We use gravity, magnetic, seismic reflection, well, and outcrop data to determine the three-dimensional shape and structural features of south-central Alaska’s Susitna basin. This basin is located within the Aleutian-Alaskan convergent margin region and is expected to show effects of regional subduction zone processes. Aeromagnetic data, when filtered to highlight anomalies associated with sources within the upper few kilometers, show numerous linear northeast-trending highs and some linear north-trending highs. Comparisons to seismic reflection and well data show that these highs correspond to areas where late Paleocene to early Eocene volcanic layers have been locally uplifted due to folding and/or faulting. The combined magnetic and seismic reflection data suggest that the linear highs represent northeast-trending folds and north-striking faults. Several lines of evidence suggest that the northeast-trending folds formed during the middle Eocene to early Miocene and may have continued to be active in the Pliocene. The north-striking faults, which in some areas appear to cut the northeast-trending folds, show evidence of Neogene and probable modern movement. Gravity data facilitate estimates of the shape and depth of the basin. This was accomplished by separating the observed gravity anomaly into two components—one representing low-density sedimentary fill within the basin and one representing density heterogeneities within the underlying crystalline basement. We then used the basin anomaly, seismic reflection data, and well data to estimate the depth of the basin. Together, the magnetic, gravity, and reflection seismic analyses reveal an asymmetric basin comprising sedimentary rock over 4 km thick with steep, fault-bounded sides to the southwest, west, and north and a mostly gentle rise toward the east. Relations to the broader tectonic regime are suggested by fold axis orientations within the Susitna basin and neighboring Cook Inlet basin, which are roughly parallel to the easternmost part of the Alaska-Aleutian trench and associated Wadati-Benioff zone as it trends from northeast to north-northeast to northeast. An alignment between forearc basin folds and the subduction zone trench has been observed at other convergent margins, attributed to strain partitioning generated by regional rheologic variations that are associated with the subducting plate and arc magmatism. The asymmetric shape of the basin, especially its gentle rise to the east, may reflect uplift associated with flat-slab subduction of the Yakutat microplate, consistent with previous work that suggested Yakutat influence on the nearby Talkeetna Mountains and western Alaska Range. Yakutat subduction may also have contributed to Neogene and later reverse slip along north-striking faults within the Susitna basin.


Knowledge of the morphology and structural features of convergent margin basins informs earthquake hazard assessments, evaluation of natural resources such as water, gas, or petroleum, and other geologic applications. The results of such studies in turn form a starting point for better understanding the associated tectonic regime and its influence on the surrounding region. The shape of and structures within the basin are impacted by the size of the accretionary prism, the rate and obliquity of subduction, the slope of the downgoing slab, and the degree of arc volcanism (e.g., Dickinson, 1973; Karig, 1974; Dickinson and Seely, 1979; Fuller et al., 2006; Ridgway et al., 2011; Noda, 2016); additionally, major structures may reflect strain localization or partitioning in response to spatial variations in rheology and/or the presence of inherited structures (Jarrard, 1986; Diament et al., 1992; Beanland et al., 1998; Lallemand et al., 1999; Noda, 2016). Although estimating shape and structure has many applications, mapping associated bedrock can be challenging because the sediments and sedimentary rock that fill basins can also hide faults, folds, and the basin bottom. The most practical imaging approaches are through geophysical methods such as seismic reflection, gravity, and aeromagnetic surveys.

South-central Alaska’s Susitna basin, located inboard of the northeastern end of the Alaska-Aleutian subduction zone and immediately north of the Cook Inlet forearc basin (Fig. 1), is of interest because of earthquake hazard potential and possible oil and gas resources (Barnes, 1966; Hackett, 1977; Ehm, 1983; Merritt, 1986; Kirschner, 1988; LePain et al., 2015; Silwal et al., 2018; Stanley et al., 2018). The basin was known to contain a sequence of nonmarine sedimentary and volcanic rocks of Cenozoic age overlying Mesozoic crystalline basement (Barnes, 1966; Hackett, 1977; Reed and Nelson, 1980; Merritt, 1986; Trop and Ridgway, 2007), but limited information was published about the detailed strata, morphology, and structure of the basin until a recent campaign to better understand the basin. This effort included studies of outcrops, well cuttings, and well logs (Stanley et al., 2014; Le Pain et al., 2015), analyses of legacy (1970s) two-dimensional industry seismic reflection data near the center of the basin (Lewis et al., 2015), collection of gravity data in the southwestern part of the basin (Saltus et al., 2016), and analyses of InSAR topography and earthquake focal mechanisms to consider modern deformation (described by Haeussler et al., 2017). In some cases, ambiguity in interpretation has arisen from the vintage and quality of the older data, and much of the basin outside the seismic and well coverage area remains poorly known.

We analyzed aeromagnetic and ground gravity data in the context of previously interpreted seismic reflection and well data to image the three-dimensional shape and structure of the Susitna basin and its surrounds as part of this larger effort. These analyses provide linkages between sparser reflection and well data that facilitate interpretation outside of their area of coverage. We find that aeromagnetic data provide information regarding structure because layers of volcanic rock within the sedimentary strata, when uplifted or offset by faults and folds, produce measurable magnetic anomalies. Magnetic anomalies due to vertical movement of volcanic layers from folding and fault displacement have been observed in the Seattle basin in Washington (Blakely et al., 2002) and the Albuquerque basin in New Mexico (Grauch, 2001) and have been illustrated through modeling by Betts et al. (2004). Over sedimentary basins, gravity methods have been used for decades to estimate depths of sedimentary basins because anomalies are typically generated by differences in density between the sedimentary fill and the underlying basement (e.g., Cordell, 1973). Analyses and interpretation of gravity data collected over the Susitna basin illuminate various aspects of the basin’s shape, which in turn reflects the region’s tectonic characteristics and history.


The Susitna basin lies ∼40 km northwest of Anchorage, Alaska, and ∼300 km inboard of the Alaska-Aleutian megathrust (Fig. 1), where both the Pacific plate and Yakutat microplate are being subducted obliquely beneath the North American plate. The basin forms a region of flatlands and low rolling hills that are largely covered by vegetation and a thick mantle of Quaternary deposits. It is bounded by the Talkeetna Mountains to the east, the Alaska Range to the north and west, and the Beluga Mountain fault and Castle Mountain fault to the southwest and southeast, respectively. Some of these features, including the Talkeetna Mountains and the Western Alaska Range, are believed to have been exhumed by flat-slab subduction of the Yakutat microplate, which is inferred to be a thick, buoyant oceanic plateau; the Yakutat is also thought to have inhibited the development of arc volcanoes north of the Cook Inlet forearc basin and west of the Susitna basin (Plafker, 1987; Eberhart-Phillips et al., 2006; Hoffman and Armstrong, 2006; Fuis et al., 2008; Haeussler, 2008; Christeson et al. 2010; Benowitz et al., 2011; Bleick et al., 2012; Arkle et al., 2013; Chuang et al., 2017; Terhune et al., 2019; Trop et al., 2019). If not for the absence of arc volcanoes on its perimeter, the term “forearc basin” might otherwise be appropriate for the Susitna basin given the north-northeast trend of the volcanoes bounding the Cook Inlet basin.

Multiple basins in the region have been identified primarily from gravity anomalies (Hackett, 1977; Meyer 2005) because surface geologic maps show mostly Quaternary sediments (Fig. 2; Barnes, 1966; Reed and Nelson, 1980; Wilson et al., 2012). Neighboring the Susitna basin are the Cook Inlet forearc basin to the south, the Beluga basin to the southwest, and the Peters Hills basin (also referred to as the Yentna depocenter) to the northwest. The Peters Hills basin is separated from the main part of the Susitna basin by a proposed Broad Pass fault, but the location of this fault is poorly constrained (Haeussler, 2008; Haeussler et al., 2017). The mostly continuous topography and surface geology from the Susitna basin to the Cook Inlet basin might suggest that the Susitna basin is a simple northward extension of the Cook Inlet basin (Merritt, 1986), but several lines of evidence suggest that this is not the case. Broad gravity lows are associated with both basins, but these lows are not continuous (Hackett, 1977). The two basins are separated by the steeply northwest-dipping, right-lateral, strike-slip Castle Mountain fault, which is known to be seismogenic (Haeussler et al., 2000), and surface geologic studies suggest up to 130 km of right-lateral offset along the fault since the Jurassic (Trop et al., 2005). The general shapes of the two basins are notably dissimilar. Lastly, the stratigraphic sequences of the basins are notably different (Stanley et al., 2014; LePain et al., 2015).

Within the Susitna basin, outcropping bedrock is limited to scattered Cenozoic nonmarine sedimentary, intrusive, and volcanic rocks (Fig. 2). Mesozoic metasedimentary rocks of the Kahiltna flysch sequence are exposed along the northern and western flanks of the basin and volcanic and intrusive rocks along the southwestern flanks. Analyses of cuttings and logs from several wells in the region (Fig. 2B) reveal an interval of early Eocene to Quaternary nonmarine strata (mostly sandstone, siltstone, and coal) overlying late Paleocene to early Eocene volcanic and sedimentary rocks (Stanley et al., 2014). The well data (Fig. 2B) show that deposition has not been continuous throughout the region. The Trail Ridge Unit 1 well, which lies within the thickest part of the basin east of the Skwentna fault as identified from the seismic data (Figs. 2A and 2C), contains early Miocene to Quaternary units that are 2483 m thick and unconformably overlie Eocene strata. Only 28 km to the northeast, the Pure Kahiltna Unit 1 well has Miocene and younger strata that are only ∼90 m thick (erosionally truncated) and unconformably overlie Eocene rocks. In contrast, the Sheep Creek 1 well, ∼12 km from the west side of the Talkeetna Mountains, penetrated late Miocene and younger strata that are 369–383 m thick and rest unconformably on late Paleocene to early Eocene basalt. Further information regarding Miocene and younger rocks could not be obtained from seismic reflection data in this area (Lewis et al., 2015). East of the Kahiltna River fault, where the Pure Kahiltna Unit 1 well is located, seismic reflections corresponding to strata younger than Eocene age were not identifiable. This is due to a combined lack of vertical seismic coverage in the area and poor quality of the seismic data at travel times of 0.2 s or less.

Recent analyses of more than 40 two-dimensional, seismic reflection profiles collected in the western part of the Susitna basin have revealed key structural features within the basin (Lewis et al., 2015; Figs. 2 and 3). Using the above well data as constraints, these analyses show an asymmetric basin that is deepest to the west with a sedimentary package more than 4 km thick overlying volcanic layers (Fig. 2). The deepest area is bound by the north-striking Skwentna and Bulchitna Lake faults to the west and east, respectively, and the northwest-striking Beluga Mountain fault to the southwest. The northern boundary of the deep area is not delineated by the seismic data. The reflection data show that basin shallows east of the Bulchitna Lake fault with about half the sedimentary fill, and again east of the Kahiltna River fault, with even less sedimentary fill present. They also indicate that the Skwentna and Bulchitna Lake faults are west- and east-dipping reverse faults, respectively. Few seismic lines cross the Beluga Mountain fault, but limited seismic imagery suggests it is southwest dipping. Analyses of gravity data suggest possible underthrusting of low-density material to the southwest beneath the Beluga Mountain fault (Saltus et al., 2016). Comparisons between the seismic reflection data and preliminary magnetic data analyses also suggest other north-striking faults, a northeast-striking fault, and miscellaneous folds (Lewis et al., 2015). Additional analyses of these data suggest some of the north-striking faults are likely associated with fault-propagation folds (Potter et al., 2016).


Magnetic Data

Aeromagnetic data covering the Susitna basin and surrounding area were collected during several surveys in 2002 and 2012 (data downloadable from https://mrdata.usgs.gov; see also U.S. Geological Survey, 2002). The surveys were flown at a nominal height of 150–300 m above ground level with NW-SE flight lines spaced 800–1600 m over most of the study area (Fig. 3). Additional proprietary data covering Cook Inlet, flown in 1997 at a nominal height of 150 m with E-W flight lines spaced 800 m (see Saltus et al., 2001), were also analyzed.

Each of these aeromagnetic surveys has a flight-line spacing that is relatively wide compared to many modern surveys (e.g., Reid, 1980) because the surveys were designed to cover a broader region more economically. This wide spacing presents data-processing challenges. The flight-line spacing is notably wider than along-track sampling, which typically ranged from 8 to 12 m. Gridding these data at the flight-line spacing would likely filter out short-wavelength anomalies of interest, but finer grids can contain aliasing artifacts. We thus gridded the data at 200 m and applied a decorrugation filter (Urquhart, 1988), which removed aliasing artifacts very effectively. We acknowledge, however, that if smaller (<800-m-wide) features are oriented parallel to the flight-line spacing, they will not be visible in these maps. The different surveys were stitched together using standard methods that perform level shifts to match overlapping areas yet also preserve the characteristics of the broader, regional magnetic anomaly. The result was reduced to the pole, but this had little impact given the steep (∼75°) inclination of the magnetic field in this region. Derivatives and filtered anomaly maps described below were calculated from this grid (Figs. 4 and 5).

The late Paleocene to early Eocene volcanic rocks (intermediate to basaltic) within the Susitna basin are likely to contain magnetic minerals and thus act as magnetic sources (e.g., Clark et al., 1992). In contrast, most of the overlying Cenozoic strata, comprising sandstone, mudstone, sand, clay, and coal should contribute little to the observed field. Faulting and folding of the magnetic volcanic strata will bring some of the magnetic sources to shallower depths, which in turn can generate corresponding anomalies (e.g., Grauch, 2001). Such anomalies are likely to be lower in amplitude and thus more subtle than those from basement sources as those sources can be several hundreds of meters thick. Calculations that separate anomalies due to shallower sources from those due to deeper sources are thus needed. Since shallower magnetic sources generate anomalies of shorter wavelength, filtering of anomalies by wavelength can assist such separation.

To approximate a separation of anomalies due to volcanic layers versus deeper crystalline basement rocks, we applied matched band-pass filtering using the method of Phillips (2001). This approach involves applying band-pass filters for wavelengths corresponding to different “equivalent” source depth ranges. We note that if a magnetic source is shallow but relatively wide, and if there is little local variation of magnetic properties within the source, it is theoretically possible that the source will be described as deeper than it is in actuality. This is a rare scenario, however, because magnetic properties of a given rock unit tend to vary over short distances. The band-pass parameters are chosen based on linear best fits to the log of the magnetic anomaly frequency spectrum. For the Susitna basin, we examined filtered anomalies corresponding to two depth ranges (Figs. 5B and 5C), one centered at 960 m and covering the surface to the upper few kilometers depth (representing the basin) and the other a half space at 7670 m depth (representing the basement). This separation is analogous to that calculated for the Cook Inlet basin by Saltus et al. (2001).

The marked difference between the survey flight-line spacing and along-line sample interval also suggests that additional short-wavelength information can be obtained directly from flight-line data before they are gridded. The combined well and seismic data interpretations suggest that basement rocks lie between 1 and 5 km depth (Stanley et al., 2014; Lewis et al., 2015). We thus applied a one-dimensional high-pass filter to the flight-line anomalies using a convolution method (Fraser et al., 1966) with a cutoff wavelength of 5 km, sufficient to image features near the bottom of the basin (Fig. 5D). The resulting anomalies were then gridded to aid visualization. We also applied the same method using a rolling window of 640 m to highlight narrower anomalies having sources closer to the surface, providing an additional view of anomalies due to even shallower sources. This wavelength was chosen because it is long enough to highlight anomalies due to sources near the surface for sensor altitudes that range from 120 m to ∼500 m, but also short enough to filter out or dampen anomalies due to sources that are more than a few hundred meters deep, such as the volcanic layers near the base of the basin. This width is less than the flight-line spacing for all surveys (minimum 800 m), so the results are more easily visualized by plotting line profiles directly on maps (“wiggle plots”) instead of as processed grids (Fig. 6). These flight-line filtering approaches are analogous to that employed by Shah et al. (2012), who used spectral methods to obtain high-pass filtered flight-line anomalies for a widely spaced survey. We note that the resulting grids are not magnetic fields in the mathematical sense because they do not satisfy Laplace’s equation. They are thus used only for qualitative (not quantitative) analyses.

Gravity Data

Gravity station data, reported by Saltus et al. (2008) and Saltus et al. (2016) throughout the study area (Fig. 3), allow us to approximate shape of the basin by identifying the thickness of the low-density basin fill in areas away from seismic reflection coverage. This approach can be complicated, however, by the presence of basement density heterogeneities that contribute to the observed gravity anomaly. A separation of anomalies due to these sources is thus needed to estimate the elevation of the basin bottom and the thickness of basin fill.

We applied a separation approach that is similar to that of Jachens and Moring (1990). This approach, which has been used in numerous locales in the western United States (e.g., Saltus and Jachens, 1995; Brocher et al., 2001; Langenheim et al., 2010; Shah and Boyd, 2018), separates the gravity anomaly into two components: a “basin” gravity anomaly representing basin fill and a “basement” gravity anomaly representing density variations within crystalline basement. An initial basement anomaly is calculated by gridding observed gravity values over the region where basement rocks are exposed and surround a basin defined from geologic maps, using the gridding algorithm (typically minimum curvature) to estimate a preliminary anomaly over the basin. The initial basement anomaly is likely to be misfit near the basin edges; therefore, the separation is iteratively improved through forward modeling of the basement and basin anomalies. Once a final basement and basin anomaly are calculated, the basin thickness is then derived from the basin anomaly using a prescribed layered density structure that is estimated using the seismic reflection and well data as constraints.

A modification of this approach is needed within the study area because the boundaries of the Cook Inlet, Susitna, Peters Hills, and Beluga basins are covered by Quaternary sediments and thus cannot be identified from surface geology alone (Fig. 2). We first defined the exposed basement as rocks that are Paleocene and older, late Paleocene and early Eocene volcanic rocks, crystalline rocks of the Wrangellia terrane, metasedimentary rocks of the Kahiltna flysch sequence, and plutonic rocks that intrude the Wrangellia terrane and Kahiltna flysch sequence (based on maps by Wilson et al., 2012). The basin fill corresponds to the Cenozoic sedimentary layers that rest upon these rocks. The basement rocks are present as outcropping rock for a number of the gravity stations. We generated an initial basement gravity anomaly grid by sampling the complete Bouguer anomaly (CBA; calculated assuming a crustal density of 2.67 g/cm3; Fig. 7A) at basement outcrop locations and then re-gridding just those stations over the region, analogous to the Jachens and Moring (1990) approach. This basement gravity grid was then subtracted from the CBA grid to produce a preliminary basin gravity anomaly, representing the contribution from sediments and sedimentary rock. Positive gravity values in this preliminary basin gravity are likely to represent dense basement sources; therefore, we removed these values from the basin grid and added then to the basement gravity grid (Figs. 7B and 7C). Unlike Jachens and Moring (1990), iterative improvement near the basin edges was not conducted because the basin edges are not well defined from the geologic maps.

Once the basement and basin anomalies are separated, a density structure is needed to determine the elevation of the basin bottom (or the thickness of the basin fill). This was derived by estimating a basin bottom elevation from seismic reflection data, where available (described below), and then comparing the results to the gravity data. We then estimated depth as a function of the basin gravity anomaly using a least-squares best fit between the gravity station data and the basin bottom grid (Fig. 7D). This approach essentially extends the basin bottom elevation interpretation away from the seismic coverage (Fig. 7E).

The presence of reverse faults suggests that it is possible to have sediments or sedimentary rock thrust beneath basement rock (Saltus et al., 2016). These buried strata will contribute to the observed gravity values where the surficial geology indicates basement rock, but the gravity values will appear to represent a low-density basement heterogeneity (i.e., a low basement gravity value), not basin fill. To highlight such areas, a second-pass separation moved some remaining low gravity values from the basement gravity anomaly into the basin gravity anomaly, and the basin thickness was recalculated. This “modified basin thickness” depicts a nonzero sedimentary layer in areas where there is outcropping basement rock, bringing attention to areas where there may be underthrusted sedimentary rock (Fig. 7F).

Seismic Reflection Data

More than 40 1970s-era, industry seismic reflection profiles were published by Lewis et al. (2015). This study revealed a set of faults (Fig. 2) that separate the central part of the basin into several “blocks” and numerous folds. Locations of these structural features serve as constraints for the interpretation of the basin magnetic anomalies (Fig. 8A).

Combining these data with constraints from well logs and cuttings allows an estimate of the basin bottom depth, which is needed for the gravity data analysis. For fault blocks containing wells, we estimated a basin bottom using common time-depth conversion methods. This involves calculating a synthetic seismogram from the well log and tying seismic sections expressed as two-way travel time to the wells. For fault blocks not containing a well, a pseudo-average velocity function was determined by computing a least-squares function from the Pure Kahiltna 1 and Trail Ridge 1 time-depth pairs. Depth estimates for the individual horizons were then computed for the seismic coverage area by gridding the two-way time horizons and applying the corresponding velocity function.


Magnetic Anomalies

The magnetic anomalies, when filtered to highlight anomalies associated with sources within different depth intervals, exhibit notable differences according to those depth intervals. Here, we separately examine filtered magnetic anomaly and/or derivative maps corresponding to sources in the basement, within the basin, and near the surface.

Basement Sources

The total field magnetic anomaly over the Susitna basin (Fig. 4A) shows highs and lows of several hundred nanoteslas. These highs and lows resemble those of the matched band-pass filtered anomalies for the deeper 7670 m depth half-space (Fig. 5B). The seismic reflection data suggest these depths represent crystalline basement. This similarity and the magnitude of the total field anomalies strongly suggest that the total field represents mostly basement sources. The specific rock types associated with the various high and low anomalies are not known because basement-penetrating drill-hole data were not available, but a comparison to the regional magnetic field of southern Alaska shows that they lie within a broad band of magnetic highs and lows of similar magnitude extending from the Canadian border in the Wrangell Mountains to the Bering Sea Shelf. This band spatially coincides with the Alaskan Peninsular and Wrangellia terranes and is thought to represent mafic igneous rocks (Saltus et al., 2007).

The magnetic anomalies are more subdued where other data suggest the magnetic basement is deeper. For example, the highs and lows become more subdued in the northern part of the basin over the metasedimentary rocks of the Kahiltna flysch sequence, which is unlikely to contain magnetic sources, and over the Cook Inlet basin. These variations are illustrated by the total gradient magnitude (i.e., analytic signal magnitude), which is higher where igneous basement rocks are shallow or crop out, and lower over the Kahiltna flysch sequence (Figs. 2 and 4B). The total gradient also shows a sharp decrease from the Susitna basin to the Cook Inlet basin along the Castle Mountain fault (following the greater basement depth within the Cook Inlet basin estimated by Hackett, 1977) and sharp linear contrasts across the Beluga Mountain, Skwentna, and Kahiltna River faults. Changes in the thickness of the sedimentary package across each of the Susitna basin faults are observed in the seismic reflection data (Lewis et al., 2015). A sharp linear contrast is also observed along the Susitna River west of the base of the Talkeetna Mountains, where seismic coverage is unavailable.

In many areas, total gradient highs are associated with magnetic anomaly highs but with two primary exceptions (Fig. 4, starred locations). In these areas, magnetic anomaly highs sharply abut magnetic lows (with contrasts as high as 1800 nT over a few kilometers), whereas the total gradient is consistently high over the magnetic high-low pair. This indicates that the total field lows are unlikely to represent a lack of magnetization because low magnetization would generate a low total gradient. It is more likely that the high-low pair represents a body with a remanent magnetization component whose direction is different from Earth’s local magnetic field (noting that the total gradient is independent of magnetization direction). This can occur either by rotation of a magnetized body or tectonic transport of material that acquired its magnetization in a different region (e.g., Blakely, 1996).

Basin Anomalies from Filtered Magnetic Data

The band-pass filtered anomaly centered at 960 m (Fig. 5B) shows a series of linear highs and lows that are mostly between 5 and 50 nT but occasionally reach up to 200 nT. This map is qualitatively similar to that of the high-pass (<5 km), convolution-filtered, flight-line data (Fig. 5C), strongly suggesting that the latter map also reflects sources within the basin. The convolution-filtered anomaly map more clearly shows breaks and changes in the lineaments, probably because the filter was applied before the flight-line data were gridded, rather than the other way around as for the band-pass filters.

Magnetic lineaments in the high-pass filtered flight-line data and matched filtered maps are oriented mostly to the northeast, with various discontinuities (Figs. 5B and 5C). A few north-trending lineaments are observed within the deeper part of the basin where north-striking faults were mapped by Lewis et al. (2015), including the Skwentna fault, Bulchitna Lake fault, Kahiltna River fault, and northeast of the Kahiltna River fault (Fig. 5D). The magnetic lineaments are more subdued and wider between the Kahiltna River and Skwentna faults, where seismic reflection and well data suggest the basin is deepest. They are also more subdued, but not as wide, in the northern part of the basin, suggesting a weaker source in that area. The lineaments exhibit a marked change in character across the Castle Mountain fault; south of this fault, they are thinner and lower amplitude. This is most evident in the eastern part of the study area where a single aeromagnetic survey crosses the fault (farther west, the fault is close to a boundary between surveys).

A comparison to the seismic reflection data shows a correspondence between the linear magnetic highs and layers imaged by the reflection data that have been uplifted due to folding and/or faulting (Fig. 8A). Most of these folds lie on magnetic highs in the filtered anomaly map. This relation can further be visualized by comparing the profiles directly to the band-pass filtered data (Fig. 9). Occasionally, the magnetic anomalies correspond to changes in thickness of the volcanic layer, showing highs in those regions, but this is much less common, and those highs have limited lateral extent (marked in Figs. 8A and 9).

Shallowest and Surficial Magnetic Sources

Applying the convolution high-pass filter with a 640-m-wavelength cutoff highlights anomalies due to the shallowest sources in the area (Fig. 6). The high-pass filtered flight-line data show distinct zones with short-wavelength anomalies and other zones where the magnetic field varies little. Most of these zones correspond directly with surface geology (Figs. 6B–6E). Some of the highest magnitude short-wavelength anomalies occur over or near outcropping igneous and low-grade metamorphic rocks such as in areas southwest of the Beluga Mountain fault, west of the Skwentna fault, along the western flank of the Talkeetna Mountains, and Lone Ridge in the western Cook Inlet basin. High values are also observed over uplifted areas where the sedimentary cover is expected to be relatively thin, such as east of the Kahiltna River fault. Anomalies also occur over sedimentary layers shown to have elevated magnetic susceptibilities such as the Arkose Ridge, West Foreland, and Talkeetna Formations, probably due to volcanic content within those strata (Altstatt et al., 2002).

In the Cook Inlet and southeastern Susitna basin, anomalies are observed in areas with glacial deposits (Fig. 6). These anomalies are most pronounced in the vicinity of late Cenozoic arc volcanoes along the western shore of Cook Inlet and may represent glacially transported fragments of volcanic material, which is likely to contain magnetic minerals (Figs. 6B and 6C). Just south of the Talkeetna Mountains, slightly weaker anomalies also correspond to glacial sediments (Figs. 6D and 6E). In this area, the magnetic anomalies continue to the base of the Talkeetna Mountains and may represent glacial fragments of igneous and metamorphic rocks from the mountains. The spatial distribution of areas with magnetic anomalies shows more complex patterns in this area. Some anomalies appear to follow small topographic slopes (marked in Figs. 6E and 6F) that form boundaries between glacial and estuarine, alluvial, or silt deposits. In other areas, these anomalies seem to be grouped in zones.

These shortest-wavelength anomalies do not show a change in character across the Castle Mountain fault. This distinguishes these shortest-wavelength anomalies as due to shallowest (and likely younger) sources associated with sedimentary depositional processes, while the anomalies identified in the gridded data are instead due to deeper, older sources that have been deformed by tectonic processes, especially along the Castle Mountain fault.

Gravity Analysis

The complete Bouguer anomaly (CBA) shows lows associated with the Peters Hills basin, the Susitna basin, the Cook Inlet basin, and the Beluga basin (Fig. 7A), consistent with studies by Hackett (1977), who interpreted similar anomalies as lower densities associated with sedimentary basin fill. Of these basins, the Cook Inlet basin shows the lowest CBA values, consistent with a deep basin, while the Beluga basin shows the highest values. The CBA shows a marked asymmetry within the Susitna basin: A pronounced low in the western part of the basin has steep slopes to the south and west and a more gradual slope to the east. The Susitna, Peters Hills, Beluga, and Cook Inlet basins have a common characteristic—at least one sharply defined northeast-trending basin margin.

The separation of the CBA into basement and basin components (Figs. 7B and 7C) shows broad basement gravity highs associated with the Talkeetna Mountains, the edge of the Alaska Range, and isolated areas including the Yenlo Hills. Lowest basement values are observed in a roughly 10-km-wide area just southwest of the Beluga Mountain fault where Cretaceous metavolcanic rocks are at the surface. Similar to the CBA, the basin gravity anomaly shows clear lows associated with the Susitna, Cook Inlet, Peters Hills, and Beluga basins.

A comparison between basin gravity anomaly and the basin bottom depth determined from seismic reflection and well data (where available) suggests a multi-layer density function for the Cenozoic basin fill (Fig. 7D). Assuming a two-layer density model and using the slope of a best-fit two-segment line through a basin gravity versus depth scatterplot at gravity station locations yields density contrasts of −0.45 g/cm3 and 0.16 g/cm3 between the sedimentary rocks and basement for upper and lower layers, respectively. We note that the basement gravity does not exhibit variations from the Peters Hills basin to the Susitna basin and thus assume that the density-depth relation derived from the Susitna basin is also valid for the Peters Hills basin.

Applying this two-layer density model to the basin gravity anomaly grid suggests the Susitna basin bottom elevation reaches 4 km below sea level (and thus contains over 4 km of sedimentary fill), with steep sides except to the east in the area north of the Lockwood Lake fault (Fig. 7E). Outside of a ∼25-km-wide area, the basin bottom is generally shallower than 2 km below sea level. The Peters Hills basin and the Beluga basin reach nearly 2 km and 1.2 km below sea level, respectively. In contrast, the northernmost Cook Inlet basin is deeper than 8 km below sea level. The southwestern, western, and northern boundaries of the Susitna basin are notably steep, with the southwestern and western boundaries corresponding to the Beluga Mountain fault and the Skwentna fault, respectively. The eastern boundary is, in contrast, much gentler over most of the basin, with exception of the area south of the Lockwood Lake fault.

Overthrusting of basin fill along the Beluga Mountain fault or elsewhere would produce low values in the basement anomaly that actually represent buried sedimentary rock. Keeping a basement gravity lower bound of −80 mGal and moving the difference to the basin anomaly produced a relatively smooth basement anomaly (we acknowledge, however, that alternative values are possible). This modified basin gravity anomaly has lows immediately southwest of the Beluga Mountain fault near its intersection with the Skwentna fault. We define a corresponding “effective basin thickness” in this area, which includes basin fill that may reside beneath surficial basement (metavolcanic) rocks (Fig. 7F), consistent with models by Saltus et al. (2016), who considered a single profile across the Beluga Mountain fault. This map-view analysis, however, suggests that the overthrusting is localized near the intersection of the Beluga Mountain fault with the Skwentna fault. The modified basin anomaly associated with the Peters Hills basin is also wider than in the initial anomaly, with possible underthrusting of sedimentary rock along its southwestern boundary.


Usage of Magnetic Flight-Line Data

The Susitna basin aeromagnetic data set provides a case study of how magnetic data can contain multiple layers of magnetic sources within a single region. Here the magnetic source layers lie beneath the basin (as igneous and metamorphic basement rock), within the basin (following changes in the depths of volcanic deposits), and near the surface (often corresponding to glacially transported sediments). Interpretations of these layers are discussed below. Imaging of these different layers was greatly facilitated by the consideration of flight-line data because magnetic anomalies generated by shallow or surficial geologic sources might otherwise be filtered out during gridding of the magnetic anomaly. The flight-line data analyses show that the shortest wavelength anomalies have direct relations to geologic features. For sources at depths within the basin, both the gridded and flight-line magnetic data were analyzed to provide maps of structure and other features, but the flight-line data provide a much “crisper” image of lineaments associated with structural features (Fig. 5C versus Fig. 5D).

This approach can thus be especially helpful for analyzing survey data with a line spacing considered less than ideal; e.g., Reid (1980) recommends a flight-line spacing that is close to the vertical distance between the magnetic sensor and source. This scenario is not uncommon in remote areas where survey costs are higher or for regional programs such as the U.S. National Uranium Resource Evaluation (Hill et al., 2009). There are two important considerations to this approach, however. First, the resulting filtered data are not reduced to the pole. This should not be problematic in areas where the magnetic inclination is steep, but it could lead to significant errors in regions closer to the equator. Additionally, as noted above, flight-line data analyses may neglect to image features that are parallel to the flight-line direction. Comparisons to complementary data sets such as seismic reflection data, gravity data, geologic maps, light detection and ranging (lidar) elevation models, and well data thus play an important role in the data interpretation. For the Susitna basin, these data indicate that most structures of interest are oriented obliquely to the aeromagnetic flight lines, supplying confidence to our analyses.


The combined gravity, magnetic, seismic reflection, and well data demonstrate that the sedimentary fill is thicker than 4 km within a ∼45 km (north-south) by 25 km (east-west) area in the western part of the basin, and less than 2 km thick farther east (Figs. 7 and 8D). The deeper parts of the basin are fault bounded, resulting in an asymmetric shape with steep slopes observed to the west coincident with the Skwentna fault, to the southwest along the Beluga Mountain fault, and to the north along a possible Broad Pass fault (Fig. 8D). In the area where the Beluga Mountain fault meets the Skwentna fault, there may be underthrusting of material, suggesting basin shortening due to horizontal compression in that area. To the east, there is a gentle rise north of the Lockwood Lake fault, but south of the Lockwood Lake fault, the basin also appears to be fault bounded. In the easternmost part of the basin, where the slope is more gentle, seismic reflection data show that stratigraphic layers are west- or northwest-dipping (lines 12, 23, 25, and 26 in Lewis et al., 2015). This suggests relative uplift in that part of the basin.

We consider several contributing causes for the depth asymmetry and west-dipping reflectors in the eastern part of the basin. Jadamec et al. (2013) found that models of dynamic topography generated by the density of the subducting Pacific slab predict subsidence (as negative dynamic topography) beneath the Cook Inlet and Susitna basins. They note, however, that these models also predict negative dynamic topography beneath the Talkeetna Mountains. Apatite thermochronology studies show evidence for exhumation of the Talkeetna Mountains in early Miocene time (Hoffman and Armstrong, 2006); thus, Jadamec et al. (2013) suggest that conditions leading to uplift of the mountains may have been different from those in modern times. They also suggest the higher elevations may be driven by subduction of the Yakutat plateau beneath this region, which would reduce the amount of negative dynamic topography (see also Fig. 1). Processes leading to uplift and/or exhumation of the Talkeetna Mountains could have also contributed to uplift of the eastern part of the Susitna basin since it is an immediate neighbor of the Talkeetna Mountains. Furthermore, uplift and erosion of the eastern part of the basin during the Miocene are consistent with the absence of strata between the early Eocene to mid-Miocene observed in the Sheep Creek 1 well in the eastern part of the basin (presumably due to erosion), but not in the Pure Kahiltna Unit 1 or Trail Ridge Unit 1 wells farther to the west (Fig. 2B). Regardless, a mechanism is needed for renewed deposition in middle to late Miocene to Quaternary time, perhaps related to the complexities of Yakutat slab subduction.

An additional possible source of basin deepening to the west could be a flexural response to loading by the Western Alaska Range, similar to the manner in which foreland basins may show asymmetric tilting toward topographic loads formed by fold and thrust belts (DeCelles and Giles, 1996). The flexural response to surrounding topographic loads depends on multiple factors ranging from the size and density of the load to the strength of the underlying lithosphere. An estimate of lithospheric strength is beyond the scope of this effort, but we can consider relative load contributions from the elevations of the surrounding mountain ranges because the Susitna basin is surrounded by high topography not only to the west but also to the east and to the north. The Talkeetna Mountains to the east have elevations that are only somewhat lower than those of the western Alaska Range (up to ∼2200 m versus up to ∼3000 m), but the elevations to the north (forming the northern Alaska Range) are much greater than either, reaching up to 5000 m above sea level. All three mountain ranges contain both mafic and felsic rocks; therefore, a substantial difference in bulk density between them is not expected. A flexural response to topographic loading from adjacent mountain ranges would then seem to favor shallowing to the south rather than to the east due to the load of the central Alaska Range. This is not consistent with the reflection data. Thus, while loading by adjacent mountain ranges is likely to have contributed to the basin shape, it does not appear to act as a dominant process.


Aeromagnetic Anomalies and Structural Features

The correspondence between linear magnetic highs and volcanic layers that have been uplifted due to folding and/or faulting strongly suggests that the Paleocene volcanic rocks are the source of the magnetic lineaments (Figs. 8 and 9). Volcanic rocks, especially those that are mafic or intermediate in composition, typically have elevated magnetic susceptibilities, while the overlying sandstone, siltstone, claystone, and coal layers are unlikely to contribute to the observed magnetic field (e.g., Clark et al., 1992). We thus used the high-pass filtered magnetic data to interpolate between fold and fault picks identified in the seismic reflection interpretations by Lewis et al. (2015), resulting in a map showing probable structures within the Susitna basin and surrounding areas (Figs. 8B–8D; Supplemental Shapefiles1). This interpretation allows us to infer the presence of structural features in areas outside of the seismic reflection data coverage. We do expect other characteristics to also contribute to magnetic anomaly variations such as thickness of the volcanic layer (several cases were identified from the seismic data, marked in Fig. 8A), local variations in magnetic susceptibility, and/or natural remanent magnetization. However, the high correspondence between the seismic reflection and magnetic data combined with the linearity of the filtered magnetic anomalies strongly suggests characteristics contribute only secondary effects, and in most areas, an association between magnetic lineaments and structural features is valid.

The magnetic and seismic reflection data suggest numerous northeast-oriented (N35E-N60E) structures (Figs. 8 and 9). Most magnetic lineaments trend northeast and are associated with anticlinal folds (with a small percentage of these related to faults); thus, we interpret the northeast-oriented lineaments outside of the seismic coverage to also represent folds. There is one exception: along the northern part of the basin, where gravity data show a sharp gradient (Fig. 7), there is a series of slightly curved, northeast-trending magnetic lineaments extending from the northern end of the Skwentna fault toward the Talkeetna Mountains (Fig. 8). These lineaments lie near but south of the Broad Pass fault proposed by Haeussler et al. (2017), who noted a wide possible zone for the location of this fault. These anomalies and the gravity gradient thus suggest a more precise location for a Broad Pass fault.

In the deeper part of the basin, some magnetic lineaments are oriented north-south (N5W-N8W) and aligned with reverse faults, representing uplift or a doubling (vertical juxtaposition) of the volcanic layers due to slip on those faults. Using magnetic data to interpolate the locations of the faults between the seismic lines, we find that several of these faults, including the Kahiltna River and Bulchitna Lake faults, appear to be offset or segmented in a left-stepping sense (Fig. 8B). The Beluga Mountain fault also appears to be segmented or offset to the left near its intersection with the Bulchitna Lake fault. The magnetic and seismic data suggest that these faults disrupt the northeast-trending folds in several areas. For example, east of the Kahiltna River fault and south of the Lockwood Lake fault over an uplifted area that trends northward and bends to the southeast, some of the lineaments are offset across the eastern edge of that feature (marked in Fig. 8A), but with a slight offset, consistent with the idea that the lineaments were once continuous but then were disrupted by an uplift event. In the eastern part of the basin, where there is no seismic coverage, several northeast-trending magnetic lineaments show discontinuities or terminations that can be traced from north to south. These are interpreted as additional (probable) north-striking faults (Fig. 8B).

There are strong differences in magnetic character and the magnitude of the total gradient across the Skwentna fault and the southern part of Kahiltna River fault (Fig. 8C), even though neither of these faults bounds outcropping basement rock. Gravity and seismic reflection data in these areas suggest steep changes in thickness of the sedimentary fill (Fig. 6). A similar difference in the total gradient is observed along a linear zone ∼15 km from the western edge of the Talkeetna Mountains, following the Susitna River, and outside of the seismic data coverage area. This linear transition is interpreted as a fault, herein named the Susitna River fault (Fig. 8C). We note that gravity stations are sparser in this area, but those data also suggest shallowing of the basement (Fig. 6). The high-pass filtered magnetic data also show contrasts in character along the southeastern edge of the Peters Hills basin, suggesting the presence of a fault (Fig. 8B). The most dramatic contrasts in magnetic and gravity data are observed, however, on either side of the Castle Mountain fault (Figs. 7 and 8).

While most of the magnetic lineaments are likely to be associated with deformation of volcanic layers, some of the northeast-oriented lineaments are associated with outcropping intrusive rocks or are observed in areas where well data indicate an absence of volcanic layers at depth. Examples include just southwest of the Beluga Mountain fault, where intrusive rocks are mapped at the surface, and immediately north of the Castle Mountain fault, where the Red Shirt Lake 1 and Fish Creek 1 wells penetrated granite near the crests of apparent anticlinal structures (Stanley et al., 2017). This suggests that the magnetic lineaments are in some cases associated with deformation of intrusive rocks. In other areas, however, such as the higher mountains farther west (comprising igneous and metamorphic rocks), the short-wavelength anomalies become more chaotic, and lineaments cannot be discerned.


It is difficult to estimate the timing of the deformation that produced the northeast-trending folds, partly because information regarding the basin before this deformation is difficult to extract from the magnetic, gravity, or seismic reflection data. However, east of the Kahiltna River fault, seismic data show that horizons from the late Paleocene to the early Eocene are parallel to horizons within the early Eocene, indicating that those folds occurred after the early Eocene. Younger units could not be delineated clearly in this area; so it is unclear whether that deformation occurred even later. Miocene and younger layers are observed in seismic data between the Bulchitna Lake and Skwentna faults, where the basin is deepest, but in this area, most folds are associated with north-striking faults. In the northern part of the basin within the Talkeetna quadrangle, Reed and Nelson (1980) mapped northeast-trending folds in exposed sedimentary rock described as Pliocene by Haeussler et al. (2017), which would suggest more recent northeast-trending folds in that area. Combining these results, we interpret the northeast-trending folds as taking place after the early Eocene, and possibly as recently as the Pliocene or later.

The north-trending faults appear to cut the northeast-trending structures. These faults have likely accommodated Neogene shortening, evidenced by fanning dips of Miocene reflectors, especially in the hanging wall of the Skwentna fault (see Lewis et al., 2015, their figure 34). Shortening along these faults is consistent with the modern east-west sigma 1 (most compressive) and vertical sigma 3 (least compressive) directions (Flores and Doser, 2005; Haeussler et al., 2017). The seismic reflection data strongly suggest that the north-striking faults have dips steeper than 45°, indicating they are reverse faults.

A comparison to surface elevation (Fig. 10) shows little to no correspondence with the northeast-trending folds. In the eastern part of the basin, InSAR data show a northeast-trending feature forming an exception; but this is rare (Haeussler et al., 2017). Some correspondence between the north-striking faults and topographic features within the Quaternary cover is visible in the central part of the basin, noting however that these data are not of fine enough resolution to image scarps, and glacial processes have modified the topography. These comparisons suggest that northeast-trending folds are mostly buried by undeformed Quaternary sediments and are probably experiencing minimal deformation in modern times, but there may be continued motion on the north-striking faults in modern times.

In summary, the reflection, magnetic, and well data suggest that NE-trending folding occurred after the early Eocene and possibly as late as the Pliocene, with more recent activity observed mainly in the north. Slip along north-striking faults has occurred from the Miocene possibly through the Quaternary. These data provide little information regarding the original formation of the basin beyond the oldest observed fill of Paleocene volcanic and sedimentary rocks.

Relation to the Regional Framework

The Aleutian-Alaskan convergent margin is undergoing complex tectonic processes such as flat-slab subduction and oblique convergence that influence regional variations in structure. Insight regarding structures within the Susitna basin may be gained from comparisons to the orientations of structures within the neighboring Cook Inlet basin (Fig. 11). Haeussler et al. (2000) and Haeussler and Saltus (2011) used seismic reflection, well, and aeromagnetic data to delineate anticlinal folds throughout the Cook Inlet. South of ∼60°N, the folds are oriented to the northeast (about N40W-N50W). Farther north, within upper Cook Inlet, the folds are oriented mostly north-northeast (about N5W to N15W), except in the eastern part of the basin close to the Castle Mountain fault where the folds are parallel or subparallel to the Castle Mountain fault. Farther north in the Susitna basin (and distal from the Castle Mountain fault), the present study shows that folds trend to the northeast (N35E to N60E), although these folds are cut by north-striking faults.

The folds in the Cook Inlet and Susitna basins are parallel to the strike of the subduction zone boundary (Fig. 11B), defined by the surface expression of the Alaska-Aleutian megathrust trench and its continuation beneath the surface as the Wadati-Benioff earthquake zone (Jadamec and Billen, 2012; Hayes et al., 2018). The trench and Wadati-Benioff zone exhibit several changes in orientation from the northern part of the Aleutian Islands toward the Canadian border. In the western part of this area, it is oriented to the northeast, but farther east, the megathrust curves to become more northerly. East of that area, near the western edge of the Yakutat microplate, the surficial thrust zone veers to the northeast near the Transition fault as the plate boundary becomes more complex. In most areas, the trench and folds are oblique to the current plate motion direction, which varies from ∼N26W to ∼N21W over the study area (Kreemer et al., 2014; Altamimin et al., 2016; https://www.unavco.org/software/geodetic-utilities/plate-motion-calculator/plate-motion-calculator.html, accessed March 2019).

Trench-parallel deformation in the upper crust and at the surface has been observed in other areas where the plate convergence direction is oblique to the subduction trench, including the Sunda arc, the Ryukyu arc near Taiwan, and the Hikurangi margin near New Zealand, and in analogue models (Jarrard, 1986; Diament et al., 1992; Beanland et al., 1998; Lallemand et al., 1999; Haq and Davis, 2010). This deformation has been attributed to strain partitioning, where trench-parallel features arise from regional-scale differences in rheology of the arc and trench. Because stresses associated with oblique subduction can be very complex, with much local variation, strain partitioning causes many structures, both parallel and oblique to the trench, to undergo transcurrent motion.

We interpret the fold-associated structures in the Susitna basin and neighboring Cook Inlet to have developed with similar trench-parallel strain partitioning. This strain will accommodate complex stresses due to both oblique Pacific plate subduction and shortening across Cook Inlet associated with Yakutat microplate collision and subduction (Bruhn and Haeussler, 2006). Other structures in the region must then accommodate other components of strain not parallel to the trench. The Castle Mountain, Denali, Bruin Bay, and Border Ranges faults are probable candidates as numerous studies have found geologic evidence of repeated Mesozoic and Cenozoic reactivation of these faults; this motion is mostly strike-slip but sometimes includes transcurrent motion (e.g., Detterman and Reed, 1980; Pavlis and Crouse, 1989; Redfield and Fitzgerald, 1993; Haeussler et al., 2000; Pavlis and Roeske, 2007; Betka et al., 2017). We note one small difference between the Susitna basin–Cook Inlet system and the other arc systems described above: the horizontal distance between the Alaska-Aleutian trench and the associated folds is much greater for the Susitna and Cook Inlet basins (300–400 km versus ∼200 km).

Recent (Miocene and later) preferred slip along the north-striking faults in the Susitna basin, however, cannot be explained by trench-parallel strain partitioning because of their orientation. Possible clues to favored motion along these faults come from comparing the timing of these faults to that of subduction of the Yakutat microplate. Geologic and tomographic studies suggest Yakutat subduction initiated during Eocene to Oligocene time, with surface effects estimated to occur as early as the Eocene to Miocene, depending on the region, and magmatism initiating in the Oligocene (Plafker, 1987; Eberhart-Phillips et al., 2006; Hoffman and Armstrong, 2006; Abers, 2008; Haeussler, 2008; Finzel et al., 2011; Trop et al., 2012, 2019; Brueseke et al., 2019). Additionally, the Susitna basin overlies the southwestern boundary of the subducted Yakutat microplate (at ∼50–100 km depth) as defined by seismic tomography analyses (Eberhart-Phillips et al., 2006; Figs. 1 and 11). It thus seems possible that the processes that led to exhumation of mountain ranges throughout the region also influenced fault slip within the Susitna basin. We note that reverse slip along these faults is favored in the current stress regime: Analyses of crustal (<30 km depth) earthquake focal mechanisms beneath the basin suggest that the modern maximum compressive stress is east-west and that the least compressive stress is vertical (Flores and Doser, 2005; Haeussler et al., 2017). Additional analyses of stress variations within the region could assist further examination of these possibilities.

Additional Observations from the Aeromagnetic Data

High-Low Anomalies: Rotated Blocks or Translated Terranes

Several locales in the study area exhibit magnetic high-low anomalies associated with magnetic gradient highs (Fig. 5; Fig. S12), which most likely represent a horizontal remanent magnetization. Similar “freckled” high-low anomalies have been observed over a 1000-km-long belt in southern Alaska (Saltus et al., 2007), so the detailed study here allows initial interpretation of this widespread and important magnetic feature. This region was referred to as the “Southern Alaska Magnetic High” by Saltus et al. (2007) because applying long-wavelength filters to these anomalies suggests a broad magnetic high. A similar high-low anomaly paired with a total gradient high was observed over a Jurassic pyroxenite body to the southwest near the Pebble porphyry deposit (Shah et al., 2013), also within that belt.

One of these anomalies is adjacent to the Skwentna fault. Here, lineaments of the high-pass filtered anomaly are oriented N20E-N25E instead of N35E-N60E as observed elsewhere in the basin (Fig. 8; Fig. S1, “1”). Both the more northerly orientation of lineaments and the presence of a magnetic high-low anomaly suggest that this block has been rotated from an earlier position. Additionally, Late Jurassic to Early Cretaceous sedimentary rocks exposed in the nearby Yenlo Hills (Hults et al., 2013) may have experienced uplift related to this rotation, an idea supported by the presence of an associated local basement gravity high (Fig. 7B). Unfortunately, this area is covered with nonmagnetic sedimentary rock and Quaternary sediments; so the magnetic source is buried and not easily subject to testing.

Two of the other areas appear to be associated with older rocks (Fig. S1, “2” and “3” [footnote 2]). One, near the western base of the Talkeetna Mountains, is covered by Quaternary sediments but close to basaltic to andesitic metavolcanic rocks with ages described by Wilson et al. (2015) as Triassic(?) to Pennsylvanian(?) (Fig. 8; Fig. S1, “2”). The other, on the western shore of Cook Inlet, is associated with outcropping volcanic rocks of the Upper Triassic(?) and Lower Jurassic Talkeetna Formation and Cretaceous and Jurassic intrusive rocks (Fig. S1, “3”). These two units are older than the period associated with hypothesized oroclinal bending, which is estimated to have started by the Late Cretaceous to early Paleocene and ceased before the late Eocene from paleomagnetic studies (Thrupp and Coe, 1986; Coe et al., 1989), and may have been rotated during that event. We note that both the aeromagnetic and paleomagnetic data do not distinguish between rotation at the scale of a continent and rotation at the scale of smaller blocks; therefore, implications for oroclinal bending are uncertain. Alternatively, the bodies may belong to accreted terranes that formed farther south and were transported north (e.g., Trop and Ridgway, 2007).

The short-wavelength highs and lows within the “southern Alaska magnetic high” described by Saltus et al. (2007) may then contain high-low magnetic anomalies associated with rotation (at a local or broader scale) or magnetization acquisition at equatorial latitudes. Near Yenlo Hills, the high-pass filtered magnetic data suggest the rotation of 30–40 km-wide blocks.

Low-Magnitude, Shortest Wavelength Anomalies: Glacial Deposits and Sediment Sorting

The shortest wavelength anomalies appear to reflect the presence of glacial deposits containing fragments of volcanic, intrusive, or metamorphic rock that are capable of generating small magnetic anomalies (Fig. 6). The varying strength of these anomalies is likely related to sedimentary provenance and sorting processes: The anomalies are strongest in proximity to the arc volcanoes on the western side of Cook Inlet and along the edges of the Talkeetna Mountains, but few anomalies are observed near the base of the Kenai Mountains where glacial deposits are also present. The Kenai Mountains comprise mostly metamorphic rocks that have much weaker magnetic properties than the young volcanic rocks on the western side of Cook Inlet, and maps of aeromagnetic data (Haeussler and Saltus, 2011) show the magnetic field in this area is subdued compared to that of the arc volcanoes. Glacial sediments in this area are thus less likely to produce magnetic anomalies. There is also less correspondence between the magnetic anomalies and glacial features in the northwestern part of Susitna basin, perhaps because the nominal flight altitude was greater for those surveys, reducing the amplitudes of the anomalies, or perhaps because those glacial deposits are more distal from magnetic source rocks and have weaker magnetic properties.

South of the Talkeetna Mountains, the anomalies show a correspondence with both surficial geology and topographic features. This area is influenced by fluvial and tidal currents of the Cook Inlet estuary. The distribution of these anomalies may then be a function of sediment sorting by glacial, tidal, and fluvial processes. Sediments with elevated magnetic susceptibility (often containing magnetite, hematite, ilmenite, and other minerals) tend to be denser than less magnetic sediments such as quartz and can produce measurable magnetic anomalies when concentrated (Shah et al., 2012). In this area, the gridded high-pass anomalies (Fig. 5) show lineaments oriented in multiple directions that are not aligned with folds and faults mapped by Haeussler and Saltus (2011) (those structures were mapped using a combination of magnetic and seismic reflection data). The gridded anomalies in the Cook Inlet basin thus appear to show anomalies due to both Tertiary volcanic and Quaternary sedimentary sources, consistent with preliminary observations of the Cook Inlet aeromagnetic data set by Saltus et al. (2001).


Gravity, magnetic, seismic reflection, and well data show that the Susitna basin is an asymmetric basin bounded by faults to the south, southwest, and north, with a mostly gentle rise to the east. The basin sedimentary fill reaches depths of more than 4 km below the surface over the western part of the basin, but outside of an ∼45 km × 25 km fault-bounded area, the basin is less than 2 km deep. This depth asymmetry and the presence of tilted strata in the easternmost part of the basin suggest flexural uplift of the basin in that area. Deformation associated with flat-slab subduction of the Yakutat microplate is consistent with uplift of strata to the east, unconformities in the stratigraphic layers observed via well data, and exhumation of the Talkeetna Mountains, as suggested by previous work. Additional loading from the surrounding mountain ranges may also contribute to flexural deformation, but probably to a lesser degree.

Comparisons between magnetic and seismic reflection data suggest that folds and faults within the basin cause local uplift of late Paleocene to early Eocene volcanic rock layers that underlie the sedimentary fill, which, in turn, generate subtle magnetic anomalies that can be visualized with targeted filtering. The magnetic and seismic reflection data suggest two primary structural directions within the basin—one is northeast-southwest and corresponds mostly to folds but also some faults, and the other is north-south, corresponding mostly to faults, some of which are associated with fault-propagation folds. The northeast-trending folds appear to have formed between the middle Eocene and early Miocene, with motion in areas to the north possibly continuing into the Pliocene or later. The north-trending faults show Miocene to Pliocene activity and probably continue to be active in the present. These broad structural trends of the Susitna basin and also of neighboring Cook Inlet basin suggest relations to regional tectonic processes. Deformation within these basins is consistent with trench-parallel strain partitioning based upon a correspondence between the strike of the three-dimensional plate boundary and the orientation of folds within the basins. Within the Susitna basin, deformation from collision and subduction of the Yakutat microplate may have led to reverse slip being more recently favored on north-striking faults.


We thank Rick Saltus for providing gravity station data used in the analyses and for thoughtful discussions regarding the magnetic data. We thank Rick Blakely, Peter Betts, Bob Gillis, and several anonymous reviewers for providing helpful suggestions that improved the manuscript. This effort was funded by the U.S. Geological Survey Energy Resources Program Alaska Petroleum Systems Project. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

1Supplemental Shapefiles. Providing locations of faults and folds determined from magnetic and seismic data, or just magnetic data, with projection NAD27-UTM5N. Please visit https://doi.org/10.1130/GEOS.S.12218162 to access the supplemental material, and contact editing@geosociety.org with any questions.
2Supplemental Material. Figure S1 showing locations of magnetic high-low anomalies that exhibit total gradient highs, numbered 1–3 as described in the text with: (A) simplified surface geology (colors as in Fig. 2), arrow marks an area near metavolcanic rock described as Triassic to Pennsylvanian? by Wilson et al. (2015); (B) reduced-to-pole magnetic field; and (C) total gradient and thick black lines delineate faults interpreted in this study and by Haeussler and Saltus (2011). Please visit https://doi.org/10.1130/GEOS.S.12218234 to access the supplemental material, and contact editing@geosociety.org with any questions.
Science Editor: Andrea Hampel
Guest Associate Editor: Jeffrey A. Benowitz
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.