Abstract

Extensive Quaternary glacial cover and a lack of dense geophysical data within the Cook Inlet basin (CIB) of south-central Alaska make locating and determining the geometry of the Border Ranges fault system (BRFS), a major feature of the Alaska-Aleutian forearc region, difficult. We use recently collected gravity data, available aeromagnetic data, and other geophysical information as constraints to develop plausible two-dimensional cross-section models that better image the BRFS and related geologic structures of the CIB. Our integrated models show a thick sequence of late Mesozoic sedimentary rocks and the Peninsular terrane basement (6–20 km depth) overlying a serpentinized body at a depth of 16–34 km. The late Mesozoic rocks and serpentinite are interpreted as possible sources of the south Alaska magnetic high over the CIB. The eastern boundaries of the CIB are characterized by gravity and magnetic highs of the emplaced Border Range ultramafic and mafic assemblages (BRUMA). Formation of the BRUMA may be related to the serpentinized rocks that composed a Jurassic oceanic arc. Our models constrain the BRFS as a structural boundary between the overthrusted BRUMA and the Chugach terrane to the east. The BRFS dips 50°–70° toward the west-northwest and extends to at least 15 km. The BRFS may penetrate steeply or shallowly to a form a décollement at greater depths. A model that includes underplated sediments at the base of the accretionary complex (12–40 km) is consistent with the observed gravity low over the Chugach Mountains (Chugach terrane). The underplating may be associated with the subducting and shortening of the Yakutat microplate in south-central Alaska.

INTRODUCTION

The Border Ranges fault system (BRFS) has played a major role in the development of the forearc basin system of south-central Alaska since Jurassic time (Pavlis and Roeske, 2007), serving as the boundary between the forearc of Cook Inlet basin (CIB) and the Chugach terrane accretionary complex (Fig. 1; Shellenbaum et al., 2010). Surface exposure of the BRFS is limited, and only a few geophysical surveys (e.g., Burns, 1982; Fisher and von Huene, 1984) have been conducted across it. We know little about the subsurface structure of this feature or how it has influenced the past and present-day tectonics of the region. The goal of this study was to use new gravity data collected across the BRFS to develop integrated geophysical and geological models of regional structure that could be used as a starting point for understanding how the BRFS has influenced regional tectonics, and for planned three-dimensional (3D) modeling studies.

The focus of this study is along the inferred trace of the BRFS extending from the central Kenai Peninsula to the Castle Mountain fault (Fig. 1), including the western Kenai Peninsula, the Anchorage area, and the eastern Matanuska-Susitna Valley. Throughout much of the region, the BRFS is marked by a pronounced change in topography from the CIB to the Chugach Mountains. However, no seismicity or recent fault scarp appears to be associated with the BRFS (Pavlis, 1982; Fisher and von Huene, 1984; Pavlis and Roeske, 2007).

In this study we used a new dense set of gravity data collected between 2009 and 2011 and existing aeromagnetic data to model plausible geometries for the inferred BRFS. We first produced new corrected anomaly maps to distinguish different geologic features based on density and magnetic susceptibility properties. We then used the updated geophysical potential field data to construct 2D density-magnetization cross-section models. Other geological and geophysical data (e.g., well logs, seismic tomography, and surficial geology) were used to help constrain the model parameters and address problems of non-uniqueness related to tradeoffs between modeled thicknesses, depths, densities, and magnetic susceptibilities in order to minimize model uncertainties (Saltus et al., 2001). We also compared our results to geological processes found in other, well-studied subduction zones, such as serpentinization and underplating of sediments, to help explain the possible causes of anomalies present in our new interpretations.

BACKGROUND GEOLOGY

The BRFS spans the region between the eastern CIB and topographically higher Kenai and Chugach Mountains. The CIB is bounded to the north by the Castle Mountain fault and to the west by the Bruin Bay fault (Figs. 1 and 2) (Fisher and von Huene, 1984; Shellenbaum et al., 2010). The three major subregions of the CIB are the Anchorage Lowland, Kenai Lowland, and Cook Inlet. The CIB is filled with continent-derived sediments ranging in age from Mesozoic to Quaternary that collectively define the forearc basin stratigraphy and the cover sequence of the Peninsular terrane (Plafker et al., 1994; Trop et al., 2005). The small-scale topography of the basin is affected by multiple Quaternary glaciations and recent alluvial and tidal processes (Swenson, 1997; Haeussler and Saltus, 2011). Glacial deposits have an average thickness of 180 m (Plafker et al., 1989) and may reach a thickness of 1200 m in the basin axis (Swenson, 1997).

The Quaternary glacial deposits unconformably overlie Late Eocene to Late Pliocene sedimentary rocks, termed the Kenai Group, a terrestrial section deposited in a forearc setting (Fig. 1; Haeussler and Saltus, 2011; Swenson, 1997). According to borehole data, the Kenai Group contains cross-bedded to massive sandstones, siltstones, claystones, and shale with an estimated total thickness of ∼2 km near the basin axis and thins radically to both the basin edges (Plafker et al., 1989). Five nonmarine formations are classified, including the Sterling, Beluga, Tyonek, Hemlock, and West Foreland Formations (Swenson, 1997). The Kenai Group overlies Paleogene nonmarine facies that record an initial Tertiary uplift and cessation of Mesozoic depositional patterns (Swenson, 1997). Seismic reflection profiles of Tertiary formations show folded and faulted structures that initially formed during Early Eocene to Late Oligocene time (Little and Naeser, 1989), some of which may be currently active (Haeussler et al., 2000). Paleogene sequences are underlain by late Mesozoic sequences to the west, but onlap crystalline basement to the east.

The late Mesozoic sequences are widely recognized as a succession of shallow-marine sedimentary rocks of Early Jurassic–Cretaceous age with an approximate thickness of ∼8500 m (Plafker et al., 1989). The late Mesozoic sedimentary rocks extend well east of the present CIB across the trailing edge of the Peninsular terrane. At least 3000 m of clastic, volcanic, and volcaniclastic rocks of Late Triassic to Early Jurassic Talkeetna Formation are underneath this succession, and compose the volcanic cover coeval with the Early Jurassic intrusive assemblages of the Peninsular terrane basement (Trop et al., 2005; Saltus et al., 2007; Clift et al., 2012). In the Anchorage area the Mesozoic rocks, as well as their late Mesozoic–Paleogene cover, emerge from beneath Neogene sediments and are well exposed to the east in the Matanuska Valley and northern Chugach Mountains (Pavlis and Roeske, 2007).

The Border Ranges ultramafic-mafic assemblage, or BRUMA, is identified as the Peninsular terrane basement rocks along the eastern forearc boundary (DeBari and Coleman 1989; Plafker et al., 1989). It is stratigraphically overlain by Middle Jurassic to Paleogene sedimentary sequences to the east of the CIB, but the primary depositional relationships are complicated by younger structures (Trop and Ridgway, 2007). The BRUMA is composed primarily of plutonic rocks that range in composition from gabbro to tonalite with local occurrences of ultramafic rocks (e.g., Burns, 1982; Plafker et al., 1994) and a fragmented crustal section of an Early Jurassic oceanic arc system (Plafker et al., 1989; Pavlis and Roeske, 2007). The ultramafic rocks of the BRUMA are interpreted as the upper mantle roots of the arc overlain by lower crustal mafic rocks (e.g., Burns, 1982; DeBari and Coleman 1989) that pass upward to early Jurassic volcanic cover of the Talkeetna Formation along the Chugach Mountain fronts (e.g., Pavlis and Roeske, 2007). Outcrops in the northern Chugach Mountains and previous aeromagnetic data show that the BRUMA forms a 5–20-km-wide outcrop belt (Burns, 1982; Burns et al., 1991) that is mostly concealed, except for a small part along the boundary between the eastern CIB and the western Kenai and Chugach Mountains.

The BRFS separates the BRUMA from the Chugach terrane throughout the Kenai and Chugach Mountains (Nokleberg et al., 1989). The Chugach terrane represents a subduction complex that was accreted by northwest-directed subduction (modern coordinates) along the margin (Plafker et al., 1994), but the BRFS is overprinted complexly by younger fault systems (Pavlis and Roeske, 2007). The Chugach terrane consists of two major lithotectonic assemblages separated by the Eagle River thrust fault: the older McHugh Complex mélange assemblage, and the younger Valdez Group metasedimentary rocks (Fig. 2) (Pavlis and Roeske, 2007). Tertiary intrusions that are related to Eocene ridge subduction are found within the Chugach terrane but are limited to small intrusions in the Anchorage area (Hill et al., 1981; Fuis and Plafker, 1991; Plafker et al., 1994). The younger accreted deep-sea fan complex that forms the Prince William terrane is separated from the Chugach terrane by the Contact fault system (Fig. 1) (Plafker et al., 1989; Fuis and Plafker, 1991).

Border Ranges Fault System

MacKevett and Plafker (1974) first defined the BRFS as a major structural feature extending more than 1300 km from Kodiak Island on its southwestern end to Baranof Island on its northeastern end. The southwestern segment of the BRFS is the focus of this study, and is defined as an arc-forearc boundary of the modern Alaska-Aleutian subduction zone (Pavlis and Roeske, 2007). The BRFS is exposed extensively in the western and northern Chugach Mountains (Pavlis and Roeske, 2007). The buried location of the BRFS beneath the Cenozoic fill has been inferred from shallow and deep geophysical data (e.g., Shellenbaum et al., 2010). Previous studies in the Cook Inlet area suggested that the BRFS forms a 2–10-km-wide suture zone (Burns 1982; Fuis and Plafker, 1991) that dips steeply to vertically toward the north or northwest (Fisher and von Huene, 1984).

The history and sense of fault movements on the BRFS from the Jurassic to middle Cenozoic are debated because of the complexities of the deformational history of the southern Alaska subduction complex (Little and Naeser, 1989; Fuis and Plafker, 1991). The Jurassic history of the BRFS is poorly defined, in large part due to overprinting and a Middle Jurassic period of subduction erosion (e.g., Clift et al., 2005). Recent work (e.g., Amato and Pavlis, 2010) indicates episodes of accretion and subduction erosion during the Jurassic and Early Cretaceous, followed by middle Cretaceous accretion of part of the McHugh Complex, followed by accretion of the Valdez Group. Cretaceous subduction accretion was followed by dextral motion on the BRFS in the northern Chugach Mountains that appears to have been transferred westward to the Castle Mountain fault in early Cenozoic time (Clift et al., 2005; Pavlis and Roeske, 2007). The timing of the lateral motion may be associated with the onset of oroclinal bending and right-oblique subduction of the Kula plate in southern Alaska (Little and Naeser, 1989; Pavlis and Roeske, 2007). Most recently, high-angle normal movement toward the northwest appears to have occurred on the southwesternmost segment of the BRFS (Pavlis and Bruhn, 1983; Little and Naeser, 1989), correlated to late Neogene forearc basin development (Pavlis and Roeske, 2007). An absence of evidence for Neogene reactivation of the BRFS suggests that it did not act as a reverse fault after the Late Cretaceous Period (Fisher and von Huene, 1984). Haeussler and Saltus (2011) presented two observations based on contrasts of topography and magnetic anomalies across the BRFS to support the hypothesis of Pavlis and Bruhn (1983), that the BRFS was reactivated by normal faulting in the Neogene. However, inferred normal faulting remains controversial (Haeussler and Saltus, 2011); faults are not well imaged on several noisy seismic reflection profiles across the BRFS, and no evidence for normal fault offsets has been observed on exposed fault scraps (Haeussler and Anderson, 1997).

GEOPHYSICAL DATA COLLECTION AND PROCESSING

We collected ∼1400 gravity stations across the BRFS within the Kenai and Anchorage lowlands from 2009 to 2011 using a Lacoste and Romberg Model G gravimeter (Fig. 2). Associated elevation and coordinate control information was collected by a Topcon GB-1000 global positioning system receiver with processing implemented in steps to achieve <0.1 m elevation control. The errors associated with readings taken by this gravity meter include those due to reading resolution, repeatability, and nonlinear instrument drift. The nulling dial has an accuracy of 0.01 mGal. The repeatability is ∼0.02–0.03 mGal based on numerous readings at base stations during the data acquisition stage. All acquired gravity values were tied to known local absolute gravity stations and corrected to free air and simple Bouguer anomalies. These recent data were combined with ∼3000 existing land and marine regional gravity observations from U.S. Geological Survey and University of Texas at El Paso databases collected before 2000 (Fig. 2).

We employed aeromagnetic data (compiled and reprocessed from Saltus and Simmons, 1997) to show magnetic intensities related to magnetic material distributions and structural features across the BRFS and south-central Alaska. The selected aeromagnetic data are based on four separate surveys conducted between 1954 and 1977 with variable flight directions (north-south and east-west), altitudes (120–760 m), and flight line spacings (1600 m to 16,000 m). The Cook Inlet survey was covered by a coarser line spacing range and data sampling grid compared to the land surveys. The flight spacing ranges and altitudes of the original survey grids were adjusted to minimize differences at the boundaries by applying upward or downward continuation and converted from level to drape as necessary to produce a consistent survey specification of 305 m above ground (Saltus and Simmons, 1997).

We used Geosoft Oasis montaj software to process reduced gravity and aeromagnetic anomalies with the same grid size of 1000 m and applied a minimum curvature gridding technique. A standard density of 2670 kg/m3 was chosen for the Bouguer correction (Burger et al., 2006) to remove the gravity slab effect (Fig. 3). A terrain correction was not applied to the gravity data, because this study is focused on deeper geologic structures related to long-wavelength signals that will serve as starting models for the next step in our analysis, a 3D inversion of the free air gravity data. In the 3D analysis we will build a model that explicitly incorporates topography with surficial variations in geology and density, eliminating the assumption of an average near-surface density that is used in most terrain correction algorithms. Total intensity aeromagnetic data were reduced-to-pole filtered to view all magnetic sources produced vertically and symmetrically (Blakely, 1995) with an inclination of 73° and declination of 25° (Fig. 4), presumable averages of values over the 1954–1977 time interval (Saltus and Simmons, 1997). A horizontal gradient analysis was performed to delineate geologic contacts or faults across which rock density or magnetic susceptibility differs (Figs. 5 and 6) (Blakely, 1995). We also used fast Fourier transform Gaussian low and high pass filtering analyses to distinguish residual and regional magnetic sources at different depths (Figs. 7 and 8).

INTERPRETATION OF RESULTS

Gravity Anomaly Interpretations

Northeast-striking gravity anomalies show two prominent features over the study area: (1) gravity lows within the CIB, and (2) gravity highs within the surrounding mountains. The deep gravity lows (–100 to –160 mGal) extend from the Kenai and Anchorage lowlands into Cook Inlet, corresponding well with the boundaries of the CIB (Fig. 3). The minimum gravity lows are within northern Cook Inlet and the northwestern Kenai Peninsula (Fig. 3). The maximum gravity highs (19 mGal) are found in a belt along the eastern margin of the CIB (Fig. 3). This belt likely represents buried high-density rocks of the BRUMA (Burns, 1982). Strong gravity gradients are observed at the northwestern edges of the CIB that are correlated to the locations of the Bruin Bay and Castle Mountains faults (Fig. 5). Even stronger gravity gradients are observed to the east of the CIB that may be related to possible locations of the BRFS (Fig. 5).

The BRFS appears to be associated with two regions of strong gravity gradients between the CIB and the BRUMA, and between the BRUMA and McHugh Complex (Figs. 5 and 9). The gravity anomaly over the accretionary complex generally increases southeastward toward the Aleutian Trench (H in Fig. 3), but decreases (L in Fig. 3) northeastward toward the central Chugach Mountains (Fig. 3). The change between higher gravity over the Kenai Mountains and lower gravity over the Chugach Mountains occurs near Turnagain Arm (Cook Inlet; Fig. 3). We suggest that this observed change in gravity anomaly over the same accreted terrane may be related to the southwestern edge of the subducted Yakutat microplate (SEY in Fig. 9), the location of which is inferred by the tomographic studies of Eberhart-Phillips et al. (2006) (Figs. 3 and 9). This gravity low is also seen on a free air anomaly map obtained from airborne gravity studies of the Cook Inlet region (GRAV-D Science Team, 2012), supporting the idea that this low is not an artifact of sparse data coverage or improper data correction in this region.

Magnetic Anomaly Interpretations

Magnetic intensity lows striking northeast-southwest are subparallel to the structural fabric of the CIB and strike in the same direction as the gravity anomalies, but they have different anomaly characteristics. The magnetic anomaly map expresses an abnormal feature for basin fill with prominent magnetic highs over the basin, termed the south Alaska magnetic high (Godson, 1984) (Fig. 4). The low pass filtered map shown in Figure 7 supports the hypothesis that the south Alaska magnetic high may correlate with fluid serpentinization of the altered lower forearc crust and/or mantle at 16–34 km depth (Saltus et al., 2001; Hyndman and Peacock, 2003). Shallow sediments and rocks (<15 km) are not a likely source of the magnetic highs because high pass filtered maps show no broadly strong magnetic highs (Fig. 8). The magnetic intensity highs to the west of Cook Inlet are likely related to intrusive and extrusive bodies associated with the active volcanic arc (Figs. 1 and 4) (Saltus et al., 2001). A linear contact between the long-wavelength highs of Cook Inlet and these smaller highs of the arc occurs at the position of the Bruin Bay fault. The eastern flank of the CIB is bordered by narrow magnetic highs (200–230 nT), termed the Knik Arm anomaly (Fig. 5; Grantz et al., 1963; Fisher and von Huene, 1984), that are in the same locations where gravity highs are observed. The BRUMA is presumably responsible for the Knik Arm anomaly. This hypothesis is supported by exposed ultramafic bodies along the mountain front between the Knik River and the Eagle River (KR and ER in Fig. 2) (Burns, 1982). The width of the Knik Arm anomaly based on the magnetic intensity gradient is narrower than the width estimated from gravity gradient analysis (Figs. 5 and 6). This difference may indicate that there is a less magnetic source, such as a mafic assemblage, located at the edges of the BRUMA (Burns, 1982). Strong gradients on the eastern flank of Knik Arm correlate to the eastern edge of the BRFS, which is interpreted from gravity data (Figs. 5 and 6). Relative magnetic lows observed over the topographically high regions of the Chugach terrane reflect fewer and/or no magnetic source rocks of the accretionary complex (e.g., Saltus et al., 2007).

DATA CONSTRAINTS FOR 2D INTEGRATED FORWARD MODELS

We used the Geosoft GM-SYS software package to produce 2D forward models of the geologic structures over the study area at both local and regional scales. Four subparallel local transects (profiles A–A′, B–B′, C–C′, and D–D′ in Figs. 3, 4, and 9) were selected for the 2D forward modeling shown in Figure 10. All profiles strike northwest-southeast in a direction perpendicular to geological structures in order to illustrate the subsurface BRFS and related geologic features. Most profiles, except the northernmost (D–D′), start from the western flank of the forearc basin and extend eastward across the CIB, the BRFS, and end at the Chugach and Kenai Mountains. We modeled structures to a depth of ∼50 km and assumed homogeneous bodies extending orthogonal to the profiles to distances effectively of infinity (±30,000 km). Because input data observations were derived from gridded simple Bouguer gravity anomaly and reduced-to-pole aeromagnetic anomaly data, the 2D integrated forward models discussed here are termed 2D integrated gravity and magnetic (2DGAM) models. We also modeled two longer regional transects (profiles E–E′ and F–F′ in Fig. 1) that cross the entire subduction zone from the trench to the arc and extend to depths of 120 km (Fig. 11). E–E′ is a southern profile where more normal subduction of the Pacific plate is occurring, while F–F′ is a northern profile that crosses the region of flat slab subduction of the Yakutat microplate (Fuis and Plafker, 1991).

To minimize errors of non-unique solution between the observed and calculated potential fields, the modeling programs require reasonable initial estimates of model parameters such as topography, depth, body shape, density, and magnetization of suspected sources. Several geologic maps from the U.S. Geological Survey database compiled by Wilson et al. (2009) were used for geologic contacts and fault constraints. Two digital elevation model data sets, the National Elevation Data set (last updated by Gesch et al., 2002; Gesch, 2007), and the Alaska Coastal Digital Elevation Model (Lim et al., 2009), were used for topographic and bathymetric constraints, respectively. Published geophysical cross-section models from Ehm (1983), Fisher and von Huene (1984), Fuis and Plafker (1991), Ye et al. (1997), Haeussler et al. (2000), Saltus et al. (2001, 2007), and Romero (2011) were used to guide the initial depth and thickness estimates for shallow and deep geologic features constructed in our 2DGAM models.

Table 1 provides information on density and magnetic susceptibility variations used in the 2DGAM models. A density of 2100–2200 kg/m3 was selected for late Tertiary sediments and sedimentary rocks of the lowland, and 2700–2740 kg/m3 was chosen for metamorphic rocks of Chugach terrane based on an estimation method outlined in Mankhemthong et al. (2012) and measurements of collected hand rock samples. Shallow densities for the basin deposits were derived from density logs obtained from the Alaska Oil and Gas Conservation Commission (2002). These gave a density of 2400–2500 kg/m3 for early Tertiary sedimentary rocks within the CIB. Previous studies (Fisher and von Huene, 1984) gave a density of 2670–2700 kg/m3 for late Mesozoic rocks. Gardner et al.’s (1974) velocity-density conversion formula was used to convert from the seismic velocity models of Fuis and Plafker (1991) and Fisher and von Huene (1984) to rock densities for deeper geology. For example, values of 2900–3000 kg/m3 were used for the BRUMA body, values of 2550–2600 kg/m3 were chosen for a low-velocity density zone, and 3300 kg/m3 was used for unaltered mantle. Based on several previous studies (e.g., Christensen, 1966; Coleman, 1971), densities of 2750–2800 kg/m3 were chosen for a serpentinized body beneath the Peninsular terrane.

Magnetization values used in the 2DGAM models were based on measured magnetic susceptibilities for the Cook Inlet (Altstatt et al., 2002; Saltus et al., 2005; Saltus et al., 2007) and Talkeetna Mountains regions (Sanger and Glen, 2003), and geophysical magnetic surveys covering the Kenai Peninsula (Table 1) (e.g., Burns, 1982; Saltus et al., 2001). Altstatt et al. (2002) classified magnetic source rocks into three groups based on susceptibility measurements, including low (<0.01 SI), moderate (0.01–0.10 SI), and high (>0.10 SI) magnetic sources. Thus high sources were used for a serpentinized body beneath the CIB (0.03–0.13 SI; Godfrey and Klemperer, 1998; Carlson and Miller, 2003). Moderate magnetic sources were correlated with the BRUMA (0.010–0.05 SI) and the volcanic arcs (0.025 SI). Magnetic susceptibility lows were associated with the metamorphosed rocks of the McHugh Complex and Valdez Group (0.001–0.006 SI). Very weak magnetic or nonmagnetic sources, such as stratified sedimentary formations, were assigned values of 0–0.001 SI. Among the basin sedimentary groups the late Mesozoic sediment sequences contain the highest magnetization; this high value may be related to its high amount of mafic-ultramafic mineral components (e.g., Burns, 1982).

The interpretations of integrated gravity and magnetic anomaly data yield nonunique solutions between rock density and magnetic susceptibility and modeled geometry. Several model families can be constructed to fit the observed data. The most geologically reasonable models that exhibit the least structural complexities and minimum misfit values were selected for the final models shown in Figure 10. Figure 12 shows alternative models used to test the sensitivity of the modeling process to variations in geologic structure or magnetic susceptibility. In this study we accepted a root mean square misfit error on the 2DGAM models when it was not >4 mGal for gravity and 30 nT for magnetic forward models (Figs. 10, 11, and 12).

2D CROSS-SECTION RESULTS

The 2DGAM models across the central CIB show Cenozoic and Mesozoic sedimentary rock sequences as deep as ∼10 km filling the forearc basin along the southern profiles and ∼6 km along the northern profiles (Fig. 7). Based on previous geological and geophysical studies (e.g., Ehm, 1983; Haeussler et al., 2000; Green, 2003; Shellenbaum et al., 2010) and well log analysis, we modeled three sedimentary rock formations with different densities: (1) late Tertiary, (2) early Tertiary, and (3) late Mesozoic rock sequences (Table 1); the thicknesses of these units in the center of the CIB are ∼1800, ∼4000, and ∼4500 m, respectively. Thickness changes observed on well logs provide evidence for normal faulting and reverse faulting associated with anticlinal structures (Shellenbaum et al., 2010), but cannot be modeled accurately on the 2DGAM models because of the coarse grid size (1000 m). These sedimentary rock sequences are modeled as overlying the basal crust of the Peninsular terrane (Figs. 10 and 11).

An ultramafic (BRUMA) block is added to match the gravity highs adjacent to the eastern flank of the CIB and ultramafic surface exposures in the northern Chugach Mountain fronts (C–C′, Fig. 10) (Pavlis and Roeske, 2007) and to the eastern flank of the Chugach terrane (E–E′ and F–F′, Fig. 11). Misfits between observed and calculated anomalies over the ultramafic body, especially along the eastern margin of the body, may be related to mixed assemblages of mafic-intermediate composition rocks, changes in the thickness and/or composition of overlying sedimentary rocks, and/or fault thickening/thinning of units (Burns, 1982; Little and Naeser, 1989; Haeussler and Saltus, 2011).

A serpentinized body is added to constrain a broad south Alaska magnetic high over the CIB, as inferred from models by Saltus et al. (2001). According to our models, the serpentinized body is also part of a lithospheric section of an ancient oceanic arc system, as is the adjacent BRUMA block. We modeled the serpentinized body as a half-wedge shape with a thickness of >20 km under the eastern CIB and ∼10 km under the basin center (A–A′ and B–B′, Fig. 10). Due to westward thinning of the serpentinized body, it probably pinches out under the western edge of the CIB (D–D′, Fig. 10). Our serpentinization model is consistent with models of Blakely et al. (2005) that suggest the source of the magnetic highs continues farther eastward. From the southern profile (A–A′, Fig. 10) to northern profile (D–D′, Fig. 10), the serpentinized body was modeled as decreasing in size in order to constrain the observed decreasing magnetic intensity amplitude toward the northern CIB (D–D′, Fig. 10). We also considered the possibility that the Talkeetna Formation could be part, or all, of the source of the high magnetic anomaly (Ab–Ab′, Fig. 12).

The 2DGAM models show that the Chugach terrane is composed of two metamorphosed assemblages with slightly different densities and magnetic susceptibilities, i.e., the McHugh Complex and the younger Valdez Group. The two assemblages are separated by the 20°–50° dipping Eagle River thrust fault (D–D′, Fig. 10; F–F′, Fig. 11). A shallow intrusion is included on profile C–C′ to decrease misfits between observed and calculated anomalies that may be caused by heterogeneous lithologies and multiform granitic intrusions in the accretionary complex (Pavlis and Roeske, 2007).

A homogenous low-velocity density zone on profiles C–C′ and D–D′ is required at the base of the accretionary complex to match observed gravity lows along the northern profiles (Figs. 10 and 11). The top of the low-velocity density zone extends to within ∼12 km of the surface on C–C′ and gently dips downwards to the northwest and southeast. The thickness of this layer is uncertain; based on the seismic cross sections of Ye et al. (1997), the low-velocity density zone appears to be at least 10 km thick (Fig. 11). A trenchward increase in the gravity anomaly along profiles A–A′ to D–D′ (Fig. 10) is constrained by a gently dipping oceanic slab with density highs beneath the low-velocity density zone and the accretionary complex.

DISCUSSION

Geometry of the BRFS

Our integrated 2D gravity-magnetic models suggest that the concealed BRFS is a northwest-dipping, high-angle fault between the Peninsular and Chugach terranes. Two strong potential field gradients that bound anomaly highs are associated with the BRFS (Figs. 5 and 6). The eastern gradient corresponds with a major fault located along the Chugach Mountains and Kenai Mountains fronts (Fig. 9). This interpretation is consistent with previous geophysical models (e.g., Burns, 1982; Fisher and von Huene, 1984; Green, 2003) and geological cross sections (Wilson et al., 2009). The western gradient may be another branch of the main fault or represents normal reactivation of the fault that is related to Late Cretaceous or Neogene uplift of the accretionary complex (Arkle, 2011).

The dip angle of the BRFS is controversial; there are insufficient geophysical data to image its deeper geometry. Our modeling results shown in profiles A–A′ to D–D′ (Fig. 10) suggest that the fault dips steeply (∼70°) toward the west–northwest and penetrates to at least 12 km, based on 2D gravity and magnetic cross-section profiles (Fisher and von Huene, 1984) and seismic tomography models (Fuis and Plafker, 1991). Due to the inherent non-uniqueness of modeling gravity and magnetic data, we tested an alternative (Aa–Aa′, Fig. 12) for the BRFS with a dip of 50° toward the northwest along profile A–A′ based on a magnetotelluric (MT) model presented by Green (2003). This model of the BRFS also provided a reasonable result with acceptable misfit errors, as shown in Figure 12. Thus, we cannot choose between these models without further geologic or geophysical constraints.

It is also difficult to constrain the distance that the faults penetrate the lower crust. Based on the seismic models of Fisher and von Huene (1984), the dip angle of the BRFS seems to decrease with depth to become a décollement at ∼16 km depth, with many steeply dipping fault branches. According to our 2DGAM results, the BRFS cuts through ultramafic rocks and serves as the structural boundary between serpentinized and unaltered rocks. Another possibility is that the BRFS penetrates steeply to the base of the crust and is along the western edge of the BRUMA. However, no recent geophysical data support this interpretation.

Serpentinized Body

All 2DGAM models show that materials with high magnetic susceptibilities are between depths of ∼16 and ∼34 km beneath the CIB, and support the south Alaska magnetic high. Hyndman and Peacock (2003) and Blakely et al. (2005) suggested that the fluidization process of forearc mantle and crust causes abnormal magnetic highs. The subducting slab drags down sediments and fluid during formation of the accretionary wedge. When the slab reaches a temperature of ∼350 °C at depth, a large volume of fluid is released and rises toward the surface (Fig. 13). At these temperatures, the released fluid reacts with preexisting ultramafic rocks and alters them to form hydrous minerals. Serpentinization produces a residual iron oxide, typically magnetite (Hyndman and Peacock, 2003). The magnetite typically imparts a strong magnetic susceptibility to serpentinites, where its value is proportional to the degree of serpentinization and amount of iron derived from source rocks (Toft et al., 1990).

The serpentinized body interpreted in our 2DGAM models corresponds to a region of conductivity highs from MT profiles (Green, 2003) and low P-wave velocities from seismic tomography models (Eberhart-Phillips et al., 2006). The tops of the conductivity highs shallow and reach the surface at the Knik Arm anomaly, which suggests that both the serpentinized body and ultramafic rocks of the BRUMA are associated with the source of the magnetic high (Saltus et al., 2007). P-wave velocity analyses are often used to estimate a degree of the fluidization. Based on studies by Coleman (1971) and Carson (2003), P-wave velocity lows of ∼6.2 m/s associated with the regional serpentinized body at depths of 16–34 km correspond to ∼40% serpentinization. This result is greater than the 20% serpentinization calculated by Hyndman and Peacock (2003) based on seismic tomographic data in the subducted forearc located south of our study area, but close to ∼50% serpentinization formed in the forearc of the Cascadia subduction zone (Bostock et al., 2002).

Two hypotheses could explain the deep source (>15 km) of the apparent serpentinization: (1) forearc mantle of the recent subducting slab (Hyndman and Peacock, 2003; Haeussler and Saltus, 2011) or (2) forearc mantle of the accreted Jurassic oceanic arc (Pavlis and Roeske, 2007). We suggest that the serpentinized body is most easily explained as altered Jurassic forearc mantle in the basement of the Peninsular terrane (Fig. 10). These rocks are observed as fault slices within the BRUMA and must extend beneath the CIB because the basin has behaved as a relatively stable block since Early Jurassic time. A serpentinized body formed from the recent forearc mantle model is less plausible because the top layer of the serpentinized body appears to be at ∼16 km depth, which is much shallower than the estimated Moho depth of 34 km based on the receiver function studies of Romero (2011). Admittedly, it is not obvious what a Moho seismic discontinuity represents in this tectonic setting, but in the broader context of the entire geophysical data set, the lower crust from depths of 16–34 km could represent altered mafic rocks and/or serpentinized ultramafic rocks.

Saltus et al. (2007) suggested that Mesozoic volcanic and volcaniclastic rocks of the Talkeetna Formation and associated plutonic rocks at depths of ∼6 km beneath the eastern CIB and exposed in the northern Chugach Mountain fronts, and the basement rocks of the Peninsular terrane, could be a shallow source for the south Alaska magnetic high. We tested this alternative (Ab–Ab′, Fig. 12) by increasing the magnetization of these late Mesozoic rocks and the basement rocks to a value of 0.27 SI (Saltus et al., 2007), while keeping other parameters (e.g., densities and geometries) fixed at the same values and decreasing the magnetization of the serpentinized body until a minimum misfit was obtained. The resulting alternative model still requires a magnetic susceptibility of 0.04–0.08 SI (compared to values of 0.07–0.13 SI for our preferred model in Fig. 10) for the deeper serpentinite body (Ab–Ab′, Fig. 12). This lower value corresponds to the value used in the magnetic cross-section models of Saltus et al. (2007). We also note that if the magnetic highs were solely related to rocks of the Talkeetna Formation, we would expect to observe the magnetic highs increasing at the northeastern edge of the CIB within Matanuska Valley where this formation crops out (Figs. 4 and 9). Consequently, these results suggest that the south Alaska magnetic high is a complicated feature related to both shallow and deep sources.

Underplating of Sediments

Our 2DGAM models (profiles C–C′ and D–D′ in Fig. 10) present one possible model to support the abnormal gravity lows beneath the central Chugach Mountains. The shallow upper crustal rock cannot be the source of the gravity lows due to the dense nature of these accreted rocks and the continuity of surface exposures of the homogeneous rock assemblage, the Valdez Group (Mankhemthong et al., 2012). Thus, to fit the gravity data we need to add a lower density material, here termed a low-velocity density zone, beneath the Chugach terrane and above the subducting slab (Fig. 13). The modeled low-velocity density zone is consistent with P-wave velocity lows observed in the original gridded velocity-depth model of Eberhart-Phillips et al. (2006, their seismic tomography study), and seismic refraction cross-section models (Byrne, 1986; Ye et al., 1997) in the Alaska-Aleutian subduction zone, as well as in other subduction zones around the world such as the Cascadia (Calvert et al., 2011), Makran (Kopp et al., 2000), and Nankai subduction zones (Park et al., 2010).

We suggest that an underplated sediment model explains the low-velocity density zone features observed over the metamorphosed accretionary assemblages. The underplated sediments formed the base of the uplifting accretionary assemblage that was associated with the subduction process in south-central Alaska (e.g., Pavlis and Bruhn, 1983; Moore et al., 1991). The subduction caused shortening of the accretionary wedge during the Cretaceous (Moore et al., 1991) with later rapid lateral growth trenchward during the Paleogene. Presumably the underplated sheets are relatively young and had to be stacked to form simple duplex structures in order to maintain a stable taper in the accretionary wedge (Fig. 10) (Ye et al., 1997), and underplating is consistent with evidence to the east and south that extensive sediment subduction is associated with the Yakutat terrane collision (Pavlis et al., 2012). Due to less compaction of mixed fluid and seafloor sediment, turbidite fan sediments, and small continental fragment lithology, the underplated sediments are expected to have lower densities (Kopp et al., 2000; Park et al., 2010) that match the observed density and velocity lows over the metamorphosed accretionary assemblages (Figs. 9 and 13).

The fluidization process also may decrease the densities of underplated formations (Park et al., 2010), as in the accreted Jurassic oceanic arc. In contrast to serpentinization processes, fluidization does not increase the magnetic susceptibility of underplated formations and the accretionary complex because the complex has minimal ferromagnesian minerals that can be altered to produce magnetite-rich rocks (Housen, 1997; Blakely et al., 2005). Magnetic anomaly lows from the low pass filtered magnetic analysis shown in Figure 7 support the idea that the hydrous underplated sediments were originally less magnetic (Housen, 1997).

Arkle (2011) suggested that underplating is the principal mechanism driving deformation and exhumation in the western Chugach Mountains (Fig. 13). New thermochronometry age dating techniques used to address the Tertiary exhumation history of the accreted terranes indicate that this terrane complex has had more than one episode of uplift since 44 Ma in the western Chugach Mountains (Arkle, 2011). The Border Ranges fault system and Contact fault system may control the locus of this first phase of exhumation, ca. 10–20 Ma (Enkelmann et al., 2008). Pavlis et al. (2012) studied the exhumation in the St. Elias orogen and suggested that uplift due to thrust duplex structures may be related to shortening of the subducting Yakutat microplate. Due to the northwestward subduction of the Yakutat microplate underneath the Chugach Mountains, we suggest that the Yakutat microplate carried its cover beneath the margin and these sediments were underplating in late Neogene time, contributing to the observed uplift and exhumation event at 3–6 Ma (Enkelmann et al., 2008; Arkle, 2011). However, the relationship between the occurrence of underplated sediment and regional tectonic processes of the subducting slab and Yakutat microplate are still too poorly understood to draw any more firm conclusions.

CONCLUSIONS

We provide detailed 2D integrated forward models for the BRFS and related geologic structures based on new gravity data with constraints from existing aeromagnetic, seismic, and well log data. These integrated models show that the CIB was filled with ∼6 km (northern basin) to ∼10 km (central basin) of Cenozoic and Mesozoic sedimentary rock overlying the Peninsular terrane basement. The large gravity lows and magnetic highs over the basin may correlate with the deep serpentinized body (16–34 km) and/or shallow Mesozoic sedimentary rocks of the Talkeetna Formation and mafic rock assemblages of the Peninsular terrane basement (6–20 km). We suggest that altered forearc mantle beneath an accreted Jurassic oceanic arc that composes the basement of the Peninsular terrane is a source of the fluid serpentinization. The eastern boundaries of the CIB are characterized by gravity-magnetic highs that are related to a 5–15-km-wide slice of the emplaced BRUMA. The BRUMA may be related to the formation of serpentinized rocks that compose a Jurassic oceanic arc.

The eastern gradient of the BRUMA anomaly along the Kenai and Chugach Mountains front is interpreted as the primary location of the BRFS. According to our 2DGAM models, the BRFS represents a buried geologic structure separating the Chugach terrane and the overthrusted BRUMA. One 2DGAM model suggests that the BRFS dips steeply (∼70°) toward the west-northwest. However, a more gently dipping (∼50°) BRFS is an alternative model that also fits our observations. The BRFS may cut through the ultramafic assemblages and serve as the structural boundary between serpentinized and unaltered ultramafic rocks or penetrate steeply to the base of the crust along the eastern edge of the BRUMA.

Gravity lows are locally observed over the western Chugach Mountains where the accretionary complex of Chugach terrane is observed at the surface. A model with underplated sediment at the base of the accretionary complex (12–40 km) and above the subducted slab is one explanation for this low and appears to be associated with a low-velocity zone imaged in previous seismic tomography studies. We term this body the low-velocity density zone. The proposed underplated sediments may have formed rapidly to maintain a stable taper at the base of the accretionary complex, and drove deformation and exhumation of the accretionary complex.

We are grateful to M. Baker, G. Kaip, B. Eslick, S. Jones, and P. Budhathoki for help in collecting gravity data in 2009, 2010, and 2011. We thank P. Haeussler for discussions related to the geological background and geophysical modeling of the Cook Inlet forearc basin and the Border Ranges fault system and for housing the field crews during the 2010 and 2011 field seasons. We also thank M. Hussein, E. Patlan, and C. Montana for advice regarding the modeling techniques and resolving modeling software problems. This research was funded through the American Chemical Society Petroleum Research Fund (grant 48312-AC8 to Doser).