High-resolution three-dimensional terrain models are used to evaluate the Ragged Mountain fault kinematics (Katalla, Alaska, USA). Previous studies have produced contradictory interpretations of the fault’s kinematics because surface ruptures along the fault are primarily steeply dipping, uphill-facing normal fault scarps. In this paper, we evaluate the hypothesis that these uphill-facing scarps represent extension above a buried thrust ramp. Detailed geomorphic mapping along the fault, using 20-cm-resolution aerial imagery draped onto a 1-m-resolution lidar (light detection and ranging) elevation model, was used to produce multiple topographic profiles. These profiles illustrate scarp geometries and prominent convex-upward topographic surfaces, indicating significant disturbance by active tectonics. A theoretical model is developed for fault-parallel flow over a thrust ramp that shows the geometric relationships between thrust displacement, upper-plate extension, and ramp dip. An important prediction of the model for this study is that the magnitude of upper-plate extension is comparable to, or greater than, the thrust displacement for ramps with dips greater than ∼45°. This model is used to analyze profile shapes and surface displacements in Move software (Midland Valley Ltd.). Analyses of scarp heights allow estimates of hanging-wall extension, which we then use to estimate slip on the underlying thrust via the model. Assuming a low-angle (30°) uniformly dipping thrust and simple longitudinal extension via normal faulting, variations in extension along the fault would require a slip gradient from ∼8 m in the north to ∼22 m in the south. However, the same north-south variation in extension with a constant slip of 8–10 m may infer an increase in fault dip from ∼30° in the north to ∼60° in the south. This model prediction has broader implications for active-fault studies. Because the model quantifies relationships between hanging-wall extension, fault slip, and fault dip, it is possible to invert for fault slip in blind thrust ramps where hanging-wall extension is the primary surface manifestation. This study, together with results from the St. Elias Erosion and Tectonics Project (STEEP), clarifies the role of the Ragged Mountain fault as a contractional structure within a broadly sinistral shear system in the western syntaxis of the St. Elias orogeny.


The St. Elias orogen of southern Alaska (USA) was the focus of the St. Elias Erosion and Tectonics Project (STEEP), and a significant part of that effort was devoted to understanding the western syntaxis of the orogeny, just east of the Copper River (Fig. 1). New lidar (light detection and ranging) data with accompanying aerial photography were acquired as part of STEEP to help resolve the nature of extensive surface ruptures known from previous work (e.g., Carver and McCalpin, 1996; Bruhn et al., 2004), but the attributes of the surface ruptures were poorly understood. A particular focus of the lidar survey was the trace of the Ragged Mountain fault, previously mapped as a thrust (e.g., Miller, 1961) but interpreted as a Holocene normal fault by Tysdal et al. (1976). Distinguishing between these end members is obviously critical for understanding the regional tectonics. Similar problems occur on other active fault systems where surface ruptures through extension occur in contractional settings. Thus, resolving this issue is fundamental.

In this paper, we use the high-resolution topography from the lidar survey together with aerial photography to evaluate the tectonic geomorphology associated with surface ruptures of the Ragged Mountain fault (Video S1, grayscale digital elevation model in the Supplemental Material1). The lidar data allow us to extract topographic profiles across the scarp traces. These profiles illustrate the curvature of the topographic surfaces adjacent to the scarp system and allow us to measure scarp heights and evaluate their origin. We use these observations to hypothesize that the extensional surface ruptures are produced by flexure above a thrust ramp. We then develop a simple kinematic model to test the hypothesis and conclude that the extensional scarp structures examined by Tysdal et al. (1976) represent a combination of erosionally modified features (glacier flutes and glaciofluvial and glaciomarine sediment) overprinted by flexural extension above a thrust system. Analysis of scarp heights along the fault trace shows an increase from north to south. Based on the model, this variation can be explained in two ways: (1) by increasing slip from north to south on a thrust of constant dip, or (2) by an increase in fault dip at approximately constant slip. We conclude by assessing the broader implications of our fault kinematic models to general studies of fault-scarp systems where thrust systems display hanging-wall extension, and consider regional tectonic implications of our results.


Southern Alaska is dominated by actively deforming collisional processes within a long-lived subduction zone (Plafker et al., 1994). North-directed subduction of normal oceanic lithosphere in the Aleutian arc transitions eastward into flat-slab subduction and then into the Fairweather–Queen Charlottes transform in southeastern Alaska (Fig. 1B). The flat-slab region was produced when the buoyant Yakutat terrane, an oceanic plateau, was carried into the subduction zone (Plafker, 1987; Pavlis et al., 2004; Worthington et al., 2012). This collision has driven broader Neogene deformation throughout southern Alaska and concentrated deformation from the eastern Chugach and St. Elias Mountains into the Fairweather Range of southeastern Alaska (e.g., Plafker et al., 1994; Koons et al., 2010).

The internally undeformed Yakutat microplate (Fig. 1B) is bounded on all sides by actively deforming zones. First, to the east, the microplate is a slip-partitioned, dextral-transpressional orogen with strike slip localized along the Fairweather fault and with an associated contractional flower structure along the strike-slip system (Bruhn et al., 2004; Doser et al., 2007). Second, to the southwest, the Transition fault separates the Yakutat microplate from the Pacific plate and is a Cenozoic strike-slip fault weakly reactivated during the collision (e.g., Gulick et al., 2007, 2013; Christeson et al., 2010). Third, the northern margin of the Yakutat microplate is an intricate fold-and-thrust belt that varies in compressional wave velocities markedly along strike (Worthington et al., 2010, 2012; Pavlis et al., 2012; Chapman et al., 2012). Fourth, the Chugach–St. Elias and Ragged Mountain faults, the most conspicuous active faults in the western part of the St. Elias orogen, form the collisional suture. This suture separates Paleogene subduction-related assemblages of the Orca Group from Cenozoic sedimentary cover rocks of the Yakutat terrane (e.g., Plafker, 1987; Bruhn et al., 2004; Pavlis et al., 2004, 2019). Bruhn et al. (2004) and Pavlis et al. (2004) concluded that the Ragged Mountain fault marks the western boundary of a structural syntaxis that initially trended approximately east-west but was deformed by vertical-axis rotation and refolding of an earlier generation of folds (Fig. 1).

Geodetic surveys of the region show that the Yakutat microplate is moving northwest to north-northwest at ∼45–49 mm/yr relative to Alaska’s interior (Elliott et al., 2010, 2013). The movement creates considerable seismicity along the Aleutian megathrust, including the 1964 CE Great Alaskan earthquake (Plafker, 1969; Plafker et al., 1994; Shennan et al., 2008, 2015; Li et al., 2010; McCalpin et al., 2011). Plafker (1969) and Tuthill and Laird (1966) reported that the rupture of the Aleutian megathrust during the M9.2 1964 earthquake caused coseismic uplift within the western St. Elias orogen. This rupture also triggered many landslides, snow avalanches, and widespread surficial deformation of Quaternary deposits. Elliott et al. (2013) determined velocity gradients within the interior of the microplate, which they interpreted as strain accumulation above buried faults that are also potential seismic sources. Block models of the geodetic data indicate complex interactions within the western part of the St. Elias orogen, consistent with the observed surface geologic complexities (Elliott et al., 2013). Similarly, seismic studies by Ruppert et al. (2008) and Doser et al. (2007) used earthquake focal mechanism solutions to indicate a predominantly mixed strike-slip to thrust faulting stress regime in the western St. Elias orogen, which is consistent with a northwest- to west-trending maximum horizontal compression axis.

Glacial erosion during the Pleistocene sculpted the landscape, forming elongated mountain blocks surrounded by flat-floored valleys filled with glaciofluvial and marine sediments and ice (McCalpin et al., 2011). This landscape is analogous to an archipelago where the sea between the islands (highlands) partially filled with young glaciofluvial and glaciomarine sediment. Some true islands (Kayak Island and Wingham Island) persist, however, because sedimentation has not yet filled accommodation space produced following sea-level rise at the close of the Last Glacial Maximum (LGM).

The complex bedrock geology of the Katalla region (Miller, 1961; Bruhn et al., 2004), together with geodetic studies, seismicity, and the first-order topographic features discussed above, indicate the region is actively deforming. However, the surface manifestation of this deformation is poorly understood. Although numerous surface ruptures are recognized in this region, most of these structures are gravity-related sackungen (Li et al., 2010; McCalpin et al., 2011). The Ragged Mountain fault system, however, displays one of the most continuous surface ruptures in the orogen (Bruhn et al. 2004). Its origin is crucial to understanding the neotectonics of the orogen (McCalpin et al., 2019).


Quaternary surficial deposits dominate the Ragged Mountain study area (McCalpin et al., 2019). The deposits include: (1) extensive, variably active cobble to boulder talus along the eastern flank of the mountain range; (2) unconsolidated alluvial sand and gravel in inactive stream channels; (3) landslide deposits; (4) glacial drift; and (5) loess. These unconsolidated deposits rest on complexly deformed rocks that include (1) the Eocene Stillwater and Tokun Formations of the Yakutat terrane, and (2) the metavolcanic unit of the Orca Group, which is a pre-collisional accretionary complex developed along the Paleogene northern Cordilleran margin (Plafker et al., 1994). The Ragged Mountain fault (Figs. 1 and 2) separates the Orca Group rocks from the deformed cover of the Yakutat terrane. It thus comprises the collisional suture, but the contact has been extensively reactivated, including via the Quaternary motion that is the subject of this paper (e.g., Plafker et al., 1994; Pavlis et al., 2004).

Miller (1951) initially mapped the Ragged Mountain fault, and Tysdal et al. (1976) documented Quaternary reactivation along at least 30 km of the fault’s trace. Tysdal et al. (1976) suggested that the Ragged Mountain fault reactivated during the Holocene as a low-angle normal fault with 180 m of backsliding toward the west. The style and amount of faulting are based on uphill-facing scarps in the hanging wall and the width of a shallow trough in the footwall block. Since the Tysdal et al. (1976) study, no evidence has arisen to further support this hypothesis. Extensive studies as part of STEEP, as well Carver and McCalpin (1996), recognized swarms of extensional scarps throughout the Katalla area, but the vast majority of these features are surficial features (sackungen) generated during major earthquakes (e.g., Li et al., 2010; McCalpin et al., 2011). A notable exception is a system of east-west–striking, sinistral-normal fault scarps just east of the Ragged Mountain fault that are demonstrably tectonic (Pavlis and Bruhn, 2011; McCalpin et al., 2011). These sinistral-normal structures are compatible with contemporary stress regimes estimated from earthquake seismology (e.g., Ruppert et al., 2008) but are incompatible with a hypothesis that the Ragged Mountain fault is extensional today. Ruppert et al.’s (2008) data and regional tectonics both indicate that the Ragged Mountain fault should be contractional, challenging the Tysdal et al. (1976) hypothesis unless the Ragged Mountain fault is recording an entirely surficial process.

McCalpin et al. (2019) recently presented evidence that the Tysdal et al. (1976) hypothesis is flawed, based on geomorphic evidence and paleoseismic studies that included a trench (Fig. 2B) excavated across a section of the fault that they had identified as a likely thrust scarp. We expand on that work, examining further details of the geomorphology and its relationship to the underlying geologic structure.


Image Processing and Georeferencing

This study began with georeferencing National Center for Airborne Laser Mapping (NCALM) high-resolution aerial photography with ∼15 cm ground resolution along with the unfiltered 1 m lidar digital elevation model (DEM) (Fig. 2A) (Videos S1 and S2 in the Supplemental Material [footnote 1]). Details of the lidar data set and data processing are contained in Pavlis and Bruhn (2011, their table 1) and the metadata with the data archive at www.opentopography.org (https://doi.org/10.5069/G9R20Z92).

The high-resolution aerial photography was collected during lidar data acquisition, but the digital photographs required both georeferencing and image processing to provide clarity. The aerial photography was delivered as photo files along with a flight-line map, but individual images were not georeferenced. The imagery was of variable quality and recorded as unprocessed TIFF images. Thus, image processing and data sorting were required.

The ∼14,000 images contained in 70 flight-line files had to be manually viewed to determine image quality, limiting the scope of this study. Most images were underexposed and illegible as raw files, and thus we were forced to make an automatic brightness and contrast adjustment as an initial step. After this step, we were able to recognize flight lines and sort the data for imagery needed for this study. After this sorting, many images were still unrecognizable or visibly blurred and so were discarded. Photos from densely forested areas where the ground surface was masked were eliminated. Chosen images were then manually georeferenced using spatial referencing properties in ArcGIS 10 software and orthocorrected. Ground control points were obtained from natural objects visible on both the photo imagery and shaded-relief imagery constructed from the unfiltered lidar DEM. The images were not systematically georeferenced in numerical order because some of the study areas were densely vegetated and flight lines had significant overlap. In general, when manually georeferencing an image from ground control points, at least four points are needed, but it is ideal to have more than four. In some cases, every third image from a catalog allowed full coverage because of overlap, but this did not always apply, requiring image-by-image selection. In some areas, image quality was too poor for full resolution, and we were forced to use low-quality imagery, although these effects are minor.

After the bulk of this study was completed (Heinlein, 2013), a satellite image with a resolution of ∼0.5 m (Fig. 2B) became available. It allowed us to extend the work to a larger area without the need for the orthocorrection of the NCALM images. These lower-resolution satellite images combined with the higher-resolution NCALM imagery further constrain our interpretations of the geomorphology.

This study used high-resolution elevation data and aerial and satellite photography to investigate and interpret the geomorphology and surficial structure of the Ragged Mountain fault system. Data collected from remote-sensing data sets were analyzed using GIS software, primarily ArcGIS 10 but also QGIS 2.x for interpretations of high-resolution satellite imagery (Fig. 2). We also used Move software (Midland Valley Ltd.) for visualization and mapping as well as cross-section work.

Topographic Profile Analyses

Filtered slope maps were used to enhance the identification of fault scarps on the lidar DEMs (see detailed description in Heinlein, 2013). The Image Classification toolbar in ArcGIS 10 was used to develop a histogram of a slope, which was then used to develop a color ramp denoting ranges of slope angles. This application enhances data visualization of slope and topographic features and improved our ability to identify the trace of the Ragged Mountain fault in areas where it is obscured in conventional hillside shaded-relief images. The five-level red-to-blue color-ramp map was used to image the spatial distribution and continuity of scarps (Fig. 3). The profiles generated using the terrain models were also used to extract the scarp height and slope angle for each identified scarp and determine the shape of the scarps (convex- or concave-upwards topographic topography up- and downslope of the scarps), which is essential for our analysis below. Measurements between the crest and the toe or base of the fault scarp were used to determine the scarp height.

Topographic profiles extending across the Ragged Mountain fault provide measurements of fault offset and the geometry of the adjacent slopes (Fig. 2B). Short profiles ranged from 6 m to 34 m in length and were used to characterize the morphology of uphill-facing scarps and measure scarp height (Figs. 2B and 4). Profiles ranging from 60 m to 2000 m in length were used to explore the morphology of the entire fault scarp area (Figs. 2 and 5). A subset of profiles was also extracted from the DEM along the eastern flank of Ragged Mountain to examine secondary and possible flexural slip scarps.

Fault Flexure Model for the Extensional Scarp

To further test the thrust faulting hypothesis, we developed a series of forward and reconstruction fault-parallel flow models using two-dimensional section tools in Move 2013–2016 software. We used the kinematics of these models to derive a relationship between fault slip, fault dip, and hanging-wall extension (see below).


Nature of the Ragged Mountain Fault Scarp

The Ragged Mountain fault system forms a regional convex-eastward trace (Fig. 2). We mapped 37 scarp segments with individual segments ranging from 41 m to 1420 m in length and scarps from 0.2 m to 11.3 m in height.

The northern half of the fault is marked by a series of discontinuous, uphill-facing scarps developed on east-facing topographic slopes of Ragged Mountain (Fig. 6). Many of these scarps are visible on lidar-generated shaded-relief images, but vague on aerial photography. These discontinuous scarps could indicate that the ruptures are relatively old because some alluvial fans, talus cones, and glacial deposits cross the fault trace uninterrupted. Nonetheless, scarps are also visible on younger active talus fans, suggesting that rupturing of the surface may be discontinuous along strike.

The fault trace increases in elevation southward, where it reaches a higher-altitude, upland surface above the local tree line, and crosses the topographic divide between the Katalla and Martin Lake valleys (Fig. 2). In this section of the fault, two distinct geomorphic features mark the fault trace (Figs. 711). One of these geomorphic features is a topographic trough that is mostly filled by permanent snowfields (Fig. 7). It includes segments occupied by streams and talus cones impinging on the west side of the trough. The second geomorphic feature is a system of uphill-facing scarps that trend subparallel to the trough and lie 100–150 m upslope to the west of the trough (Figs. 810). Tysdal et al. (1976) mapped the fault trace along the trough, but the most prominent surface ruptures are the uphill-facing scarps located in talus and landslide deposits above the trough. The trough is relatively irregular, with rough glacial fluted terrain immediately to the east of the trough, including fluting of the ridge that forms the trough’s eastern edge (Fig. 10). The fault scarps upslope from the trough can be followed southward to similar uphill-facing scarps that lie upslope from the trench site where McCalpin et al. (2019) exposed thrust fault splays (Fig. 11). Thus, the uphill-facing scarps located above the trough are presumably secondary structures related to movement on the underlying thrust fault, as described at the trench site by McCalpin et al. (2019). Together these observations indicate that the trough is either: (1) a much older fault scarp system that predates the latest ice cover of this region, or (2) merely an erosional feature developed along fault rocks of the Ragged Mountain fault system. Fleisher et al. (1999) dated glacial drift from the mountain pass just east of the trough and showed that ice retreated from this region ca. 10 ka, placing a minimum age for the trough and maximum age for the uphill-facing scarps along the Ragged Mountain fault. Following the reasoning of McCalpin et al. (2019), we interpret the trough as an erosional feature.

To the south of the topographic divide, the trough disappears for ∼5 km (Fig. 12). In this section of the fault, the eastern slope of Ragged Mountain abuts a relatively flat, glacially carved surface with minor modification by post-glacial stream erosion. The uphill-facing scarp continues through this segment, ∼100 m west of the slope break, and its sharp geomorphic expression indicates that the rupture must be relatively young. That is, the rupture cuts young talus cones with only local evidence of burial by talus accumulations. In the southern portion of the study area, the scarp system changes character in the 7–8 km of the fault trace visible before it disappears offshore (Fig. 12). Some details are lost in the lidar bare-ground model due to insufficient filtering of low brush from the DEM, producing a surface texture that obscures details. Even so, a fundamental distinction in this part of the fault trace is that the prominent uphill-facing scarp observed in the central section of the fault becomes the dominant scarp. That is, although some smaller scarps are present, the fault trace becomes readily traceable as a significant, west-side-down (uphill-facing) scarp that locally ponds drainages to produce small lakes and deflects drainages along the trace (Fig. 12). The main scarp in this segment reaches heights of as much as 11.3 m.

Two observations stand out from our analysis of the topographic profiles across the Ragged Mountain fault (Figs. 2, 4, 6, 7, and 8):

  • 1. The northern two-thirds of the fault trace is characterized by one to two small uphill-facing scarps that consistently lay upslope from the base of the eastern slope of the Ragged Mountain (Figs. 4 and 6). These surface ruptures all occur within the hanging wall of the Ragged Mountain fault. In contrast, the southern one-third of the fault trace is marked by a much larger uphill-facing scarp that lies near the base of the topographic slope, but this scarp remains in the hanging wall of the principal fault (Figs. 4 and 6).

  • 2. The slope of the mountain front beneath the uphill-facing scarps is convex upward in several areas. Convex-upward topographic profiles are unusual at the base of mountain slopes unless faulting, folding, differential erosion, or landsliding have disturbed the slope. Along most of the trace, however, this convex shape appears to be a result of surface warping rather than erosion, an observation that is relevant to our analysis.

Kinematic Model for the Ragged Mountain Fault System

Based on our analysis of the Ragged Mountain fault scarp and the results from McCalpin et al. (2019), who exposed a thrust in the central segment of the fault, we propose that the uphill-facing normal fault scarps along the Ragged Mountain fault trace represent flexural extension above a buried ramp in a thrust. Flexure-related normal faults are known from many thrust earthquakes (e.g., Philip and Meghraoui, 1983; Sloan et al., 2010; Arrowsmith and Zielke, 2009; Arrowsmith et al., 2017). Determining the amount of flexure-related extension, however, depends on the flexural model. For example, the hanging-wall stretch is zero in a classic fault-bend fold (Suppe, 1983) but varies among blind thrusts models (Ekstrom et al., 1992) and is significant in models like the fault-parallel flow model (Egan et al., 1997, 1999).

In this case, we evaluate our hypothesis that the uphill-facing normal fault scarps represent flexural extension using a fault-parallel flow model for thrusting (Egan et al., 1997, 1999) over a ramp at a depth that transfers to a flat fault near the surface (Fig. 13). In fault-parallel flow (Fig. 13A) material flow lines in the hanging wall parallel the template of the fault, which forces the material to extend as it moves through the principal axial plane of the ramp anticline above the ramp (Egan et al., 1997, 1999; Ziesch et al., 2014). This hanging-wall extension is fixed by the geometry of the hanging wall on the buried footwall ramp, with the main variables shown in Figure 13B. Here, we assume that the uphill-facing scarp records this hanging-wall extension and that the scarp forms along the leading axial plane of the ramp anticline (Figs. 13B and 13C). This assumption fixes a geometric relationship between the ramp dip (θ) and the location of the top of the ramp, which allows us to relate the hanging-wall extension to the amount of thrust motion through a trigonometric solution (Figs. 13B and 13C). From this geometry, we first consider the length of a segment of the hanging wall that moves through the axial surface during a displacement (d). From Figure 13B, we can calculate the initial length (Li) using the law of sines (Fig. 13B):

The angles γ, τ, β, ∈, and α, shown in Figure 13B, are interrelated:

  • • τ is the angle between line BD (representing Li) and line BE (which represents the length h measured along the axial surface);

  • • τ is the angle between line BE (h) and line BF (displacement d in the horizontal direction in advance of the axial surface);

  • • β is the angle between line DE (displacement dp parallel to ramp dip trailing the axial surface) and line BE (h);

  • • ∈ is the angle between line BF (d) and line EF (final length, Lf); and

  • • α is the surface slope angle.

These angles are related by:
because the sum of angles in triangle ABC must be 180°, and by:

by opposite adjacent angles.

To calculate Lf, we need to know the length h, which through law of sines is:
Once h is determined, Lf is resolved using the law of cosines for a fourth intermediate obtuse triangle created along the Li triangle (the green axial surface reference line element in Fig. 13B), which is:
Substituting Equations 4 and 2 into Equation 1 produces:

From Equations 6 and 7, we can determine the change in length (Δl = LfLi) and the stretch (S = Lf / Li) of the fault.

Equations 6 and 7 show that the magnitude of the hanging-wall stretch is related to three parameters: θ, d, and α. The surface slope α can be measured directly from the DEM. The displacement d, however, can only be indirectly estimated as a minimum value from surface deflection because weathering and erosion processes would have reduced the surface deflection and d is also dependent on θ. Finally, θ is an independent parameter, and variations in θ constitute a significant control on the stretch. Thus, Equations 6 and 7 allow us to evaluate the upper-plate extension for combinations of d and θ.

Figure 14A illustrates the model predictions for the upper plate. The key observations from this figure are: (1) the hanging-wall extension is modest (<30% of the thrust displacement) for a thrust with a low dip (<20°–30°) but increases at steep dips, such that at dips of 65° or more, and (2) the hanging-wall extension may exceed the slip on the thrust (Fig. 14B). The effect of the surface slope is also evident in the sets of model curves (Fig. 14A), with the somewhat counterintuitive effect of decreasing extensional strain (shown as stretch in Fig. 14A) with increasing surface slope at high ramp dips (Fig. 14A). Note that this effect is a finite-strain effect related to the orientation and length of the line elements relative to the fault-parallel flow, similar to variations in the stretch of line elements in shear zones. Thus, when this strain is shown as change in length per unit thrust displacement d (Fig. 14B), a more meaningful field parameter, the relationships are more straightforward.

The results of Figure 14 can be considered in another way to estimate thrust slip from the hanging-wall extension using the model (Fig. 15). For example, if the hanging-wall extension is 4 m, then the thrust displacement that generated this extension would be 4.3 m for a thrust with a 60° ramp and 30° slope, but for a 30° ramp and the same slope, the thrust would have to have 10 m of slip to produce the same extension.


Slip-Sense Problem for the Ragged Mountain Fault Scarp Based on Previous Studies

Tysdal et al. (1976) proposed that 180 m of normal slip, or a slip rate of ∼1.8 cm/yr, occurred on the Ragged Mountain fault based on the assumption that the hanging wall of the fault had slid westward from the eastern edge of the trough. This slip rate seems unlikely given that it is a significant fraction of plate convergence rates, and geodetic as well as seismicity studies predict contraction, not extension (Elliott et al., 2010, 2013; Elliott, 2011; Enkelmann et al., 2015; Pavlis et al., 2019). Nonetheless, the presence of numerous smaller-scale extensional scarps made the hypothesis allowable.

McCalpin et al. (2019) summarized the geomorphic expression of the trough based on STEEP field studies and analysis of the lidar data. They concluded that the trough was not fault related, but rather an erosional feature. We add to their observations that the U-shape of the trough (Figs. 7 and 10) suggests glaciers at least partially carved it during the LGM when ice passed over and began to retreat at ca. 10 ka (Chapman et al., 2009, 2011; Sirkin and Tuthhill, 1987; Denton, 1974). In addition, fluted glacial terrain with landforms that cross over the geomorphic trough (Figs. 7 and 10) indicate that even if the trough records movement, there is no Holocene offset. Thus, any inference of timing or rates cited by Tysdal et al. (1976) is not relevant, even if the feature were extensional.

Collectively, our observations and those of McCalpin et al. (2019) indicate that the trough described by Tysdal et al. (1976) is irrelevant to the local active tectonics because it is neither a surficial landslide-related feature nor a tectonic scarp. Thus, although it is one of the most striking landforms in the study area, it has no bearing on the tectonic problem.

Slip-Sense Problem for the Ragged Mountain Fault Scarp Based on Geomorphology

A fundamental observation from this study and McCalpin et al. (2019) is that the only evidence that the Ragged Mountain fault is active along its entire trace is the array of uphill-facing normal faults, upslope from the mapped position of the fault. The only direct evidence that the structure is a thrust is limited to a short, ∼1-km-long surface rupture with classic thrust-scarp morphology and unequivocal evidence of thrusting in the trench that crosses the fault (McCalpin et al. 2019). Thus, an allowable hypothesis is that the thrust scarp described by McCalpin et al. (2019) is a surficial feature (e.g., landslide toe), and that the fault system as a whole is extensional.

However, there is other evidence that supports the thrust fault hypothesis. First, regional stresses based on earthquake focal mechanisms (Ruppert et al., 2008) are favorable for reverse and strike-slip faulting (Ruppert et al., 2008), not extension. Ruppert et al. (2008) showed that the maximum horizontal stress direction inferred from earthquake focal mechanisms shows various orientations and high variance values, indicating that the stress field is heterogeneous. Bruhn et al. (2012) expanded on this interpretation, inferring approximately east-west P-axes (pressure axes) and north-south T-axes (tension axes), consistent with contraction on the Ragged Mountain fault and McCalpin et al.’s (2011) evidence for strike-slip movement along scarps east of the Ragged Mountain fault. Similarly, geodetic modeling (Elliott et al., 2013; Enkelmann et al., 2016) predicts contractional strain across the Ragged Mountain fault.

Second, although thrust fault scarps are confined to only a few hundred meters of the ∼30 km trace of the fault (McCalpin et al., 2019), short surface ruptures or nonexistent surface ruptures (blind) are common in active thrust systems. For example, the Suusamyr earthquake (Kyrgyzstan) in 1991 CE was a magnitude 7.3 earthquake with a rupture length of 40 km determined from seismology (Ghose et al., 1997). Nevertheless, the surface rupture of the earthquake was limited to a very short, ∼300 m segment of the fault (Ghosh et al., 1997). Thus, the observed short thrust-rupture segment at Ragged Mountain is consistent with thrusting but that most of the thrust did not rupture to the surface.

Third, field photos (Fig. 8) and many of the extended topographic profiles (Fig. 5) reveal a convex-upward surface slope above (Fig. 8A) and below the uphill-facing scarps of the Ragged Mountain fault system (Figs. 8B and 8C). For sites where convex-upward profiles occur above the uphill-facing scarp (e.g., Figs. 5 and 8A), the slope curvature probably reflects motion on a curved normal fault, dipping toward the slope. For sites with convex-upward profiles below the uphill-facing scarp, the topographic profile is inconsistent with a simple erosional slope, but rather is characteristic of slopes above a curved thrust fault. Indeed, our kinematic models (Fig. 13B) suggest that convex-upward topographic profiles should be characteristic of slopes formed above thrust ramps. We suggest that collectively these observations imply that these convex-upward landforms are the result of a thrust at the toe of the slope, but that thrust is mainly blind except near the trench site. We suggest further that the extensional, uphill-facing scarps are the result of flexure above this mostly blind thrust trace.

Application of the Fault Flexure Model for the Extensional Scarp

To further test the thrust hypothesis, we developed a series of forward (Fig. 16) and reconstruction models (Fig. 17) for the Ragged Mountain fault system. In particular, we evaluated the use of the extensional scarp development to infer thrust slip development of the convex-upward topographic profiles (Fig. 16).

The forward model’s primary purpose was to analyze the theory and use the models as a guide for reconstruction. Figure 16 shows a simple forward model we used in this analysis: a 400 × 600 m rectangular model, a fault ramp dip angle of 45°, and a surface slope of 29° before faulting (Fig. 16A). Note that the exact dimensions are irrelevant in this model other than ramp dip, which is similar to that of natural thrust systems. We also stipulate a large slip (100 m) on the fault, which is intentionally exaggerated to illustrate the model geometry clearly. The forward model includes two steps, which coincide during a seismic event: step 1 imposes slip on the thrust (Fig. 16B), and step 2 imposes slip on a listric normal fault (Fig. 16C) with a slip magnitude consistent with the theory’s predictions (Fig. 13). This model illustrates how the uphill-facing scarps with a convex-upward topographic profile can form through this process. Upslope from the extensional scarp, convex-upward profiles can form from a combination of flexure above the ramp (Fig. 16B) and flexure along a curved normal fault (Fig. 16D). Similarly, if this model were to include curvature at the top of the thrust ramp, the leading edge of the thrust would also show a convex-upward profile typical along thrust scarps, as described by McCalpin et al. (2019).

Restorations (Fig. 17) provide additional insight. Two simple examples (Fig. 17) of a restoration model are applied to a topographic profile in the central portion of the study area. Based on the model theory, the restorations for this profile require two thrust ramps to account for two distinct uphill-facing scarps upslope from the thrust scarp. We show two cases: a 45° fault ramp dip (Fig. 17A) and a 30° thrust ramp (Fig. 17B). In both restorations, we restore 10 m of extension in the hanging wall (left panels) followed by restoration of slip on the thrust consistent with the theory: 27 m for the 45° ramp, and 16 m for the 30° thrust ramp. Note that this net slip in the restoration is far greater than paleoseismic slip estimates for the most recent seismic events reported by McCalpin et al. (2019) because our estimate is an attempt to restore the topographic profile to remove the convex-upward shape. Thus, our restoration presumably records longer-term accumulated slip than is recorded in the trenches. Indeed, our extension estimate of 10 m is probably a maximum because it is the fault slip required to restore the normal faults, which is an overestimate unless the faults bottom in a detachment (as shown). Despite this simplified assumption, the model produces a reasonable restoration of the topographic profile by applying the theory to both ramp-dip models (Fig. 17). However, the 30° model shows a slightly more realistic restoration of the topographic profile to a concave-upward slope (Fig. 17B). Given these assumptions, we conclude that the application of the model provides a reasonable result given that deposition and erosion are not accounted for in the reconstruction. This model assumes that all of the extension is transferred as slip to the uphill-facing scarp.

Elsewhere in the study area, the thrust motion presumably was blind, yet the model should still be applicable. We extracted normal-slip estimates from the topographic profiles and averaged them in three segments: north, central, and southern. In all cases, we used only profiles that contained minimal evidence of erosional modification, such as minimal smoothing at the scarp crest. In the southern section, we eliminated profiles that crossed the area where McCalpin et al. (2019) interpreted a landslide. With these criteria, we used six profiles in the north, yielding an average slip of 6.5 m; four profiles in the central segment, with an average slip of 11.1 m; and two profiles in the south, with an average slip of 17.7 m. These slip estimates should be an upper bound for hanging-wall extension because the offset would imply that all of the slip was transferred to slope-parallel extensional strain, which would require the faults to bottom in a detachment above the thrust. Alternatively, we can assume a simple extension across a 60°-dipping normal fault (cos [60]) where cos [60] refers to the slope parallel component of the slip or half the slip magnitude transferred to extensional strain. This value is the elongation parallel to slope and should serve as a reasonable lower bound for the extension. Figure 18 shows a plot of how hanging wall extension can convert to thrust slip under different fault dip conditions in the model (Fig. 13). The crucial observation from this analysis is that based on the model, the increase in extension seen from north to south could be produced by increasing slip to the south, or more likely, a simple increase in the dip of the fault at relatively constant net slip. Assuming the latter, we can estimate a net thrust fault slip ranging between ∼7 and 17 m dependent on the model parameters. Note that this net slip estimate is consistent with the 16 m of thrust slip estimated in the reconstruction in Figure 17, assuming a 45° dip on the ramps.

Fault Slip Rate on the Ragged Mountain Fault

One outcome of the modeling results in Figures 13B and 18 is an estimate of the Holocene net slip on the thrust, which cannot be estimated from the McCalpin et al. (2019) trench site itself. Specifically, if we assume the range 7–17 m of slip as the post-glacial offset, we can use the Fleisher et al. (1999) age of deglaciation of ca. 10 ka to calculate a slip rate between 0.7 and 1.7 mm/yr. This rate is modest in an orogen that absorbs as much as 45 mm/yr of shortening from the Yakutat collision. This estimate also is significantly greater than the slip estimates of McCalpin et al. (2019) for the trenched thrust segment, where they recognized three slip events of <1 m over at least 17 k.y. based on optically stimulated luminescence dating of trenched strata. This discrepancy is not surprising, however, given the short length of the thrust scarp segment and the evidence that the thrust is blind elsewhere. Indeed, this hypothesis can explain the distinction between times of slip on the uphill-facing scarps versus thrust faults seen in the McCalpin et al. (2019) study, and probably means that earthquake recurrence times are considerably shorter than the estimates from the thrust trench. For example, if individual thrust events had slip of ∼1 m, there should have been seven to 17 events in the last 10 k.y. based on our net slip estimate.

We suggest that thrust ramp–hanging-wall extension modeling provides a potentially new approach to assessing slip rates on thrust systems that are partially to entirely blind. If hanging wall–normal fault systems can be reconstructed to quantify the extension, then these data can be used to indirectly determine slip on the main thrust, even if it is blind or only ruptures the surface at erratic intervals. More examples need to be examined to test this concept, but there are many areas where the method could be applied. For example, Arrowsmith et al. (2017) studied the 1911 CE Mw 8.02 Kebin earthquake in Kyrgyzstan, a large thrust earthquake that ruptured a zone over a hundred kilometers long. In that study, they recognized uphill-facing scarps in the hanging wall of the thrust, similar in scale to the features we see at Ragged Mountain but from a single event. In that case, there is a thrust scarp where Arrowsmith et al. (2017) were able to assess net slip from the event, and this result could be directly compared to dimensions of the extensional scarps using the methods we use here.

Regional Tectonic Implications

The orientation of the Ragged Mountain fault relative to the convergence vector between the Yakutat microplate and North America suggests sinistral oblique convergence along the fault (McCalpin et al., 2019). Nevertheless, like earlier studies (e.g., Bruhn et al., 2004; Pavlis et al., 2004; McCalpin et al., 2019), we found no direct evidence for oblique slip on the Ragged Mountain fault system. One solution to this problem is that the fault reflects local stresses related to the vertical-axis refolding of folds seen just to the east (Bruhn et al., 2004). That process is driven by partitioned slip with dextral strike slip to the north, driving a sliver westward toward the Ragged Mountain fault (Pavlis et al., 2004). However, geodetic work during STEEP showed no evidence for active strain accumulation indicative of strike slip along the “backstop” to the collisional fold-and-thrust belt (Elliott et al., 2013), raising major questions on the partitioned-slip hypothesis.

Figure 19 shows a potential explanation for apparent approximately east-west contraction along the Ragged Mountain fault in the collisional process without the need for a dextral strike-slip boundary driving a sliver westward. Key in this figure is that a band of sinistral vorticity is centered on the Ragged Mountain area. In this study, a band of sinistral vorticity in this orientation would produce a contraction and extension axis, as shown in the yellow arrows. When added to regional contraction, this secondary deformation would presumably lead to a significant counterclockwise rotation of the local contraction axis, producing approximately east-west shortening recorded as thrusting along the Ragged Mountain fault. Perhaps more significant, however, is that this process’s long-term persistence would lead to counterclockwise rotation of line elements across the zone. Thus, over time, the process would lead to vertical-axis rotations accommodated by the second-phase folds described by Bruhn et al. (2004) as well as the larger-scale rotation of the suture to form the north-south–trending segment of the suture at Ragged Mountain.

This conclusion has broader implications for collisional tectonic studies. The geometric style of the Katalla area with the right-angle bend of the suture is strikingly similar to the western syntaxis of the Himalaya (e.g., Koons et al., 2013). This similarity implies a similar tectonic driver, as illustrated in Figure 19, and is consistent with the conclusion that the geodetic strains match model predictions for indenter margins (e.g., Koons et al., 2013; Nettesheim et al., 2018). Despite this similarity, however, there is an essential distinction in the Katalla region relative to other syntaxes. These other areas are now well known as sites of intense, focused erosion with some of the youngest cooling ages on Earth (e.g., Koons et al., 2013; Enkelmann et al., 2015; Nettesheim et al., 2018), yet the western syntaxis of the St. Elias orogen does not show this characteristic. Regional, very young apatite He ages are well documented (Berger et al., 2008). However, there is no evidence for localized, intense exhumation like that seen in the eastern syntaxis of the orogen (e.g., Enkelmann et al., 2015). Understanding the origin of this distinction between the eastern and western syntaxes is a vital problem for further studies. However, it probably results from a combination of differences: underthrust basement and the presence of a normal subduction zone to the west, in addition to the effects of the trailing edge of the colliding block (Transition fault) near the region.


The Ragged Mountain area is an excellent natural laboratory for applying geospatial techniques that use high-resolution aerial photography and lidar data to investigate active faulting. Virtual mapping of the fault scarps and adjacent topography along the Ragged Mountain fault system, together with elevation profiles across it, reveal an array of uphill-facing extensional fault scarps in the hanging wall that increase in height from north to south, with convex-upward slopes both upslope and downslope of the extensional scarps. Analysis of the geomorphic trough used by Tysdal et al. (1976) to infer extensional slip on the Ragged Mountain fault reveals that the trough is an erosional feature unrelated to Holocene motion on the fault.

We develop model relationships for hanging-wall extension over a thrust ramp using a fault-parallel flow model to determine that uphill-facing scarps develop by hanging-wall extension above a ramp. Restorations using this model replicate the morphology of the Ragged Mountain fault scarps, which, together with regional observations, suggests strongly that the Ragged Mountain fault system is a thrust fault and that the uphill-facing, extensional scarps are related to extension above a fault ramp at depth. The absence of conspicuous thrust scarps aside from the immediate area of the STEEP 2006 trench remains a problem, but our observations indicate that the thrust fault rupturing is primarily blind. Based on the model, the dominance of blind thrusting could indicate that the thrust ramp is relatively steep over most of the area, but more work is needed to test this hypothesis. Using surface offsets, we use the model to estimate a Holocene slip rate on the thrust system of 0.7–1.7 mm/yr. Our analysis suggests that this method can be used to estimate net slip on thrust systems where the surface ruptures are primarily extensional, subject to limitations of the model related to interrelationships of thrust dip and estimates of extension magnitude.


This research was supported primarily by U.S. National Science Foundation (NSF) awards EAR-0409009, EAR-0735402, and EAR-1009533 to T.L. Pavlis; and EAR-0409009 and EAR-1009584 to R.L. Bruhn. The first author also obtained significant support from NSF GK-12 awards DGE-0947992 and DGE-0538623 and a PIRE-Kamchatka award to the University of Alaska Fairbanks (award 0530278). The University of Texas at El Paso (UTEP) Department of Geological Sciences provided critical software and computer facilities for the project; the UTEP CyberShARE aided the study’s visualization components. ConocoPhillips and the Vernon G. and Joy Hunt Endowed Scholarship Fund also provided support for this research. We thank Peter Koons for discussions of the broader tectonic issues and for providing the information for constructing Figure 19. Thank you to Jim McCalpin for sharing his insight and knowledge of the area. We want to thank two anonymous reviewers for their comments, which greatly improved the manuscript.

1Supplemental Material. Video S1: Grayscale digital elevation model generated from high-resolution lidar data illustrating surface expressions at the 1 m to tens of meters scale. Video S2: False-color digital elevation model generated from high-resolution lidar data illustrating surface expressions at the 1 m to tens of meters scale. Please visit https://doi.org/10.1130/GEOS.S.13151009 to access the supplemental material, and contact editing@geosociety.org with any questions.
Science Editor: Shanaka de Silva
Associate Editor: Gregory D. Hoke
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.