregg and Fink, 1995; Gregg et al., 1996]. We show below that the pre-existing slope and terrain did not significantly affect the morphology of the 2005–2006 flows. Additionally, petrographic and geochemical analyses of volcanic glass samples suggest that the lava rheology remained relatively constant throughout the course of the eruption. Samples from the 2005–2006 flows are relatively homogeneous compositionally and are mostly aphyric (≪5% plagioclase and olivine) regardless of morphologic type [Soule et al., 2007; Goss et al., 2010]. Thus local flow rates, primarily controlled by lava effusion rates, remain the key controls on surface morphologies developed during the 2005–2006 eruptions. Using the empirical relationship between flow rate and morphology, we conclude that the 2005–2006 event erupted at higher volumetric rates than those estimated for the 1991–1992 eruption [Gregg et al., 1996]. Additionally, flow rates were highest and lava output was most voluminous within the 9°51′N flow lobe, further suggesting that this was the locus of the diking event that fed the main phase of the eruption.

5.1. Influence of Pre-existing Slope and Terrain on Surface Morphology

[28] The EPR at ∼9°50′N has been studied for ∼20 years by near-bottom imaging [e.g., Haymon et al., 1991; Kurras et al., 2000], providing a substantial volume of data that can be used to assess changes caused by seafloor eruptions, and to infer the processes responsible for creating the lava flow surface morphologies. To assess whether pre-eruptive surfaces influenced the extent and morphology of the 2005–2006 eruptions, information collected from the recent TowCam surveys was augmented by four studies of the area prior to the 2005–2006 eruptions. These included: (1) camera-tow surveys of the seafloor in 1997 and 1999 [Kurras et al., 2000], (2) bathymetric data collected by WHOI's Autonomous Benthic Explorer (ABE) in 2000 [Fornari et al., 2004], (3) bathymetric data collected by the 30 kHz EM300 multibeam system in November of 2005 [White et al., 2006], and (4) side scan sonar data collected in 2001 [Fornari et al., 2001, 2004; Schouten et al., 2002; White et al., 2002; Escartín et al., 2007].

[29] Kurras et al. [2000] analyzed camera-tow surveys collected along and across the axis of the EPR prior to the 2005–2006 eruptions. Four of these deep-towed camera survey lines traversed areas that were subsequently covered by the 2005–2006 flows, and were conducted in areas in close proximity to post-eruption survey lines. The pre-eruption surveys were reanalyzed using the classification scheme we adopted, and used in conjunction with pre-eruption side scan sonar data [Fornari et al., 2004] to investigate whether the morphology of the pre-existing terrain correlated with or possibly influenced the morphology produced during the recent eruption. EM300 multibeam bathymetry data [White et al., 2006] and ABE Imagenex 120 kHz scanning altimeter bathymetry data collected in 2000 by ABE [Fornari et al., 2004] were used to determine the pre-existing slope of the terrain prior to the 2005–2006 flows [e.g., Gregg et al., 1996]. The EM300 survey covered the entirety of the area subsequently covered by 2005–2006 flows; however, it is gridded at a lower resolution than the ABE survey, which covered a smaller area at 9°50′N. The local pre-existing slope surrounding each image location was derived from the 30 m × 50 m gridded EM300 depth measurements and 5 m × 5 m gridded ABE depth measurements by calculating the average gradient between the grid cell and its four nearest neighbors at each image position.

5.1.1. Pre-Existing Slope Analysis

[30] We derived slopes of ∼0°–11° from the EM300 bathymetry and ∼0°–19° from the ABE bathymetry within the area of the new flows. Analysis of lava morphology as a function of topographic slope derived from both gridded depth determinations indicates that the distribution of slopes for each of the morphologies, with a possible exception in the distribution of pillows, correlates with the overall distribution of slopes in the survey area (Figure 11). These results contrast with observations from the Puna Ridge [Gregg and Smith, 2003], where a strong partitioning of morphology as a function of slope is observed (although the range of slopes is much larger). Relative to the overall slope distribution, pillows in this study trend toward slightly higher pre-existing slopes calculated from both the EM300 and ABE bathymetry data (Figure 12). However, we note that few pillows were identified in these surveys relative to other flow morphologies (2%) and we are therefore cautious in interpreting this as evidence that pillows form on higher slopes. Additionally, some pillow flows were observed just outside of the AST and on the floor of the AST where slope is essentially flat. The slope in the data appears to be higher in these areas due to the gridding method of the bathymetric data, which could account for the small peak of high-slope pillows. As a further check on whether slope strongly correlates with seafloor morphological variations, we analyzed lava morphology as a function of topographic slope derived from near-bottom TowCam depth profiles (auxiliary material). Unlike gridded bathymetry data, which may underestimate the actual slope of steeply dipping faults or flow fronts due to their small aerial extent relative to grid cells, TowCam bathymetry profiles provide a more faithful representation of seafloor slope. In addition, these profiles take into account topography generated by the 2005–2006 eruption, although in most cases we do not see significant differences between pre- and post-eruption bathymetry [Ferrini et al., 2009], which further confirms the relatively thin (<1–2 m) flow thickness. Regardless, these results support our conclusion of very limited to no correlation between topography encountered by the 2005–2006 flows and the resulting lava morphology.

Details are in the caption following the image
Lava morphology plotted as a function of pre-existing slope as calculated from the bathymetric surveys conducted with the EM300 multibeam system and ABE. The local pre-existing slope surrounding each image location was derived from the 30 m × 50 m gridded EM300 depth measurements and 5 m × 5 m gridded ABE depth measurements by calculating the average gradient of the four nearest grid cells at each image position to account for uncertainties in navigation. All four morphologies can be found over the entire range of pre-existing slopes, indicating pre-existing slope was not a dominant controlling factor of the 2005–2006 EPR eruption surface morphology.
Details are in the caption following the image
The distributions of the four main morphologies of the 2005–2006 eruptions correlate with the global distribution of slopes in the survey area, with a slight exception in the distribution of pillow lavas at higher slopes. (a) The relative frequency of each morphology type as a function of the maximum pre-existing slope calculated from the EM300 data [White et al., 2006]. (b) The delta frequency, relative to the overall slope distribution, plotted as a function of the maximum pre-existing slope shows there is a shift in slope ranges where pillows exceed the average distribution of slopes toward slightly higher slope values.

5.1.2. Pre-Existing Morphology Analysis

[31] The four TowCam surveys used in this study conducted prior to the 2005–2006 flows were analyzed in areas where they overlapped the mapped 2005–2006 flows boundary (Figure 13). These morphology observations were compared to morphologies of the 2005–2006 flows located within 100 m of the pre-eruption surveys. The 100 m buffer accounts for navigational uncertainties in the layback calculations. In the 1997 surveys, lobate lavas were ubiquitous over the area surveyed and accounted for 74% of the observed morphologies. Hackly lavas comprised 11% of the morphologies observed, sheet lava was less abundant comprising only 8% of the morphologies in the surveyed area, and 7% of the lava was pillows. Areas that were predominately lobate in the 2005–2006 main flow lobe prior to the recent eruptions were overprinted by sheet and hackly flows during the 2005–2006 eruptions. The post-eruption surveys indicate that, within the 100 m buffer zone of the pre-eruption surveys, the abundance of lobate, sheet, and hackly morphologies changed significantly after the 2005–2006 eruptions. This is the first well-documented study in which variations in seafloor lava morphology can be directly related to known eruption episodes where effusion rate varied significantly. In recent surveys within the 100 m buffer zone, lobate lava was much less abundant than in the pre-eruption surveys, accounting for only 34% of the morphologies observed. Sheet flows are the dominant morphology type observed, accounting for 65% of the morphologies within the study area, 34% of which are hackly and 31% lineated or ropey sheet. Only 1% of the morphologies observed are pillow. The results of this comparison indicate that the pre-eruption morphology was not a dominant controlling factor or predictor of the morphology of the 2005–2006 eruption.

Details are in the caption following the image
(a) Pre-eruption camera surveys on the EPR crest [Kurras et al., 2000] were analyzed in areas where they overlapped with the mapped 2005–2006 flows. These surveys indicate that the pre-existing terrain was largely characterized by lobate flows. (b) Morphology of the 2005–2006 EPR flows within the same region as the pre-existing camera surveys. The black lines represent a 100 m buffer surrounding each pre-eruption survey line to account for navigational uncertainties. The percentages of sheet and hackly flows are significantly higher in the 2005–2006 EPR eruptions.

[32] The lack of correlation between pre-eruption seafloor slope and the morphology produced suggests that local flow rate is likely controlled by eruption rate. The comparison of the pre- and post-eruption morphologies shows a substantial increase in the presence of sheet and hackly flows as well as the presence of many new lava channels over a large portion of the 2005–2006 lava flows. Comparison between the pre-eruption near-bottom side scan in the 9°28′–55′N EPR crestal area [Fornari et al., 2004; Escartín et al., 2007] and post-eruption side scan over the 2005–2006 flows [Soule et al., 2006] confirms that prior to the 2005–2006 eruption, this area was characterized by lobate flows with few lava channels, which changed dramatically with the emplacement of the 2005–2006 flows that were dominated by lava channels. This further supports the conclusion that the 2005–2006 eruptions produced lava at a higher effusion rate than what is typical for this portion of the EPR [Soule et al., 2007].

5.2. Distribution of Morphology Along Axis

[33] The volume of lava erupted during 2005–2006 at the EPR axis varied substantially along the ∼18 km long eruptive fissure system. In some locations, lava flows reached <0.5 km from the AST, while in other locations flow lobes extend as far as ∼3 km from the AST. The most voluminous output of the 2005–2006 eruptions occurred within the 9°51′N lobe, coincident with the highest density of active high-temperature hydrothermal venting [Fornari et al., 2004] and the highest MgO lava compositions within the eruption area [Goss et al., 2010]. Relative to the rest of the newly emplaced flows, the 9°51′N lobe contains the highest percentage of high-flow rate morphologies (∼70% sheet and hackly flows) and is the locus of where lava was transported farthest from the AST. The 9°51′N area is also the inferred site of the primary diking event that fed the 2005–2006 eruptions based on seismic evidence [Tolstoy et al., 2006]. However, this region is not the only place that experienced higher eruption rates than those produced by the 1991–1992 eruptions. While the 9°51′N lobe represents the most voluminous and greatest abundance of high-effusion flows, we see an increase in sheet and hackly flows within almost all of the 2005–2006 flow lobes relative to pre-eruption abundances. We conclude from these observations that the entire length of the axial eruption experienced higher eruption rates than previous studied eruptions, regardless of variations in the volume of material erupted along individual fissures.

[34] The relative distributions of the four morphology types along axis further indicates that the most recent eruption was focused along fissures within the AST at 9°51′N. Although all morphologies are observed along the length of the 2005–2006 main flow unit, the relative abundance of sheet and hackly flows is greatest in the central portion of the flow and especially around the 9°51′N lobe, whereas the relative abundances of lobate and pillow flows increase with along-axis distance from this region (Figure 6). The relative abundance of high-flow rate morphologies appears to correlate with the volume of material erupted, with the greatest flow rates and flow lengths at the center of the eruptive fissures (near 9°51′N) and the lowest flow rates and flow lengths at the northern and southern ends of the eruptive fissures (near 9°55′N and 9°47′N, respectively, Figure 4).

5.3. Distribution of Morphology Across Axis

[35] The formation of the four morphologies classified in this study is dependent on flow rate. The general across-axis trend that we observe in the 2005–2006 flows is that axially erupted lavas formed sheet and flat-lying lobate morphologies near the vent and less coalesced lobate and pillow morphologies with greater distance from the axis. The majority of pillow flows imaged in the surveys (>80%) were found within 100 m of the flow terminus (regardless of distance), indicating that distal flow fronts were advancing at relatively slower rates. As the flow surface morphologies likely reflect the flow conditions at the time of solidification, this spatial characteristic in morphologies suggests that flow rates were higher during the initial stages of the eruptions and waned with increasing time (i.e., distance traveled across the seafloor). The greater abundance of pillows at the margins of flow lobes should help us identify the separate pulses that comprised the 2005–2006 eruptions [e.g., Rubin et al., 2006, 2008], but there is no substantial evidence of internal pillow fronts in the survey images. Flow fronts produced during the 2005–2006 eruptions, however, often terminate as lobate flows rather than pillow fronts, making it more difficult to distinguish these separate pulses from one another within the interior of the flow.

5.4. Distribution of Other Volcanic Features

[36] In addition to flow morphology, we examined the distribution of volcanic collapse, kipukas, faults and fissures in the new flows. The results of our examination of volcanic collapse is very similar to previous studies on the EPR [Engels et al., 2003], with the key feature of the distribution being a concentration of collapse pits near the eruptive vents and a rapid decrease with distance down flow. It is useful to note that our accounting of collapse is conducted within a single eruption, whereas previous studies have looked at a portion of the EPR ridge crest that likely included the products of many eruptions, most of which resulted from lower effusion rates. The similarity in collapse distribution suggests that these features are not particularly sensitive to eruption rate, and depend more on the ability of the flow to inflate and deflate. Thus proximity to eruptive vents where drainback is common is the main controlling factor, as has been suggested by previous workers [Engels et al., 2003; Chadwick, 2003; Fornari et al., 2004]. Likewise, the low abundance of faults and fissures we observe within the new flow agrees with previous interpretations that volcanism overprints small tectonic features formed within ∼1–2 km of the axis, other than those within the AST [e.g., Escartín et al., 2007; Wright et al., 1995].

[37] Finally, we examined the presence of kipukas, which have not before been systematically mapped on mid-ocean ridges. The presence of kipukas throughout the 2005–2006 lava flows supports the idea that the flows were thin (1–2 m [Soule et al., 2007]). In addition, the greater abundance of kipukas at the distal end of the flows, where low-flow rate morphologies are more abundant, suggests that kipukas may be more likely to be maintained during periods of slower flow advance.

6. Conclusions

[38] This study provides the most comprehensive analysis of imagery data over a well-defined and mapped MOR eruption. From our analyses and interpretations of deep-towed camera surveys, and high-resolution bathymetric profiles, we draw the following conclusions:

[39] 1) Surface morphology of the 2005–2006 flows, and by inference flow rate, was not strongly influenced by the pre-existing slope or lava rheology, nor was pre-eruption morphology a strong predictor of the morphology that developed during this eruption. We conclude that seafloor lava morphology is primarily controlled by eruption rates at the vent and their variability along axis and with time.

[40] 2) The dynamics of the 2005–2006 eruptions were substantially different than previous eruptions along this portion of the EPR. The increased abundance of lava channels and high-flow rate morphologies suggest that the most recent eruption was characterized by higher effusion rates than previous eruptions in addition to being more voluminous than the 1991–1992 EPR eruptions. This is the first well-documented study in which variations in seafloor lava morphology can be directly related to known eruption episodes where effusion rate varied significantly.

[41] 3) The 2005–2006 eruption is characterized by 52% lobate flows, 31% sheet flows, 14% hackly flows and 2% pillow flows. Over 80% of the pillow morphology observations occurred within 200 m of the flow boundary, indicating decreased flow rates at the flow margins, presumably due to late-stage advance after supply from the vent had ceased. In addition, we find higher concentrations of low-flow rate morphologies at the northern and southern extent of the fissure system where we also find significantly lower volumes of erupted material.

[42] 4) The greatest abundance of high-flow rate morphologies is found within the area covered by the 9°51′N flow. This portion of the EPR crest also produced the greatest volume of lava and created numerous lava channels, which served as distribution pathways that allowed the lava to travel as far as ∼3 km off axis. The AST in this area also hosts the majority of high-temperature hydrothermal vents along this second order segment and is the location of the most primitive lava compositions throughout the flow. These observations indicate that the 9°51′N area in the AST served as the source of the 2005–2006 eruptions.

Acknowledgments

[43] We thank the officers and crew of the R/V Knorr, R/V New Horizon, R/V Atlantis, and R/V Alvin shipboard operations groups, who all contributed to the success of our field studies of the 2005–2006 East Pacific Rise eruptions. The chief scientists of the cruises, J. Cowen, K. Von Damm, and T. Shank, generously made time available for TowCam surveys to be accomplished, and we thank the science parties of the cruises for their cooperation in collecting the TowCam data. Marshall Swartz has been instrumental in the development and operation of TowCam and his contributions to this work, both in the field and on shore are gratefully acknowledged. We thank Ken Rubin, Adam Goss, and Dorsey Wanless for helpful discussions about this study as well as Scott White and an anonymous reviewer for constructive and careful reviews. This work has been supported by NSF grants OCE-0525863 (Fornari and Soule), OCE-0732366 (Soule), OCE-0138088 (Perfit), and the Woods Hole Oceanographic Institution Vetlesen Foundation Funds (Fundis, Fornari and Soule).