This paper demonstrates how a seismic refraction model of the central Aleutians (Murdock 1967, 1969) might explain the evolution of the Aleutian Island arc.

Therefore, as I am becoming convinced I will never get the form right, I have decided to use the World-Wide Web to distribute the concepts, and this paper was written specifically for that purpose. It gives enough background in the Introduction so that those not familiar with the Aleutians can become acclimatized, and for those familiar with the Aleutians, the Summary is designed to give an overview of the concepts, with references therein to supporting arguments and illustrations in the main body of the text.

Return to Table of Contents

Return to Table of Contents

This paper demonstrates how a seismic refraction model (M-1) of the central Aleutians (Murdock, 1967, 1969), derived from data of the 1964 Aleutian Island Seismic Experiment, might explain the evolution of the forearc of the Aleutian Island arc (map, Fig. 1). M-1 appears to provide an insight into how production of the forearc basin in the central Aleutians might evolve to a mountain range in the easternmost Aleutians. Prevailing viewpoints of the evolution or tectonics are maintained by Dickinson and Seely (1979), Pavlis and Bruhn (1983), Lonsdale (1988), Plafker et al. (1989), Ryan and Scholl (1989), and Plafker et al. (1994), all of whom view the evolution or tectonics from the perspective of strict Plate Tectonics (SPT). Distinctions are made between the prevailing interpretations and the candidates presented herewith. Typically, the differences can not be resolved with existing data, hence, the paper can be viewed as outlining some of the important uncertainties that need to be addressed. A critical test to distinguish the models is recommended.

As stated in the foreword, for readers familiar with the background literature of the Aleutians, or for those interested primarily in the general arguments, the Summary (Section 7) gives an overview of the major concepts of the paper.

1.2.1 Discussion of M-1, a seismic refraction model of the central Aleutians. Figure 2 depicts M-1 in the framework of converging plates. M-1 represents the only hypothesis of the central Aleutians that has attempted to resolve both the Pn seismic refraction data and the earthquake hypocenters (i.e, three-dimensional locations). The hypothesis proposes two parallel large linear structural depressions of the crust-mantle boundary, one beneath the arc ridge and the other beneath the forearc (Fig. 2). A zone of hypocenters beneath the ridge-terrace border separates the two, with the shallow events marking a thrust fault, and as proposed by Engdahl and Scholtz (1977), with the deeper events marking stresses in the downgoing plate. The two structural depressions contrast markedly with conventional concepts of the crust-mantle boundary, that are not based on Pn data, but rather, are largely conjectural. They show one giant trough extending from north of the volcanoes almost all of the way to the trench (Dickinson and Seely, 1979, and Scholl et al., 1986, for examples). This trough is sometimes referred to as the "arc massif" (e.g., Scholl et al., 1986).

In the framework of convergence of the plates, M-1 indicates that the entire forearc shelf of the central Aleutians is being transported islandward by the Pacific plate and thrust northward steeply beneath the North American plate at the ridge-terrace border (Fig. 2), about 70 to 100 km south of the volcanoes of the arc. (In the context of the central Aleutians, when I use the term forearc shelf, I mean the region between the ridge-terrace border and the trench and above the downgoing plate--it corresponds approximately to the accretionary wedge of SPT.) The upthrust south limb of the ridge trough corresponds spatially with the +150 mGal free air gravity anomaly (Scholl et al., 1983, pg. 417, Grow, 1973) of the forearc, but it is not known whether this anomaly relates mainly to the upthrust limb or to shallower features in the zone of the gravity high, perhaps as exemplified by the basaltic extrusive and intrusive rocks exposed on Amlia Island described McLean et al. (1983).

At the time M-1 was presented, support for the model was overwhelmed by the newly developing concept of Plate Tectonics. Its main thrust fault (Grow, 1973, Scholl et al., 1986, Taber et al., 1991) is at the trench (T-thrust), almost 100 km oceanward of the M-1 thrust fault at ridge-terrace border (RTB-thrust). Displacements on the T-thrust are postulated by proponents of SPT to balance the reported amount of sea-floor spreading given by Minster and Jordan (1978) and DeMets and coworkers (1990), i.e. about 80 km/Ma, whereas subduction of the contemporary forearc shelf, implied by M-1, severely limits the amount of acceptable convergence if transport and subduction of the forearc shelf is a steady-state process (Section 3 below).

The position of the RTB-thrust was estimated from earthquake activity recorded for one month in 1964 and from a dozen one-ton shots detonated in the forearc (Murdock, 1967, 1969). These data were recorded at as many as six island stations (five for the shots) from Atka-Amlia to Amchitka, a distance of 400 km. Subsequently other investigators have reported the seismic activity from additional island networks in the central Aleutians. These include Engdahl (1977) who reported activity observed at 8 stations for 2 1/2 years, beginning in 1970. These stations are often referred to as the Amchitka network and they were sited essentially in the western region of the former 1964 network. The 1977 report also included observations for the first 1 1/2 years of operation of the 11-station Adak network. It began full operation in 1975, after the Amchitka network had been closed, and was sited in the central and east-central region of the former 1964 network. Gross and Kisslinger (1994) reported a longer interval of the Adak network. Both Engdahl and Gross and Kisslinger demonstrated that the intense activity is indeed concentrated at the ridge-terrace border region, as had been indicated by the 1964 experiment. Furthermore, the dip of the zone of activity is akin to the dip of the RTB-thrust, as may be seen from the locations determined by Engdahl which are abstracted in Figure 3A-3E.

The general earthquake locations by the Adak network (Fig. 3D, E) were confirmed by data of an onshore-offshore experiment. It consisted of 8 ocean-bottom seismometers that were deployed in the forearc off Adak for 6 weeks in 1978 (Frohlich et al., 1982) and their data were analyzed in conjunction with data of 7 of the stations of the Adak network (inset, Fig. 4A). Results of this experiment, Figure 4A, confirmed onshore locations made from the Adak network, albeit some of the onshore-offshore hypocenter locations are somewhat shallower than the locations made only from data of the onshore stations. Furthermore, the pattern of hypocenters of the onshore-offshore experiment is very similar to that observed during the 1964 experiment, as may be seen from the overlays of Figure 4B. Thus it would seem that the RTB-thrust might be well substantiated, although the fault has never been widely accepted by the scientific community, perhaps because subduction of the forearc shelf will not accommodate the amount of reported sea-floor spreading, noted above.

In 1987 Engdahl and Gubbins studied earthquakes of the Adak area that had P waves well-recorded at distances greater than 30 degrees. The investigators estimated new hypocenters for these teleseisms (earthquakes recorded at great distances) by inverting the P-wave times simultaneously for location and complicated model velocity. (This model attempts to compensate for effects of the downgoing plate that might be seen at the distant stations.) The estimated depth of occurrence (focal depths) of the shallow (h < 40 km) events were fixed by indicators of the depth seen on the seismograms at the distances greater than 30 degrees or by data of the local networks (Engdahl and Gubbins, 1987, pg. 13,856). For these shallow events of fixed focal depths, the data of the island stations were not used to constrain the new solutions for the map locations (Engdahl and Gubbins, 1987, pg. 13,856).

The teleseismic locations demonstrated a somewhat different pattern of hypocenters (abstracted in Fig. 5A) from that of the local networks (Figs. 3A-E and 4). It extends at least 20 km farther trenchward from the pattern determined locally for roughly the same area (Fig. 3E, which is overlain on the teleseismic locations in Fig. 5B), albeit not for the same time interval. This displacement relates to the shallow teleseisms with map locations which were not constrained by the local data. Here it is noteworthy that the hypocenters determined locally by Engdahl (1977) are stated to have been estimated by using both P and S data (1977, pg. 265). If so for all of them, I would be surprised that the map locations could be considered largely erroneous because displacing shallow events by 20 km away from the Adak network would correspond to 2 to 3 seconds in the S minus P interval, whereas Engdahl (1977, pg. 265) stated a perceived accuracy of 0.03 sec for the timing of P and S arrivals. Furthermore, the general locations determined by data of the onshore stations were confirmed by the onshore-offshore experiment discussed above.

Engdahl et al. (1989) showed a larger set of teleseismic locations, but with somewhat less restricted depth control because some of the focal depths were fixed arbitrarily as they discussed, pg. 15,482. Also, the 1989 work included aftershocks of the large (M=8.0) May 7, 1986 Andreanof Island earthquake, but these earthquakes may not have inordinately affected the results as Engdahl and coworkers concluded (pg. 15,497) that the pre- and post-earthquake spatial patterns for the main thrust zone (seemingly the T-thrust in their view) are roughly equivalent. The 1989 presentations show events as much as 30 to 40 km trenchward from the local pattern, somewhat farther than reported from the smaller data set in 1987.

From the viewpoint of SPT, the locations of the shallow events at the ridge-terrace border and islandward, estimated from data of the local networks, are caused by deformation in the upper plate, and displacements between the plates are indicated by the locations of the shallow events farther oceanward, estimated from data of the teleseismic network, and these locations define the T-thrust (Taber et al., 1991, pg. 31). However, as suggested above, bias in the map locations can not be ruled out, especially since some of the locations for shallow events have been moved as much as 50 km oceanward (Engdahl and Gubbins, 1987, their Fig. 5), with respect to conventional teleseismic determinations.

Solutions for the source parameters do not differentiate the RTB-thrust from the T-thrust either. The solutions for fault planes have a range of dips, from 40 to 50 degrees for estimates made from P waves recorded nearby (LaForge and Engdahl, 1979, pg. 1521) to roughly 25 degrees for solutions with teleseismic data (Ekstrom and Engdahl, 1989, their Fig. 5). Whereas I have shown the RTB-thrust as steeply dipping (about 50 deg., Fig. 2), it might be more realistically regarded as a steeply dipping zone of thrust faults, many of which may display somewhat smaller dip. Thus, although the dips of the fault planes estimated by Ekstrom and Engdahl are perhaps what one would expect for the T-thrust, they also are what one might expect for the RTB-thrust zone as well.

In the central Aleutians, if the source solutions by teleseismic data were consistent with the plate models, the arguments by SPT for the T-thrust would have more credibility. As DeMets and coworkers (1990, pg. 460) discussed and showed (their Fig. 41), in contrast to the region of the Alaska Peninsula and Kodiak Island, the slip directions indicated by the source solutions and the calculated slip directions between the plates are in discord by as much as 30 degrees or more, and for the region west of the Alaska Peninsula, the solutions were not used to estimate the slip direction for their model. DeMets and coworkers suggested that the geology of the central Aleutians may be somewhat different from what they assumed, hence the discrepancy with the source solutions that generally show slip more normal to the arc than does the plate model. Interestingly, as suggested above, the solutions for the central Aleutians might be for the RTB-thrust zone, not for the T-thrust as their model initially assumes.

In addition to the 1964 experiment, two other major seismic refraction experiments have been conducted in the central Aleutians, but they do not distinguish M-1 from SPT either. One was by Shor (1964) and his data was for the trench, the islands, and the Bering Sea. He did not have data in the forearc. His results were used to construct M-1 as were data of a minor north-south line conducted by Texas Instruments Incorporated in the Adak region and discussed by me (Murdock, 1967, 1969). The second major experiment was by Texas Instruments (Linville et al., 1968) who reported an 800-km NE-SW seismic refraction line centered on Amchitka. In the area of Amchitka Island and the ridge-terrace border their experiment gave largely inconclusive results. However, in the region of the Bering Sea and of the terrace of the forearc, they reported arcward dip of the crust-mantle interface, in accord with both M-1 and SPT.

1.2.2 Other major seismic experiments defining the crust-mantle interfaces of the Aleutians. In the Gulf of Alaska east of the Kenai Peninsula, Brocher and coworkers (1994) reported the megathrust of SPT and the upper part of the Pacific plate by using seismic reflection and refraction data of the Trans-Alaska Crustal Transect (TACT). They displayed the Pacific plate and (what they interpret to be) the megathrust with modest (about 10 deg.) dip as the previous conceptual models showed (e.g., Plafker et al., 1982, Pavlis and Bruhn, 1983), but with unexpected minor undulations of the downgoing plate.

Inland, on the Kenai Peninsula, the general position of the downgoing plate is perhaps best indicated by the hypocenter locations of Stephens et al. (1983) that also suggest about 10 degrees of dip, Figure 6. The dip of the zone of hypocenters begins to steepen near Cook Inlet, apparently denoting the area where the plate begins to plunge into the asthenosphere.

The EDGE deep seismic reflection line (Moore et al., 1991) adjoined the TACT experiment on the southwest. It consisted primarily of a generally NW-SE profile that began at the trench, traversed a pass between Kodiak Island and the Kenai Peninsula, and extended across Cook Inlet. The geometry of the Pacific plate it demonstrated is somewhat more in accord with previous conceptual models than that of TACT. Although EDGE traversed Cook Inlet, the experiment did not clearly define the crust-mantle interface there, and the position of the interface or its configuration has not been given in detail by seismic reflection or refraction data of the region, but rather must be inferred from the location of earthquake hypocenters.

Further justification for entertaining alternates to SPT is provided by Stevenson and coworkers (1983). They studied the sedimentology of the mid-Cenozoic Zodiac fan. It is oceanward of the Alaska Peninsula and eastern Aleutians (Fig. 1). They reported Alaska for the likely source of the sediments, with no more than about 500 km of displacement of the fan relative to the Aleutians, while SPT requires 1500-3000 km of relative displacement, as Stevenson and coworkers discussed (pg. 271-272). They suggested rafting of the Aleutians northward to account for the imbalance. However, from the paleomagnetic data of the Aleutians, Harbert (1987) estimated essentially no poleward displacement of the arc during the mid- and late Cenozoic, and he argued anew for huge convergence. In contrast, Carey (1988) presented another scenario for deposition of the fan, but with no convergence of the plates at all. He proposed sea-floor spreading compensated by expansion of the Earth, and from his viewpoint, the general paleomagnetic data support his hypothesis, albeit he cited none from the Aleutians, which ordinarily indicate northward displacement (e.g, Hillhouse and Coe, 1994, their Figs. 1 and 2). On the other hand, Hillhouse and Coe (1994) seem to view the (not voluminous) paleomagnetic data from the perspective of SPT, and any potential fit to an expansion hypothesis is not discussed.

Although the possibility of smaller than expected convergence of the plates is entertained for one of the two models presented herewith, speculations on the fundamental cause for any such discrepancies are beyond the scope of this paper. I maintain that the amount of sea-floor spreading and the probable amount of convergence of the plates should be considered separate entities, so that any likely imbalance between them can be investigated.

This paper discusses the major features of the structure of the Aleutian arc. First, the paper correlates the major features of the easternmost Aleutians to those of the central Aleutians, in the framework of M-1. Second, for the earthquakes of the ridge-terrace border of the central Aleutians, instead of minor activity in the upper plate, this paper argues that the earthquakes are manifestation of production of the forearc basin. Third, it provides models for the evolution of the forearc basin of the central Aleutians to the mountains of the easternmost Aleutians, with provisions for both existence and non-existence of the T-thrust. Fourth, by using differences in the spatial patterns of the seismic activity, the boundary area between production of the forearc basin and deformation of it is estimated. Fifth, in a brief discussion, the time-dependent aspects of the model are compared with the known history. Finally, in the central Aleutians, the paper notes that existence of the T-thrust is easily tested, and such a test is recommended.

Return to Table of Contents

On Kodiak Island and the Kenai Peninsula of the easternmost Aleutians, geologic maps and cross sections (von Huene et al., 1979, Plafker et al., 1982) show a Mesozoic zone of primarily melange (Uyak and McHugh complexes), with slivers of rock as large as one km (Connelly, 1978). As depicted by the cross-sections, the zone ranges up to ten kilometers thick and the oceanward part dips 40 to 50 degrees toward the volcanic arc. It was described by Connelly (1978), who studied it on Kodiak Island, as an abandoned subduction zone, and this interpretation was supported by Little and Naeser (1989) who studied it in the northeastern part of the Kenai Peninsula. The Border Ranges fault (Fig. 1) bounds the zone on the northwest. (Hereafter when I mention the Uyak-McHugh complex, I mean to include the Border Ranges fault.) The dip, thickness and deformation of the zone are what one might expect of the RTB-thrust zone, which is defined by a band of earthquakes about 30 km wide that dips steeply toward the volcanoes, as demonstrated by data of the local networks of the central Aleutians (Figs. 3A-E, 4).

While the similarities between the RTB-thrust and the Uyak-McHugh complex suggest that the RTB-thrust might be the active equivalent of the abandoned subduction zone, the data are insufficient to test the actual conjoining or the contiguity of the RTB-thrust to the complex. A candidate region for the conjoining is the Shumagin Islands as it will be argued in Section 5 that this area is within a transition zone of the tectonics, with a steeply dipping zone of earthquakes west of the Shumagins marking the active RTB-thrust complex. In the Shumagin region, the intense activity of the shallow earthquakes of the west (Fig. 3F) appears to extend oceanward as far as 150 to 160 km from the volcanoes, about 20 km farther oceanward than the outermost Shumagin Islands. The extent is perhaps best shown by Fig. 7 (west) of Taber and coworkers (1991), which displays the hypocenters of Hudnut and Taber for a narrow region adjacent to the islands. (Taber and coworkers show roughly 35 deg. northwest dip for the primary zone of activity which is about 10 km wide, as does Abers, 1994, for the region west of the Shumagins.) The landward limit of this steep zone of intense activity appears to project on the surface roughly 10 km oceanward of the outermost islands, corresponding to about 140 km from the volcanoes.

The nearest outcrop of the Uyak-McHugh complex is near the northwest shore of Kodiak Island, about 400 km northeast of the Shumagin Islands. It is not clear where the subduction complex might project to the Shumagins, nor is it clear whether it is active and manifest by the earthquakes, or still inactive as for the region farther east. Instead of 140 km oceanward from the volcanoes as for the steeply dipping zone of intense shallow activity, Plafker and coworkers (1994) speculated a position for the bounding Border Ranges fault (shown in their Fig. 6) that is only about 60-80 km oceanward from the volcanoes. If their concept is correct, the primary RTB-thrust and the Uyak-McHugh complex might conjoin westward, perhaps west of Unimak Island where the gravity anomaly seemingly associated with the RTB-thrust in the central Aleutians (Section 5) begins to bear away to the east from the arc. If the Uyak-McHugh complex and the main RTB-thrust conjoin near Unimak Island instead of in the Shumagins, perhaps an active branch of the RTB-thrust has splayed off eastward near Unimak Island. Such a splay could explain the steeply dipping zone of shallow earthquakes, and the splay might correspond to any one of several faults shown by Plafker et al. (1994, their Fig. 6) oceanward of the Uyak-McHugh complex on Kodiak Island and offshore thereof. One of them is their proposed westward extension of the Contact fault that they display about 70 km oceanward of the Border Ranges fault, as for the relationship of the steeply dipping zone of earthquakes in the Shumagins to the projection of the Border Ranges fault discussed above.

While I suggest association of the Uyak-McHugh complex to the RTB-thrust, SPT typically represents the Uyak-McHugh complex as an abandoned T-thrust (Connelly, 1978, Little and Naeser, 1989), i.e., the megathrust of SPT that cropped out at an ancient trench. This correlation to the T-thrust is postulated in spite of the fact that the features of the Uyak-McHugh complex are not at all akin to today's features at the trench. For example, today's T-thrust is proposed by SPT to have modest dip at the trench, contrasting sharply with the steep dip of the Uyak-McHugh complex. Moreover, as discussed in Section 3.3, today very little deformation is seen at the trench, and this again contrasts sharply with the Uyak-McHugh complex that, as described, displays chaotic deformation over a broad zone. Indeed, the size of the slivers of rock in the zone are equivalent to 1/4 or more of the height of today's islandward trench wall, while seismic reflections (e.g. Scholl et al., 1982) show no evidence of such deformation at today's trench. Thus, it seems reasonable to entertain the hypothesis that the Uyak-McHugh complex is not the ancient equivalent of the T-thrust, in view of the inconsistencies relating the two.

Although the conjoining of the Uyak-McHugh complex to the RTB-thrust can not be demonstrated, given the correspondences between them, and moreover, the incongruities in the competing model, hereafter I will assume that the RTB-thrust and the Uyak-McHugh complex indeed represent the same fault zone, with the active RTB-thrust system (by system, I mean faults that are genetically and physically related, such as secondary splays) being demonstrated by today's earthquakes west of the Shumagins. Beginning in Section 3, I will consider two possible alternatives, namely (1) that the RTB-thrust (and by association, the Uyak-McHugh complex) is a primary fault of plate convergence and subduction, and (2) that the RTB-thrust (and by association, the Uyak-McHugh complex) is only a secondary feature, with convergence of the plates being accommodated by the T-thrust (i.e., the megathrust of plate tectonics). While the first alternative will accommodate only a small fraction of the sea-floor spreading that has been reported, the second alternative will accommodate all of it.

A well-developed linear basin, with sedimentary layers ranging from early Mesozoic (or older) to latest Cenozoic, and with a total thickness of more than ten kilometers, occurs beneath Cook Inlet between the outcrop of the Uyak-McHugh complex and the volcanoes (Kirschner and Lyon, 1973, Plafker et al., 1982). As discussed in Section 1.2.1, in the central Aleutians, a linear structural trough appears to occur on the arc ridge between the RTB-thrust and the volcanoes (Fig. 2). In the context of this paper, as the Uyak-McHugh complex corresponds to the RTB-thrust, the linear basin beneath Cook Inlet corresponds structurally to the linear trough of the arc ridge. However, the sedimentary basins likely were not contemporary throughout most of the Mesozoic, as the central Aleutian arc might have formed only in the latest Cretaceous to earliest Cenozoic (Scholl et al., 1983, pg. 429).

I do not mean to imply that a basin with a thick sedimentary sequence similar to that of Cook Inlet exists on the ridge of the central Aleutians. Seismic reflection profiles (Scholl et al., 1983) provide no evidence of it although the fine structure of the upper crust is known to a depth of as much as 2 to 3 km (Fig. 2 of Scholl et al., 1983). To these depths essentially undeformed marine sedimentary layers as old as mid- to late Miocene are indeed shown in the central part of the ridge, but they unconformably overlie rocks dominantly of a different composition that crop out on Atka and Amlia islands. (Amlia Island is southeast of Atka Island and about 20 km north of the southern edge of the ridge.) Amlia Island is part of the crest of a wave-eroded antiform (Scholl et al., 1983) and the antiform corresponds spatially approximately to the upthrust south limb of the ridge trough of M-1. As Scholl and coworkers discussed (pgs. 419-421, 429), the exposed antiform consists of probable mid- to upper Eocene mainly volcanic and volcaniclastic rocks, with lesser amounts of sedimentary rocks; and except for the folding, the layers generally are little deformed. As the investigators noted, rocks of similar composition occur on Atka Island. McLean and coworkers (1983) discussed the igneous rocks of Amlia Island also. They stated (pg. 1024) that these rocks occur as massive and pillowed flows, as dikes, as sills, and as intrusive plugs, with the flows ranging in composition from basalt to rhyolite. They also stated (pg. 1025) that the volcanic-flow rocks composed of basaltic andesite and andesite are the major rocks volumetrically of the island.

Whereas the rocks described above for the central Aleutians certainly do not correlate directly to today's well-developed basin of Cook Inlet, they might correspond to its lowermost sequences, perhaps the early Mesozoic or older early phase of Kirschner and Lyon (1973, pg. 398) that included ophiolite deposits. (The early phase is the stage before the dioritic rocks of the adjoining batholiths to the northwest became important sources of the basin sediments in the mid-Jurassic, Kirschner and Lyon, pg. 400.) The early Jurasic Talkeetna formation is part of the early phase, and in a review Barker (1994, pg. 572) described it as containing both sedimentary and volcanic rocks including flows and sills of basalt and andesite. Thus, in contrast to the upper sequences, the lower sequences of the Cook Inlet basin are not dissimilar from the basement rocks reported on Atka and Amlia islands.

The Cook Inlet basin is probably better developed than the ridge-trough of the central Aleutians due (1) in part to likely older age of the Cook Inlet basin, (2) in part to the mountains of its northwest border, that probably provided more than 10 kilometers of Mesozoic mainly marine clastic sediments, derived primarily from the diorites and related rocks (Kirschner and Lyon, 1973), and (3) in part to the highlands on its southeast border that restricted growth of the basin oceanward during the Cenozoic (Kirschner and Lyon, 1973), and, as discussed by Kirschner and Lyon, together with the northwest borderland and interior Alaska provided many more kilometers of almost entirely non-marine sediments during this time.

While I correlate the basin beneath Cook Inlet to the ridge trough, from the perspective of SPT, it corresponds to the forearc (Dickinson and Seely, 1979, their Fig. 6, Pavlis and Bruhn, 1983, their Figs. 1 and 2, Little and Naeser, 1989), i.e., the region adjoining the SPT accretionary complex of trench and islandward from it. This difference in interpretation results from SPT having only one structural depression in the island arc, not two as for M-1.

As described in Section 1.2.1, M-1 shows a second huge linear east-west structural depression which is parallel to the ridge trough but trenchward of it and of the RTB-thrust. As such, it is beneath the forearc shelf. In the central Aleutians, Scholl and coworkers (1983, visited further in Section 3.1) show a giant (75-km wide) linear basin in this area of the forearc. However, in the easternmost Aleutians, the area oceanward of the the Uyak-McHugh complex is uplifted and severely deformed as exhibited by exposures on Kodiak Island and the Kenai Peninsula. Indeed, on the Kenai Peninsula the area is a mountain range (the Kenai Mountains). The correlation of the Uyak-McHugh complex to the RTB-thrust of M-1 suggests that the rocks oceanward of the outcrop of the Uyak-McHugh complex are forearc basin deposits laid down in situ (in an area corresponding to the forearc shelf region of the central Aleutians) and deformed there. Broad in situ deposition is in accord with the postulates of Kirschner and Lyon (1973, their Fig. 6), who show the deposits as turbidites seemingly derived from Alaska.

While in the context of M-1 the area immediately oceanward of the outcrop of the Uyak-McHugh complex is a deformed forearc shelf-basin, in the context of SPT it is either a deformed trench (Moore, 1973) or deformed accreted assemblages that have been transported many hundreds of kilometers (Dickinson and Seely, 1979, Pavlis and Bruhn, 1983, Plafker et al., 1989, Plafker et al., 1994). Again, the difference in the interpretations results in part from M-1 having two parallel structural depressions, not only one as for SPT.

Return to Table of Contents

While M-1 suggests that the entire forearc shelf region is being transported islandward by the Pacific plate, thus severely limiting the amount of convergence during the mid- to late Cenozoic, the deep (h > 60 km) earthquakes beneath the islands, apparently marking the downgoing plate as discussed by Engdahl and Scholtz (1977), suggest that at least a few hundreds of kilometers of convergence of the plates must have occurred at some time, even if one were to postulate sea-floor spreading compensated by expansion of the Earth, e.g. as by Carey (Section 1.3). In view of the ages of the rocks of the forearc, together with the geometry of the forearc basin, not even the relatively small amount of convergence indicated by the deep earthquakes can be accommodated by transport of the existing forearc.

In my conceptual model, assuming continuous transport applies, almost all of the convergence suggested by the deep earthquakes occurred before the forearc shelf that evolved into the basin was produced. (This earlier stage of evolution might be exemplified today by the island arcs with inordinately deep trenches, such as the Mariana arc.) Conceptually, after the early Cenozoic, convergence of the plates ceased (or slowed sufficiently) so that a sedimentary shelf could be built in the forearc. Thereafter, convergence by the Pacific plate subducted the forearc shelf as shown in Figure 7, producing the forearc basin. As suggested in the illustration, the south limb of the basin is produced by transport of the Pacific plate and the north limb is produced by displacements of the RTB-thrust. Folding of the upper layers of the basin is occurring above the RTB-thrust, and production of melange, corresponding to that of the Uyak-McHugh complex (Section 2.1), is occurring below the floor of the ocean in the zone of intense earthquake activity.

While M-1 suggests that the forearc basin is manifestation of a subducted sedimentary shelf, in the view of SPT (e.g., Ryan and Scholl, 1989) the south limb of the basin is produced by underplating (a product of tectonic accretion) that elevates the former sub-horizontal (perhaps oceanward dipping) sedimentary layers to produce islandward dip, and the trenchward inclination of the north limb is controlled by the geometry of the underlying basement rocks.

The easternmost Aleutians is the representative region for deformation of the forearc. As described in Section 2, the region consists of the ridge-trough (the basin of Cook Inlet), an apparent abandoned subduction fault zone (the Uyak-McHugh complex), and the deformed forearc trough (the shelf-basin of Section 3.1) immediately oceanward of the fault complex.

3.2.1 The general concept. In the conceptual model of continuous transport, when the forearc shelf-basin can no longer be subducted, transport by the Pacific plate compresses the shelf against the North American plate both locking the RTB-thrust and producing pronounced deformation and uplift oceanward of the thrust. In the easternmost Aleutians, as discussed in Section 2, the locked RTB-thrust is postulated to be manifest by the Uyak-McHugh complex and the deformed forearc shelf-basin is manifest by the rock formations and mountains immediately oceanward of the complex.

The vast deformation requires separation of the forearc shelf from the downgoing plate, and the separation is accomplished by a blind decollement (Fig. 8). This decollement, with its associated deformation front, propagates oceanward as transport of the forearc continues (Fig. 8). Uplift associated with deformation allows erosion of the rocks of the forearc shelf-basin. This erosion produces sediments for a new shelf (derived shelf, Fig. 8) that grows the forearc oceanward, and the new shelf itself may eventually be deformed by the propagating decollement as suggested in the illustration.

The differences between the model of Figure 8 and SPT should be recogonized. First, the model of continuous transport implies propagation of the arc seaward by sedimentation, and this contrasts sharply with SPT which dictates propagation primarily by tectonic accretion, see, for instance, Dickinson and Seely (1979), Pavlis and Bruhn (1983), Lonsdale (1988), Plafker et al. (1989), Plafker et al. (1994). In the view of the latter four, some of the accreted terranes have been transported hundreds or thousands of kilometers. Also, whereas they and Little and Naeser (1989) propose that the observed uplift was produced, at least in part, by underplating, continuous transport suggests it was produced by compression as described above.

3.2.2 The earthquakes of the easternmost Aleutians. The earthquakes of the easternmost Aleutians can not be used to distinguish SPT from the model of continuous transport. Nevertheless, several implications need to be discussed.

As mentioned in Section 1.2.2, a seismic network was operated in the area of the outcrop of the Uyak-McHugh complex on the southern Kenai Peninsula (Stephens et al., 1983), see Figure 6. The network began recording in 1980 and hypocenters for 2 1/2 years of operation are shown in the illustration. Although the pattern of hypocenters is somewhat similar to the teleseismic locations of the central Aleutians, the pattern (Fig. 6) differs markedly from that of the local networks there: While the intense activity of the central Aleutians dips steeply and extends almost to the surface, the intense activity of the easternmost Aleutians shows only moderate (about 10 deg.) dip 40 km beneath the outcrop of the Uyak-McHugh complex, compare Figure 3A-E with Figure 3H, or indeed Figure 4 with Figure 6. This lack of steeply dipping intense seismic activity seems to clearly demonstrate that the RTB-thrust is not active in the easternmost Aleutians.

Whereas the dip of the zone of earthquakes of Figure 6 is what might be expected of the decollement of the model presented herewith or of the T-thrust of SPT, neither fault can be demonstrated by the source mechanism studies. The zone of earthquake activity in Figure 6 is as much as 20-30 km thick and the earthquakes thereof range in magnitude from about 0 to 4. In studying the polarity patterns of the small earthquakes, Stephens and coworkers (1988) found no evidence of thrust faulting, but this type of faulting in the zone can not be excluded as only a small sample (about two dozen) of the total number of events well-located during the entire experiment (about 800) were studied (Stephens, personal communication, 1996). Although they made no formal analyses, Stephens and coworkers (1988) stated that the earthquakes of the upper part of the zone appear to be of normal faults, as indicated by the polarity patterns noted above. Assuming dominantly normal faults, two possibilities seem likely for their origin (1) they manifest bending of the downgoing plate below the thrust that might be either locked or creeping or (2) they manifest tension in the material above the thrust, i.e., above the downgoing plate. The first possibility (with a locked fault) was preferred by Stephens et al. (1988) and the second possibility seems likely in view of the after effects of the rupture of the great 1964 Prince William Sound earthquake. It occurred in the far northeast part of the Kenai Peninsula and was recently described by Holdahl and Sauber (1994) and Johnson et al. (1996). Its rupture (or ruptures) produced NW-SE tensional deformation (as much as 0.0001 strain) across the peninsula and subsidence thereof (e.g., Holdahl and Sauber, 1994, their Fig. 1). Perhaps tensile stresses, which have yet to be compensated by the converging plates, are producing the normal faulting. (Although not their preferred interpretation in 1988, this possibility was mentioned by Stephens et al., 1983, pg. 20, and they noted that the tension axes of the several source solutions made at that time were generally E-W or NW-SE.)

That tensil stresses have yet to be compensated might be demonstrated by the inordinately high rate of uplift of the peninsula suggested by Cohen et al. (1995). They measured the cummulative amount of uplift since 1964 in 1993. Cohen and coworkers stated the cummulative uplift would require average horizontal displacements of about 100mm/yr, roughly twice that of the plate models, and they anticipate further, but diminishing, uplift in the future. Also, except for the region of Kodiak Island where thrust sources have been reported (e.g. as reviewed by DeMets et al., 1990), the earthquakes throughout the easternmost Aleutians typically do not show thrust mechanisms (Page et al., 1989, Page et al., 1991). However, instead of ongoing regional compensation, Page and coworkers suggest the earthquakes primarily manifest stresses in the downgoing plate, apparently a normal condition, as seemingly is the opinion of Stephens et al. (1988) for the earthquakes of Figure 6.

Thus, disappointingly, the reported locations and source parameters of the earthquakes of the Kenai Peninsula can not be used to precisely define the major thrust fault between the plates, but rather this thrust must be inferred from the geodetic data of the Kenai Peninsula (noted above) and Kodiak Island (Section 1.3) together with the source parameters of the earthquakes of the island.

The seismic reflection profiles of the central Aleutians show virtually undeformed sedimentary layers of the trench fill extending islandward several kilometers or more beneath the north trench wall (Scholl et al., 1982, their Figs. 6-8, Scholl et al., 1983, their Figs. 8, 10, McCarthy and Scholl, 1985, their Figs. 2, 5-7), and McCarthy and Scholl (1985) postulated they were thrust there, with the forearc being propagated seaward primarily by tectonic accretion. However, the position of the layers can be explained by burial which is depicted schematically in Figure 9: The trench is the toe of the sedimentary shelf and sedimentation, not thrust faulting and tectonic accretion, propagates the north trench wall over the older layers of the trench. Burial would explain the absence of chaotic deformation at the trench that one might expect for tectonic accretion or that one might expect in a subduction zone as indicated by deformation of the Uyak-McHugh complex (Section 2.1).

If the decollement and its associated deformation front propagate to the trench, the trench could become the leading edge of tectonic deformation as for SPT, or the trench could be completely destroyed. Because the forearc thins trenchward, dominant regional folding near the locked RTB-thrust might translate to dominant imbricate faulting (Figs. 8 and 10) as the traveling front of the decollement approaches the trench. It is noteworthy that, in contrast to SPT, there is not any genetic relationship between major thrust faulting and the trench, hence there is not any requirement that the decollement crop out there. However, if the decollement does crop out at the trench, the fault might be indistinguishable from the T-thrust of SPT, both structurally and seismically.

The decollement could have produced the intense deformation proposed (Moore et al., 1991) immediately landward of the Aleutian trench on the EDGE seismic line. However, if M-1 applied to their respective island arcs, two clearer examples of the deformation stage likely are the Nankai trough area (G.F. Moore et al., 1990) and the Barbados trench area (J.C. Moore et al., 1988). Both have decollements only tens of meters thick verified by drilling (contrasting to as much as 10 km for the Uyak-McHugh complex, thought the ancient subduction zone, Section 2.1), both are associated with seemingly propagating deformation fronts, and the decollements of both project beneath the trench as the model of the propagating decollement suggests might happen.

Although the model implies convergence of the plates as the driving mechanism for the tectonics, as inferred in Section 3.1, the geometry of the forearc basin in the central Aleutians, together with the age of the rocks deposited therein, can accommodate no more than several tens of kilometers of convergence since perhaps the mid-Cenozoic, whereas the conventional models of sea-floor spreading (Minster and Jordan, 1978) suggest many hundreds of kilometers of convergence since this time. If the model presented herewith is valid, and assuming the reported spreading rates (Minster and Jordan, 1978, DeMets et al., 1990) also are valid, SPT would require fundamental modification, if not world-wide, at least in the north Pacific. Modification of SPT world-wide would require explaining rapid convergence of the plates today that the geodetic data clearly indicate (e.g., Smith et al., 1990, 1994), but imbalance between the amounts of sea-floor spreading and convergence over the long-term. For instance, if expansion of the Earth is proposed as a candidate to account for the imbalance in the Aleutians, the amount of such expansion very likely must be proposed time dependent, with today's rapid convergence of the plates being not representative of the geologic past, except perhaps for the early Cenozoic in the central Aleutians (Section 3.1).

There is yet another puzzling aspect of the model of continuous transport: A region of outcrop for the RTB-thrust in the central Aleutians has not been identified. While pure dip slip might be accommodated entirely by arc-parallel folding of the surface rocks, with production of melange subsurface in the zone of earthquakes, ideally I would expect to see indications of the strike-slip component of the faults at the surface, especially off Amchitka where the plate models indicate slip about 50 degrees (DeMets et al. 1990) from normal to the arc. (It might be pertinent that slip directions of fault-plane solutions are more normal to the arc than are the model displacements, as discussed in Section 1.2.1.) Although the seismic reflection data show evidence of strike-slip faulting behind the trench and immediately behind the island arc (Scholl, personal communication 1995), there is no clear evidence of major faulting near the ridge-terrace border, either off Adak (Scholl, personal communication, 1995), or off Amchitka (Scholl, personal communication, 1981), and Scholl and coworkers (1983) show none off Atka-Amlia as well.

The absence of clearly displayed faults is indeed perplexing as off Amchitka intense seismic activity at the ridge-terrace border has been reported as shallow as 5 km or less (Fig. 3A-C). Also, the onshore-offshore seismic experiment suggested some of the earthquakes are very shallow at the ridge-terrace border off Adak as well, Figure 4A. Evidence of major faults on the ocean floor, relating to the earthquakes, might have been masked by sedimentation contemporaneous with displacements. Alternatively, Geist et al. (1988) have proposed microplates with strike-slip faulting both north of the volcanoes and between Hawley ridge and the trench. Perhaps the strike-slip component is manifest by displacements of lids such as these plates, with the northward protrusions of seismic activity seen on Figures 3A, B, D and E possibly being evidence of the displacements.

On the other hand, the earthquakes above the downgoing plate at the ridge-terrace border might be indicative of only secondary faulting as suggested by Taber et al. (1991), with the major displacements between the plates occurring on the T-thrust. Whereas the local networks documented little seismic activity beneath the forearc oceanward of the ridge-terrace border, even if the networks are adequately sampling the seismic activity, the hiatus could be evidence of locking on the T-thrust there, not absence of the fault as I assumed for M-1.

Return to Table of Contents

As noted in Section 1.3, others have proposed locking of the T-thrust as a mechanism for deformation and the concept appears to be widely entertained by those studying the geodesy of convergent margins. A model for Cascadia was given by Savage et al. (1991, their Fig. 10). It shows a locked zone near the trench, with a transitional zone landward of the locked zone, and the completely unlocked zone landward of the transitional zone beneath the northeast portion of the Olympic Peninsula. This simple model was discussed in some detail by them, and although it might not provide a unique fit to the observed deformation, Savage and coworkers concluded (pg. 14,505) that the model is reasonably successful in explaining both the instrumental observations and the geologically observed subsidence. Thus, entertaining locking appears to be amply justified.

In my concept of production of the forearc basin by discontinuous transport, this shelf basin is formed during times when the T-thrust is locked, and the Pacific plate carries the forearc shelf islandward and into the mantle (Fig. 11). As for continuous transport, the oceanward limb of the basin is produced by transport by the Pacific plate and the islandward limb of the basin is produced by displacements on the RTB-thrust far below the floor of the ocean.

The primary difference between production of the forearc shelf-basin by continuous and discontinuous transport is the means of propagation of the forearc oceanward: In the model of continuous transport the forearc is propagated oceanward by sedimentation, whereas in the model of discontinuous transport it is propagated oceanward primarily by tectonic accretion as for SPT.

During the production of the forearc shelf-basin, the absence of observed activity on the T-thrust has different implications from the model of SPT: Proponents of SPT usually interpret aseismic areas to indicate aseismic slip (e.g., Taber et al., 1991, their Fig. 3), whereas discontinuous transport suggests the aseismic areas might indicate locking, as stated above.

Also, in contrast to SPT, discontinuous transport does not require underplating to produce the geometry of the basin. However, the model does not exclude underplating having a contributing role.

As for production of the shelf-basin, the deformation of it during discontinuous transport is similar to the mechanism for continuous transport: When the forearc shelf can no longer be subducted by the Pacific plate that is locked to it, transport islandward by the plate locks the RTB-thrust (as for continuous transport, the RTB-thrust is manifest by the Uyak-McHugh formation) and thereafter transport without subduction compresses the forearc shelf-basin deforming it (Fig. 12). However, rocks oceanward of the deformed basin are primarily accreted sediments (Fig. 12), similar to the concepts of SPT, and the forearc is propagated oceanward primarily by tectonic accretion, not by sedimentation in situ followed by deformation as for continuous transport (Fig. 8). Stated another way, in the model of discontinuous transport as for SPT, the T-thrust always exists near the leading edge of propagation of the forearc.

The primary difference between SPT and discontinuous transport is the region of origin for the rocks immediately oceanward of the Uyak-McHugh complex. As noted in Section 2.3, a popular concept of SPT stipulates that these rocks were transported as large assemblages for hundreds or thousands of kilometers and were accreted to the North American plate (e.g., Plafker et al., 1994), whereas for the models presented herewith (both continuous and discontinuous transport) the rocks were deposited in the forearc shelf-basin of the region where they now exist and were deformed there.

In the model of discontinuous transport, because the T-thrust exists, the tectonics of the trench is as for SPT (e.g., McCarthy and Scholl, 1985). Namely, the layered trench fill beneath the islandward trench wall was underthrust, not buried as for continuous transport. Although chaotic deformation is not observed at the trench, the model stipulates that tectonic accretion occurs there and propagates the forearc oceanward as for SPT.

Here it is noteworthy that none of the three competing models (SPT, continuous transport, or discontinuous transport) can demonstrate chaotic deformation (i.e. akin to the Uyak-McHugh complex) that might be expected where major faults of the models are postulated to exist. Hence, fault-associated deformation can not be used to distinguish the models, especially since continuous transport provides a credible explanation for the undisrupted layered fill beneath the islandward trench wall.

Return to Table of Contents

In the portion of the Shumagin network west of the Shumagin Islands, although the very shallow zone of activity (Fig. 3F) is broader than in the central Aleutians, the zone of intense activity seems to dip steeply as for the zone in the central Aleutians. However, in the Shumagins and eastward, the zone of intense activity seems to show somewhat smaller dip (Fig. 3G, from Hudnut and Taber, 1987, their Fig. 3). This smaller dip is intermediate between that typical of the central Aleutians and Figure 3H, the zone beneath the Kenai Peninsula.

Abers (1994) used a subset of the Shumagin network data to invert for hypocenter location and model velocity. In the western part of the network, he also showed steep dip (his Fig. 6) as for Hudnut and Taber, but it must be noted he has poor control for shallow events and few of them. Nevertheless, as a general trend, the dips of the zones of hypocenters appear to progress to smaller values eastward, as for the more general depictions of Hudnut and Taber, but Abers showed no shallow hypocenters east of his Shumagin Island section.

Hypocenters restricted to the region of the Shumagins were given by Taber et al. (1991, their Fig. 7), who plotted two subsets of the data of Hudnut and Taber. The two subsets display the zones of shallow hypocenters for 40 km on either side of the Shumagin Islands (Taber et al., 1991, their Fig. 6). Their cross-sections depict a 10-km wide zone of intense activity of steep dip (they suggest 35 deg.) to the west, but a shallower zone of intense activity with poorly-defined modest dip (they suggest 10 deg.) occurs east of the islands, and it extends farther oceanward than the zone of steep dip.

The subtle changes in the pattern of shallow hypocenters shown by all three sets of investigators, together with the correspondences to the other networks and differences therefrom, suggest that the Shumagin Islands-western Alaska Peninsula might mark the region where the RTB-thrust system is being locked, hence marking the territory where the deformation of the forearc, so dramatically shown 400 km eastward on Kodiak Island (von Huene et al., 1979), begins.

There might be yet another indication of the transition. The region immediately west of the Shumagins shows two subparallel zones of deep earthquakes (Fig. 3F), with the steeper of the two zones displaying the greatest activity. However, the region to the east shows only one zone (Fig. 3G), with a dip similar to that of the zone of moderate activity to the west. Hence, the proposed locking might be affecting stresses in the downgoing plate as manifest by the patterns of the deep earthquakes.

Seismic reflection profiles made in the Shumagin region (e.g., Bruns et al., 1987) show a series of subbasins in the area between the peninsula and the trench. In the west part of the region, in the area between Unimak Island and the Shumagin Islands, the subbasins appear offset by a series of northwest striking transcurrent faults in a zone about 30 to 40 km wide. Whereas these faults extend almost all of the way across the forearc, transcurrent faults limited essentially to the islandward trench wall were reported throughout the region of the Shumagins (Lewis et al., 1988), with the strike of the faults varying somewhat systematically from northwest (in the western part of the region) to northeast (in the eastern part of the region).

As reported by Bruns and coworkers (1987), in the eastern part of the region, in the area between the Shumagin Islands and the Semidi Islands, what seems to be basement on the border of the subbasins shoals to as little as 0.5 km below the bottom of the ocean, but the basin fill shows essentially horizontal layers perhaps ranging in age from Quaternary to latest Miocene (the upper series of Bruns et al. 1987, pg. 163). The horizontal layers of the basin fill, together with the absence of pronounced deformation indicated by the geodetic data of the islands (Larson and Lisowski, 1994), suggest that the region has been remarkably stable throughout the latest Cenozoic.

Thus, whereas the transcurrent faults might be evidence of the transition region, the seismic reflection data appear to be useful mainly in an exclusory manner: Deformation similar to that reported on Kodiak Island seems to have not occurred in the Shumagin Islands region, even as far oceanward as the trench, at least during the latest Cenozoic. The remarkable stability suggests that the Shumagin Islands themselves might indeed lie landward of the active RTB-thrust system, as considered in Section 2.1, with contemporary deformation occurring primarily oceanward of the system as shown by the transcurrent faults between the islands and the trench.

Although addressing the details of the igneous petrology of the arc and forearc is beyond the scope of this paper, it should be recogonized that the proposed locking of the RTB-thrust system also seems to correspond to a transition zone in the general petrology of the arc: As Fournelle et al. (1994, pg. 730) have noted, whereas the lavas and pyroclastics are dominantly basalts and basaltic andesites from Mount Veniaminof (120 km northeast of the Shumagins) westward, to the east they are dominantly siliceous. The correspondence to the proposed locking suggests that the unlocked RTB-thrust system might now be, or might have been, a contributing conduit for basaltic magmas to the west, but during injection the conduit is not available to the east because it has been closed by compression. If so, conduits with associated plutons could explain, at least in part, the portion of the ~+150 mGal gravity anomaly north of the RTB-thrust system in the central Aleutians that follows the ridge and extends eastward, terminating east of Mount Veniaminof (Fig. 13, also shown by Fig. 1 of Abers, 1994). The concept of intrusives is consistent with the interpretations of McLean et al. (1983) who stated the metamorphism of Amlia Island suggests shallow plutons beneath the island, and a precedent for proposing the RTB-thrust system as a conduit for basaltic magmas might be given by the maps of von Huene et al. (1979) and Plafker et al. (1982). These show ultramafic and gabbroic rocks in several places in the Uyak-McHugh complex, and the rocks are mapped as lower Mesozoic, the time before the fault complex was locked by compression (in the context of this paper). However, all of the exposures are shown as fault bounded and hence not necessarily injected into the melange. Indeed, in the context of SPT the mafic rocks were formed at the base of the crust of an island arc and exposed by thrust-rotation of the entire crust upon which the basin of Cook Inlet was formed (e.g., Plafker et al., 1994, their Fig. 16); however, those studying the structure and stratigraphy of the basin (e.g., Kirschner and Lyon, 1973) have not independently proposed such vast rotation.

In contrast to the central Aleutians where a positive 150 mGal anomaly is associated with the RTB-thrust, there is only a small gravity peak (about 30 mGal) associated with Uyak-McHugh complex of the easternmost Aleutians. This anomaly was shown by Fisher and von Huene (1984) primarily for the area adjacent to and northwest of the Border Ranges fault. Perhaps the relatively small anomaly of the easternmost Aleutians might imply that the intrusives during the Mesozoic were minor compared to those during later times when the RTB-thrust was locked east of the Shumagins and injection of magma occurred in the unlocked portion west of the Shumagins.

While one part of the gravity anomaly roughly follows the ridge, but begins to bear off from it near Unimak Island and seems to fade northeast of the Shumagins, the other part of the gravity anomaly bears off sharply oceanward from the ridge near the Shumagins (Fig. 13), seemingly in association to the proposed locking as for its termination on the ridge. It is not clear what the sharp change in bearing signifies, but one possibility might be that new faults and conduits were developed in the forearc as a consequence of the locking, and these might have controlled emplacement of basaltic magma, after the conduits of the RTB-thrust system east of the Shumagins had been closed.

Return to Table of Contents

Return to Table of Contents

Proponents of strict Plate Tectonics (SPT) base evolution of the forearc of the Aleutian Islands (map, Fig. 1) on a giant thrust fault that supposedly crops out at the trench (T-thrust, the megathrust of conventional models). Conventionally, displacements on the T-thrust are proposed to balance the amount of sea-floor spreading, but a small group, perhaps led by Carey (1988), proposes the amount of sea-floor spreading to be balanced by expansion of the Earth. Whereas the geodetic data seem to clearly indicate convergence of the plates today approximately in accord with conventional plate models (e.g., Smith et al., 1990, 1994), it might be incautious to assume that this process has continued throughout the geologic past. Indeed, Stevenson and coworkers (1983), in studying the sedimentology of the Zodiak fan (Fig. 1), suggested no more than about 500 km of convergence of the plates since the mid-Cenozoic, whereas convention dictates 1500 to 3000 km since that time (Section 1.3).

Although this paper argues that the tectonics of the Aleutian forearc is driven by convergence of the plates, the paper makes no requirement that the amount of sea-floor spreading be balanced by the amount of convergence, in contrast to convention. Two scenarios are considered: Whereas one will not balance the amount of sea-floor spreading, the other will if specific conditions are met. Both of the scenarios are based on a model (Fig. 2, Section 1.2.1) developed from seismic refraction data and earthquake hypocenters of the central Aleutians. It shows underthrusting of the entire forearc shelf beneath the ridge at the ridge-terrace border (the RTB-thrust of Fig. 2), and both scenarios include this underthrusting which indicates transport of the forearc by the Pacific plate. (In the context of the central Aleutians, when I use the term forearc shelf, I mean the material between the ridge-terrace border and the trench, Fig. 2, and above the downgoing plate, as such it corresponds approximately to the accretionary wedge of SPT.)

One of the two scenarios entertains continuous transport of the forearc shelf, i.e., continuous underthrusting of the shelf (Section 3). This concept is in accord with the absence of intense earthquake activity beneath the forearc shelf of the central Aleutians. (The absence appears to be clearly demonstrated by local seismic networks both onshore, Figures 3A-3E, and onshore-offshore Figure 4A.) Transport of the entire forearc shelf also explains the undisrupted layers of trench fill beneath the north trench wall that seismic reflection profiles (e.g., Scholl et al., 1982) clearly show: The layers were buried (Fig. 9), not underthrust as is usually interpreted (e.g., McCarthy and Scholl, 1985).

However, while the model of continuous transport explains the near absence of seismic activity beneath the forearc shelf and the lack of deformation at the trench, it has its own puzzling features. Namely, due to the ages of the rocks of the forearc shelf and the geometry of the forearc basin described by Scholl et al. (1983), the model of continuous transport for the central Aleutians will accommodate only a few tens of kilometers of convergence of the plates since perhaps mid-Cenozoic, whereas the plate model (Minster and Jordan, 1978) suggests many hundreds of kilometers of convergence since that time. Also, seismic reflections reportedly show no evidence of major faulting at the ridge-terrace border (e.g. Scholl et al., 1983), even though very shallow earthquake hypocenters have been located there (Figs. 3A-3C, 4A). Although there are possible explanations for the absence of evidence of faulting (Section 3.5), such as sedimentation contemporaneous with deformation, one must consider that the RTB-thrust might be only a secondary fault and that the lack of intense activity beneath the forearc shelf of the central Aleutians might be due to temporary locking of the T-thrust that crops out at the trench, with the absence of pronounced deformation at the trench not withstanding (e.g., the reflection profiles of Scholl et al., 1982, noted above).

Discontinuous transport of the forearc shelf (Section 4) is the second of the two scenarios considered to produce thrusting at the ridge-terrace border of the central Aleutians. In this conceptual model, the RTB-thrust is only a secondary fault, with the primary fault being the T-thrust of SPT. Transport occurs when the T-thrust is locked (i.e., when the forearc shelf is locked to the downgoing plate). This type of transport of the forearc shelf could accommodate the proposed (Minster and Jordan, 1978) amount of convergence of the plates if decoupling on the T-thrust is dominant. Furthermore, intermittent transport implies that the forearc shelf would be propagated oceanward mainly by tectonic accretion as for SPT, not by sedimentation as for continuous transport, and this would be in accord with the postulates of many investigators (e.g., McCarthy and Scholl, 1985), albeit none of them advocate subduction of the entire forearc shelf as I do. Also, existence of the T-thrust, required for intermittent transport, would be in accord with the teleseismic locations of the earthquakes by Engdahl and coworkers (1987, 1989) who generally place the map locations more oceanward than do the local networks, but bias in the teleseismic solutions for the map locations cannot be ruled out (Section 1.2.1).

The models of continuous and discontinuous transport of the forearc shelf differ from the conventional concepts in three important ways. First, convention dictates only one synform in the central Aleutians, whereas the models presented herewith show two parallel troughs, one beneath the ridge and another beneath the forearc shelf (Figs. 2,, 7, and 11). The one of the forearc shelf (the linear forearc basin) is clearly shown by seismic reflection profiles of Scholl et al. (1983), and the south limb of the linear trough on the arc ridge seems to be indicated by the reflections as well (Section 2.2). The two troughs are separated by a thrust fault (the RTB-thrust) that produces the south limb of the ridge trough and the north limb of the forearc shelf-basin, both by differential displacements in the subsurface, see Figs. 7 and 11 for concepts for the two different scenarios considered. The two parallel linear synforms impact on the interpretation of the structures of the easternmost Aleutians. According to convention, the region northwest of the Border Ranges fault (Fig. 1), i.e. the volcanoes and the basin of Cook Inlet, corresponds to an entire island arc, and the region oceanward of the fault corresponds to the trench or to accreted assemblages that have been transported hundreds of kilometers (e.g., Plafker et al, 1994). In contrast, this paper argues that Cook Inlet basin corresponds to the trough on the arc ridge of the central Aleutians, and the region oceanward of the Border Ranges fault (e.g., Kenai Peninsula and Kodiak Island) corresponds to the forearc shelf-basin region of the central Aleutians (Section 2.2).

Second, according to convention, the south limb of the forearc basin of the central Aleutians is produced entirely by underplating, whereas according to concepts presented herewith (Sections 3 and 4), this limb of the shelf-basin is produced as the Pacific plate carries the forearc shelf downward toward the mantle. Figures 7 and 11 show this concept for the two different modes of transport. Underplating near the trench could exist only for the condition of intermittent transport, and although underplating is not required, it may play a contributing role in the islandward tilt of the south limb.

Third, convention provides no means to deform the forearc shelf-basin, and indeed some investigators regard it as part of an arc massif (e.g. Scholl et al., 1986), whereas this paper suggests that the forearc shelf-basin will be deformed when it can no longer be transported into the mantle by the Pacific plate (Sections 3 and 4). While production of the shelf-basin is occurring in the central Aleutians and is demonstrated by seismic activity of the RTB-thrust (Figs. 2, 3A-3E, Fig. 4A), deformation of the forearc shelf-basin is occurring in the easternmost Aleutians and is manifest by a high rate of uplift on the Kenai Peninsula (Cohen et al., 1995, Section 3.2).

Translation from production of the shelf-basin to deformation of it relates to abandonment of the RTB-thrust: When the forearc shelf can no longer be subducted, transport without subduction locks the RTB-thrust by compression. In the easternmost Aleutians, a Mesozoic zone of chaotic deformation (the Uyak-McHugh complex), up to 10-km wide, with steep northwest dip, crops out in the northwest part of the Kenai Peninsula and Kodiak Island (Section 2.1). It is delimited on the northwest by the Border Ranges fault, and when I mention the Uyak-McHugh complex I mean to include this fault. The Uyak-McHugh complex appears to be a subduction zone abandoned as such in the early Cenozoic (Little and Naeser, 1989) or perhaps earlier, and I postulate it corresponds to the locked RTB-thrust, i.e., the fault systems are one and the same: The RTB-thrust is recogonized as the Uyak-McHugh complex in the easternmost Aleutians.

Oceanward of the Uyak-McHugh complex, across the entire Kenai Peninsula and Kodiak Island, the rocks are faulted and severely folded as shown by the cross-sections of von Huene et al. (1979) and Plafker et al. (1982). I propose that this vast deformation was accomplished by transport of the forearc shelf to, and compression of it against, the North American plate which is delimited by the Border Ranges fault: As stated above, when the forearc shelf can no longer be subducted, transport without subduction severly deforms the shelf. Whether the transport is continuous, producing a propagating blind decollement (Fig. 8, Section 3.2) as the forearc shelf is removed from the downgoing plate, or intermittent by short-term locking of the T-thrust (Fig. 12, Section 4), a concept similar to those who are studying the geodesy, is not resolved.

Regardless of the mode of transport, the three-dimensional patterns of seismic activity (e.g., Taber et al., 1991) suggest that the RTB-thrust system is being locked in the area of the Shumagin Islands (Section 5). This means that the Shumagin Islands might mark the region of transition between production of the forearc shelf-basin to the west and deformation of it to the east. The largest part of the positive gravity anomaly (Fig. 13, Section 5.3) bears off oceanward in the Shumagins, whereas to the west it is generally parallel or sub-parallel to the ridge, and in the central Aleutians the largest positive values (about 150 mGal) are immediately islandward of the RTB-thrust. The sharp change in the heading in the Shumagin region might be associated with the proposed change in the tectonics: The unlocked RTB-thrust system of the model might have acted as conduits for magma, emplacing basaltic intrusives in the ridge of the central Aleutians eastward to the Shumagins; and farther eastward, where the conduits of the RTB-thrust system might have been closed by compression, the gravity anomaly might be related to basaltic magma injected into other faults and/or conduits developed as a consequence of the locking (Section 5.3).

In review, so that the differences in the faults, features, and mechanics of the three models (SPT, discontinuous transport and continuous transport) can be clearly discerned, they are shown in Figures 14 and 15. In Figure 14, note the differences between SPT and the other models for production of the forearc shelf-basin discussed above. In Figure 15 note the difference in interpretations for the area of the volcano also discussed above. For SPT assumption of an entire island arc for this region typically requires tectonic accretion to produce the huge thickness of material oceanward of the abandoned T-thrust of its model (the Uyak-McHugh complex); while for the models of this paper, the huge thickness represents the forearc shelf-basin region oceanward of its RTB-thrust (also the Uyak-McHugh complex).

For either mode of transport, the model sequence of deformation of the forearc appears correct (Section 6): The geologic record indicates that uplift of the region oceanward of the Uyak-McHugh complex began approximately contemporaneously with abandonment of the complex as a subduction zone, and eposodic uplift has continued to the Recent.

Return to Table of Contents

In the central Aleutians, both SPT and discontinuous transport dictate rapid underthrusting of the trench fill beneath the north trench wall or accretion to it, hence the maximum age of the fill in the trench is determined by its residence time there. On the other hand, the model of continuous transport suggests that the trench fill will be buried as the forearc shelf propagates oceanward by sedimentation (Fig. 9). In both instances, the oldest layers of the fill occur at the base of the column at the foot of the north trench wall, and the seismic reflections show that the basal layers can be traced beneath the wall, as discussed (Section 3.3).

At the foot of the wall off Amlia Island, by using seismic reflections, Scholl et al. (1982) estimated the maximum thickness of the fill to be about 4 km. As Scholl and coworkers discussed (1982, pg. 124), assuming the long-term rapid convergence of the plates at the T-thrust (as for SPT or discontinuous transport), the residence time in the trench of the sediments is only 0.3 Ma. However, by using arguments based primarily on the age of the sediments of the easternmost Aleutian trench, they estimated the maximum age to be somewhat more, about 0.5 Ma. Apparently to obtain the required rate of sedimentation, they (pg. 124-125) proposed the eastern Aleutian trench, 1200 km away, as the source, with transport therefrom by a turbidity current driven by the gradient of the trench floor.

Because the trench fill of the model of continuous transport is buried by the propagating forearc, the age of the oldest layers beneath the north wall relate to the rate of propagation of the forearc oceanward, not to the rate of underthrusting as for SPT and discontinuous transport. Hence, the oldest layers, at the base of the 4-km column of fill, likely are much older than the 0.5 Ma dictated by the competing models.

Herein lies the means to distinguish SPT and discontinuous transport from continuous transport. If the basal fill can be shown to be only 0.5 Ma as proposed by Scholl and coworkers, clearly the paradigm of rapid convergence of the plates would be strongly supported, and the model of discontinuous transport (or indeed, SPT) would prevail over the model of continuous transport. Moreover, challenges to the paradigm from those such as Carey (1988) and his colleagues would be muted. On the other hand, if the fill can be shown to be likely an order of magnitude or more older, the model of continuous transport would prevail as only a small amount of convergence of the plates would be inferred, at least for the late Cenozoic of the central Aleutians. To shed light on these issues, the fill should be drilled as soon as practical to determine its ages.

Return to Table of Contents

The 1964 Aleutian Island Seismic Experiment, which produced data for the root model, was conducted by Don Tocher (deceased, in 1964 he was Director of the former Advanced Seismic Experiments Group and later of the former Earthquake Mechanism Laboratory, San Francisco), James Taggart, and John Hoffman. In an earlier unpublished version of the current paper, David Scholl provided many helpful comments, even though he prefers alternate hypotheses. Support for the earlier (1981) work was also provided by Allen T. Chen who reported the seismic activity of the Aleutians and who also maintained a competing model. The comments of Michael Fisher in 1985 must be noted as they have impacted on this paper. Illustrations were drafted by Chris Krohn. I thank all of those who made constructive remarks, and especially the readers whose published concepts the paper challenges. I am indebted to Larry Jaksha who has made comments on several of the papers. I thank Simon Klemperer who made a number of strategic remarks on my talk at the recent (1995) AGU Fall meeting.

Return to Table of Contents

Comments and questions should be directed to James N. Murdock, 611 Green Valley Dr. SE, Albuquerque, NM 87123, USA, (phone: 505.299.5254).