Dynamic response to strike-slip tectonic control on the deposition and evolution of the Baranof Fan, Gulf of Alaska

The influence of tectonics on sedimentary processes has been researched in a wide variety of contexts and scientific disciplines, including tectonic geomorphology, basin modeling, and climate-tectonic interactions. Here we focus on the influence of regional plate tectonic motion on the deposition and evolution of a deep-sea fan, separated from its sediment supply by a strike-slip fault. Our study of the Baranof Fan in the Gulf of Alaska addresses questions about large-scale sedimentation along the margin, and the Baranof Fan serves as a natural laboratory for examining the influence of strike-slip tectonics on deep-sea fan sedimentation patterns. There are few instances of large-scale strike-slip tectonics influencing deep-sea sedimentation in the modern world, but there is evidence for the process in the geologic past. For example, the Zodiac Fan, the largest of the Gulf of Alaska deep-sea fans, is located along the Aleutian Trench, hundreds of kilometers removed from its sediment supply (Stevenson et al., 1983). In this study we show that the Baranof Fan is a type example illustrating the extent to which plate motion can influence deposition of large-scale sedimentary fans.

Sedimentary fans, including architectural elements such as channels, overbank deposits, lobes, and avulsions, have been studied at length (e.g., Mutti and Normark, 1991; Piper and Normark, 2001). Many publications discuss sediment distribution mechanisms such as turbidity currents and shelf canyon systems (e.g., Normark and Carlson, 2003; Piper and Normark, 2009) and how they are related to the broad range of downslope deposits (Normark and Piper, 1991; Piper and Normark, 2009). Early and substantial analysis of sedimentary distribution systems and associated deposits has helped to inform our analysis of the Baranof Fan, which has not been examined in detail since 1987 (Stevenson and Embley, 1987).

In this study we redefine the Baranof Fan boundaries, area, volume, and depositional controls, and provide new constraints on its age and evolutionary history. Through use of two-dimensional (2D) seismic reflection and multibeam bathymetry data, we suggest a tectonic influence on Baranof Fan deposition, including northwest to southeast channel avulsions and strike-slip–driven translation of shelf point sources along the fan’s landward edge. We also show the paleopathway of the previously unstudied Baranof channel, describe its influence in Baranof Fan evolution, and suggest the near-term formation of a new channel at the Dixon Entrance.

In addition to the Baranof Fan, the other two deep-sea fans in the Gulf of Alaska (Fig. 1) are the late Oligocene Zodiac Fan, located along the Aleutian Islands (Stevenson et al., 1983), and the younger, Pliocene–Pleistocene Surveyor Fan to the northwest, which is currently still active (Reece et al., 2011). The Baranof Fan is similar in area and volume to the other two Gulf of Alaska deep-sea fans (Stevenson and Embley, 1987). The Gulf of Alaska fans have had both fluvial and glacial inputs; the Zodiac Fan predates Pleistocene glaciation in North America, but the Surveyor and Baranof Fans have transitioned to become primarily glacially fed (e.g., von Huene and Kulm, 1973; Stevenson and Embley, 1987; Dobson et al., 1998; Reece et al., 2011).

Baranof Fan sediment derives predominantly from the Coast Mountains along the southeast Alaska margin (Plafker et al., 1994). Rivers and glacial streams erode the mountains near the coastline; the associated fluvial or glacial systems then carry the eroded sediment out to the shelf edge, where turbidity flows distribute the sediment to the deep seafloor via channel-levee environments (e.g., Ness and Kulm, 1973; Manley and Flood, 1988; Dowdeswell et al., 1996; Lopez, 2001). Currently the main conduits for sediment transport across the continental shelf to the Baranof Fan are glacial sea valleys adjacent to the Dixon Entrance and Chatham Strait, as well as a shelf canyon between these sea valleys (Fig. 2). The sea valleys, or shelf-crossing troughs, are ∼30-km-wide features representing the pathways of recent glacial advances (Carlson et al., 1982, 1996). Slope canyon systems associated with the channel heads and sea valleys are largely absent; rather, sediments are transported downslope via gully systems (Stevenson and Embley, 1987; Normark and Carlson, 2003).

The Baranof Fan overlies the Tufts Abyssal Plain and its channels weave through the Kodiak-Bowie Seamount Chain, which dominates both the seafloor (Fig. 2) and the subsurface in the Baranof Fan area (Morley et al., 1972). These volcanic edifices typically are 2–3 km above the surrounding crystalline basement and are thought to have been generated by a hotspot at the Pacific–North America–Juan de Fuca triple junction to the southeast (Silver et al., 1974). Smaller seamounts in the chain are buried by sediment but are still clearly visible in the subsurface seismic data.

During the Baranof Fan evolution since the late Miocene, depositional processes have been influenced by the 4.4 cm/yr right-lateral motion of the Pacific plate relative to North America (Elliott et al., 2010). Northwestward motion of the Pacific plate along North America is accommodated by the Queen Charlotte fault, a strike-slip fault located along the southeastern margin of Alaska, linking with the Fairweather fault to the northwest (von Huene et al., 1979; Carlson et al., 1988). The Pacific plate’s changing position relative to sediment sources on North America has caused both disruptions to sediment supply and evoked the creation of new channel systems in the Baranof Fan’s history (Bruns et al., 1984; Stevenson and Embley, 1987; Dobson et al., 1998).

As a late Miocene high-latitude fan, the Baranof Fan system has likely been influenced by glacial cycles, which typically correspond with periods of higher sedimentation (e.g., Vorren et al., 1989, 1991). There is evidence for post-Miocene global cooling (Mathews and Rouse, 1963) as well as several glaciation events that likely had a strong influence on fan sedimentation. Glacial periods in the northern Pacific include an alpine and tidewater glaciation event ca. 5.5 Ma (Lagoe et al., 1993), hemisphere-scale glacial intensification at 2.56 Ma (Lagoe et al., 1993; Raymo, 1994; Farley et al., 2001), and a transition to ∼100,000 yr glacial-interglacial cycles following the mid-Pleistocene transition at 0.7–1.2 Ma (Clark et al., 2006; Berger et al., 2008). In particular, intensification of glaciation ca. 2.6 Ma may have spurred isostatic uplift in the Coast Mountains (Farley et al., 2001) that was a positive feedback for increased glacial erosion rates during this period.

The Horizon and Mukluk channels (Fig. 2) are two of the longest deep-sea channels in the Gulf of Alaska and the most notable morphological features of the modern Baranof Fan. Merged bathymetry data from this study indicate that the total length of the Horizon channel is ∼800 km, and that of the Mukluk channel is ∼750 km. Both channels extend nearly 500 km onto the Gulf of Alaska abyssal plain from the shelf edge. Analysis of new seismic data shows channel fill deposits of 2–7 km width, commonly composed of several smaller kilometer-scale channels characterized by high-amplitude reflections (e.g., Deptuck et al., 2003) and arranged in channel complexes. These channel complexes are so named because they represent a collection or complex of smaller channel deposits (Abreu et al., 2003) that have formed via different phases of fill (Deptuck et al., 2003).

The Horizon and Mukluk channels are the primary modern depositional pathways for Baranof Fan sediment. Previous work acknowledged the existence of a third, unnamed channel system north of the Horizon and Mukluk channels (Stevenson and Embley, 1987), though the extent of its influence was previously unpublished. We provide new constraints on this third system and name it the Baranof channel. Sediment in the channels is carried downslope by turbidity currents and deposited in the lower gradient basin, a process typical of submarine fans (Piper and Normark, 2001, 2009). Herein we analyze downslope processes and present data showing that the channels appear to have evolved in a manner typical of many submarine fans, with lobe switching and avulsion events (Damuth and Flood, 1983; Manley and Flood, 1988). We consider the crust on which the sediment is deposited, the longevity of exposure to sediment sources, tectonic controls on sediment supply, and regional controls on sediment supply such as glaciation events; these are all important factors when developing a conceptual framework for sedimentation processes (Mutti and Normark, 1987).

The primary seismic data used in this study are from a June 2011 high-resolution data set collected aboard the R/V Marcus G. Langseth (the MGL1109 cruise); the data were acquired by the University of Texas Institute for Geophysics (UTIG) and the U.S. Geological Survey (USGS) as part of the U.S. Extended Continental Shelf (ECS) project, which is aimed at determining the full extent of the U.S. continental shelf maritime zone. The MGL1109 cruise collected 3260 km of multichannel seismic (MCS) data in 17 profiles. Data were acquired using an 8 km streamer towed at 9 m depth with 636 channels spaced at 12.5 m. The source was a 6600 in³ 36-airgun array, with 50 m shot spacing for most MCS profiles and 150 m spacing for two MCS lines coincident with ocean-bottom seismometer (OBS) stations. Common depth point (CDP) spacing is 6.25 m with a maximum fold of ∼80. The data sampling rate was 2 ms, and record length was 16 s. The reflection data were processed to poststack time migration using Paradigm’s FOCUS software (http://www.pdgm.com/solutions/seismic-processing-and-imaging/seismic-processing) utilizing the following processing flow: SEG-D convert, geometry definition, trace editing, 3-7–100-125 Hz bandpass filter, multichannel gap deconvolution, CDP sort, velocity analysis, spherical divergence correction, water-bottom mute, normal moveout (NMO) correction, stretch mute, trace balancing, stack, and F/K migration (using constant 1500 m/s velocity).

Other seismic data sets used for Baranof Fan analysis include earlier seismic reflection profiles, including USGS surveys S-6-79-GA (1979), L-6-81-NP (1981), and F-7-89-EG (1989). Survey F-7-89-EG was collected as part of GLORIA (Geological Long-Range Inclined Asdic), a survey conducted by the USGS and the Institute of Oceanographic Sciences (now the University of Southampton National Oceanography Centre) in an effort to better define the U.S. exclusive economic zone (EEZ) (EEZ-SCAN Scientific Staff, 1991). GLORIA surveys primarily aimed to obtain sidescan sonar coverage of the Gulf of Alaska, but 2-channel, single-airgun seismic data were also collected and these data have good coverage over the Baranof Fan. Processing of survey F-7-89-EG included trace editing and balancing, muting, and bandpass filtering (Reece et al., 2011). Surveys S-6-79-GA and L-6-81-NP are older (1979 and 1981, respectively), deep-water 2D USGS seismic surveys.

Bathymetry data in the Baranof Fan region include MGL1109 multibeam acquired coincident with the seismic data lines, and a high-resolution (∼100 m²) multibeam survey collected as a part of the U.S. ECS project (Gardner et al., 2006) that covers much of the southeast Alaska continental margin. The MGL1109 multibeam data were processed by UTIG and USGS using the CARIS HIPS and SIPS (http://www.caris.com/products/hips-sips/) software package. The MGL1109 and ECS multibeam data have been merged with the ∼1 km²-resolution ETOPO1 global bathymetry grid (Amante and Eakins, 2009), provided by the National Oceanic and Atmospheric Administration (NOAA) National Geophysical Data Center. The integrated bathymetric data are displayed in Figure 2.

After processing MGL1109 MCS data, we imported all 2D lines into Halliburton’s Landmark OpenWorks (https://www.landmarksoftware.com) interpretation software; we completed the bulk of the seismic interpretation using the DecisionSpace Desktop module. We gridded bathymetry data using the Generic Mapping Tools (GMT) software package (http://www.soest.hawaii.edu/gmt), rendered the data using ESRI ArcGIS mapping software (http://www.esri.com/software/arcgis), and imported the rendered data into DecisionSpace Desktop to be used in conjunction with the seismic 2D lines.

We mapped regional seismic unconformities in DecisionSpace Desktop, including the seafloor and oceanic basement. In addition to seafloor and basement, a regional stratigraphic downlap surface we call the Base Baranof horizon is observed as a mappable, high-amplitude seismic reflector throughout new MGL1109 and older USGS seismic transects (e.g., Fig. 3). Channels observable below the Base Baranof surface are generally smaller and less developed than channels above the surface, suggesting that channels above the surface dominated deposition of the fan (hence the name Base Baranof).

We gridded the seafloor, basement, and Base Baranof surfaces on a 0.01° grid (block size ∼1 km²) using GMT and visualized the result with Quality Positioning Services BV (QPS) Fledermaus software. From these grids, we generated two-way traveltime (TWTT) thickness grids: one for the thickness between the seafloor and the Base Baranof surface (Fig. 4A) and the other for the thickness between the seafloor and oceanic basement (Fig. 4B). Over the fan area, we calculated sediment volume both between the Base Baranof surface to the seafloor as well as between the mapped basement to the seafloor using these TWTT isopach grids. In areas where sediment thickness could not be picked or interpolated because of sparse data coverage, we calculated an average thickness in TWTT from existing grid cells (∼650 ms) and used this value to then calculate sediment volume.

There are no core data to constrain seismic velocities in the Baranof Fan, though stacking velocities derived from MGL1109 seismic reflection processing are well constrained. In order to simplify the calculation and account for the fact that MGL1109 data only cover more distal areas of the fan, we applied a constant sediment velocity of 2000 m/s to convert sediment thickness from TWTT to meters and thereby estimate sediment volume. The average acoustic sediment velocity of 2000 m/s is the same value that Reece et al. (2011) used for the adjacent Surveyor Fan, the value derived from Deep Sea Drilling Project Leg 178 cores (von Huene and Kulm, 1973) located within the Surveyor Fan. The 2000 m/s value is also consistent with preliminary MGL1109 OBS velocity models in the Baranof Fan (Reece et al., 2012).

Channel complex deposits were mapped along with modern seafloor channel deposits and smaller (∼1 km) channelized features. The location of these channels relative to modern seafloor channel pathways was used to assist in interpreting paleopathways and channel avulsions. We interpret levee deposits, lenses of sediment built up by sediment overbank, adjacent to channel complex deposits.

Adjacent to each channel complex, we identified and mapped 2–3 levee overbank deposits, mapping surfaces on the basis of high seismic reflectivity, regional continuity, and/or discontinuous stratigraphy. The uppermost levee reflector was mapped for each channel complex, representing the latest stage of active channel deposition. These surfaces were then correlated across 2D seismic reflection lines and assigned relative ages based on their stratigraphic relationships. Levee relationships with parent channels also allowed interpretation and interpolation of paleochannel pathways. In this manner we were able to determine a relative sequence through time of channel-levee deposits and thereby construct a history of channel avulsions within the fan (Figs. 5–8).

Five distinct channel pathways (and associated levees for each) were mapped throughout the subsurface fan, in addition to the Horizon and Mukluk channels on the modern seafloor. Utilizing the principle of superposition and therefore assigning relative ages to the channel-levee systems, we were able to determine a sequence of channel avulsions through time. In order to visualize the changing position of the channels relative to sediment sources onshore, we built a simplified tectonic reconstruction (Fig. 9; see also Animation 1). The reconstruction and subsurface channel relationships together allowed us to estimate the approximate timing of channel initiation, avulsion, and beheading. Given the lack of age data, the reconstruction model was built primarily as a qualitative visualization tool rather than a quantitative age constraint. We assume, for example, channels were initiated after passing major sediment pathways on the shelf such as Dixon Entrance sea valley and Chatham Strait, and were later beheaded when sediment supply across the shelf was cut off by Baranof Island (Fig. 2).

To build the tectonic reconstruction model, we used the GPlates open-source software package (www.gplates.org), superimposing the channel reconstruction on the modern coastline and the tectonic plate boundary (represented by the Queen Charlotte fault) in a reference framework where North America is fixed. We also included the Yakutat block and the Kodiak-Bowie Seamount Chain in the reconstruction, two tectonic features that have influenced channel initiation and morphology. The total reconstruction was built at a resolution of 0.5 m.y. and begins at 8 Ma. The plate boundary remained a dextral transform interface during this period (e.g., Atwater, 1970; Hyndman and Hamilton, 1993; Prims et al., 1997), so we assume constant plate direction and rate of motion in our reconstruction.

On the abyssal seafloor, the Baranof Fan is between the Surveyor Fan to the northwest and the Scott-Moresby sedimentary system to the southeast (Mammerickx and Winterer, 1970; Morley et al., 1972; Stevenson and Embley, 1987). The relationship between the Baranof Fan and the Scott-Moresby Fan is poorly constrained, but we are able to provide new insight into the relationship between the Baranof and Surveyor Fans using new data from the MGL1109 survey. The intersection of the Baranof Fan with the Surveyor Fan is apparent at the Baranof Fan’s northwestern edge in three seismic data transects, including two of the new MGL1109 MCS transects (Fig. 3). The Surveyor Fan overlies the Baranof Fan, onlapping Baranof sediment that in turn pinches out in the subsurface beneath Surveyor sediment, meaning that the Surveyor Fan is younger than the Baranof Fan in this area. The Chirikof channel is clearly visible in the seafloor of the Surveyor Fan (Fig. 3), showing that it is entirely distinct from Baranof Fan channel systems. Truncations of Baranof Fan reflectors into the Baranof-Surveyor boundary are visible as well (Fig. 3). The boundary between the two fans is also visible due to a difference in seismic facies; acoustic amplitudes in the Surveyor Fan are generally higher as opposed to the lower amplitude Baranof reflectors. We interpolated the boundary between transects where it is visible and thereby constrain the northwestern and northern edges of the Baranof Fan.

The unconformity between the Baranof and Surveyor Fans (Fig. 3) suggests erosion of the Baranof Fan before or at the time of Surveyor Fan deposition. Because of this erosional relationship, it is difficult to assign an age to this surface, though we can say that Surveyor sediment is, at the oldest, equivalent to the age of Baranof Fan erosion here. Much of the Surveyor sediment has been deposited in the past 1 m.y. (Reece et al., 2011), suggesting that the Baranof Fan’s northern region, which we interpret as being the oldest lobe of the fan, is >1 m.y. old. The younger southern part of the Baranof Fan, however, could be equivalent in age to the Surveyor Fan.

Because the acoustic facies of the Surveyor Fan are higher amplitude than the more transparent Baranof Fan reflectors (Fig. 3), we suggest that the fans consist of different sediment types and therefore possibly derive from different sediment sources on the shelf. Where much of the sediment in the Baranof Fan is sourced from the Coast Mountains, Surveyor Fan sediment has largely and most recently been supplied by glacial erosion in the Chugach–St. Elias orogen (Fig. 1; Reece et al., 2011). This difference may be in part because of the geometry of the Yakutat block, an oceanic plateau and microplate (Christeson et al., 2010; Worthington et al., 2012) that has translated northward along North America, transcurrent with the Pacific plate, over the past 20 m.y. (Plafker, 1987). We interpret that the Yakutat block acted to separate sediment source regions for the more recent (∼7 m.y.) history of the Baranof and the Surveyor Fans, with the Baranof Fan forming in the wake of the passing Yakutat block. In the Surveyor Fan, tidewater glaciation events transport sediments from their source across the wide shelf, locally formed by the Yakutat block, to the deep-sea fan (Reece et al., 2011). The separation of fan sediment sources due to the Yakutat block was suggested by Stevenson and Embley (1987) and our observations of fan seismic facies and regional tectonics support this interpretation.

Seafloor channels visible in new merged bathymetry data assist in defining more distal Baranof Fan boundaries and channel morphology. Major deep-sea channels are visible in bathymetry as much as 500 km from the shelf edge. In the very distal fan, the Horizon and Mukluk channels terminate into a series of abyssal ridges (Fig. 2), interpreted to be at the Baranof Fan’s southern edge and a part of the most distal lobe. A third, sinuous channel is observable in the distal bathymetry (Fig. 2), though we cannot constrain its existence with available geophysical data and therefore do not include it in the discussion. Because the channel is proximal to other channel systems in the Baranof Fan, however, it is included in area and volume estimates for the fan. Close to the shelf where seismic data are sparse or of poor quality, fan extents are interpreted in between Baranof Fan channels and channels of neighboring systems, the Surveyor Fan’s Chirikof channel (Reece et al., 2011) and the Scott-Moresby channels to the south and southeast (Mammerickx and Winterer, 1970; Morley et al., 1972).

The fan boundaries we observe in the available bathymetry and 2D seismic data, or otherwise based on interpolation between data points, give us a new estimate of 323,000 km² for the area of the Baranof Fan. The area estimate is likely a minimum because data are sparse at the more distal edges of the fan, and as it only includes seafloor extents, the estimate does not include sediment pinching out beneath the Surveyor Fan (Fig. 3). The wide shape of the fan is affected by the presence of the Kodiak-Bowie Seamount Chain, which obstructs sediment pathways to the deep seafloor (Fig. 1).

Small channel deposits exist below the regionally mapped Base Baranof surface (Fig. 3), though channel complex deposits above the surface are much larger; therefore, we interpret the surface as representing the onset of organized Baranof Fan deposition. Sediment below this reflector likely represents some combination of smaller scale channelization and pelagic and hemipelagic sedimentary processes, existing prior to initiation of large, organized channel systems. Because of this, the gridded Base Baranof horizon can be thought of as an approximation to the paleoseafloor at the onset of major channel formation within the fan. Using a fan area of 323,000 km² and isopach grids generated from the regional seafloor, Base Baranof, and basement surfaces (Fig. 4), we calculate a sedimentary volume of 209,000 km³ for the seafloor–Base of Baranof isopach (Fig. 4A) and a volume of 301,000 km³ for the entire sediment column within the fan (seafloor–basement; Fig. 4B). These new estimates for Baranof Fan volume are both larger than the previous estimate of 200,000 km³ (Stevenson and Embley, 1987). We reiterate that these values are minimum estimates as they do not include Baranof sediment pinching out beneath the Surveyor Fan. This is because Baranof sediment beneath the Surveyor Fan is only constrained along three 2D seismic transects and depends on the accuracy of fan boundaries, which may be in doubt due to a lack of good geophysical data coverage in the distal Baranof Fan. Given these new size estimates, we show that the Baranof Fan is comparable in size to the Mississippi Fan, and therefore among the largest deep-sea sedimentary fans in the world (Barnes and Normark, 1985; Sømme et al., 2009).

Subsurface sediment consists primarily of large channel-levee systems that are mostly buried by recent sedimentation. Channels appear as U- or V-shaped unconformities in the seismic reflection data, with shoulder-shaped levee deposits on either side. These channel complexes and channel-levee systems are thought to have been deposited via typical downslope processes such as turbidity currents (e.g., Ness and Kulm, 1973; Stevenson and Embley, 1987; Dowdeswell et al., 1996; Mohrig and Marr, 2003). The majority of subsurface and surficial channel complex deposits are 2–7 km wide (e.g., Figs. 5 and 6), containing channel fill from several iterations of smaller channels (∼1 km width). In many cases, overbank deposition due to thick overflow of turbidity currents causes 1-km-wavelength sediment waves within the levees (e.g., Normark et al., 2002; Posamentier, 2003; Babonneau et al., 2012) that are visible in 2D seismic and bathymetry throughout the fan (e.g., Fig. 5). Sediment waves have been observed in several deep-sea fans throughout the world, including the Indus, Amazon, and Monterey Fans (Normark et al., 2002; Fildani et al., 2006). Many levees we observe are also asymmetric, with the higher side of the overbank occurring on the western sides of the channels (Figs. 6 and7); this asymmetry is likely due to the ocean currents from the counterclockwise Gulf of Alaska Gyre (Rea and Snoeckx, 1995; Bart et al., 1999; Keevil et al., 2006) or Coriolis force (e.g., Cossu and Wells, 2010). In seismic images, channel-levee systems appear to be dominantly aggradational, with some erosional transitions (e.g., Fig. 6), similar to channels observed in the Amazon and Surveyor Fans (e.g., Ness and Kulm, 1973; Damuth and Flood, 1983; Manley and Flood, 1988; Reece et al., 2011).

A notable pattern within the Baranof Fan is that the oldest channel-levee deposits are in the northwestern fan, and deposits become progressively younger to the southeast. In addition to the well-mapped Horizon and Mukluk channel systems, there is at least one and possibly two additional channel systems visible in subsurface seismic data near the northern extent of the Baranof Fan. The oldest channel-levee deposits (which we call “oldest Baranof deposits”) were difficult to interpret due to the quality of the data; only one new seismic line crosses that region (Fig. 5) and the Kodiak-Bowie Seamount Chain reduces our confidence in correlating reflectors between seismic lines. The oldest Baranof deposits cannot be connected to other known channel systems (i.e., the Baranof, Horizon, and Mukluk) in the fan. The lack of correlation between the oldest Baranof deposits and other known systems could be due to poor seismic data quality, or because the oldest Baranof deposits are a part of a separate channel system. Regardless, because the oldest Baranof deposits are now buried under ∼0.25 km of sediment (Fig. 5) and also topped by Surveyor Fan sediment at their northern extent, we are confident that this is the oldest channel system in the Baranof Fan, now beheaded and inactive.

The next oldest channel system (which we call the Baranof channel), directly south of the oldest Baranof deposits, can be mapped clearly in new 2D seismic data (Figs. 6 and 7) and is visible in high-resolution bathymetry data nearest to the shelf and on the older, abyssal seafloor (Fig. 2). Although the Baranof channel has a modern seafloor expression, the seafloor channels are less than half the width of previous iterations of the channel, and sediment has nearly aggraded to the top in some places (Fig. 7). For this reason, we interpret the Baranof channel to be recently beheaded after passing Chatham Strait and therefore waning in deposition.

The Baranof channel has been discussed in literature as a possible third channel system in addition to the Horizon and Mukluk channel systems (Stevenson and Embley, 1987). Our results support the existence of the Baranof channel as a third, distinct channel system. The Baranof channel is similar in scale, at ∼700 km length, and in subsurface character to the ancestral Horizon and Mukluk channel deposits. In addition, the Baranof channel system appears to be the primary routing system for approximately half of the sediment deposited in the northern and western parts of the fan. In the 2D seismic data, we map two distinct buried channel complexes that seem to feed into the same levee overbank deposits, suggesting that these channels were separate systems active at the same time for at least a portion of their history (Fig. 7). We interpret one of these buried channels to be an early iteration of Baranof channel, currently inactive as it is buried by ∼100–200 m of sediment. The other buried channel’s uppermost levee reflector can be traced to the base of the modern Horizon channel (Fig. 7), which we interpret to be an avulsion of the Horizon channel; therefore, we interpret this buried channel complex to be a paleo–Horizon channel.

The traces of the two youngest channels, the Horizon and Mukluk channels, are largely constrained by bathymetry data, especially in the distal Baranof Fan where seismic data are sparse. Both of these channels have subparallel seafloor pathways as they curve through the Kodiak-Bowie Seamount Chain, terminating into a series of abyssal ridges among the Patton-Murray Seamount Chain (Fig. 2), seemingly blocked by the topography. The Horizon and Mukluk channels are ∼800 and 750 km in length, respectively, and extend nearly 500 km from the shelf edge onto the abyssal seafloor. We can see the relationship of the modern Horizon channel to the modern Mukluk channel in 2D seismic data from the more proximal fan (Fig. 8). The Mukluk channel fill extends deeper into the sediment than the modern Horizon channel fill, which might suggest that it is older than the Horizon channel. We do not map an avulsion of the Mukluk channel, however, whereas the Horizon channel has undergone at least one avulsion (Fig. 7). We therefore propose that the Mukluk channel is younger than the Horizon channel, though we cannot be certain due to relatively poor data quality and coverage over the Mukluk channel.

We interpret the Baranof Fan to be a reactive system as described by Covault et al. (2010, 2013), with high sediment flux (interpreted based on large fan volume and glacial interaction) allowing for external tectonic forcing to be visible in the sediment record. Our clear observation of consistently southward-avulsing channels and channel-levee deposits younging southward suggests that the development of sediment pathways was influenced by the translation of the Pacific plate past sediment point sources on the shelf, such as the bathymetrically imaged glacial sea valleys or shelf-crossing troughs (Carlson et al., 1982, 1996; Vorren and Laberg, 1997). The sequence of channel system formations and beheadings supports this tectonics-driven depositional pattern in the fan, as the oldest channel systems in the north are now beheaded and the youngest (Mukluk) channel is the farthest south. This interpretation supports similar results from previous studies (Stevenson and Embley, 1987; Dobson et al., 1998).

Because of lack of age control, there is still a question of when deposition of the Baranof Fan initiated and how quickly its channel systems developed. The timing of initiation also implies a position of the Pacific plate relative to North America sediment sources, an important factor to consider. We hypothesize that major channel systems formed only when adjacent to major sediment sources such as the Dixon Entrance (Fig. 2), and that sediment supply was cut off where pathways to the seafloor are blocked, namely, in the south at the Haida Gwaii Islands and in the north by Baranof Island (Fig. 2). We know that the oldest channel systems in the northern Baranof Fan, now beheaded, must have been active when they were adjacent to sediment sources between Haida Gwaii and Baranof Island. The timing of activation of the oldest channel systems, however, could vary. In a small-offset model, the Baranof channel, possibly the oldest channel system in the fan, could have initiated when it was adjacent to the Chatham Strait sea valley ∼100 km southeast of its modern position, with the Horizon and Mukluk channels forming subsequently as they received sediment via a shelf canyon system and the Dixon Entrance, respectively. In a longer offset model, the Baranof channel initiated when it reached the Dixon Entrance ∼300 km southeast of its modern location, the other channels forming later as new parts of the Pacific plate became exposed to this sediment source.

Our observations favor a long-offset (∼300 km) model, with the Baranof channel initiating at the Dixon Entrance and the Horizon and Mukluk channels forming sequentially as they in turn reached the Dixon Entrance. A short-offset model requires a southward-sequential pattern of channel development despite the three channel systems being simultaneously exposed to shelf sediment pathways. Long-offset-aided sequential exposure seems more likely than simultaneous exposure given that the seismic reflection data clearly show southward-younging deposits in the subsurface, and channel avulsions filling available accommodation space in the south. In addition, a shorter offset model implies a much smaller time frame for fan development (∼2 m.y., assuming constant plate motion). The Surveyor Fan’s Chirikof channel system is visible in the seafloor on top of older, northern Baranof Fan sediment (Fig. 3). A short-offset model suggests full development and beheading of the Baranof channel system as well as formation of the young Chirikof channel system over the past ∼2 m.y., less likely than a long-offset solution that allows more time for Baranof channel evolution.

Based on a long-offset model, we provide an approximate age for the onset of Baranof Fan deposition using tectonic reconstruction, despite lack of age constraints from cores. Our tectonic model is constrained using previous regional analysis. Dextral strike-slip motion has dominated the Pacific–North America plate boundary along the Queen Charlotte fault for the past ∼20 m.y. (Atwater, 1970; Hyndman and Hamilton, 1993); a discrete clockwise rotation in the Pacific plate motion vector ca. 6 Ma (Doubrovine and Tarduno, 2008) caused oblique convergence along the southern fault (Hyndman and Hamilton 1993). Today, the Queen Charlotte fault is seismically active and has undergone large strike-slip and some oblique-thrust events (Lay et al., 2013).

Our tectonic reconstruction assists with visualization of channel formation, avulsions, and beheadings through the Baranof Fan’s history (Fig. 9; see also Animation 1). We superimposed the reconstruction on the modern coastline to best emphasize the plate offset and create a frame of reference. We also include the outline of the full extent of the Yakutat block, which has translated north along with the Pacific plate for the past ∼20 m.y. and is currently undergoing flat-slab subduction beneath North America (Eberhart-Philips et al., 2006). We use recent GPS measurements from Elliott et al. (2010) to provide a relative dextral offset rate of 4.4 cm/yr between the Pacific and North America plates. Assuming a constant 4.4 cm/yr rate and that sedimentary deposition initiated when the northernmost channels (the oldest Baranof deposits) were at the southernmost sedimentary source (the Dixon Entrance), the 300-km-long shift of the northernmost (now buried) channel from the south end of the Dixon Entrance to its modern position must have taken ∼7 m.y. This calculation means that the oldest sedimentary deposits in the Baranof Fan are late Miocene, which is consistent with the 12 Ma basement rock underlying the fan sediment (Berggren et al., 1985). The timing of channel formation and beheading is based strictly on location relative to sea valleys on the shelf, with channels initiating as they pass the Dixon Entrance and beheading after passing Chatham Strait.

Although sediment pathways south of the Dixon Entrance are blocked by Haida Gwaii (Fig. 1), we acknowledge that the Queen Charlotte Sound south of Haida Gwaii may also have served as a source for older sediment in the Baranof Fan early in its history (Yorath, 1987), providing a very-long-offset model possibility. However, the now-distal Baranof Fan may have been at the Queen Charlotte Sound ca. 12–14 Ma, and cut off from sediment sources until ca. 7 Ma as it passed by Haida Gwaii. The age of the basement beneath the proximal Baranof Fan is only 12 Ma, making a very long offset model unlikely; however, older sediment beneath the Base Baranof surface in the more distal fan may derive from the Queen Charlotte Sound if it is older than ca. 7 Ma.

Assuming that deposition of the Baranof Fan occurred over the past ∼7 m.y., overlapping with several major periods of Northern Hemisphere cooling (including the Pleistocene Epoch beginning ca. 2.6 Ma), glacial events must have influenced sedimentation. The Surveyor Fan, north of the Baranof Fan, has been strongly influenced by periods of glaciation throughout its depositional history (Reece et al., 2011). Without the necessary core data to better constrain the ages of the sedimentary deposits in the Baranof Fan, however, it is not possible to match channel formation events or periods of accelerated deposition to specific glacial periods. Given the high latitude, the glaciation of the area, and the dynamic capability of glaciers to carry massive amounts of sediment (e.g., Dowdeswell et al., 1996; Reece et al., 2011), we think that glaciation over the past 7 m.y. has likely had a significant influence on the sedimentation rates and flux to the Baranof Fan, particularly since glacial intensification and increased uplift of the Coast Mountains and the St. Elias orogen ca. 2.56 Ma (Farley et al., 2001; Enkelmann et al., 2009).

It is interesting that there is no new channel forming at the mouth of the Dixon Entrance, despite the Mukluk channel having passed it by; however, we hypothesize that there is either a channel poised to form at the mouth of the Dixon Entrance sea valley, or that all of the sediment in the area is still being funneled to the Mukluk channel. The lack of new channel supports the idea that the Baranof Fan is a sea-level lowstand fan system, and that perhaps a new channel may form at the Dixon Entrance during a near-term lowstand event.

In both scale and downslope channel-levee dynamics, the Baranof Fan is similar to lower latitude fluvial fans like the Amazon Fan (Damuth and Flood, 1983; Manley and Flood, 1988; Lopez, 2001), though its high latitude, glacial valleys on the shelf, and thick sediment near the shelf edge would suggest a component of glacial influence (Carlson et al., 1982; Dowdeswell et al., 1996; Laberg and Vorren, 1996). The Gulf of Alaska deep-sea fans differ from high-latitude sedimentary deposits observed elsewhere (e.g., Norway, Greenland), however, which are typically referred to as trough-mouth fans (TMFs) (Vorren et al., 1989). The Gulf of Alaska fans have well-developed channel distributary systems similar to river-fed fans (e.g., Damuth and Flood, 1983), whereas TMFs tend to be composed of debris flow lobes (Vorren et al., 1989; Laberg and Vorren, 1996; Vorren and Laberg, 1997); in addition, the Gulf of Alaska fans are as much as 1–2 orders of magnitude larger than most TMFs. The Baranof Fan may have had sedimentary inputs from both fluvial and glacial sources, though there is evidence for more recent glaciation on the seaward shelf edge.

Though the Baranof Fan has few, if any, modern analogues at a similar scale, we find similar sedimentary processes when comparing it to smaller scale fan systems affected by strike-slip motion. There is evidence that strike-slip motion along the San Andreas fault in southern California, a fault similar in scale and offset to the Queen Charlotte fault (Carlson et al., 1988), has caused sequential lobe switching in adjacent fans such as the Monterey Fan (e.g., Normark, 1998). The upper Monterey Fan, which consists of channel-levee systems similar to the Baranof Fan, has undergone shifts in sediment source and changes in channel geometry due to tectonic influence (Normark, 1998; Fildani and Normark, 2004). However, both the volume of the upper turbidite sequence in the Monterey Fan (∼100 km³) and the time required to deposit it (500 k.y.) are several orders of magnitude smaller than those of the Baranof Fan.

Based on our interpretation of seismic and bathymetry data, we provide new constraints on the depositional history of the Baranof Fan. First-order mapping of bathymetry, channel deposits, and regional seismic horizons provide estimates of the area, shape, and volume of the Baranof Fan; the area of the fan is 323,000 km², with a total sediment volume of 301,000 km³. Organized fan deposition comprises a sediment volume of 230,000 km³ above the regionally mapped Base Baranof downlap surface. These size constraints are larger than previous estimates (Stevenson and Embley, 1987), making the Baranof similar in size to the Mississippi Fan. The intersection of the Baranof Fan with the Surveyor Fan to the north helps define the northern extent of the Baranof Fan. In this area, the Baranof Fan is older than the Surveyor Fan; much of the Surveyor Fan was deposited in the past 1 m.y., implying that this oldest lobe of the Baranof Fan is at least 1 m.y. old.

The Horizon and Mukluk channels are ∼800 and ∼750 km in length, respectively, curving sharply to the south at the Kodiak-Bowie Seamount Chain, creating a wider rather than longer fan shape. There is strong evidence for one or two now extinct or recently beheaded channel systems north of, and older than, the modern Horizon channel system. We identify one of these as the Baranof channel system in seismic data, observing it as a major channel-levee system distinct from the Horizon and Mukluk systems. It is difficult to map individual channel deposits in the sparse and lower quality seismic data in the northernmost fan, so it is unclear if the oldest Baranof deposits (Fig. 5) represent an early phase of the Baranof channel system, or if there was actually a fourth channel system, older than all of the other known channel systems.

We observe that the relative ages of channel-levee deposits in the Baranof Fan are consistently younger southward. Based on this observation, we conclude that channel avulsion, formation, and beheadings have progressed from north to south within the fan over the past ∼7 m.y. Given the northwestern motion of the Pacific plate relative to sediment sources on the North America plate, it is likely that this tectonic motion has been the dominant influence on channel development throughout the Baranof Fan history.

We thank the crew, officers, and science party aboard the R/V Marcus Langseth MGL1109 cruise in June 2011 for assistance in collecting data. Thanks to the reviewers and editors of Geosphere for constructive comments that considerably strengthened, clarified, and broadened the impact of the text. We also thank Jed Damuth, Peter Haeussler, Brian Horton, and David Mohrig for helpful scientific discussion and feedback. This project was funded by the U.S. Geological Survey in support of the U.S. Extended Continental Shelf (ECS) Project and by the NSF Continental Dynamics study the St. Elias Erosion/Tectonics Project (STEEP), NSF grants EAR-0408584 and EAR-1009986. This is University of Texas Institute for Geophysics (UTIG) Contribution 2709.