INTRODUCTION

Some time ago, the authors realized that their combined unpublished marine seismic data covered large areas in the central and eastern Gulf of Alaska not covered by previously published data. These data show some unreported features and confirm or modify some tectonic zones inferred from the theory of plate tectonics. In addition to our own data, other seismic and magnetic data were available from NAVOCEANO, NOAA, and DSDP. A compilation

End_Page 273------------------------

of the seismic reflection records was used by Silver et al (1974) for a study of the Kodiak-Bowie seamount chain, but did not cover the continental margins which will be discussed here. We will relate regional late Cenozoic tectonic features to a more detailed plate-tectonic model than presented in previous publications. The simplicity of the general global plate-tectonic model diminishes rapidly as the scales of plate diagrams are expanded. The scale of Atwater's pioneering plate diagram (1970) and the Gulf of Alaska diagram of Richter and Matson (1971) differ by almost an order of magnitude. A simplified sketch (Fig. 1) based on these diagrams and one by Rogers (1977) indicates three types of plate boundaries: (1) the eastern transform boundary, (2) the western convergent boundary, nd (3) the oblique convergent boundary between them. Richter and Matson speculated that the central oblique convergent boundary may involve a 300-km-wide section of the continent, taking up relative motion of the Pacific and North American plates in a series of concentric tectonic zones.

DEVELOPMENT OF THE PLATE-TECTONIC MODEL

Twenty years ago, St. Amand (1957) proposed an offshore fault along the coast of southeastern Alaska connecting the Queen Charlotte Islands Fault on the south and the Fairweather Fault on the north. Both these faults are easily recognized from their strong physiographic expression, and because both have been the sites of great earthquakes (Sykes, 1971), the continuity was appealing despite the lack of marine geologic data. St. Amand drew the fault just off the straight trace of the rugged outer coast of Baranof and the Chichagof Islands and through the single volcano along it, Mt. Edgecumbe near Sitka.

The north Pacific plate-tectonic model required a transform fault to separate the Pacific and North American plates between the spreading ridges off British Columbia and the Aleutian Trench. Concepts at the time made it more appealing to infer the extension of the Queen Charlotte Islands Fault along a morphologic boundary of greater significance, the linear continental slope, since plate boundaries were commonly thought to correspond with continental slopes. But this principle had to be violated in order to connect the northern end of the inferred fault with a line drawn across the continental shelf to the end of the Fairweather Fault. Data collected during this study indicate such a fault across the shelf near Cross Sound. Figure 2 summarizes the present tectonic knowledge of the sou heastern Alaska continental margins (including data presented here) and also shows the distribution of some aftershocks accompanying great earthquakes. The zone defined by relocated epicenters of aftershocks of the 1949 Queen Charlotte Islands earthquake (Tobin and Sykes, 1968) narrows the possible location of the extension of the Queen Charlotte Islands Fault, and

Fig. 1. Diagram of late Cenozoic tectonic plate boundaries, Gulf of Alaska, principally after Atwater (1970), Richter and Matson (1971), and Rogers (1977). The Denali, Totschund, Fairweather, Chatham Strait, Chichagof-Baranof, and Queen Charlotte Islands faults are indicated by D, T, F, CS, CB, & QCI, respectively. Location of DSDP site 178 is indicated by a dot.

End_Page 274------------------------

first-motion studies confirm its right-lateral strikeslip nature. Additional constraint in the location of a fault off Baranof Island was provided by the aftershocks of the 1972 Sitka earthquake (Page, 1973). The northern end of this aftershock zone was shown to correspond to an offshore fault just north of Sitka (Fig. 2). This fault is traced farther north on a series of seismic reflection records reported here.

The coastal onshore geology of the central Gulf of Alaska has been discussed (Stoneley, 1967; Plafker, 1967 and 1971), but offshore geology was poorly known until the preliminary findings of a study on

Fig. 2. Major faults and aftershocks from three major earthquakes, southeast Alaska and British Columbia, after Beikman (1975), Tobin and Sykes (1968, including only most accurately located epicenters), and Page (1973). Tracklines show location of seismic records used in this study: solid lines locate records shown in Figure 3, with K series and C from SIO, numbered series and D from NOAA, line A reinterpreted from von Huene et al (1972); the dashed track lines are from institutions as noted, and show part of the network of records used to follow the Pliocene-Pleistocene boundary from DSDP site 178 to Baranof Fan, and Miocene-Pliocene boundary to record D. Numbers along refraction lines indicate locations of stations given in Table 1. Hachured area shows location of attenuated magneti anomalies. Bathymetry after Chase et al (1970).

End_Page 275------------------------

the shelf by Bruns and Plafker (1975) were reported. Our data in this area are scanty but sufficient to establish the continuation of a pattern of deformed and undeformed segments on the shelf to the continental slope. Recently Rogers (1977) has published sketch maps suggesting a similar segmentation.

In this discussion zones of tectonism are proposed that may correspond to eastern plate boundaries and their tectonic style is outlined. The structure at the junction of the transform and the seaward edge of the oblique plate boundary is then dealt with and finally the structure along the offshore part of the oblique boundary is discussed. In the summary discussion, observations and the plate model for late Cenozoic time are integrated and some consequences of the model for earlier periods of geologic history are pointed out.

FAULT ZONE BETWEEN FAIRWEATHER AND QUEEN CHARLOTTE ISLANDS FAULTS

Major transform faults can look relatively unimpressive in seismic reflection records (for example, the San Andreas fault just off the Golden Gate). The records in Figure 3 are probably sufficient to identify major faults in sediments of the continental slope and rise, but these records are insufficient to show tectonic features beneath the continental slope. It is generally difficult to obtain clear subsurface information on steep rugged slopes, and these records were made with relatively low energy sources such as 90- to 160-kJ sparker or 40-in³ airguns. In addition, the high vertical exaggeration at which they were recorded diminishes the maximum dip that can be resolved. Therefore, the subsurface structure of the slope is poorly defined, and it becomes a place to infer hidden fault when one cannot be found elsewhere.

The continental rise between the Queen Charlotte Islands and Cross Sound is a series of coalescing deep-sea fans collectively called the Baranof Fan. Most of the rise is of Pleistocene age (1.8 MY) based on the Pliocene-Pleistocene reflector, which can be traced through a network of seismic reflection records (partly shown, Fig. 2) from DSDP site 178 (noted in Fig. 3, lines 4, 5, 7, and 9). We have a ± 0.1 sec (about 80 m) level of confidence in the

Fig. 3. Line drawings of seismic records, Cross Sound to Queen Charlotte Island. Vertical exaggeration normalized to approximately 15:1. Location of ships tracks shown in Figure 2. P is location of Pliocene-Pleistocene boundary traced from DSDP site 178.

End_Page 276------------------------

position of the seismic reflector inferred to be equivalent to the 1.8 MY time line, which is the maximum error of closure about multiple paths in the network of seismic lines. The volume of Baranof Fan is unquestionably great; the maximum measured thickness is at least 3 km (assumed velocity = 2 km/sec), and individual turbidity current channels are 10 km or more wide and 1,000 km or more long (Figs. 2 and 3, lines K-4, 5, 7, Hamilton, 1967). The origin of this great volume of material is probably the glaciated terrain of southeast Alaska. Off the Queen Charlotte Islands, the Pleistocene fan is much thinner, because sediment was intercepted by a large trough behind the islands on the continental shelf (Shouldice, 1971). Therefore, at the base of the slope off the Queen Charlotte Isla ds, the glacially derived sediment is thin in contrast to the thick accumulations along the margin north of Dixon Entrance (Chase and Tiffin, 1972).

At the base of the continental slope north of Dixon Entrance, six of the eight records in Figure 3 (4, K-6, 5, 6, 7, 9) show a filled trough. Absence of the trough in two of the records may be due to insufficient penetration to reach the base of the sedimentary section and igneous basement. The trough is formed by depression of the oceanic crust, and it is not only shaped like a trench but has dimensions similar to those of the Aleutian Trench. Such a trough is not conceptually associated with transform plate boundaries. Chase and Tiffin (1972) imply local tectonism seaward of the continental slope off the Queen Charlotte Islands, possibly associated with the development of the trough. They show a fault in the rise west of Dixon Entrance (Chase and Tiffin, 1972, Profile A, Km 12, Fig. 7) but we found no faults in our nearby records (K-4 and 7, Fig. 3). To the north, between Dixon Entrance and Cross Sound, our records show no major faults in Pleistocene sediments of Baranof Fan. However, the faults generally found on the landward side of such a buried trough are not seen because our records contain little subsurface information at the head of Baranof Fan and under the continental slope. Although the origin of this trough, and the volcano, Mt. Edgecumbe, are puzzling in terms of a plate-tectonic model, this is not the first report of a trough along a proposed transform plate boundary (for example, western Aleutian Trench, Puerto Rico Trench).

Tectonism along the continental slope is largely inferred from physiography. Despite the great amount of sediment that must have passed across the slope north of Dixon Entrance in Pleistocene time, an uncovered irregular surface is seen in records across the slope (lines K-4, 6 and 7, Fig. 3), suggesting that the topography has a tectonic origin. The rate of vertical tectonic displacement may have exceeded the rate of glacial sedimentation, or most of the vertical relief may be of post-glacial age, when the slope was sediment-starved. A tectonic origin is proposed because farther south, off the Queen Charlotte Islands, such features are faults, as indicated by more complete bathymetric and seismic data (Southerland-Brown, 1968; Chase and Tiffin, 1972), whereas the seismic lines off Pr nce of Wales Island make this point with less certainty. It is possible that the fault-controlled topography on the continental slope off the Queen Charlotte Islands extends north of Dixon Entrance to the area off Chatham Strait, approximately where the Chatham Strait fault is projected onto the continental slope.

North of Chatham Strait, the topography of the slope is relatively smooth in records running downslope. Records parallel to the slope show numerous small canyons; however, no major ridges and troughs parallel to the slope are indicated. Therefore, if faulting occurs beneath the continental slope, displacement with major surface expression has been masked by Pleistocene sediment, or relief from postglacial deformation has not been as pronounced as to the south. Furthermore, the earthquake epicenters recorded by a local network of seismometers show tectonic activity at the edge of the shelf, but not the foot of the slope (Fig. 2).

The earlier data combined with those reported here are consistent with the fault inferred previously along southeast Alaska. A fault zone has been established with reasonable certainty by previous studies along the continental slope off the Queen Charlotte Islands (Southerland-Brown, 1968; Tobin and Sykes 1968; Chase and Tiffin, 1972) and on the outer continental shelf off Baranof Island (Page, 1973). But the continuity of faulting between southern Baranof Island and Dixon Entrance is uncertain because of insufficient data. The main evidence for faults is good physiographic expression as shown in our records. As this stretch is about 130 km of a 1,050-km-long active offshore fault system, the Queen Charlotte Islands Fault is here inferred to continue along the slope of Chatham Strait nd to join with the fault outlined by the aftershocks from the 1972 Sitka earthquake. The Chatham Strait Fault, a prominent tectonic boundary, appears to separate an area to the south where faulting is now dominant along the slope from an area to the north where faulting is now dominant along the edge of the shelf.

STRUCTURE AT JUNCTURE OF TRANSFORM AND OBLIQUE CONVERGENT PLATE BOUNDARIES

Early in the study, an opportunity was recognized to define structure across the oblique convergent plate boundary at its juncture with the transform plate boundary. A projection of the oblique boundary intersects the transform boundary on the continental shelf just south of Cross Sound (Fig. 1). Thus, the zone of oblique convergence, which generally corresponds with the continental slope, might be studied on the shelf, where conditions are much more favorable for resolution with the reflection technique.

This line of investigation involved first acquiring seismic reflection lines across the shelf to define locations of faults corresponding to the transform boundary. Then the zone of convergence was examined seaward of the transform with reflection lines

End_Page 277------------------------

across its presumed tectonic fabric (see Fig. 1).

Reflection records that cross the shelf between Sitka and Cross Sound (Figs. 2-5) all show faults. In addition, the records confirm that some sharp bathymetric linear features, defined by soundings and suspected to be fault scarps, are faults with surface scarps in reflection records (Figs. 4 and 5). The fault scarps occur only off Chichagof Island and in Cross Sound, and they may not be apparent to the south because the heads of canyons have destroyed bathymetric continuity. From these data we interpret a continuous fault zone between Sitka and the Fairweather Fault (Fig. 2). The northern 75 km of the fault are well located, with the exception of a break in Cross Strath where post-glacial sedimentation partly masks it. The fault zone appears to offset the southeast bank of Cross Stra h right-laterally along one or perhaps two faults (Fig. 4). The offset is estimated to be 400 ± 150 m along the well-defined offset and 300 ± 150 m along the questionable one. This bank has been eroded deeply (Fig. 4F), probably by a glacier that occupied Cross Strath and formed the moraine at its seaward end (Fig. 4E). Although the measurement of the lateral component of faulting cannot be clearly separated from vertical or erosional effects, the possible amount of offset along a glacial feature indicates rapid tectonism. The local fault scarp in the thick post-glacial sediment of Cross Strath also indicates a high rate of slip (Fig. 5). The fault has been named the Chichagof-Baranof

Fig. 4. Location of selected seismic records along and across the northern Chichagof-Baranof fault, Cross Sound area. Lettered lines are illustrated in Figure 5. Seismic records northwest of Cross Strath are from Molnia et al (1978). Inset shows data in area of offset glacial bank. Dots are location of soundings from USC & GS H-4529.

End_Page 278------------------------

Fault (CB fault) (Beikman, 1975).

The most obvious trace of the CB Fault cannot be followed directly into the Fairweather Fault. A projection of the surface scarps from where the fault is last recorded, about 4 km off Icy Point, intercepts land about 2 km west of the mapped Fairweather Fault (Fig. 4). However, this mismatch may be more apparent than real, because the width of the fault zones is not shown in Fig. 4. The CB Fault is generally seen as two faults with local subsurface deformation in a zone 1.5 km to perhaps 8 km wide (Fig. 4). Similarly, the mapped traces of the Fairweather Fault and the thrust fault offshore of Icy Point (Plafker, 1967) are 2 km apart. Therefore the area of juncture is tectonically more complex and broader than indicated by the presently mapped fault traces. The strike of the Fairweather Fault is in line with the strike of the Peril Strait Fault, and since the inferred connection (Beikman, 1975) crosses a rough glaciated ocean floor, it can only be shown convincingly by a detailed survey. The juncture of the CB Fault and the Fairweather-Peril Strait Fault is also in an area of rough ocean floor and is probably difficult to follow. The CB Fault could also be related to the large offshore thrust fault subparallel to the Fairweather (Plafker, 1967). Such a relation is suggested by the physiographic lineaments just west of the CB Fault (Fig. 4).

Although the relation between the CB and Fairweather Faults is unclear, there is no reason to believe that they are not part of the same tectonic system. Both are tectonically active, with recorded seismicity and strong surface expression. The displacement of glacial features on the Fairweather (Plafker, 1976) is similar to the displacement of a glacially formed submarine bank along the CB Fault. No other faults on land are known to have Holocene displacements of this magnitude, nor have any other faults of similar continuity been traced on the continental shelf. Post-glacial displacement on the CB Fault is perhaps as great as 4 to 5 cm/yr if one assumes that the southeast bank of Cross Strath ceased being eroded by glaciers 10,000 years ago. This is about the same amount as the propo ed relative motion of the Pacific and North American plates. For all these reasons we interpret the CB Fault to be part of the present transform boundary between the Pacific and North American plates.

The central Gulf of Alaska continental slope and the corresponding oblique convergent plate boundary projects across the seaward end of Cross Strath, and it should meet the CB Fault off Chichagof Island (Fig. 4). Here the continental slope changes trend through 40°, expressing the juncture of the oblique convergent and transform margins of the plate model, but no strong morphologic evidence of plate convergence is seen. Two reflection records which parallel the CB Fault where the plate boundaries

Fig. 5. Tracings of selected seismic reflection records along the Chichagof-Baranof fault and across the continental margin off Cross Sound. CB fault indicated by CB. All lines keyed to Figure 4 except I, which is off Sitka (line 4, Figure 2).

End_Page 279------------------------

meet, one of which is shown in Fig. 5 (line F), also show no strong subsurface convergent deformation. The records look much the same as records elsewhere across the shelf off Chichagof Island. Therefore, if subduction is occurring across the obliquely convergent plate boundary it is either occurring at a slow rate that does not deform upper (0.5 sec) reflecting horizons, or deformation of those sediments is concentrated close to the base of the continental slope.

An alternate explanation is that an undetected transform fault paralleling the CB Fault runs beneath the rugged continental slope from the Queen Charlotte Islands Fault to the area off Cross Sound and joins with the oblique convergent zone. Such a fault zone is queried in Figure 4 and a similar fault has been shown by Plafker et al (1975). The unusual position of the CB Fault at the edge of the shelf rather than in a more conventional position under the slope lends support for the hidden fault. A reflection record across a smoother part of the continental slope shows unfaulted sediments below the sea floor and possible deformation about 430 m (0.5 sec at 1.7 km/sec) below this (arrow, line E, Fig. 5). The possible age of the unfaulted section is suggested by analogy with a similar set ing off Kodiak Island at DSDP site 182, where early Pleistocene sediment of the downslope apron was recovered from 212 m depth (Kulm et al, 1973).

Because compressive deformation does not affect the upper beds across the oblique convergent plate boundary at its juncture with the transform boundary, the rate of deformation is inferred to be too slow to show through Pleistocene sedimentation. We do not propose inactivity, because some oblique convergence further along the boundary off Fairweather Ground is indicated by seismicity (Gawthrop, et al, 1973), and recent local convergent structure elsewhere along the boundary is described in a following section.

CRUSTAL MODELS ACROSS THE QUEEN CHARLOTTE-FAIRWEATHER TRANSFORM

Transform plate boundaries are generally modeled as simple vertical faults where the plates slide laterally past one another without significant accretion or consumption. Yet previously constructed crustal models based on deep geophysical data show the Queen Charlotte Islands fault zone as a wide boundary with an associated trench. The first crustal model through Dixon Entrance by Shor (1962) emphasizes the buried trough at the foot of the slope. Shor did not have sufficient data to continue velocity units across the transition between oceanic and continental crust. Couch, (in Dehlinger et al, 1970) modeled a gravity transect using the constraints from Shor's work, and in his interpretation the continental-oceanic boundary dips seaward, which suggests a wide zone of faulting across th margin. The crust-mantle boundary in Dixon Entrance was later refined by Johnson et al (1972) on the basis of new refraction information along the coastal mountain and fiord area. From these models it is uncertain if the Queen Charlotte Islands Fault is a simple vertical fault separating oceanic and continental crust.

New information has become available since the above-mentioned crustal models were constructed, and we constructed a series of models using data along line 7 (Fig. 2) where gravity, magnetic, and seismic reflection data were measured simultaneously. The line of new data crosses the slope where topography is relatively smooth, and thus the gravity data (made with a stable platform rather than gimbal instrument and with satellite navigation) are correspondingly smoother than Couch's data in Dixon Entrance. The analysis of bulk densities from DSDP Leg 18 cores (Kulm et al, 1973) has provided a general density gradient in upper Tertiary sediments of the Gulf of Alaska.

The two models in Figure 6 illustrate a vertical and a 45° landward-dipping continental-oceanic crustal boundary, thereby showing a possible latitude of interpretation within the constraints of the geophysical data. Both models indicate that the trough at the foot of the slope seen in seismic records extends 20 km or more under the slope and that it may contain a sediment section from 5 km to 8 km deep. The landward end of the trough is suggested by the inflection in observed gravity values and by a magnetic anomaly peak. The continental-oceanic boundary is not a simple break, but probably a zone of faults, as is shown by the topography off the Queen Charlotte Islands. The sediment-starved margin off the Queen Charlotte Islands has thinner sediment than must exist off the Dixon En rance because the calculated gravity values off Dixon Entrance are difficult to match to an inflection in the observed values without including a thick block with sediment densities at the edge of the shelf (km 120-140, Fig. 6). Therefore, steps in the continental slope could be obscured by sediment north of the Queen Charlotte Islands.

From two-dimensional gravity modeling of the continental-oceanic boundary, with relatively good seismic data on either side, we conclude that a 20- to 30-km-wide tectonic zone separates the trough off Dixon Entrance from the crystalline continental crust. If a single feature in the zone is the dominant boundary, its dip is unknown. This general structure

Fig. 6. See caption on page 281.

End_Page 280------------------------

Fig. 6. Two-dimensional gravity models through Dixon Entrance and across the continental margin along line 7, Figure 2. Landward values of gravity (dashed line) projected from Couch (in Dehlinger et al, 1970). In model A, vertical boundary between oceanic and continental crust is assumed; in model B, 45° boundary. Seismic refraction data from Shor (1962), and at km 122 from Milne (1964). Arrow indicates landward limit of seismic reflection control on oceanic basement from line 7 (Fig. 3). Refraction velocities in km/sec are indicated in italic type, assumed densities in g/cm³ indicated in roman type. Gravity measured with a LaCoste-Romberg stable platform instrument; positions controlled by navigional satellite. Observed gravity shown by line, computed gravity by dots.

End_Page 281------------------------

is more like that of a trench boundary than a transform boundary. If one accepts the model of Grow and Atwater (1970), it was a trench 30 million years ago, but since that time a large amount of transform motion is thought to have occurred and thus the traces of an older trench could be found far to the north.

OBLIQUE CONVERGENT PLATE BOUNDARY

Early plate-tectonic models were not explicit about the oblique convergent plate boundary between Cross Sound and the Aleutian Trench; however, since then, studies have outlined the major structure along the continental shelf there. The structure of the continental shelf has been described by Bruns and Plafker in publications preliminary to a more complete summary discussion (Bruns and Plafker, 1975, 1976). They establish a significantly different structural style on the shelf between the oblique folded area west of and the less complex folded area east of Icy Bay. Rogers (1977) developed this point further, relying more heavily on data from the western Gulf of Alaska. He proposes that the convergent deformation of the slope and trench off Kodiak Island continues into the continent be ween Kayak Island and Icy Bay as the oblique-trending folds of Bruns and Plafker, which merge with onshore splays from the Fairweather Fault. Rogers also believes that in the oblique convergent zone on the shelf, folding has developed sequentially from earlier at Kayak Island to later off Icy Bay, or generally from west to east. He contrasts this with the less complexly folded shelf area between Icy Bay and Cross Sound, in which Bruns and Plafker (1975) describe a broad basin flanked by a shelf-edge structural high (Fairweather Ground). The explanation given by Rogers for this structural pattern is that the continental shelf is moving more slowly than, but in the same direction as the Pacific plate, converging with the continent along the oblique fold belt from Kayak Island to Icy Bay. T e data from the continental slope in the eastern Gulf of Alaska and the floor of the Gulf of Alaska, presented here led us independently to a similar interpretation. We have both seismic refraction and reflection data bearing on this problem.

Seismic refraction stations were established off Icy Bay and Dry Bay by the Scripps Institution of Oceanography, on Leapfrog Expedition in 1961, and preliminary results were presented by Shor (1965); results presented here deviate somewhat from that presentation because of re-analysis taking into account reflection as well as refraction data. Field methods were generally as discussed by Shor (1963); corrections and structural interpretation followed the methods given by Ewing (1963), which require the assumption that layers are plane and of constant velocity. Cases where data obviously deviate from these assumptions are noted below.

Stations LF27 and LF28 (Figs. 2, 8) yield a "compound profile," with short split profiles across each receiving position and a reversed profile between the two stations. LF27 was received on the fan at the foot of Pamplona Seavalley. The reversed profile was shot up the seavalley, and across the outer part of the continental shelf. LF28 was received on the shelf, and the short run landward was shot to the entrance of Icy Bay. As there is more than 2 km difference in water depth between LF27 and LF28, and the slope is far from smooth, the plane-layer assumption is suspect here; without additional field data, however, it is the best that can be done.

Refracted arrivals from the shallowest sedimentary layer were observed on both runs for station 28. Scatter from a straight line suggests that there may be lateral variations in velocity, but that the value is at least 1.66 km/sec. A second sedimentary layer, with velocity near 3.0 km/sec, is well observed on both runs on LF28; strong second arrivals over a considerable range of distance are evidence that it is a discrete layer with a sharp interface, rather than representing a continuous increase of velocity with depth in the sediments. No arrivals from the upper of these sedimentary layers were observed at LF27 and only one arrival from the lower layer, so the same velocities were assumed here as observed in LF28. The next deeper layer, with velocity near 3.9 km/sec, and a fourth la er with velocity 4.7 km/sec may well be one continuous formation with a velocity gradient. The major volume of the continental slope and rise is comprised of this material, the lower part of which appears on reflection line B (just southeast of LF27) to be folded sediments. The plane-layer solution becomes a bit strained here, where literal application results in a solution with negative thickness of the 4.7 km/sec layer at LF27. As layer thicknesses beneath receiving stations are always extrapolations (due to ray offset), having a negative thickness at one end of a reversed profile does not reflect a physically impossible solution. The deepest observed layer has a velocity in excess of 6 km/sec; the value of the velocity is so dependent upon the large topographic corrections and upon th plane-layer assumption that no particular importance should be given to the exact value. Arrivals from this layer received at station 27 were all from shots fired on the shelf; arrivals from the same layer received at LF28 (on the shelf) were all from shots fired on the slope, and in both cases the topographic corrections were so large that the value of the calculated velocity can be changed significantly if one changes the assumption as to which interface represents the surface topography. The depth to the 6-km/sec layer at LF27 is very nearly the same as that to the oceanic crust under stations to the southwest under the Alaskan abyssal plain, which have velocities close to 6.7 km/sec. If one were to assume that this is the true velocity, the travel times could be satisfied by refract ons from a layer that is horizontal beneath the continental slope, about 5.5 km below sea level, and then dips down steeply towards land beneath the shelf as shown in Figure 8 by a dashed boundary. LF27-28 crosses the shelf where structure is simple, and the results are generally consistent with sonobuoy station data taken

End_Page 282------------------------

later along strike (Bayer et al, 1977). The layers with velocity 3.9 to 4.7 km/sec are anomalously thick and of low velocity for continental crust; the velocity structure is similar to the convergent margin structure along the Aleutian Trench at Kodiak (Shor and von Huene, 1972).

A series of end-to-end ("leapfrog") profiles were shot offshore from the portion of the coast between Yakutat Bay and Cross Sound. The first pair, LF29-LF30, was shot at the foot of the continental slope from Yakutat Bay to Dry Bay; the three reversed profiles from LF30 to LF33 were perpendicular to the slope off Dry Bay; the final reversed pair, LF33-LF34 was again parallel to the slope from Dry Bay to Cross Sound.

On the reversed pair LF29-LF30, refracted arrivals were observed from sediments of relatively high velocity at the seafloor (fan deposits?) on LF29, and from a deeper sediment layer with velocity 3.6 km/sec on both runs. A deeper layer, with velocity of 5.6 km/sec was observed on both shooting runs, with a linear refraction travel-time plot. The layer with velocity of 5.6 km/sec may be either oceanic basement or high-velocity folded sediments similar to those seen on LF27-28 and on data from stations on the landward wall of the Aleutian Trench near Kodiak (Shor and von Huene, 1972). Curiously, this layer, which appears clearly on both LF29 and LF30, at the base of the continental slope, cannot be identified clearly on profiles from the contiguous station pair LF30-LF31 normal to the s ope. The oceanic crust, with velocity slightly high at 7.1 km/sec, is well determined on both runs of the reversed pair. Mantle arrivals are observed on numerous shots on both runs, but an anomalously high mantle velocity of 8.8 km/sec may be an artifact created by irregularities in the thickness of the sedimentary layers where the shooting run comes close to the foot of the continental slope near LF30.

Stations LF30, LF31, LF32, and LF33 in combination yield three short reversed pairs perpendicular to the slope, and pair LF33-34 is parallel to the coast to the west; all show consistent results. High-velocity sediment arrivals from the seafloor are seen on a few shots on the first profile; thereafter no seafloor refractions are seen, and it is probable that the composition of the seafloor sediments changes sufficiently that they can be more closely approximate by a lower velocity (2.15 km/sec or less), previously determined by wide-angle reflection studies in the Alaskan abyssal plain to the southwest (Shor, 1962). Strong refracted arrivals were received on all runs from a higher velocity sedimentary layer about one km below the seafloor. The top of this layer corresponds approximate y to the base of the Pliocene on reflection line D (Fig. 7). Travel-time plots on all stations from LF30 to LF34 break over directly from the sediment arrivals to refracted arrivals from the oceanic crust, with velocity near 6.8 km/sec. Strong reflections are observed from the apparent base of the sedimentary layer on profiles LF31 to LF34, and the arrivals refracted from oceanic crust are generally a little later than they should be if they were coming from this reflector. A few second arrivals with velocity 5.0 km/sec are observed on LF30-out. Basement reflections on LF30 are considerably shallower than on LF31 to LF34 and disappear at short range before the basement-refracted arrivals appear; we suspect that the basement deepens abruptly just offshore of the receiving position for LF3 and so have not tried to reconcile the solutions for the two reversed profiles that join at this point. Solutions for data from all stations from LF30 to LF34 have been calculated on the assumption that a masked basement layer is present, and that the travel-time curve is tangent to basement reflections observed on LF31 to LF34. Where no refracted arrivals were observed, a velocity of 5.55 km/sec was assumed. Results are shown in Figure 8 and Table I. This basement layer is considerably thinner than normal. Mantle refractions may be present on a few of the farthest shots on each of the lines between LF30 and LF33, but were not observed over a sufficient distance to be considered reliable for determination of velocity or depth. Good mantle arrivals are detected on the longer reversed pro ile parallel to the coast, LF33-34, and show mantle at nearly normal depth.

The general appearance of the section in Figure 8 is similar to ones across a trench margin, from the outer swell to the trench floor. LF29-30, at the foot of the slope, would correspond to a line in a similar position on the inner wall of a trench, except for the presence of the thick layers of fan sediments.

Tectonic structure of the continental slope is shown in seismic reflection records across the eastern, central, and western parts of the margin (Figs. 2, 4, and 7). The eastern and western records show a rather typical Pacific continental margin with a mildly deformed slope beginning at a shelf-edge structural high that merges with a downslope sediment apron and ends at a current-structured sediment wedge of the ocean basin. The continental slopes are deformed along their lower parts (at the arrows, lines A and C, Fig. 7). Uplifts on shelf-edge structural highs are recorded by erosion at the highs, and also by tilting of reflecting horizons in the shelf basins. At the foot of the slope, in record A, is a broad channel that probably developed from turbidity currents associated with a t ough from the Bering Glacier. A buried turbidity-current channel is seen at the foot of the slope in record C.

Between these records is one that indicates much less tectonic deformation in the upper sediments of the slope (B, Fig. 7). The upper continental slope is a thick wedge of sediment without folds or faults that looks much like a passive continental margin. This wedge is at least 2.5 to 3 km thick along the shelf edge, thinning to 0.2 km near the foot of the slope. The Yakataga Formation of early Miocene to Pleistocene age is about 4.3 km thick directly onshore from this line, suggesting that the wedge is at least of Miocene age (Plafker et al, 1975). The wedge is underlain by a highly diffractive ridge with a rough upper surface, and, as shown on nearby refraction profiles LF27-28, has seismic velocities that

End_Page 283------------------------

Fig. 7. Tracings of seismic records across oblique convergent margin, central Gulf of Alaska. Line A from yon Huene et al (1972); Line B, von Huene et al (1975); line C, SIO; line D, NOAA. Arrow indicates deformed area at foot of slope. Depth sections constructed using velocity curve derived from refraction stations. Projection of refraction stations indicated along line D.

Table 1. Refraction Station Data

End_Page 284------------------------

are typical of consolidated (probably deformed) sediment. From this record, it appears either that the earlier period of intense tectonism has been followed by a present period of reduced compressional tectonism or that any present deformation is confined to a ridge at the foot of the slope.

Evidence from the adjacent deep ocean floor also suggests a period of reduced tectonism. In this area the deep ocean basin has magnetic anomalies that are highly attenuated (Fig. 2; Naugler and Wageman, 1973; Taylor and O'Neill, 1974). One cause of anomaly attenuation is depression of oceanic crust to greater than normal depth, as shown in refraction lines LF30-34 where the crust is depressed beneath a thick wedge of sediment. This sediment wedge is also seen in a seismic reflection line west of refraction station LF29 (D, Figs. 2 and 7) but basement is not seen in the thickest part of the wedge. In the time section the body is not as clearly wedge-shaped as in the refraction data, but a depth section indicates greater wedging than in the overlying Pliocene and Pleistocene continental rise deposits. Perhaps a basement irregularity modifies the shape of the body as suggested in Figure 7. Part of the wedge might be an older buried deep-sea fan or a filled trough.

The age of the buried fan or trough can be estimated by tracing reflections for 275 km from the age boundaries at DSDP site 178 along a seismic record made by the Glomar Challenger, and then along the refraction line shown in Figure 8. The original age boundaries have recently been revised on the basis of radiometric age determinations on ash layers (Hogan et al, 1978). The Pliocene-Pleistocene boundary at site 178 can be followed easily, but the revised Miocene-Pliocene boundary can be traced with less confidence (possibly ± 200 m) because of major interruptions by four seamounts. At the thickest part of the wedge, the pre-Pliocene section may be 3 km thick. The magnetic anomalies under the wedge are between numbers 7 and 13 (Naugler and Wageman, 1973), or 27 to 38 MY old in the age chronology of Heirtzler et al (1968); hence it could be as old as Oligocene. The position of this pre-Pliocene fan suggests that there has been less convergence across the slope than the net convergence between the Pacific and North American plates in the late Cenozoic, because at that rate an old fan would have soon been subducted. The convergent vector across a plate boundary trending parallel to the slope would be about 3 cm/yr, if one assumes a constant relative motion of 5 cm/yr between the Pacific and North American plates at this latitude.

The areas of greater and lesser deformation on the shelf seem consistent with deformation along the slope in records A and B (Fig. 7) and E (Fig. 5), but in record C (Fig. 7) the deformation along the shelf edge (Fairweather Ground) and the slope seem inconsistent with the adjacent less-deformed shelf. Our records show shelf-edge structural highs that are much broader than the structures reported on the inner shelf (von Huene et al, 1971). Bruns and Plafker (1975) distinguish two types of structure in the more deformed part of the shelf, the broad shelf-edge structures paralleling the regional physiographic trend and linear folds and thrust faults that strike oblique to the regional trend. The latter are found only in the more deformed area between Kayak Island and Icy Bay. Rogers (19 7) implies that it is the oblique-trending folds paralleling the Aleutian Trench that are genetically related to the

Fig. 8. Seismic refraction stations, across oblique convergent margin from Icy Bay across the shelf, and seaward of Yakutat Bay. Location of stations shown in Figure 2. On stations LF27-28, the dotted upper boundary is the true seafloor along the shooting line; the first solid boundary is the datum plane assumed for computation. The mode of topographic correction used involves the assumption that the first layer thickness varies linearly along the profile, and that all of the irregularity lies in the second layer. The second dotted boundary is, then, the base of the first layer of sediment calculated on this assumption. The deepest dashed boundary on LF27-28 is one that could produce the observed travel times if one assumes that the crust velocity is 6.8 km/sec.

End_Page 285------------------------

regional grain of the Kodiak shelf. Therefore, the distinction between areas of different structural style on the shelf is made on the basis of the oblique rather than the shelf-edge structures. Our data are too widely spaced to define oblique structure on the slope, of which Pamplona Ridge may be an example; however, they show a discontinuity of the shelf-edge structure. The significance of this discontinuity in relation to convergence is obscure. Although it has been proposed that such structure is a result of slow convergence (Seely and Dickinson, this volume), discontinuity of the shelf-edge structure is also reported along the adjacent Kodiak margin (von Huene et al, 1972, and this volume), where the rate of convergence is presumed to be much greater. In a broader context than th records presented here, the significance of slope areas with little recent deformation and the unsubducted pre-Pliocene trough or fan deposits is that they suggest a rate of convergence considerably less than the convergent component of relative motion of the Pacific and North American plates. The shelf and slope data do not show the same division between areas of greater and lesser deformation; this is not necessarily inconsistent, because the structure of the slope is not well enough known.

It is tempting to relate the large area of seafloor with thin basement and attenuated magnetic anomalies, and the pre-Pliocene fan, to a common geologic process. The explanation for a thin layer 2 and subdued anomalies that nearly fits this situation was proposed by Larson et al (1972), and by Lawver and Hawkins (in press). The proximity of newly generated ocean crust to a large source of terrigenous sediment is thought to have prevented extrusion of magma from the ridge and formation of sills at depth rather than surface flows. This would result in a thinner layer of basement with low velocities associated with pillow lavas, and formation of larger-grained magnetic minerals that are more easily altered. This might be cited as a supporting argument that oceanic crust now at the foot o the slope was originally formed in close proximity to a continent and that underthrusting has been minimal since early Miocene time.

SUMMARY

The main results from the study of our data are first summarized and then are fitted into a regional plate-tectonic interpretation.

1. The Queen Charlotte Islands fault zone has previously been inferred to extend from Dixon Entrance to Cross Sound. Our seismic reflection records between Queen Charlotte Islands and Chatham Strait indicate that recent tectonism is mainly along the slope. The slope physiography consists of ridges and troughs similar to the fault physiography off the Queen Charlotte Islands. North of Chatham Strait the presumably contiguous Chichagof-Baranof Fault is indicated by a narrow band of earthquake epicenters (Page, 1973) located along a fault zone which can be seen in the seismic reflection records from off Sitka to Cross Sound. The dominant zone of faulting here appears to be concentrated along the outer shelf and perhaps the upper slope. The CB Fault heads into the Fairweather Fault at 20& eg;, and their probable intersection has not been mapped. We agree with the previously inferred continuity of these three major faults, because they appear to be the most active elements of the continental margin observed in seismic data.

2. A thick wedge of continental-rise sediment (Baranof Fan) occurs along the foot of the slope between Dixon Entrance and Cross Sound. Correlation with cores from DSDP site 178 indicates a Pleistocene age. Under Baranof Fan is a buried trough which appears to have subsided as recently as Pleistocene time, as indicated by dipping beds. Corresponding buried Pliocene fans or troughs are not seen in our reflection records. These features must have been buried by Pleistocene progradation of the slope and perhaps by underthrusting.

3. The buried trough is required in modeling gravity data across the margin, and the trough may extend under most of the slope. Two-dimensional modeling indicates that the base of the trough may be as much as 8 km below sea level at Dixon Entrance. The fault zone which forms the landward wall of the trough is buried beneath the continental slope and extends as far landward as the edge of the continental shelf; however, the models fail to indicate the dip of faults in the zone. Despite the apparent simple break used in modeling, the gravity and seismic reflection data indicate a 20- to 30-km-wide zone of faults, although one fault or another could be more active than the rest at a particular time. Thus, the present position of the CB Fault at the edge of the shelf rather than under the slope may not be typical. The trough and the single volcano near Sitka suggest some underthrusting if interpreted in accord with the plate-tectonic model.

4. Along the slope of the central Gulf of Alaska, oblique convergence is predicted by the plate model. Although some parts of the slope are deformed, there are others that show little folding or faulting. At the foot of the slope there is a wedge of pre-Pliocene sediment nearly 3 km thick buried beneath a normal continental-rise deposit. This wedge may be part of a submarine fan or a trough that has not been subducted. These observations suggest that convergence across this slope is much less than the convergent vector of the motion of the Pacific and North American plates.

5. The thick pre-Pliocene wedge is associated with oceanic crust with a thinner than normal "basement" layer and with attenuated magnetic anomalies over the deep ocean basin between Cross Sound and Pamplona Ridge. The thick wedge probably originated near a continent, and the anomalous crust might also have been affected by a nearby continent during its formation.

6. Crustal structure across the continental shelf and slope just east of Pamplona Ridge consists of a normal shelf sequence underlain by a thick wedge of deformed sediments, with velocity of 3.9 to 4.7 km/sec. This type of structure is more common

End_Page 286------------------------

along convergent continental margins than along transform or passive margins.

DISCUSSION

One possible tectonic configuration that was developed to explain the previous observations is based on the idea of Richter and Matson (1971). It consists of an arc-transform junction that involves both oceanic and continental crust. Oblique convergent splays short-cut the corner of the arc-transform system as diagrammed in the inset of Figure 9. As the corner is short-cut by the first oblique zone, slip along the transform and the normal convergent zones is decreased because of convergence along the oblique zone. Two possible short-cuts are shown, one on the continental slope in the central Gulf of Alaska and the other along the splays from the north end of the Fairweather Fault. As seen in Figure 9, not all the required faults have been identified. For instance, if Pleistocene slip long the middle of the Denali Fault (Hickman et al, 1977) is related to plate motion, a communication of stress between the Totschunda and Fairweather Faults is implied. This implication has been made by other authors, although a continuous fault has not been mapped in the intervening rugged terrain.

The limited convergence along the slope in the central Gulf of Alaska indicated by our data is explained

Fig. 9. Sketch of major structures, Gulf of Alaska. Structures on land from Plafker et al (1975), offshore Icy Bay to Montague Island from Bruns and Plafker (1975), Montague to Kodiak Islands, von Huene et al (in preparation). Dashed lines represent Mesozoic plate boundaries. Diagonal lines indicate extent of Benioff zone. Inset shows relative magnitude of convergent vector along plate boundaries. Letter "a" is full Pacific-North America relative plate motion and subsequent letters indicate lesser magnitudes as oblique convergent zones splay from the transform boundary. Diagonal lines in insert indicate convergent zone along landward wall of Aleutian Trench.

End_Page 287------------------------

by the proposed plate configuration. The amount of convergence along the slope is the difference between a) Pacific-North American plate motion, and b) the slip along the Fairweather Fault. Pacific-North America relative motion is estimated to be 5 to 5.6 cm/yr (Silver et al, 1974, Minster et al, 1974). Along the Fairweather Fault Page (1969) estimates 4 cm/yr of late Pleistocene slip, and Plafker and others (1976) estimate 5 cm/yr. If these estimates are correct, the convergence would range from 0.6 to 1.6 cm/yr, or about 10- to 30 percent of Pacific-North American plate motion.

If most of the plate motion is transformed inland, the continental shelf must be moving with the Pacific plate but at a lesser rate. Rogers (1977) came to this conclusion and called the piece of plate southwest of the Fairweather Fault and its splays the Yakutat block. The western edge of the Yakutat block converges against the Kenai Peninsula at a rate equal to the slip on the Fairweather Fault (Fig. 9). This convergence probably results in the oblique linear folds and thrust faults on the shelf west of Icy Bay (Plafker, 1967; Plafker et al, 1975; Bruns and Plafker, 1975; Rogers, 1977). The thrusting on Kayak Island and in its vicinity (Plafker, 1974) is probably part of the same sequence of oblique folds and thrusts, which may first have developed in the northwest and progressively pread to the southeast (Rogers, 1977).

The tectonic system described above cannot be projected back for a long period of time without conflicting with other observations. Hudson and others (1977 a, b) state that the distribution of Tertiary plutons and the distribution of metamorphic facies along the Fairweather Fault during the past 20 MY do not require large strike-slip displacements. This statement suggests doubt that most Pacific-North American plate motion has occurred along the Fairweather Fault system for a prolonged period of time. Plafker et al (1976) suggest that present high displacement rates along the fault began about 100,000 years ago and that before this time plate motion was taken up on faults seaward of the Fairweather. A fault that may have been active earlier is the inferred fault seaward of the Fairwea her that might join the CB Fault as suggested by the linear bathymetric features between them (Fig. 4). Rogers also found it difficult to project the present tectonic system back in time, because it would require considerable subduction of the Yakutat block beneath the Alaskan continent. Bally (1977) has proposed such subduction for the Alps (A-Subduction), but the lack of particularly high mountains on the continental shelf at the western edge of the Yakutat block makes extensive A-subduction unconvincing.

A period of tectonism in the Yakutat block that preceded the present one is suggested in the reflection and refraction data. Reflection record B (Fig. 7) shows a deformed sedimentary sequence overlain by the little-deformed sequence. The lower deformed sequence probably corresponds to the thick units with velocities of 3.9 to 4.7 km/sec (refraction profiles LF27-28, Fig. 8) overlain by the normal shelf sedimentary sequence. Similar crustal structure is seen across the margin off Kodiak (Shor and von Huene, 1972), where a 4.5-km/sec layer is interpreted as sedimentary rock deformed along a convergent margin (von Huene, this volume) and overlain by much less deformed sediments of a forearc basin.

From these observations one can suggest some aspects of pre-late Cenozoic history, as well as a mechanism. As proposed by Grow and Atwater (1970) the relative motion between the Kula plate and the North American plate in the early Tertiary would have resulted in a subduction component along the entire eastern boundary of the Gulf of Alaska, creating a trench along the coast from the Queen Charlotte Islands to the junction with the Aleutian Trench. The northward movement of the Kula Ridge continuously shortened the trench from the south, replacing it with a transform fault, as the relative motion of the Pacific and North American plates is essentially tangential to this boundary. With disappearance of the Kula Plate, all of the former Queen Charlotte margin would have been converted fr m a trench to a transform fault, except for the short oblique section from Cross Sound to Kayak Island. However, the present trough along southeastern Alaska and British Columbia is probably not a vestige of the pre-30-MY-old subduction event because magnetic anomalies beneath the trough (anomalies 1 through 7) are younger than 30 MY. If the pre-30-MY trench was seaward of the transform faults as indicated in seismic records and the gravity models, it must have been rafted north on the transforming Pacific plate. This is consistent with termination of Paleozoic and Mesozoic terrains seaward of the British Columbia and southeastern Alaska coasts. Also terminated is the Chatham Strait Fault, along which the latest known large offsets are of early Cenozoic age (Ovenshine and Brew, 1972). In the present trough Pliocene and Pleistocene beds tilt landward but effects of subsidence from sediment loading cannot be easily separated from those of tectonic subsidence. Thus the trough may be a late Cenozoic feature and the pre-30-MY-old trench may have been rafted north and now be buried under younger sediment.

The layers of deformed sediments seen on the reflection records and the thick accretionary wedge shown on refraction profiles LF27-28 were probably formed by convergent tectonism. Fitch (1972) has proposed that an oblique subduction system is unstable, and that frictional drag will separate the orthogonal components into pure subduction normal to the trench and into strike-slip motion along one or more faults cutting the overlying slab. We suggest that the zone of dominant deformation has shifted landward through the seaward transform faults to the Fairweather Fault and others in the Yakutat block, to reach an eventual more stable configuration with pure transform motion along a fault parallel to the vector of relative motion of the two plates.

The present orogenic episode began in Miocene time in most areas of the Gulf of Alaska. Along the

End_Page 288------------------------

eastern Gulf Monger et al (1972) postulate dominantly transform motion since Miocene time. Along the central Denali Fault, Hickman et al (1977) identify the beginning of the present strike-slip movement in mid-Miocene time. In lower Cook Inlet, Fisher and Magoon (in press) identify the beginning of the present tectonic episode in late Oligocene to early Miocene time. In the Yakutat block, Plafker et al (1975) indicated that the present episode began perhaps in mid-Miocene time and that thrusting on Kayak Island occurred after mid-Miocene time. Kayak Island is implied to be in the oldest part of the convergent zone in the western Yakutat block (Rogers, 1977). The initial age of the present tectonic episode might also be indicated by the age of the unsubducted pre-Pliocene deep-sea fan t the foot of the Yakutat block boundary. Thus, if indications of an earlier episode of tectonism in the seismic data are correct, the present tectonic pattern involving the Yakutat block might have begun between mid-Miocene and Pliocene time. However, understanding rather than speculating about the pre-late Cenozoic plate history involves knowing the timing of major movement in 300-km-wide zone of mostly rugged and glacially covered terrain.

The foregoing has made some aspects of the ideas first proposed by Richter and Matson (1971) more specific. The implications of these ideas touch on the time-frequency analysis of great north Pacific earthquakes. Sykes (1971) made a convincing case that a great earthquake was overdue in the central Gulf of Alaska, if more frequent earthquakes on either side of this area were an indication of the earthquake recurrence interval across all of the Pacific-North American plate boundary. Page (1973) considers that the 1972 Sitka earthquake filled one of the gaps identified by Sykes. The basic premise is that if great earthquakes leave between them a similar zone in which the strain is not relieved, an earthquake should follow shortly. However, since the plate boundary along the central coas of the Gulf of Alaska is dissimilar to the plate boundaries on either side, the strain may be relieved differently and the recurrence interval for large earthquakes may differ accordingly.