Tectonic and climatic influence on the evolution of the Surveyor Fan and Channel system, Gulf of Alaska

Many studies have used accumulation rates and sediment provenance derived from submarine fan sediment cores worldwide to describe the tectonic growth and erosion on land that provided the source material for the marine record (e.g., Figueiredo et al., 2009; Hebbeln et al., 2007; Rea and Snoeckx, 1995; Ullrich, 2010). The terrigenous component of marine sediment accumulation rates serves as a proxy for changes in erosion rates on land, and therefore changes in either orogenic exhumation rates or climate. Burbank et al. (2003) showed that Himalayan erosion is dominated by tectonic uplift in spite of large variations in precipitation, but other studies demonstrate that climate can dominate erosion in glacial orogens at glacial maxima (Berger et al., 2008; Hebbeln et al., 2007).

Much work has been done to constrain the relationship between tectonics and climate in the glacial St. Elias orogen in southern Alaska (Berger et al., 2008; Rea and Snoeckx, 1995), but this study fills in a key additional component. The proximity of the St. Elias Mountains to the coastline assures that the majority of glacially eroded sediment is deposited in the Gulf of Alaska, much of it in the deep-water Surveyor Fan. A long-offset 2D seismic-reflection study, acquired in 2008, linked to previous studies of Surveyor Fan sediment cores yields information on margin processes, erosion, climate events, orogenesis, and exhumation. The results of this study will add to the existing body of work on climate-dominated tectonic systems, and demonstrate how such a system cannot only transform the orogen, but also the entire margin from source to sink.

The Yakutat terrane in the northern Gulf of Alaska is a 15–25-km-thick mafic terrane that originated as an oceanic plateau (Christeson et al., 2010; Eberhart-Philips et al., 2006). At least 600 km of Yakutat terrane has subducted at a low angle beneath North America since ∼10 Ma (Fig. 1) (Eberhart-Philips et al., 2006; Gulick et al., 2007; Plafker et al., 1994; Rea and Snoeckx, 1995). This flat-slab subduction has resulted in the still-active Chugach–St. Elias orogen, which exhibits the highest coastal relief and greatest glacial influence of any active orogen globally (Pavlis et al., 2004). Varying degrees of glacial erosion and rock exhumation in the St. Elias Range since the Miocene (Enkelmann et al., 2010; Lagoe et al., 1993; Rea and Snoeckx, 1995; Spotila and Berger, 2010) have distributed sediment into the Gulf of Alaska, leading to periodic significant increases in growth of the Surveyor Fan (Lagoe et al., 1993; Rea and Snoeckx, 1995; Stevenson and Embley, 1987), the terrigenous outwash body that comprises the majority of the Alaska Abyssal Plain (Fig. 1). Fan sediment as old as ∼20 Ma varies in provenance, quantity, and content as a result of changes in onshore exhumation rates and glacial extent (Berger et al., 2008; Enkelmann et al., 2010; Ingle, 1973; Spotila and Berger, 2010). These changes are likely represented in fan stratigraphy, morphology, and shifts in acoustic character (Ness and Kulm, 1973; Stevenson and Embley, 1987).

The first appearance of Gulf of Alaska ice-rafted debris appeared in the now uplifted and exposed marine Yakataga Formation at ∼5.5 Ma. This alpine glaciation event is referred to as glacial interval A (Lagoe et al., 1993). Glacial activity was diminished during the mid-Pliocene warm interval, but returned during the onset of Northern Hemisphere glaciation at 2.9–2.4 Ma (Raymo, 1994), referred to as glacial interval B (Lagoe et al., 1993). Glacial interval B provided the first ice-rafted debris observed in the distal fan at Ocean Drilling Program (ODP) site 887 (Fig. 2) (Rea and Snoeckx, 1995). A further intensification of glacial activity, glacial interval C, occurred at ∼1 Ma, and could have been a regional response to the mid-Pleistocene transition (MPT) (Berger et al., 2008), a change in glacial-interglacial cycles from 40 k.y. to 100 k.y. (Clark et al., 2006).

Terrigenous sediment flux in the Gulf of Alaska at ODP 887 doubled at each of the three glacial intervals (Rea and Snoeckx, 1995), with the increase at the MPT being the largest on record. Additionally, Berger et al. (2008) showed a major increase in St. Elias orogen exhumation rates associated with glacial interval C and proposed that the glacially carved Bering Trough (Figs. 1 and 2) first reached the shelf edge near the time of the MPT, providing a sediment delivery pathway from exhumed orogen to shelf edge. Ultimately, due to erosion of sediment during repeated glacial advances across the shelf, the Surveyor Fan may be the most robust long-term record of onshore exhumation and climate events.

Terrigenous turbidites, mudstone, and claystone of the Surveyor Fan overlie marine chalk and a basaltic basement of the Pacific plate crust (Shipboard Scientific Party, 1973). Previous studies divided the Surveyor Fan into two major sequences (Ness and Kulm, 1973; Stevenson and Embley, 1987; von Huene and Kulm, 1973). These sequences were termed upper and lower and were based on sedimentation rates and differences in acoustic facies imaged in 2-D seismic-reflection profiles. The boundary between the two sequences represents a shift from a lower coarser grained facies to an upper finer grained facies possibly associated with Surveyor Channel inception and its control on fan sediment distribution during deposition of the upper sequence (Ness and Kulm, 1973; Stevenson and Embley, 1987).

The Chirikof and Surveyor Channel systems (Fig. 1) dominate present-day Surveyor Fan morphology and sediment distribution (Carlson et al., 1996; Stevenson and Embley, 1987), but unlike other large deep-sea channels, the Chirikof and Surveyor Channels are not associated with a major fluvial system or submarine canyon (Ness and Kulm, 1973). The Surveyor Channel is >700 km long with three major tributaries (Carlson et al., 1996; Ness and Kulm, 1973; Stevenson and Embley, 1987). The main channel has an average width of ∼7 km, but is as wide as 27 km in some locations and up to 500 m deep at its distal section (Carlson et al., 1996; Stevenson and Embley, 1987). Ness and Kulm (1973) divided the channel into three sections (Fig. 2) based on differences in channel relief, axial gradient, deposition, and erosivity. The two upstream sections, the upper and middle channel, divide the channel length between the base of slope and the Kodiak-Bowie Seamount Chain. The last section, the lower channel, begins past the Giacomini Seamount and extends to the Aleutian Trench. Ness and Kulm (1973) interpret the upper, middle, and lower sections as being dominantly depositional, transportational, and erosional, respectively.

The interaction between changes in orogenic exhumation rates and glacial erosion are all recorded in Surveyor Fan sediment (Berger et al., 2008; Lagoe et al., 1993; Rea et al., 1993; Rea and Snoeckx, 1995). Correlating stratigraphic changes revealed in new seismic data provides supporting information about relative timing of climatic and tectonic events, their influence on each other, and is the key to understanding major changes in sedimentation and fan morphology through time. We regionally analyze the upper-lower sequence boundary across the Surveyor Fan and introduce additional boundaries and interpretations using new and existing seismic-reflection data. By correlating sequences across the entire fan, we show that they are regionally extensive deposits related to increases in exhumation on land and regional response to global changes in climate. A detailed understanding of Surveyor Fan stratigraphy provides an analogue for marine systems globally that record the complex interactions between climate and tectonics in glacially dominated systems.

A recent high-quality, multichannel seismic (MCS) data set was acquired in 2008 aboard the R/V Marcus G. Langseth as part of the National Science Foundation (NSF)–Continental Dynamics St. Elias Erosion/tectonics Project (STEEP). This survey provided ∼1250 km of MCS data recorded on an 8 km streamer with 640 channels. The source was composed of 36 bolt airguns in four linear arrays with 6600 in³ total air volume. To date, this STEEP survey has been used for studies of shelfal, Yakutat terrane-related tectonics, deformation, and stratigraphy (i.e., Christeson et al., 2010; Worthington et al., 2010).

A high-resolution MCS data set was also collected in 2004 on the R/V Maurice Ewing as part of an Integrated Ocean Drilling Program (IODP) site survey. This MCS survey utilized two 105/105 in³ generator-injector airguns that provide ∼5 m vertical resolution. Data from this survey were used in studies of deformation, stratigraphy, and glacial history (i.e., Berger et al., 2008; Cowan et al., 2010; Gulick et al., 2007; Worthington et al., 2008).

Three additional MCS surveys were collected by the U.S. Geological Survey (USGS) in the 1970s. The first of these, G175, was collected in 1975 with a 22-airgun, 1200 in³ array and a 2.4 km, 48-channel streamer. Surveys L378 and L677 utilized a 5-airgun, 1326 in³ array and a 2.4 km, 24-channel streamer. These data were used primarily in Yakutat terrane-related tectonics and fault mapping studies (e.g., Bruns, 1982, 1985; Bruns and Carlson, 1987; Gulick et al., 2007; Pavlis et al., 2004; Worthington et al., 2010). Four remaining single-channel seismic surveys (F689, F988, F186, and F789) were collected by the USGS in the 1980s in conjunction with the GLORIA sidescan sonar mapping of the newly expanded United States Exclusive Economic Zone.

We use two drill sites for control on depth, lithology, age, and accumulation rates. Site 178 of Deep Sea Drilling Project (DSDP) Leg 18 was drilled in the Surveyor Fan in 1973 ∼55 km northwest of Giacomini Seamount in the Kodiak-Bowie seamount chain (Fig. 2) (Shipboard Scientific Party, 1973). DSDP 178 provides limited age control using magnetic polarity stratigraphy (von Huene et al., 1973) and ⁴⁰Ar/³⁹Ar dating of volcanic ash layers (Hogan et al., 1978). DSDP 178 is the only source of information on depth, lithology, and age control (Fig. 3) for sequences observed in seismic-reflection data presented here. The main limitation of DSDP 178 as a data source is the poor core recovery rate, which resulted in many missing and damaged sequences (Shipboard Scientific Party, 1973).

Site 887 of ODP Leg 145 was drilled in the extreme distal part of the Surveyor Fan on the Patton-Murray seamounts in 1992 (Fig. 2) (Rea et al., 1993). In spite of the distal location, ODP 887 provides the best control within the fan for identifying relative changes in sedimentation rates and timing of climatic and tectonic events (Rea and Snoeckx, 1995).

Multibeam bathymetry data with 100 m² resolution were obtained in 2005 for the United Nations Convention on Law of the Sea (Figs. 1 and 2) (Gardner et al., 2006). Use of these data has been related to shelf-proximal studies of the Yakutat-Pacific plate boundary, the Transition fault (Gulick et al., 2007), and the Yakutat–North America deformation front, the Pamplona Zone (Worthington et al., 2008, 2010).

This study builds on previous work with identification of additional sequences and boundaries, so new nomenclature is necessary. We renamed the traditional lower and upper sequences I and II, respectively, and added a new sequence, III. We interpreted the sequences and boundaries based on observations of 2D seismic-reflection data and ordered them from oldest to youngest, with I directly overlying the basement, and III situated at the seafloor. The sequences are regionally correlative and present on the majority of seismic images in the region of Surveyor Channel influence. Sequence I is the least commonly observed sequence based on poor acoustic energy penetration on the older surveys, but is present in enough survey lines for reasonable correlation across the study area.

We interpret the Surveyor and Chirikof Channel systems to be genetically similar in character and morphology upstream of the Kodiak-Bowie Seamount Chain. Additionally, both comprise the Surveyor Fan and have been influenced by the same tectonic and climate events that fed the Yakutat shelf. Keeping that in mind, we choose to focus on the Surveyor Channel system for the purposes of this study.

Sequence boundaries defined in this study were correlated to DSDP 178 using line 13 from USGS survey F689 located 10 km away (Figs. 2–4). DSDP 178 and line 13 are not separated by significant morphological or tectonic boundaries, and are well correlated in terms of seismic facies to lithologic descriptions. DSDP 178 is the only drill site that can be correlated with available seismic data within the Surveyor Fan. We depth converted seismic line 13 using velocities published in von Huene et al. (1973) based on lithology in recovered cores from DSDP 178. ODP 887, located ∼310 km SW of DSDP 178, is too distal from land and seismic data locations to directly correlate stratigraphy and ages, but it does provide the record and timing of significant changes in terrigenous sediment flux vital to understanding onshore changes in exhumation and glacial extent. We correlate such distal flux “benchmarks” to the seismic record in the proximal fan by correlating sequences to glacial intervals A–C (Berger et al., 2008; Lagoe et al., 1993).

Processing of the 1970s USGS data used in this study included bandpass filtering, muting, normal move-out correction, and stacking. The 1980s USGS data processing included trace editing and balancing, muting, and bandpass filtering. Processing of the 2004 and 2008 seismic-reflection data included trace regularization, normal move-out correction, bandpass filtering, muting, frequency–wavenumber filtering, stacking, water-bottom muting, and finite-difference time migration using Paradigm Geophysical's FOCUS software (Berger et al., 2008; Christeson et al., 2010; Gulick et al., 2007; Worthington et al., 2010). All seismic data were interpreted in time using Geoframe, the Schlumberger seismic interpretation software. The sequence boundaries, seafloor, and basement horizons were gridded using a least-squares algorithm in Geoframe and subtracted from each other to create the two-way travel-time thickness, or isopach, maps in Figures 5 and 6. Figures 5D and 6 show seismic lines and data density utilized to construct the isopach maps. We calculated an estimate of Surveyor Fan sediment volume using the isopach in Figure 6, an average sediment velocity of 2000 m/s, and the planimeter tool in Geoframe.

Sequences I and II exhibit laminated, laterally semicontinuous reflectors consistent with turbiditic deposition (Fig. 7). Sequence III is thinly laminated and contains reflectors that are laterally continuous, flatter, and smoother than those in the other sequences (Fig. 7). Stratal relationships at the sequence boundaries are highly variable and greatly influenced by basement topography (Fig. 8). Strata appear concordant across large areas of the boundaries, but exhibit angular discordance in others. Sequence I is generally basement conformable but onlaps basement topography locally (Figs. 4 and 8A). Sequence II onlaps sequence I in the areas where sequence I exhibits topography (Figs. 7 and 8B), but is conformable in other locations (Figs. 4 and 8D). Sequence III onlaps sequence II in the proximal fan (Fig. 9) and downlaps it in the distal fan (Figs. 4 and 8C), where both sequences pinch out farther from the sediment source. Sequence III is also conformable to sequence II in some locations (Figs. 8E and 9). Reflectors in sequence I are much higher in amplitude than overlying reflectors in the distal part of the fan (Figs. 4 and 8B). In the proximal fan, the high-amplitude reflectors only appear at the top of sequence I (Fig. 10). This provides a stark contrast to the characteristically lower-amplitude reflectors at the base of sequence II (Figs. 8B and D), although sequence II reflectors also increase in amplitude upsection, providing contrast with the lower amplitudes at the base of sequence III (Fig. 9). Sequence III reflectors vary in amplitude across the fan, but generally gain amplitude upsection (Fig. 9), with the highest amplitudes at and directly below the seafloor. In general, sequence III contains the highest amplitude reflectors between the basement and seafloor.

Two-way travel-time thickness, or isopach, maps (Fig. 5) for the three sequences show a varying depositional history in the Surveyor Fan. Depocenters for sequence I are distributed in the topographic lows between basement highs. Sequence I sediment distribution shows no significant spatial variation, but merely depocenters that infill Pacific plate topography. Deposits of sequences II and III are more organized, with shelf proximal depocenters that thicken into the Yakutat slope (Figs. 5B and C). The depocenters are positioned in the same locations for sequences II and III, but they are greater in thickness and cover a much larger area in sequence III.

Correlation of seismic stratigraphy to DSDP 178 (Figs. 3 and 4) places the I-II sequence boundary at ∼5 Ma, near the beginning of glacial interval A, based on ⁴⁰Ar/³⁹Ar dating of ash layers (Hogan et al., 1978) (Fig. 3). The sequence I-II boundary occurs at ∼330 m depth in the DSDP 178 core within a section of fine-grained sand to silty turbidites and interbedded diatomaceous ooze and mud with increasing diamictite upsection (Shipboard Scientific Party, 1973). No change in lithology occurs at the I-II sequence boundary in the DSDP 178 core (Ullrich, 2010). The II-III sequence boundary is dated at ∼1 Ma based on correlation with a magnetic polarity reversal in the DSDP 178 core (von Huene et al., 1973) (Fig. 3), making it coincident with the onset of glacial interval C. At 130 m depth, the II-III sequence boundary lies 10 m above a change in fan lithology. The section from 96 to 141 m contains abundant diamictite interbedded with silty clay and diatom-rich intervals, whereas the section from 141 to 280 m contains less diamictite, much more silty clay and a decrease in diatom abundance (Shipboard Scientific Party, 1973). Both sequence boundaries are synchronous with a doubling in terrigenous sediment flux observed at ODP 887 at ∼5 Ma and ∼1 Ma. Although there is also an observed doubling of terrigenous sediment flux at ODP 887 at glacial interval B (Rea and Snoeckx, 1995), we do not observe a regional sequence boundary or change in lithology that could be interpreted as the onset of glacial interval B. Therefore, sequence II deposits include glacial intervals A and B.

Surveyor Channel fill deposits are located almost exclusively within sequences II and III, but are present in sequence I in the lower channel section (Fig. 11), defined as the section of the Surveyor Channel south of the Kodiak-Bowie seamount chain. Channel profiles in the upper and middle channel sections show the modern Surveyor Channel atop a thick section of channel fill (Figs. 11 and 12). The shelf-proximal upper channel (Fig. 2) exhibits distinct levees, reflectors that turn down into the channel flank, with some sidewall failure and reflection truncation (Figs. 10 and 12), and occurs in the thickest section of the Surveyor Fan. The middle channel, north of the Kodiak-Bowie Seamount Chain (Fig. 2), has similar sidewall character, but typically exhibits less channel fill than the upper channel and has less distinct or no levees (Figs. 11 and 13). The lower channel exhibits little to no historical channel fill, no downturned reflectors at the sidewall, no levees, and occurs much lower in the stratigraphy than the upstream channel sections (Figs. 11 and 14).

The two main tributaries of the Surveyor Channel are the Yakutat and Alsek Legs (Fig. 1). The Yakutat Leg is most closely associated with the Yakutat Sea Valley and related Malaspina and Hubbard Glaciers. The Alsek Leg is associated with the Alsek Sea Valley and Alsek glacial system. The channel-fill deposits of the Alsek Leg first appear at the I-II sequence boundary in the upper channel, demonstrating that at this location a leveed channel has been in place since the onset of sequence II deposition (Fig. 12); however, downstream of the Alsek Leg in the middle channel where the Alsek and Yakutat Legs have merged to form the main trunk, the channel-fill deposits first appear at the II-III boundary (Fig. 13A).

The Yakutat Leg is the wider of the two main channel legs and is the only section of the Surveyor Channel that exhibits significant historic lateral movement. The channel-fill deposits beneath the Yakutat Leg exhibit the same or greater width than the present-day channel, and migrate a maximum of 35 km to the southeast upsection (Figs. 7, 9, and 10). The migration of the Yakutat Leg occurs directly to the east of a substantial basement high (Fig. 12). The basement high underlies a large bathymetric ridge formed by a sediment wedge that has aggraded and grown to the southeast. The Yakutat Leg migration corresponds to depocenter growth observed on the two-way travel-time thickness maps for sequences II and III (Fig. 5). The main levee depocenter in sequence III (Fig. 5C) is significantly larger than the depocenter in sequence II (Fig. 5B), with growth toward the southeast. All other observed channel legs sit atop an aggradational stack of channel-fill deposits and evidence no significant lateral movement. The Icy East and West Legs (Fig. 1) are associated with partially buried sea valleys between Icy Bay and the Pamplona Spur, with some input from the Bering Trough and Glacier. The Icy East and West Legs exhibit a major change in depositional history, with typical high-amplitude channel fill capped by a turbidite drape that is laterally continuous with the surrounding reflectors below the seafloor (Fig. 9).

The upper and middle sections of the channel trend relatively straight southwest until reaching Giacomini Seamount in the Kodiak-Bowie Seamount Chain, north of which they coalesce into one major trunk channel (Figs. 1 and 2). Once the trunk channel passes through the seamounts, it turns to the northwest for the final leg to the Aleutian Trench. This lower section tracks into the center of a ∼400-km-wide, heavily faulted bathymetric low that funnels the channel down its center (Fig. 14). The Kodiak-Bowie and Patton-Murray Seamount Chains flank the bathymetric low to the north and south, respectively (Figs. 1 and 14).

The Surveyor Fan thins away from the Yakutat margin, southwest into the Patton-Murray Seamount Chain (PMSC, Fig. 6), implying a long-term connection between the fan and the Yakutat shelf-edge. The Patton-Murray Seamount Chain acts as a natural boundary between the Surveyor Fan and the older Zodiac Fan, located to the southwest of the Chain (Fig. 14). To the east, the Surveyor Fan shares a nondistinct boundary with the younger Baranof Fan defined by the interwoven over-levee deposits of the Chirikof (Surveyor Fan) and Horizon (Baranof Fan) Channel systems. This boundary, however, is not distinguishable in the seismic data used in this study. Therefore, we have approximated the boundary (Fig. 6) for the purpose of calculating new area and volume estimates for the Surveyor Fan. Boundaries as shown are supported by previous work (Stevenson and Embley, 1987). The area occupied by the Surveyor Fan is equivalent to 3.42 × 10⁵ km². The volume of the Surveyor Fan, defined by the sediment between the seafloor and the top of Pacific plate crust, is 6.8 × 10⁵ km³. This new estimate is significantly larger than previous estimates, and places the Surveyor Fan in the company of the largest fans in the world, roughly equivalent to the size of the Amazon Fan (Table 1).

The proximity of the upslope heads of the Surveyor and Chirikof Channels to the shelfal sea valleys (Fig. 1) is suggestive of a causal relationship, in which the sea valleys played a role in the formation of the channel system. Surveyor Channel tributaries may have developed once glacial ice first carved the sea valleys to the shelf edge, focusing fan sediment input to only a few shelf-edge locations. In this scenario, a climate event that pushed glacial ice to the shelf edge for the first time could have been responsible for the addition of a major channel tributary and additional reorganization of Surveyor Fan deposition.

The Alsek Leg originates at the mouth of the Alsek Sea Valley, supporting an apparent connection to the historic Alsek glacial system where the modern day Alsek River is present. However, considering the movement of the Yakutat terrane and Pacific plate to the northwest along the Fairweather fault over the past ∼5 Myr (Elliott et al., 2010; Kreemer et al., 2003), the glacial interval A sediment source to the Alsek Sea Valley area of the Yakutat shelf was likely not the St. Elias Range, but the Coast Ranges farther to the south (Fig. 15B).

The Alsek Leg of the Surveyor Channel appears to originate at the I-II sequence boundary based on the lower termination of the channel-fill deposit at that boundary in the upper channel section (Fig. 11A). This relationship could indicate that Alsek Sea Valley was the first carved across the entire shelf, and triggered channel formation in the Surveyor Fan during glacial interval A. However, we suggest that glacial interval A may not have been sufficiently severe to drive glaciers across the entire shelf. The early Alsek Leg was only present in the shelf proximal fan, and a sea valley may not have been necessary to form this relatively minor precursor to the modern Surveyor Channel (Figs. 15A and 15B). The I-II sequence boundary does not correlate to any change in lithology at DSDP 178, further evidence that glacial interval A exerted the greatest influence on the proximal Surveyor Fan. Strata across the I-II boundary do exhibit angular discordance and a large change in seismic amplitude; these characteristics may be associated with changes in sediment distribution caused by the genesis of the shelf-proximal proto–Surveyor Channel, demonstrated by the shift in depocenters from Figures 5A to 5B.

The sequence I-II boundary is interpreted based on changes in acoustic character and dated at ∼5 Ma based on correlation with an ash layer at DSDP 178 (Fig. 3). However, Glacial Interval A is dated at ∼5.5 MA based on the first appearance of ice-rafted debris at ODP 887 and in the Yakataga Formation. We regard that Glacial Interval A created the regional sequence I-II boundary sometime between ∼5.5 and 5 Ma and that the discrepancy in dating may be explained in part by only partial core recovery at DSDP 178 as well as the additional time that may have been necessary for Glacial Interval A to affect the more distal region of the Surveyor Fan.

Berger et al. (2008) showed that the Bering Glacier made its first full shelf transit in the mid-Pleistocene based on a preserved erosional unconformity beneath the present-day Bering Trough. All tributary leg channel-fill deposits except for the Alsek first appear in the sediment record near the sequence II-III boundary (Fig. 9) in the upper channel, and the Alsek Leg appears at II-III sequence boundary in all but the most shelf-proximal seismic lines (Figs. 10 and 12). We interpret this pattern to indicate that glacial interval C was the impetus for further channel growth and change of the sediment distribution network across the Surveyor Fan (Fig. 15C). The increased sedimentation associated with the MPT was quite extreme, shifting from ∼750 to 2000 mg/(cm²k.y.) in the distal fan (Rea and Snoeckx, 1995). The increased sediment flux resulted in downstream migration of the main channel legs (Figs. 15B and 15C) and pronounced aggradation and progradation of the channel levees in sequence III (Figs. 5B and 5C). This interpretation is supported by an observed change in elemental, magnetic, and mineralogical composition from more variable below the II-III sequence boundary, to more uniform above the II-III sequence boundary in the DSDP 178 core (Ullrich, 2010). The transition in composition could represent focusing of sediment pathways (Ullrich, 2010) (i.e., sea valley and Surveyor Channel formation) that occurred in conjunction with glacial intensification and potential ice-stream erosion associated with the MPT (Berger et al., 2008). We speculate based on the degree of increased sediment flux and the proximity of channel-fill origins to the II-III sequence boundary, that the majority of shelf sea valleys made their first full shelf transit as a result of glacial interval C. However, seismic evidence on the shelf to support this interpretation is limited by sparse data and glacial erosion.

Since we do not observe a regional sequence associated with glacial interval B, we suggest that sediment distribution during glacial interval B was not significantly different than that of glacial interval A. That is, the increase in sedimentation alone was not substantial enough to spur a major change in fan sediment distribution, such as that observed in conjunction with glacial interval C. This observation stands as further evidence for the necessity of shelf-edge glaciation for major channel formation in the Surveyor Fan, first occurring at the onset of glacial interval C.

The channel fill of the Icy Legs exhibits a major change in acoustic character, from high-amplitude channel-fill deposits to laterally continuous turbidite drapes in the channel bathymetric low (Fig. 9). We interpret the boundary between the two as a “shutting off” of the Icy East and West Legs. The sea valleys at the upslope heads of these legs are the Pamplona Troughs, which surround the Pamplona Spur, a tectonically controlled basement high (Carlson et al., 1982). The Pamplona Troughs are the smallest observed sea valleys on the shelf, a maximum of 20 km long and focused at the shelf edge. The small size is mostly due to slump infilling and tectonic deformation associated with the Pamplona Zone, the deformation front for Yakutat terrane subduction beneath North America (Worthington et al., 2008). In our interpretation, the glacial system associated with the Pamplona Troughs became extinct or merged with the Bering or Malaspina systems at some point after the onset of glacial interval C. Turbidites have since draped the channel legs, which were significantly large bathymetric lows to maintain the appearance of channels on the modern-day seafloor. The shutting off of the Icy Legs due to the loss of shelf-edge glaciers further supports the necessity of shelf-edge glaciers to begin and maintain the Surveyor Channel.

In an attempt to classify the Surveyor Fan, we compared its features and morphology to other well-studied fans globally. Although Surveyor Fan sediment is glacially sourced, Table 1 shows that the Surveyor Fan ranks better with fluvial fans by size than with glacial trough-mouth fans (TMFs). However, the Surveyor Fan, as described in this study, contrasts with the established TMF model. On the Norwegian, Greenland, and Antarctic margins, TMFs formed at the mouths of glacial sea valleys on the shelf as a result of glacial erosion. Most of these fans remain as single entities with occasional channels formed between and are studied as such (Dowdeswell et al., 2008; King et al., 1996; Ó Cofaigh et al., 2006; Vorren et al., 1998). The Surveyor Fan covered the Alaska Abyssal Plain before sea valleys were carved. Once the sea valleys reached the shelf edge, any resulting TMF did not retain individual identity within the Surveyor Fan, possibly a result of sea valley proximity combined with the high magnitude of sediment flux provided to the Yakutat margin by glacial interval C.

The terrigenous sediment flux into the Surveyor Fan and Channel system has been periodic due to the glacial-interglacial cycle, leaving large portions of the system essentially inactive during interglacial periods. This can also be said for other glacially influenced channel systems, such as the Inbis and Lofoten Channels on the Norwegian margin, with which the Surveyor Channel shares many characteristics. These systems are also constructional, with low sinuosity and tendency to avulse, and long-lived distributaries (Ó Cofaigh et al., 2006; Vorren et al., 1998). Differences also abound, with most of the North Sea channels being laterally restricted by submarine landslide deposits or TMFs and containing various levels of braiding in the lower channel sections. Perhaps the most significant difference is that the Surveyor Channel stretches more than twice the length of other glacial channels and empties into the Aleutian Trench.

Given enough accumulated active time, all glacial channel systems may eventually achieve morphology similar to a fluvially influenced channel system. For this to be true, sediment flux and time, but not method of sediment delivery to the shelf edge, play a critical role in the development of deep-sea morphology. Alternatively, method of sediment delivery to the shelf edge is important for deep-sea fan and channel morphology, and glacial systems are able to maintain unique fan and channel systems indefinitely. Yet another possibility, the Surveyor Channel may be entirely unique because of its drainage into the Aleutian Trench. Most seafloor channels empty onto the abyssal plain. Deposition may eventually block the main channel, triggering an avulsion upstream, and the deposition of a new lateral lobe or area of the fan. The increase in gradient and accommodation created at the end of the Surveyor Channel by the Aleutian Trench may preclude the necessity of major channel avulsion, allowing the channel to maintain the same pathway as long as it remains in contact with the sediment source and trench.

The sea valleys and troughs are a remnant of cross-shelf glacial transit, and reveal that at some point in the past, glaciers delivered sediment directly to the slope at a glacial maximum. In spite of their location and ability to bypass sediment to the slope, the glacial sea valleys may not be a direct proxy for fluvially influenced shelf canyons. During glacial maxima, glaciers can potentially reach the shelf edge and supply sediment directly to the slope like a shelf-edge river delta (e.g., Carvajal and Steel, 2009). During interglacial periods, the sea valleys have the potential to act as conduits for sediment to bypass the shelf, much like a fluvial canyon during highstand (e.g., Covault and Graham, 2010). However, in glacial systems, sediment flux is greatly reduced during interglacial periods compared to that at glacial maxima, whereas a fluvial system could have a relatively constant sediment flux from highstand to lowstand (Covault and Graham, 2010). We suggest that while sea valleys may bypass some sediment during interglacial times, overall they provide increased shelf accommodation space, as shown in the Bering Trough by Worthington et al. (2010). Maximum sediment delivery to the fan only occurs during glacial maxima due to the associated increase in glacial erosion. However, once the sea valleys were formed, glacial periods that did not achieve glacial shelf transit could utilize the sea valleys for shelf bypass of sediment, possibly analogous to a mid to highstand fluvial system. Therefore, the Surveyor system after glacial interval A has been built by periodic sediment pulses associated with glacial maxima, and has a shorter “active” life than a fluvial system with a coeval origin.

The basement high just to the west of the Yakutat Leg is collocated and possibly associated with activity at the Gulf of Alaska Shear Zone (e.g., Gulick et al., 2007) (Figs. 7 and 12). We suggest that the eastward direction of growth of the overlying sediment wedge and lateral movement of the Yakutat Leg may have been caused by continuous uplift and deformation of the basement at this location through time. We interpret the Yakutat Leg migration as an anomaly influenced by tectonics, with all other channel legs and sections exhibiting significantly less to no lateral migration since inception.

The bathymetric low that contains the majority of the lower Surveyor Channel is heavily faulted and regionally extensive (Figs. 1 and 14). We interpret this low as a northeast to southwest zone of extension flanked by basaltic edifices of the Kodiak-Bowie Seamount Chain to the north and the Patton-Murray Seamount Chain to the south. The zone of extension is a sufficient bathymetric low to divert the Surveyor Channel down its center to the west. The change in gradient as the Surveyor Channel travels from near the outer trench swell, through the zone of extension, and into the Aleutian Trench gives the lower channel section the highest axial gradient of any Surveyor Channel section (Ness and Kulm, 1973). Some channel fill is recorded in the erosional lower section of the Surveyor Channel (Figs. 11D and 14). This deposition could be evidence for a sediment pulse driven by glacial-interglacial cycling. Transport in the lower channel may only have the power to incise at peak flow conditions during glacial maxima. The channel fill in the lower section could be representative of sediment deposition between glacial maxima, whether from terrigenous sources or channel sidewall failures upstream. The current channel fill therefore could be an accumulation of fill since the last glacial event, or represent an amalgamation of interglacial deposits that were not fully eroded during glacial maxima.

(1)The I-II sequence boundary was created by the onset of glacial interval A at ∼5.5 Ma, and sequence II consists of deposits from glacial intervals A and B. The II-III sequence boundary was created by the onset of glacial interval C at ∼1 Ma, and sequence III contains deposits from glacial interval C to present day.

(2)The thickening of sequences II and III into the Yakutat terrane continental slope is evidence of the long-term connection of the Surveyor Fan to the Yakutat shelf. Due to Pacific plate and Yakutat terrane motion past North America on the Fairweather fault, Surveyor Fan provenance likely varies from southern Coast Range sources in older fan sediment to St. Elias Range in younger fan sediment.

(3)Glacial interval A reorganized fan sediment distribution by spurring Surveyor Channel genesis. The early Surveyor Channel shifted sediment depocenters to shelf-proximal areas at the base of the Yakutat slope. Glacial interval C built onto this, and by pushing glaciers to the shelf edge and increasing the sediment flux to unprecedented levels, extended the Surveyor Channel across the Alaskan Abyssal Plain.

(4)Correlation of glacial interval C and the MPT to the II-III sequence boundary supports the Berger et al. (2008) hypothesis that the MPT was a threshold where climate, compared to exhumation, started dominating erosion in the St. Elias orogen via glacial intensification.

(5)The Surveyor Channel system is a unique deep-water sediment delivery pathway because of its glacial source and trench terminus, both of which may contribute to the Surveyor's long-term ability to maintain a major channel and evade avulsion over long periods of time.

(6)Glacially carved cross-shelf sea valleys are not direct proxies for fluvial canyons, but do share some characteristics. During interglacial periods, the valleys serve mostly to accommodate sediment rather than bypass the continental shelf due to a greatly decreased sediment flux. During glacial maxima, deposition by shelf-edge glaciers may be comparable to that of a shelf-edge river delta, albeit with more efficient sediment transport. Sea valleys may also be analogous to fluvial canyons during glacial periods that exhibit a limited shelf transit but still provide increased erosion and sediment flux.

(7)A NE-SW zone of extension between the Kodiak-Bowie and Patton-Murray Seamount Chains creates a regional bathymetric low that funnels the lower Surveyor Channel to the Aleutian Trench.

The authors thank the crew and science party of R/V Marcus G. Langseth cruise MGL0814 for acquisition of the 2008 MCS data set, and the USGS for assistance with open-source seismic data. This manuscript benefitted significantly from discussion with John Jaeger, Ron Steel, David Mohrig, and Julia Schneider and reviews by Roland von Huene and Andy Stevenson. This project was funded by NSF grant EAR-0408584 to The University of Texas at Austin. Reece received partial support from the University of Texas Institute for Geophysics (UTIG) Ewing-Worzel Fellowship, Consortium for Ocean Leadership grant SA8-03, and a Marathon Oil Company Fellowship. This is UTIG Contribution #2346.