Geologic and geomorphic evidence for multi-phase history of strands of the San Andreas fault through the San Gorgonio Pass structural knot, southern California

The San Andreas fault (SAF) zone is uniquely complex through the San Gorgonio Pass region (SGPr) (Fig. 1) and, despite numerous investigations, remains incompletely understood. This complexity, demonstrated by multiple subparallel fault strands, bends, kinks, or offsets of individual fault strands, along with an orientation that is rotated in a counterclockwise sense relative to the overall strike of the fault system, led to previous descriptions of this region as a “structural knot” (Matti et al., 1992a). Allen (1957) recognized this complexity and questioned the continuity of faulting through the SGPr, citing the apparent absence of large displacements and the lack of youthful tectonic geomorphology associated with the most likely path of the modern SAF. Many of the questions posed by Allen (1957) persist, in spite of numerous investigations designed to better understand this enigmatic region. Among the questions remaining are how the SAF resolves itself through the SGPr, and whether there is a pathway for fault rupture to propagate through the SGPr, or if instead, this region serves as a barrier to rupture along the SAF initiating from the southeast or northwest. Additionally, it is important to know both the three-dimensional geometry and the kinematic evolution of the SAF through the region in order to fully characterize the earthquake hazards of the system.

Principal geologic structures within the SGPr landscape (Fig. 1) include (1) strands of the SAF zone, including the Mission Creek, Mill Creek, San Bernardino, Gandy Ranch, Banning, and Garnet Hill strands, that have been active at different intervals in the recent geologic past (Matti et al., 1992a; Matti and Morton, 1993; Kendrick et al., 2015); (2) the transpressive San Gorgonio Pass fault zone (SGPFZ); (3) the Crafton Hills fault complex; and (4) the Whitewater fault (nomenclatural usage for all of these faults follows Matti et al., 1985, 1992a, 2003; Kendrick et al., 2015). Throughout this study, we will refer to the strands of the SAF simply as strands without designating that they are strands of the SAF at each mention. Matti et al. (1992a) proposed that these SAF strands formed sequentially, with each new pathway produced in response to the increasingly complex deformation and offset of older strands.

Evidence for the most recent slip is located along faults, including, from west to east, the dextral San Bernardino strand, the transpressive SGPFZ, and the transpressive Garnet Hill and dextral Banning strands (Fig. 1); all of these faults are characterized by youthful primary fault features indicating Holocene activity (Matti et al., 1992a; Yule and Sieh, 2003). At the surface, the transpressive SGPFZ seems to link the Banning and the Garnet Hill strands to the southeast and the San Bernardino strand to the northwest.

In spite of clear evidence of recent faulting along these faults, there is a gap in the surface expression of through-going late Quaternary rupture between the Burro Flats section of the San Bernardino strand and the SGPFZ (shown by the dotted yellow line on Fig.1). Some studies have interpreted that the active trace of the SAF terminates just south of Burro Flats (Allen, 1957; Matti et al., 1985, 1992a; Matti and Morton, 1993), whereas other researchers have cited some limited evidence for a continuation of the active SAF (Yule and Sieh, 2003) through this gap. Yule and Sieh (2003) noted subtle pressure ridges and hanging valleys along the western margin of Potrero Creek. They interpreted that these features indicate the San Bernardino strand continues southeastward from Burro Flats. Yule and Sieh (2003) acknowledge, however, that the observed features are subtle and, thus, must indicate a diminished rate of slip, as compared to that measured along the SAF to the northwest.

The SGPr is critical to understanding the seismic potential of the SAF, which depends on the connectivity of the fault strands, and the ability for earthquake rupture to propagate through the structural complexity found there. This region recently has been the focus of considerable debate among geoscientists concerned with fault-rupture scenarios and attendant seismic risk (e.g., Carena et al., 2004; Plesch et al., 2007; Dair and Cooke, 2009; Yule, 2009, 2010; Cooke and Dair, 2011; Heermance and Yule, 2017; Fosdick and Blisniuk, 2018; Douilly et al., 2020). For example, the scenario earthquake presented to the public and frequently cited (Jones et al., 2008; Perry et al., 2008; Treiman et al., 2008) highlights a tremendous seismic hazard to the heavily populated Los Angeles Basin when SAF rupture originating from the southeast, near the Salton Sea, continues through the SGPr unhindered. If rupture were to be arrested or significantly slowed at the SGPr, the resulting seismic energy within the Los Angeles Basin would be much lower, but other regions could be more severely impacted.

This study explores the fault linkage and kinematic evolution of the SGPr as well as how the landscape is affected by slip along the San Bernardino, Banning, and Garnet Hill strands. We first present our mapping of faults in the study area. We define discrete tectonic blocks in the area, building on and modifying the work of Spotila and colleagues (Spotila et al., 1998, 2001; Spotila and Sieh, 2000). Morphometric analysis is presented that confirms that the terrains on either side of the San Bernardino strand in the vicinity of Burro Flats have experienced different uplift histories, suggesting that through-going faulting has separated these terrains during some interval in their history, and we discuss the implications for this fault strand continuity. Finally, we present observations of offset geologic units and landforms along the San Bernardino and Banning strands, estimate displacement and age of some of these features, and discuss the implications of this to fault strand switching in the SGPr.

In this study, we used geologic observations and geomorphic mapping and analysis, combined with soil description and correlation, to better constrain the fault activity along SAF strands in the SGPr. We made geologic and geomorphic observations of the study area at scales of 1:24,000 and larger, mainly using traverse-mapping techniques. From the locality where a map unit or its boundaries were initially identified or defined, we extended those features to other areas through direct outcrop observation and aerial photographic interpretation. For surficial units, we plotted bounding contacts digitally from rectified and georeferenced 2005-vintage National Agriculture Imagery Program (NAIP) aerial stereo pairs that allow location accuracy equivalent to or better than the accuracy standard for the topographic-contour base. Landslide deposits were defined based on morphologic characteristics observed on topographic maps, aerial photos, and imagery. For a calculation of landslide areal percent of each block, each landslide area was attributed to its source block.

Many high-resolution (~1 m) airborne light detection and ranging (lidar) data sets are available within the SGPr and were used to examine geomorphic features in detail. Because the lidar spatial coverage is not complete in the region, we used 10-m digital elevation models (DEMs) to assess the elevation and slope distribution in the morphometric analyses to avoid bias. To compare the landscapes within each structural block, we define a subsection area that delineates the extent of the bedrock highlands. For morphometric analyses within these subsections, we have also removed the areal extent of mapped landslides. We measured local relief and the standard deviation of slope using a 100-m-radius circular area centered on each pixel. We mapped the drainage net within the selected terrain using flow-path tools (ArcMap v. 10) and confirmed it using aerial photography.

Soil pedons were described using the definitions and abbreviations of Birkeland (1999) and the Soil Survey Staff (1951). The degree of soil development was characterized using indices of Torrent et al. (1980), Harden (1982), and Harden and Taylor (1983) and was compared to regional, dated soil chronosequences (McFadden and Weldon, 1987; Kendrick and McFadden, 1996; Kendrick et al., 2002).

Here we report on our mapping of the faults critical to understanding the recent activity of the San Bernardino and Banning strands of the SAF.

The San Bernardino strand extends over 60 km along the base of the San Bernardino Mountains from Cajon Pass southeast to the vicinity of Banning Canyon, in the SGPr (Fig. 1). About halfway between the Santa Ana River and Mill Creek (Figs. 1 and 2), the structural configuration of the San Bernardino strand becomes progressively more complex to the southeast (Matti et al., 1992a; this study), with multiple anastomosing fault strands replacing a single strand. At Pine Bench (Fig. 1), the mapped trace of the strand bends to a more southerly strike, from ~290° to the west of Pine Bench, to ~325° to the east of Pine Bench. From there, the San Bernardino strand traverses Burro Flats (Yule et al., 2001; Yule and Sieh, 2003; Yule and Spotila, 2010) and presumably passes into the Potrero Creek drainage (Fig. 2). Because of the fault's unique northwest strike along this reach, we refer to it as the Burro Flats section of the San Bernardino strand.

The San Bernardino strand continues through the southern part of Burro Flats and demonstrates clear youthful morphology as it crosses the Burnt Canyon landslide breccia, first identified and described by Allen (1957). Just southeast of this deposit (Fig. 3) is a northeast-facing topographic step in youthful alluvial deposits of Potrero Creek. This scarp-like feature (Fig. 3, inset) is most prominent in visualizations of the 2007 Earthscope lidar (Bevis et al., 2017) and appears to be fault-controlled because (1) it is on-trend with the last documented trace of the San Bernardino strand directly to the northwest with a consistent up-to-the-southwest vertical separation; (2) the two alluvial surfaces it separates have similar surface texture, including spacing and depth and orientation of fluvial channels suggesting identical or similar ages; and (3) there is an ~35° obliquity between the fluvial channels on both sides of the topographic step (165°) and the orientation of the step (130°), which makes a purely fluvial origin for the scarp less convincing. Assuming a vertical fault, this scarp indicates an ~1.5 m up-to-the-southwest throw and an unknown horizontal offset. The scarp is most clearly expressed along an ~300 m length but appears to extend up to 1300 m.

Southeast of this scarp, for ~3 km, we see no convincing tectonic geomorphology, either in the 2007 Earthscope lidar imagery or in 1930- and 1952-vintage aerial photographs or 1975 low-sun-angle aerial photographs flown for the U.S. Geological Survey by I.K. Curtis. We note, however, that the expected position of the fault is within youthful fluvial sediments of Potrero Creek, and the primary fault features could well have been removed or covered by recent fluvial processes. In Figure 3, we show the inferred trace of the San Bernardino strand as dotted where concealed, which is also how it was depicted by Yule and Sieh (2003). On the eastern side of the canyon containing Potrero Creek, near the mouth, is a fault segment that is aligned with the Burro Flats section farther to the north (labeled in Fig.3). This could be a continuation of the San Bernardino strand, or might be part of the Gandy Ranch strand, discussed below.

The Gandy Ranch fault was named and described by Allen (1957); it extends from Millard Canyon west to Hathaway Creek (Figs. 1 and 3). This fault was recognized as having active fault morphology and recent lateral displacements (Allen, 1957). Yule and Sieh (2003) included the fault in their kinematic model for the SAF through the SGPr; we agree with this interpretation and hereafter refer to the fault as a strand of the SAF.

Between Millard Canyon and Hathaway Creek, crystalline rocks of San Gabriel Mountains–type (as defined by Matti et al., 1985, 1992a) locally are traversed by zones of crushed rock having one or more discrete fault planes. These are best observed between Millard Canyon and Potrero Creek, where both Allen (1957) and Yule and Sieh (2003) associate the bedrock traces with the Gandy Ranch strand. Locally though, some zones of crushed rock could coincide with the toes of large landslide blocks. The Gandy Ranch strand also can be observed in crystalline rocks at the head of an unnamed tributary that extends northwest from the mouth of Potrero Creek (Fig. 3). Our mapping here confirms observations made originally by Allen (1957): the fault is localized along a narrow zone of crushed crystalline rock. This crush zone projects westward into the Hathaway Creek drainage, where the fault dextrally offsets bedrock ridgelines by as much as ~125 m.

Pivotal to interpreting the spatial extent of the Gandy Ranch strand are tectonogeomorphic features developed in young Quaternary alluvial deposits, particularly between Potrero Creek and Millard Canyon. We begin with the easternmost occurrence of the Gandy Ranch strand in Millard Canyon, where previous workers recognize prominent scarps traversing Holocene alluvium in the canyon narrows (Fig. 3). The scarps were discussed by Yule and Sieh (2003) and were trenched by McBurnett (2011; location NM on Fig.1). Ages for Holocene sediment are provided by McBurnett (2011), Heermance et al. (2014, 2015), Desjarlais (2018), and Heermance and Yule (2017).

West of Millard Canyon, tectonostratigraphic relations among Quaternary surficial materials are not as obvious. Along this reach, localized Holocene debris-fan deposits heading into crystalline rocks of San Gabriel Mountains–type have prograded southward onto the broad Potrero Creek fan head. Our examination of B4 lidar (bare-earth rendering of Bevis et al., 2017) at this location identified discontinuous geomorphic features that we interpret as fault scarps (Fig. 3). These features project northwestward toward Potrero Creek, where they are associated with a conspicuous area of warped ground, characterized by gently folded and uplifted surfaces, in the vicinity of the Potrero Creek narrows (Fig. 3). Allen (1957) observed springs associated with this area, and in vintage aerial photographs as recent as 1975, the zone is populated with abundant cottonwood trees and other vegetation.

We have not examined this area on the ground due to access limitations. However, various imagery sets (1930, 1938, 1952, 1975, 2005 NAIP, and B4 lidar) show that the warped ground is traversed by several west-northwest to west-trending discontinuous scarps, some obvious but others subtle. We interpret these scarps as faults and conclude that the zone of warped ground has resulted from contraction in an area where the Gandy Ranch and San Bernardino strands converge. The disruption is a broad monoclinal step in the topography, ~14 m down to the south, over a distance of 500 m.

We have no direct evidence for the age of surficial deposits in the warped-ground footprint, nor do we have ages for tectonogeomorphic features (fault scarps and folding) that deform them. The surficial materials can be subdivided into two kinds: (1) those old enough to have been deformed within the warped-ground complex and (2) those incised into the complex and postdate deformation that produced it. The older deposits have upper surfaces the geomorphic traits of which are similar to those on deposits upstream and downstream from the warped-ground area, and the deposits themselves appear to be stratigraphically continuous with those upstream and downstream. We interpret the latter as mainly Holocene in age and conclude that this deformation is Holocene (and probably older).

We interpret that the warped-ground footprint bears on the distribution of the Gandy Ranch and San Bernardino strands of the SAF, which in turn, inform the latest Quaternary record of dextral slip on the SAF zone in the SGPr. The San Bernardino strand presumably approaches the warped-ground area from the northwest and (as discussed above) displays no primary fault features in youthful alluvial deposits of the Potrero Creek lowland (Figs. 2 and 3). Scarps in the warped ground trend west and northwest, an orientation similar to—but not identical with—the west-northwest trend of the Gandy Ranch strand in the Hathaway Creek drainage. Given the apparent absence of primary fault features associated with the San Bernardino strand directly to the northwest, we conclude that scarps in the Potrero Creek narrows were formed by fault ruptures on the Gandy Ranch strand. This interpretation is supported by scarps between Potrero Creek and Millard Canyon. It is consistent with Allen's (1957) conclusion that scarps in the Millard Canyon narrows are a southeastern continuation of his Gandy Ranch strand. Yule and Sieh (2003) propose that the San Bernardino strand (Burro Flats section) merges here with the Gandy Ranch strand and therefore contributes to ground-surface deformation (“subtle pressure ridges”). We cannot rule out this hypothesis, given that geometric and kinematic relations here no doubt are complex.

Smith (1979) first applied the name “San Gorgonio Pass Fault” to a scarp in the eastern SGPr. Matti et al. (1985) independently applied the name “San Gorgonio Pass Fault zone” (SGPFZ) more regionally to a collection of Quaternary reverse, thrust, and dextral tear faults that extends from the Whitewater area westward to the Calimesa area (Fig. 1) with diminishing displacement and geomorphic expression. Smith's (1979) “San Gorgonio Pass Fault” is part of that regional fault complex. The fault system has a distinctive zig-zag expression caused by repetition of a characteristic serially partitioned fault geometry (e.g., Norris and Cooper, 2007)—an L-shaped fault segment with the longer section oriented northwestward and exhibiting dextral lateral slip and the shorter section oriented northeastward and contractional in nature (Matti et al., 1985, 1992b).

All faults of the SGPFZ have been active in late Quaternary time. Some faults in the complex may have been active only in the Pleistocene; others have been active throughout the late Pleistocene and Holocene and have generated ground ruptures as recently as a few hundred years ago (Smith, 1979; Matti et al., 1985, 1992a; Treiman, 1995; Ramzan, 2012; Wolff et al., 2013, Wolff, 2018). We separate the SGPFZ into an eastern section and a western section, with a crustal fault boundary (not a tear fault) near Banning Canyon (Fig. 1). We base this boundary on differences in characteristics between the two sections. The eastern section is structurally more complex than the western section, having multiple thrust-fault strands that together represent a long period of Quaternary contractional deformation; it is not clear whether older strands recognizable in the eastern section are present in the western section. There are longer and more clearly pronounced faults at the surface in the eastern section. Finally, there is abundant evidence for Holocene activity in the eastern section, but in the western section, this evidence is equivocal west of the Banning Bench.

The Banning strand of the SAF extends from the Indio Hills across the Coachella Valley to the eastern edge of the SGPr. To the east of the SGPr, in the vicinity of Devers Hill (Fig. 1), the trace of the Banning strand can be distinguished by a subparallel series of north- and south-facing scarps. At Devers Hill and to the southeast, the scarps face to the south; just northwest of Devers Hill, they face north. In the transition zone between these two sections, a right step occurs between two en echelon fault strands. Just southeast of State Highway 62 (Fig. 1), scarps associated with the Banning strand again face south; continuing northwest across the highway, the strand is marked by local scarps and is dotted-where-concealed beneath late Quaternary deposits of various ages.

Where it crosses Whitewater River, the Banning strand is marked by a prominent vegetation lineation. To the west of Whitewater River, the Banning strand forms the northern edge of Whitewater Hill¹, and the fault dips ~45°N. Just to the west of Whitewater Hill, the strand is characterized by dextrally offset ridges and drainages. The Banning strand is not observable farther to the west, where it intersects the SGPFZ.

The Garnet Hill strand was named “Garnet Hill Fault” by Proctor (1968) who deduced its existence in the northern Coachella Valley mainly on the basis of gravity data. Langenheim et al. (2005) refined this geophysical diagnosis by associating the Garnet Hill strand in the northern Coachella Valley with an alignment of magnetic lows and thicker basin fill. The fault is also defined by steps in groundwater levels (Tyley, 1974). Proctor's geologic map denotes the fault using a dotted-where-concealed trace. We concur: throughout most of its extent in the northern Coachella Valley, we do not observe primary fault features (scarps and/or lineaments) that would constrain the location of the fault.

In the eastern SGPr, Allen (1957) does not use the name “Garnet Hill Fault.” Along the southern base of Whitewater Hill, however, he mapped two faults that Matti et al. (1985, 1992a; Matti and Morton, 1993) associated with the Garnet Hill strand—an interpretation refined by Yule and Sieh (2003) and Huerta (2017). Although mapping interpretations in this area differ in detail, all agree that the Garnet Hill strand lies along the south flank of Whitewater Hill and projects northwestward toward the Cottonwood Creek area. There, each investigator offers a different interpretation of structural relations. In our usage, we associate with the SAF those geologic structures having high-angle dips and predominantly dextral slip; in contrast, we associate with the SGPFZ those geologic structures having low-angle dips and predominantly thrust and/or reverse slip. Using these criteria, the Banning and Garnet Hill strands are high-angle dextral faults that approach the SGPr from the southeast and gradually transition into shallowly north-dipping contractional structures that we associate with the SGPFZ—mainly west of Cottonwood Canyon. Our Figures 1 and 4 illustrate how we interpret geologic relations between the Garnet Hill strand and the SGPFZ.

The Pinto Mountain fault was named by Hill (1928), and early researchers noted that the intersection of the Pinto Mountain fault with the SAF resembled the intersection between the sinistral Garlock fault with the SAF to the northwest (Hill and Dibblee, 1953; Allen, 1957). Dibblee (1968a, 1968b) first documented that the Pinto Mountain fault is a left-slip structure, proposing that the fault in the Morongo Valley area has as much as 9.7 km of sinistral slip displacing Quaternary sedimentary materials containing quartzite and basalt clasts from recognizable bedrock sources in the San Bernardino Mountains. Subsequently, offset crystalline bedrock units of Mesozoic age and older have been cited as evidence for total left slip on the Pinto Mountain fault of 16 to19 km (Dibblee, 1975, 1992; Bacheller, 1978; Powell, 1993; Hopson, 1996). The range of geologic offsets is confirmed by geophysical observations of magnetic anomalies displaced along the fault as well as the length of narrow pull-apart basins between left steps in the fault trace (Langenheim and Powell, 2009).

In the western portion of the Pinto Mountain fault system, a major strand branches southward from the main fault in a northeast orientation (Dibblee, 1968b). Proctor (1968) named this southern branch the “Morongo Valley Fault” and noted sinistral displacements along the strand. In addition to sinistral slip, the fault has generated significant up-on-the-south displacement that has formed the north-facing bedrock escarpment bounding the south margin of Morongo Valley (Fig. 5). The Morongo Valley, between these two major strands of the fault system, has been interpreted as a pull-apart basin formed by sinistral slip along both of these faults combined with dextral slip on the SAF (Dibblee, 1982; Hopson 2012).

Kendrick et al. (2015) mapped the intersection of the Pinto Mountain fault and the Mill Creek strand and noted the latter was offset between 1–1.25 km by the former. This same amount of slip restores displaced streams in the Mission Creek drainage (Kendrick et al., 2015, their figure 12). We build on these observations by noting that the eroded mountain front of the San Bernardino Mountains, on the northwest side of Morongo Valley, is similarly offset by approximately this amount (Fig. 5). We interpret the morphology of the mountain front to indicate that this most recent slip along the Pinto Mountain fault has either reactivated a long-dormant strand within the system or has pioneered a new strand. Previous researchers have suggested that the Pinto Mountain fault was responsible for the ~15 km left bend in the Mission Creek strand (Matti et al., 1985, 1992a). Restoring the 7.1–8.7 km of slip along the Mill Creek strand (Kendrick et al., 2015), which is co-located with the trace of the Mission Creek strand to the south of the juncture with the Pinto Mountain fault, positions the Morongo fault, the southern splay of the fault zone, adjacent to the left bend in the Mission Creek strand. We interpret that the Morongo fault carried much of the total fault displacement, measured to the east of the juncture of the Morongo and Pinto Mountain faults. This interpretation is revisited below when we propose a model of fault activity for the region.

Allen (1957) applied the name “Whitewater Fault” to a north-trending structure that occurs along the east wall of Whitewater Canyon. Allen does not discuss the fault, but his geologic map shows it extending for ~3.3 km (~2 mi) from the Banning strand of the SAF north to where the fault enters the modern wash of Whitewater River. From that point, Allen (1957) does not speculate about the fault's northward continuation. At its south end, Allen indicates that the Whitewater fault is terminated by the Banning strand.

Peterson (1975) reexamined structural and stratigraphic relations associated with Allen's (1957) Whitewater fault and reported that the fault dips 60° to 80° to the east, displays both normal and reverse separations, and juxtaposes the Miocene Coachella Fanglomerate against rocks of the basement complex and Quaternary alluvium. He noted that the fault zone is usually marked by up to 2 m of basement gouge.

Our interpretation of the Whitewater fault differs somewhat from that of either previous investigator, in that we use different criteria for distinguishing this fault from older faults. As a result, our mapping of the Whitewater fault is positioned slightly to the west of the fault mapped by Allen (1957) in the southern section of the fault. This southern portion of the fault (Fig. 4) is marked by scarps in Pleistocene alluvial deposits and by a zone of sheared and crushed crystalline rocks. Shears within this crush zone dip between 65° and 35° to the east, with motion up on the east (site 1 on Fig.4). Farther to the north, the fault dips between 38° and 84° to the east (site 2 on Fig.4).

In this study, we define three structural blocks (Fig. 1), approximately within the bounds of the Morongo block defined by Spotila et al. (1998, 2001).

We apply the name “Kitching Peak block” to the eastern part of the “Morongo block” of Spotila et al. (1998, 2001). The name derives from a summit located centrally within the block's footprint (Fig. 1). The majority of the block is underlain by crystalline rocks of San Gabriel/Chocolate Mountains–type (bedrock nomenclature after Matti et al., 1983, 1985, 1992a; Matti and Morton, 1993). The rocks are characterized by pervasively layered fabrics and structures that range from textural foliation through gneissose compositional layering. Some rocks clearly are plutonic in origin and typically are granodioritic to tonalitic in composition; monzogranite is rare. Other rocks are metamorphic in origin and range from layered gneiss to augen gneiss; these rocks also are granodioritic to tonalitic, although locally mafic components are dioritic to amphibolitic. Pegmatite dikes and layers are common in some areas. The rocks locally are highly fractured and sheared, and they weather readily into colluvial slopes and cones.

Boundaries for the Kitching Peak block differ structurally and geomorphically throughout their extent (Fig. 1). To the west, the block is bounded by the Burro Flats section of the San Bernardino strand, and the contrast between the low-standing landscape of Burro Flats and the high-standing Kitching Peak terrain is dramatic. To the north, the Kitching Peak block is separated from the Yucaipa Ridge block by the Mission Creek strand of the SAF. To the south, the Kitching Peak block is bounded by thrust and reverse faults of the SGPFZ that carry crystalline rocks of the block over late Cenozoic sedimentary rocks in foothills to the south. The Banning strand marks the block's southeastern boundary. The eastern boundary of the Kitching Peak block is defined as the Whitewater fault. The northern extent of the Whitewater fault is unknown, but here we extend it to intersect the Mission Creek strand, and we map it as concealed beneath the alluvium along Whitewater River and to the east of the hill associated with Wathier Landing (Fig. 1).

We apply the name “Pisgah Peak block” to the western part of the “Morongo block” of Spotila et al. (1998, 2001). The name derives from a summit located at the north part of the block's footprint (Fig. 1). The Pisgah Peak block is underlain by foliated crystalline rocks of San Gabriel Mountains–type that are lithologically and structurally similar to those in the Kitching Peak block (described above).

The Pisgah Peak block is bounded to the east by the Burro Flats section of the San Bernardino strand and to the south by the SGPFZ (Fig. 1). Together, the Mission Creek and San Bernardino strands form a compound structural boundary that separates the Pisgah Peak block on the north side from the Yucaipa Ridge and Cram Peak blocks (Fig. 1). A western boundary for the Pisgah Peak block is open to interpretation. We note that the boundary zone coincides with the westward transition from contractional structures of the SGPFZ (Matti et al., 2015) to extensional structures of the Crafton Hills fault complex (Figs. 1 and 2). Logically, this tectonic transition marks the western boundary of the Pisgah Peak block. Unfortunately, faults that might define a western boundary for the block have not been recognized at the surface, although groundwater investigations and gravity surveys (summarized by Mendez et al., 2016) recognize groundwater barriers in the eastern part of the Yucaipa lowland that may coincide with faults that do not daylight. Acknowledging these uncertainties, we propose that the western boundary of the Pisgah Peak block coincides with the eastern margin of the Yucaipa alluvial lowland (Fig. 1).

We apply the name “Garnet Hill block” to the landscape between the Banning and Garnet Hill strands of the SAF; the name derives from the isolated outlier of late Cenozoic sedimentary rocks that occur at Garnet Hill (Fig. 1). This landscape is distinguished by two features: (1) high-standing, dome-shaped structural uplifts that expose late Cenozoic sedimentary deposits (some as young as middle Pleistocene) and (2) low-lying areas of late Quaternary surficial materials that either buttress against the domal uplifts or issue from canyons and arroyos that are aggressively eroding into the uplifts. In either case, the tectonic geomorphology of the Garnet Hill block is the most recent and obvious example in the SGPr of the interplay between tectonism and attendant landscape response.

Our primary interest is to compare the geomorphology of the Pisgah Peak and the Kitching Peak blocks in order to evaluate the continuity of the San Bernardino strand to the south of Burro Flats. Here we present geomorphic variables of the bedrock and the mapped distribution of Quaternary deposits for each of the two blocks (Figs. 6–8).

We compare the elevation distribution of the Kitching Peak and Pisgah Peak blocks (Fig. 6A). The overall relief of the Pisgah Peak block is lower, with elevations ranging from 673–1675 m, as compared to Kitching Peak block elevations, which range from 500–2415 m. The Kitching Peak block has high elevations along its northern boundary, especially adjacent to lower Raywood Flats on the Yucaipa Ridge block, as well as at Kitching Peak, decreasing to the east, south, and west from this high. High elevations within the Pisgah Peak block, while lower than those in the Kitching Peak block, are also concentrated at the northern edge, along the boundary with the Yucaipa Ridge block, and decrease to the south. The hypsometry of the two areas is nearly identical, as indicated by the hypsometric integral (Table 1) and the normalized plot of hypsometry (inset, Fig. 6A); we suggest this is likely a function of sampling similar bedrock types across large regions.

A comparison of the areal extent of landslides shows that the Kitching Peak block hosts numerous landslide deposits, whereas landslide deposits are largely lacking in the Pisgah Peak block (Fig. 8; Table 1). Approximately 15.9% of the area represented by the bedrock upland region of the Kitching Peak block is covered by landslide deposits, as compared to ~3% of the Pisgah Peak block area. Within the Pisgah Peak block are landslide deposits (Pine Bench and Burnt Canyon) that originated on the Kitching Peak block and are therefore not included in the areal total for the Pisgah Peak block.

Local relief (Fig. 6B), measured for a radius of 100 m for each pixel, shows many of the same features as does the elevation plot. This measure also highlights, within the Kitching Peak block, a zone of high local relief that extends approximately N-S, located just to the east of Kitching Peak, as well as high local relief extending east of that feature toward Whitewater River. This is a region that is also characterized by many drainage net irregularities and recent stream captures, discussed in the following section. To the north of Whitewater Hill is a region that is both low in elevation and low in local relief and slope. The southwest portion of the Kitching Peak block is also characterized by low elevation and low relief and slope measurements (Figs. 6A and 7A).

The distribution of the standard deviation of slope in each of these blocks also shows that the Pisgah Peak block is morphologically distinct from the Kitching Peak block (Fig. 7B). This metric is a useful measure of surface roughness (Grohmann et al., 2011) and highlights the variability of the shorter slopes within the drainage net of the Pisgah Peak block, in contrast with that of the Kitching Peak block. The characteristics of the elevation distribution as well as of the standard deviation of slope of the southwesternmost portion of the Kitching Peak block seems to be more similar to the Pisgah Peak block, suggesting that the boundary between these blocks may have followed a different fault system prior to the most recent fault configuration, discussed further below. Drainage density, the length of drainages as a function of area, is quite similar between the bedrock regions of the two blocks beyond an accumulation threshold of 100 m² (Fig. 8B; Table 1).

The Pisgah Peak block is characterized by thick Pleistocene through earliest Holocene accumulations of sediment in the canyons, including, from west to east, along Little San Gorgonio River, Noble Creek, and San Gorgonio River (for locations, see Fig.1). Incision into the Banning Bench by the San Gorgonio River provides a three-dimensional view of the volume of sediment accumulation, where exposed thickness of the deposit associated with the Banning Bench ranges from ~15 to >30 m. These canyons and the associated deposits divide the bedrock upland region of the block. In contrast, the Kitching Peak block is largely lacking such sedimentary accumulations. The sedimentary deposits that are present are located along the edges of the block, particularly in the Burro Flats region, and, to a lesser extent, along the eastern side of the crystalline bedrock.

To understand the history of slip along the San Bernardino and Banning strands of the SAF, as well as of the SGPFZ, we describe displacements along these faults that are indicated by our mapping.

At the northwest end of our study area, Matti and colleagues (Matti et al., 1985, 1992a, 1992b, 2003; Matti and Morton, 1993) describe deposits adjacent to the San Bernardino strand, between the Santa Ana River and Mill Creek; these deposits provide an estimate of displacement along the San Bernardino strand (Figs. 2 and 9). These deposits contain clasts that, taken together, are unique and characteristic of those in the Mill Creek drainage system, including Triassic porphyritic monzogranite, the orthogneiss of Alger Creek, and those derived from the Mill Creek Formation. Matti et al. (2003) mapped beds within the unit as dipping up to 65°SW and interpret the unit to represent alluvial fan deposits that were originally adjacent to the canyon mouth of Mill Creek. This previous study suggested that 2–3 km of dextral slip has occurred on the San Bernardino strand since the deposition of this unit. This reconstruction aligned the center of the deposit, and associated landform, with the mouth of the present-day Mill Creek Canyon.

Here we refine and modify this palinspastic and paleogeographic scenario. First, we propose that the stream system that delivered sediment to the displaced alluvial deposits was a broad catchment area on the north side of the modern Mill Creek, centered approximately on Falls Creek (Figs. 1 and 2). This drainage, which we informally name the Falls Creek drainage system, contains the unique rock types that are present in the deposits along the fault. Allen (1957) observed that “rapid erosion of [Mill Creek] canyon has left the valleys of Falls, Vivian, and High creeks hanging 1000 feet above the floor of the main canyon.” By implication, the modern floor of Mill Creek once was at an elevation where Falls Creek, High Creek, and Vivian Creek and other tributaries to the west could combine to form a trunk drainage that then coursed downstream and through the narrows at the range front—the current mouth of Mill Creek Canyon. The elevation of Falls, Vivian, and High creeks illuminates the most recent capture by the headward erosion of the modern Mill Creek; the western portion of this original drainage basin has had more time to adjust to the deep incision and lengthening of Mill Creek. The deposition of the unit adjacent to the San Bernardino strand, hereafter called Falls Creek deposits, precedes motion along the Mill Creek strand, based on the degree of soil development and comparison with soils associated with deposits that constrain the timing of slip on that strand (Kendrick et al., 2015); for further discussion, see the “Proposed Long-Term History of San Gorgonio Pass Region” section. Prior to motion along the Mill Creek strand, with the associated elongation of the drainage along that fault, this Falls Creek drainage system, with the associated catchment region, was directly across the Mission Creek strand fault escarpment from the deposits. The modern Mill Creek replaced the ancestral drainage system when tectonic uplift caused excavation of the modern Mill Creek Canyon and headward erosion of Mill Creek along the crush zone of the Mill Creek strand to the dramatic headwaters of the Mill Creek Jumpoff (Kendrick et al., 2015; see their fig. 5A). These events left the Vivian, Falls, and High Creek tributaries hanging above the modern floor of Mill Creek Canyon.

In a second modification to the previous fault displacement model (Matti et al., 2003), we propose that dextral displacement of this unit could be as much as 6.1 km, using as a piercing point gravelly and sandy sediment exposed on the incised east wall of the Santa Ana River (Fig. 9A) and considering the entire width of the canyon mouth as a potential source area. The sedimentary sequence in the offset deposit is >40–50 m thick and contains most of the clast types transported down the ancestral Falls Creek Canyon; hence, the sequence could well have accumulated directly down-gradient from the canyon mouth—again, depending on the radial orientation of streamflows associated with the alluvial-fan landform. Based on this model, we interpret that the deposit is time-transgressive, with the western part deposited prior to the eastern portion. Importantly, although the 6.1 km offset is the maximum indicated by the present configuration, the NW extent of the deposit has been eroded by the modern Santa Ana River. The degree of soil development associated with the western part of this deposit (Fig. 2) indicates an estimated age greater than 500 ka (Figs. 9B–9D; VC01 in Table 2), based on comparison with regional, dated surfaces (McFadden and Weldon, 1987; Kendrick et al., 2002). This soil contains abundant secondary pedogenic silica, present in seams and sheets throughout the pedon, as well as in grain-to-grain cementation. The degree of development of this soil pedon indicates that it was stable either prior to the initiation of slip along the Mill Creek strand (Kendrick et al., 2015) or early in the history of slip along that strand. Restoring the displacement on the Mill Creek strand positions the Falls Creek headwaters to be nearly adjacent to the mouth of the drainage system, as discussed further below.

At Pine Bench, ~1–2 km northwest of Banning Canyon and the San Gorgonio River (Figs. 1, 2, and 10A), Matti et al. (1983, 1992a) mapped an extensive landslide deposit that caps ~2 km² of the Pine Bench landscape. This depositional unit constitutes a large body of crushed and shattered crystalline rock (Figs. 10B–10D). These rocks are highly deformed rocks of the San Gabriel/Chocolate Mountains–type and were not derived from the Mojave Desert–type rocks of Yucaipa Ridge directly to the north. Although it is not possible to unequivocally argue for a cross-fault match for this deposit, we propose that a truncated landslide on the north side of the San Bernardino strand, just northwest of Burro Flats, has a similar dimension as that of the Pine Bench landslide; we interpret that this is the cross-fault source (Fig. 10A). Restoring ~4–4.9 km dextral slip and 90–230 m vertical, up on the southwest, on the San Bernardino strand aligns these landslide deposits. We do not have age control of this feature.

Whitewater Hill, to the west of Whitewater River, is a distinctive dome-shaped geomorphic feature bound on the north by the Banning strand and on the south by the Garnet Hill strand. The preserved surface of Whitewater Hill is associated with very strong soil development and the key to our geologic and stratigraphic interpretations. Observationally, the soil is strongly rubified (2.5YR hues) and has an associated duripan that extends to a depth of several meters (Figs. 11E and 11F). A shallow soil pit (depth limited to 117 cm) shows that the profile is stripped, a characteristic shared by many old soils in the region, and that numerous clay films are present, up to the designation of thick clay films (soil description nomenclature after Birkeland, 1999; Soil Survey Staff, 1951) on ped faces. We noted silica cementation present at a depth of 71 cm in the soil pit, but the limited depth of the pit prevented us from describing the full soil profile. Pedogenic processes associated with duripan formation created a very hard and resistant soil profile that accounts for extremely steep slopes.

Whitewater Hill is undergoing aggressive erosion by south-flowing streams the canyons of which expose a succession of gravelly and sandy alluvial sediment that is at least 230 m thick. Allen (1957) originally grouped all these sediments into his Cabezon Fanglomerate—a Pleistocene unit that was defined by Vaughan (1922) and crops out in the foothills of the central and eastern SGPr. We prefer a distinct nomenclature, to recognize the unique clast assemblage within these deposits that is not present in the Cabezon Formation, as described to the west. Clasts observed in the basal unit of this succession include (1) sedimentary and volcanic clasts derived from the Miocene Coachella Fanglomerate; (2) saussuritized porphyritic quartz monzonite; (3) foliated and texturally massive rocks of the Mojave Desert–type; (4) mafic tonalite rock, foliated granodiorite, and piedmontite-bearing felsic rock of the San Gabriel Mountains–type; (5) quartz biotite gneiss of Mojave Desert–type; and (6) pegmatite. These clasts suggest a source similar to that sampled by the modern Whitewater River (Matti et al., 1985, 1992a).

We classify the Whitewater Hill succession as depicted in Figure 11A. Like Huerta (2017), we recognize two units: a lower unit that we assign to Qvof1 and an upper unit defined as Qvof2 (Figs. 11E and 11F). The degree of soil development associated with the upper unit, described above, is comparable to a similar duripan developed on alluvial deposits in the Mission Creek alluvial complex that Kendrick et al. (2015, their pedon MC-2) assign to their unit Qvof2.

The sedimentary succession underlying Whitewater Hill is important because it provides evidence for the amount and timing of late Quaternary dextral slip on the Banning strand of the SAF in the SGPr. In addition to the clast compositions, reconnaissance paleocurrent measurements (streamflow toward the south, southeast, and rarely southwest) are consistent with a northern source for these deposits. Clast counts and paleocurrent measurements by Huerta (2017) generally reinforce this interpretation, although Huerta reports some southwest-oriented paleocurrent directions. To accommodate these data, Matti et al. (1985, 1992a) and Matti and Morton (1993) proposed that the Banning strand of the SAF has displaced the Whitewater Hill sequence 2–3 km from its original depositional position downstream from the ancestral Whitewater River to its current position west of the modern Whitewater River. Huerta (2017) proposed a similar offset of 1.5–2.7 km to realign the central portion of this landform with Whitewater River. In the following Discussion section, we will present various models to constrain minimum and maximum offset of the geomorphic feature and associated deposits, as well as justification for our preferred option.

Two prominent soil profiles are present on the eastern embankment of Whitewater River (Hofmann et al., 2017, 2019; Fig. 11B). The surface soil is rubified, with 7.5YR colors in the maximum B horizons and 5YR clay film colors, and has very few to few clay films on ped faces and in pores. In the maximum B horizons, the clasts are nearly completely grussified, and there is no duripan associated with this soil (pedon WW01; Table 2). The buried soil is characterized by 7.5YR colors in the maximum B horizon, thin, scarce clay films, and less weathered clasts than within the surface soil (WW02; Table 2). This buried soil profile also lacks the duripan that is present in the soil pedon associated with Whitewater Hill. These results suggest that a direct correlation cannot be made between either of these two soils and the soil developed on Whitewater Hill.

Elevation, total relief, local relief, and slope metrics all differ significantly between the Kitching Peak and Pisgah Peak blocks. In addition, the distribution of landslide and fluvial deposits indicates that these two blocks have had differing uplift histories and continue to experience different dominant surficial processes. Landslide deposits are abundant on the Kitching Peak block, which is also characterized by longer, steeper slopes, indicated by the elevation, local relief, and slope measurements (Figs. 6 and 7). Thick sedimentary packages of Quaternary alluvium are located along drainages that divide the bedrock uplands of the Pisgah Peak block but are absent on the Kitching Peak block. This pattern suggests that, because of either the amount or recency of uplift, or both, sediments generated from the bedrock uplands of the Kitching Peak block are transported beyond the boundaries of the terrain, rather than being stored along canyons within the block. Alternatively, the equivalently aged deposits within the Kitching Peak block boundaries may have been present in the past but were stripped by erosion during recent uplift. In contrast to the Pisgah Peak block, the Kitching Peak block is characterized by numerous stream captures, discussed further in subsequent sections.

Morphometric variables of these two blocks provide evidence that the SAF bisects the landscape to the southeast of Pine Bench; the Kitching Peak and Pisgah Peak blocks have experienced different uplift histories, likely over an extended period of time. The Pisgah Peak block is a landscape that has experienced either lower uplift rate or prolonged stability and increased maturity. We can use the morphometric contrasts between the Kitching Peak and Pisgah Peak blocks to infer that at some point in the past, the SAF system was well integrated, and that slip transferred between the San Bernardino and the Banning strands.

In spite of the evidence for different uplift histories for the two blocks, evidence for the bounding fault between the Pisgah Peak and Kitching Peak blocks is discontinuous at the surface. The San Bernardino strand cannot be unequivocally mapped to connect with the SGPFZ, as discussed above. One possibility for this lack of continuity may be that slip is stepping across to the Gandy Ranch strand (Yule and Sieh, 2003), to the south of the Burro Flats section. This fault has evidence of recent faulting, including scarps and small offsets. According to our mapping, this fault is the northernmost fault in the fault study of the Millard Canyon area (McBurnett, 2011; Heermance and Yule, 2017). In spite of this youthful activity, the Gandy Ranch strand lacks evidence for long-term fault offset and cannot be traced farther to the west than Banning Canyon. Another possibility is that the evidence for youthful faulting along the San Bernardino strand has been removed by fluvial activity along Potrero Creek. A short, inferred fault fragment, labeled SBS/GRS on Figure 3, has a similar strike and is approximately aligned to the San Bernardino strand where we map it to the northwest. This fault section might represent an active part of the San Bernardino strand. It might alternatively be a part of the Gandy Ranch strand, or be reactivated by recent activity on the Gandy Ranch strand.

The lack of strong geomorphic evidence for sustained, youthful faulting along the San Bernardino strand south of the Burro Flats region (Allen, 1957; Matti et al., 1985, 1992a; Matti and Morton, 1993) might be explained by one of two scenarios. One possibility is that the SAF is actively re-establishing a through-going trace following disruption of a previously active strand and the accompanying obliteration of surface expression by young fluvial processes and deposits, as mentioned above. In this scenario, the young fault trace has been lengthening and extending toward the SE from the approximate location of Pine Bench. This hypothesis follows from the model presented by Matti et al. (1992a) of the establishment of the San Bernardino–Banning–Garnet Hill route for plate-boundary slip following the disruption of, and termination of, significant slip along the Mill Creek strand. This scenario can explain why the fault segment to the south of Burro Flats lacks the tectonic geomorphology associated with a major fault system. This lack of evidence for faulting would be a result of the recency of formation; this segment would have experienced only a limited number of ground-rupturing earthquakes, and a strong, well-defined geomorphologic expression of the fault is therefore absent.

The distinct difference in morphology between the Pisgah Peak and Kitching Peak blocks suggests an alternative scenario. The differences in the landscapes on either side of the projection of the San Bernardino strand suggest that these blocks have experienced differing uplift histories over a long interval of time. This would be explained if the San Bernardino strand was well-integrated with the then-active SAF in the northern Coachella Valley at some time in the past, permitting through-going rupture along the SAF through the SGPr, and varying uplift between the two blocks as a result. In this case, more recent disruption has complicated and obscured the connectivity between the structures. This disruption might be compressional movement along the SGPFZ, as proposed by Matti et al. (1985, 1992a), or some other structural disruption to through-going slip. The eastern section of the SGPFZ, the section that demonstrates current fault slip, spans the region of the projected juncture of the San Bernardino strand with this fault, as discussed above. This is similar to the results presented by Fattaruso et al. (2016) using Boundary Element Method models to explore the fault network changes. They find a similar uplift of the Kitching Peak block and down-dropping of the Pisgah Peak block when the San Bernardino strand is active and connected with the SAF strands to the SE.

A third option would be a combination of these two models. There may have been a long-active fault connection along the path of the San Bernardino–Banning–Garnet Hill strands in the SGPr, but the Burro Flats section, with its discontinuity, might represent a newly active portion of the fault system.

An intriguing aspect of the morphologic analysis is the apparent lineament that separates the southwest portion of the Kitching Peak block from the remainder of the block (Figs. 6, 7, and 12). The measure of standard deviation of slope, calculated over a 100 m radius for each 10 m pixel, yields a fairly sharp delineation (the northernmost dashed line on Fig.12). The elevation range by pixel (Fig. 6A) also indicates a broad step in topography at this location. A second lineament is present roughly parallel to the first and is marked by short, disconnected segments of faults mapped in bedrock (Allen, 1957; Yule and Sieh, 2003; this study) and multiple large landslide deposits. The southern lineament bounds the southern edge of an unusual geomorphic landform. This landform, a circular closed depression that lies to the northeast of Lion Canyon, might represent a southward-oriented headwater region blocked by lateral translation of an elevated terrain (a shutter ridge) or have a purely landslide origin. For purposes of discussion, we refer to this unnamed landform as Hidden Flat, after a nearby benchmark. The northern of the two lineaments seems to fall along the northern edge of this closed depression, is covered by a large landslide (Fig. 12A), and has a fault mapped by Allen (1957) and Yule and Sieh (2003) along some of the reach. The projection of this lineament to the northwest might have continued along the northeastern edge of Burro Flats, and its presence could help to explain the shape of this intermontane valley. This region, in the southwest portion of the Kitching Peak block, shares morphometric similarities with the Pisgah Peak block. This suggests that a bounding structure in the past may have followed one or both of these lineaments. We informally refer to this zone as the Lion Canyon lineament zone, named for a canyon central to its trace. This zone is within a region where we lack the land access to make bedrock observations. Considering this limitation, we envision three alternative models that could explain this lineament (Fig. 12B).

Model 1 asserts that this is a previously active fault zone but acknowledges that the evidence for this fault zone is not robust, particularly toward the east. The fault traces within this lineament zone, as mapped by this study and previous researchers (Allen, 1957; Yule and Sieh, 2003), are discontinuous, but have been verified by field checking (D. Yule, 2019, written commun.). The lineament zone is partially obscured by landslides (Fig. 12A), and our reconstruction, described below, would have had fault slip occurring on this section prior to initiation of slip along the Mill Creek strand, with no recent activity; both of these factors account for the lack of clear fault morphology. This zone is within a region where we lack the access necessary to check our interpretation.

An alternative interpretation of this lineament zone is that it is a manifestation of an active footwall structure (Model 2). In their N-S cross section, Langenheim et al. (2005; their Figure 7) interpret that the presence of a wedge of low-density materials beneath the SGPFZ best matches their magnetic and gravity data. This low-density material is bound to the north by the subvertical Banning strand, which is present in the footwall of the SGPFZ, and does not extend to the surface. The broad Lion Canyon lineament zone then might reflect a brittle fabric or damage zone that has developed above this footwall structure, perhaps enhanced by the contrast in material properties of the emplaced sediments adjacent to the crystalline bedrock at depth. A shallowing of SGPFZ fault could be associated with overthrusting of the basin sediments and be a contributing factor. This lineation zone is aligned with a projected connection between the San Bernardino strand, just north of Burro Flats, and the Banning strand where it is last clearly defined (shown as small black triangles on Fig.1). This connection at depth would represent an active high-angle fault and offers an alternative explanation for the lack of surficial connection through the SGPr.

A third model to explain the presence of this lineament would be that it represents an abandoned thrust strand (Model 3). The Lion Canyon lineament would be the former position of a range-front thrust that is now uplifted and back-rotated by the development of more outward and active thrust faults. An argument in favor of either Model 2 or Model 3 is that the lineament zone lacks preserved fault offsets in the geomorphology.

For purposes of comparison, we reevaluated the geomorphic variables, using this Lion Canyon lineament zone as a boundary to the south of Burro Flats (Table 1; dashed line shown on Figs. 6–8). This southwest portion of the Kitching Peak block (SWKP) is characterized by numerous landslides, similar to that observed in the Kitching Peak block as a whole. Including the area within Pisgah Peak slightly reduces the distinction between the two blocks for this variable. Including the SWKP in the Pisgah Peak block slightly increases the difference between blocks for elevation mean and standard deviation but has a negligible effect on other geomorphic differences (Table 1).

To better address the alternate explanations for the continuation of the San Bernardino strand south of Burro Flats, we focus on what is known about recent fault activity on the San Bernardino, Gandy Ranch, and Banning strands, as well as the SGPFZ. We then turn to the evidence provided in the offsets along these strands, presented above, and explore the implications for timing of activity of these strands.

The Wilson Creek slip-rate site is to the west of the SGPr (WC; Fig.1). There, Harden and Matti (1989) described alluvial fan displacements and estimated the age of the surfaces using the degree of soil development. The alluvial fans contained distinctive clasts of the Potato Sandstone, transported along Wilson Creek. The preferred slip rates at this site are 14–25 mm/yr for the past 14 k.y., 22–34 mm/yr for the past 30 k.y., and 12–16 mm/yr for the past 65 k.y., or 90 ka.

At the northwestern corner of Burro Flats (Figs. 1 and 2), Yule and others conducted paleoseismic and slip-rate studies of the San Bernardino strand and associated secondary splays (BF; Fig. 1; Yule and Sieh, 2001; Yule et al., 2001; Yule and Sieh, 2003; Orozco, 2004; Yule, 2009, 2010; Yule and Spotila, 2010). They report fault dips of 75°SW and five paleoearthquakes in the past ~1500 years. Orozco (2004) reported slip rates ranging from 4.0 ± 1.5–6.0 ± 5.0 mm/yr over the past 4.0–4.3 k.y.

Along the Gandy Ranch strand, Heermance and Yule (2017) report a slip rate of ~1.2 mm/yr for the past 5.7 k.y. and 3.7 mm/yr for the past ~1.3 k.y. on what they refer to as the northern fault (NM; Fig.1). They estimate the amount of slip by combining vertical separation and an assumed regional stress vector to calculate lateral slip. They combine this slip determination with ¹⁰Be dating of the vertically separated terrace surfaces. These researchers estimate the number of earthquakes by assuming a vertical displacement of 1 m per event.

In the same study, along the SGPFZ, Heermance and Yule (2017) report slip rates of between 3.2 and 4.2 mm/yr, over the past 1.3–5.7 k.y., respectively (SM; Fig.1). These slip-rate sites are along the fault they refer to as the southern fault in their study. The estimation of slip rates at this location is based on the same criteria as are applied to the Gandy Ranch strand—using a similar regional stress vector combined with vertical separation. Just to the east of Millard Canyon, at Lion Canyon (LC; Fig.1), also along the SGPFZ, paleoseismic investigations have recorded the most recent earthquake to have occurred between 0.6–0.8 ka (Ramzan, 2012; Wolff, 2018), along with three to four previous earthquakes over the past ~6 k.y. (Wolff, 2018).

In the Painted Hills region, east of Whitewater River, Gold et al. (2015) constrained the age of an offset alluvial fan using U-series dating of pedogenic carbonates and ¹⁰Be cosmogenic nuclide exposure dating of surface clasts (PH; Fig.1). They used this age along with the displaced fan apex to conclude a Holocene slip rate for the Banning strand of 3.9 +2.3/−1.6 mm/yr to 4.9 +1.0/−0.9 mm/yr. Paleoseismic investigations along the Banning strand in Desert Hot Springs, to the east of the Painted Hills site, documented five paleoearthquakes that occurred in the past ~3 k.y. and eight paleoearthquakes in the past 7.1–5.7 k.y. (Castillo, 2019; DHS; Fig.1).

To the northwest of Burro Flats, a maximum measured slip of 6.1 km is required to restore the deposits adjacent to the mouth of present-day Mill Creek Canyon to their source in the Falls Creek headwaters (Figs. 1 and 9). We hypothesize that this canyon mouth is a long-lived landscape feature, previously associated with the Falls Creek drainage system, as discussed above. The displacement might be larger than this measurement, though, because the NW extent of these deposits is adjacent to the Santa Ana River, and any deposits farther to the north would have been removed by erosion along this stream course. The displacement might also have been less than this measurement, if the landform emplaced was an alluvial fan with flow to the northwest along the fault strand; we lack the paleocurrent measurements to address this uncertainty. The soil development associated with this deposit indicates that these deposits have been stable for more than ~500 k.y. (VC01; Table 2). A consequent interpretation stemming from this restoration is that the headwater region of the San Gorgonio River was aligned with the Oak Glen/Live Oak drainage system, and flowed toward the Santa Ana River by way of the San Timoteo drainage system (Fig. 2). At the same time, the downstream portion of what is now the San Gorgonio River had its headwaters in the catchment of Wood Canyon (Fig. 2). This amount of slip, as much as 6.1 km, represents the total cumulative slip observed on this fault strand within the SGPr. Although the displacement of the Vincent thrust fault (shown by thick red lines on Fig.2) cannot be measured with any accuracy, given the distances over which it is concealed by youthful sediments, the 6.1 km of restored slip on the San Bernardino strand better matches an approximation of that slip than the previously proposed 2–3 km (Matti et al., 2003).

A slightly lower amount of slip, ~4–4.9 km, restores the lower portion of the Pine Bench landslide to its probable origin, discussed above (Figs. 2 and 10). The same amount of slip aligns the northwesternmost beheaded drainage of Hathaway Canyon with the eastern edge of Wood Canyon catchment area (Fig. 2); subsequent beheaded tributary drainages to the southeast are likely to have been occupied sequentially. Each of these smaller beheaded channels was likely occupied for a relatively short interval; this would result in a size differential between these and the continuously occupied Wood Canyon headwater drainage. The headwaters of the San Gorgonio River would also have sequentially occupied drainages to the east of the Oak Glen/Live Oak drainage system (Fig. 2), including Little San Gorgonio Creek (~4.6 km offset) and Noble Creek (~2.5 km offset). The downstream portion of Millard Canyon would have aligned with Hidden Flat and was likely a much smaller drainage prior to the capture of the upper reach of the canyon.

Within Burro Flats is the Burnt Canyon Breccia of Allen (1957; Figs. 2 and 10), which he recognized as a body of fractured and crushed crystalline rock and rock rubble. He hypothesized that it was sourced from one of the tributaries to Wood Canyon, labeled Burnt Canyon on older topographic maps. We refine this interpretation by proposing that the “Burnt Canyon” unit probably was emplaced as a long-runout rock avalanche deposit (sturzstrom) that coursed down Wood Canyon and came to rest in the Burro Flats area (Figs. 2 and 10). The deposit buttresses against both crystalline rocks as well as tilted sedimentary deposits of Burro Flats and locally overlies alluvial deposits we assign to unit Qvof3 (Allen's [1957] “Heights Fanglomerate”). Since its deposition, the rock avalanche deposit has been displaced by the San Bernardino strand. Thus, the mapped distribution of the Burnt Canyon Breccia (Fig. 3) does not represent its original paleogeographic configuration. Although there is clear indication of recent faulting, how much displacement might have occurred along the San Bernardino strand here is not apparent, since the original extent of the landslide deposit cannot be determined, and reliable piercing lines are lacking.

We prefer the interpretation that the two larger offsets described above occurred on a fault system that stepped across to the Lion Canyon lineament zone, rather than continuing southeast along the Burro Flats section of the San Bernardino strand. The Lion Canyon lineament zone is on strike with SAF strands to the NW and SE, including the San Bernardino strand NW of Pine Bench and the Garnet Hill strand to the SE. The most recent movement, displaying youthful tectonic geomorphology and offsetting the Burnt Canyon Breccia, occurs along the Burro Flats section of the San Bernardino strand. Regardless of the fault system that carried the slip, it is significant that the San Bernardino strand, broadly defined, has had a much longer history than previously recognized.

Along the Banning strand, the displacement of Whitewater Hill and the associated deposits, from a position to receive clasts sourced from the Whitewater River headwaters to its current position, poses several conundrums, which we here describe and address with alternate models of slip. One question is in understanding the nature of the deposits. The depositional units that comprise Whitewater Hill might have been sourced from a central apex, forming an alluvial fan. This would require that the Banning strand defined the edge of landscape that was high to the north, so that the location of the strand served as a transition from canyon to unconfined drainage. A further consideration is that, in order for Whitewater Hill to be the remnant of a single alluvial fan, the axial drainage would need to be entrenched, and the landform able to be displaced without reworking of the deposits.

An alternative interpretation is that these deposits were emplaced by a stream system that flowed across the landscape without forming an alluvial fan. This would be expected if there were not an escarpment associated with the Banning strand at the time of deposition. We prefer this model because the observed reconnaissance paleocurrent measurements in our mapping demonstrate that the streams flowed primarily to the south-southeast and south, rather than in a radial pattern. A consequence of this streamflow model is that the deposits are time-transgressive, with the western deposits emplaced first. A second consequence, though, is that the easternmost stratigraphy of the Whitewater Hill is only displaced ~750 m from the Whitewater River at its intersection with the Banning strand. Farther to the south, near the intersection with the Garnet Hill strand, the deposits are located within 200 m of the active Whitewater River. It is difficult to conceive that this deposit, which is associated with a very strongly developed duripan, indicating stability for more than ~500 k.y., and has subsequently been uplifted and deformed, has experienced so little displacement along the Banning strand. This leads us to a second consideration—that the source of these deposits might not have been Whitewater River as we currently define it.

We propose a model that utilizes flow down a former drainage of Whitewater River, flowing along the ancestral Super Creek to emplace the units (Fig. 13). This former drainage would access the lithologies (associated with the current Whitewater River) that are present in the deposits associated with Whitewater Hill. Restoring 5.5 km of slip on the Banning strand positions the western portion of the Whitewater Hill to receive deposits on its western margin by way of a drainage that is higher in the landscape than the currently active drainage system. Note that at this time, Whitewater Hill was not an elevated landform. This drainage is defined by, from north to south, the upper portion of Whitewater River, wind gaps in the landscape to the southeast of the upper Whitewater River (also shown in Fig.4), and Super Creek (Fig. 13A). After ~2.5 km of motion along the Banning strand, restoring 3 km, the easternmost portion of the current Whitewater Hill was positioned to receive deposits from Super Creek (Fig. 13B). Following this time interval, Whitewater Hill was laterally moved away from the active drainage, the surface associated with these deposits began to stabilize, and a soil began to form. Uplift and compression along the north-dipping Garnet Hill strand began after a lengthy period of soil development, and the Whitewater Hill was folded and uplifted (Fig. 13C). The soil profile is deformed, as well as the deposits, indicating that the soil was already well developed when the folding began. After ~3.5 km of slip had occurred along the Banning strand, Super Creek continued to deposit on the eastern flank of the fold, resulting in the buried soil that today is observable in the eastern wall of Whitewater River (Figs. 11B–11D). Following this deposition, and with continued uplift and deformation of the Garnet Hill block, Super Creek became incised between the Banning and Garnet Hill strands, just to the west of the current path of lower Whitewater River. Eastward tilting of the Kitching Peak block, as described in Kendrick et al. (2015), results in initiation and development of eastward-flowing streams. These are shown in this step of development (Fig. 13C) but may have originated earlier.

By the time that ~4.7 km of displacement had occurred along the Banning strand (Fig. 13D), alignment of a terrace surface on the eastern flank of Whitewater Hill with the crush zone of the Whitewater fault and the upstream portion of the modern Whitewater River permits the integration of drainage along the current drainage course. Super Creek continues to flow, although with a diminished headwater region. The last ~450 m of displacement along the Banning strand have separated the upper and lower portions of the ancestral Super Creek drainage (Fig. 13E). The truncated upper portion of Super Creek now exits the mountain front to the east of the main channel, and forms a small alluvial fan, with a distributed drainage pattern, before finally connecting to the lower portion of Super Creek as a tributary. The lower portion of the original Super Creek has extended across the mountain front by headward erosion. Continued headward growth and capture of the drainages flowing eastward into Coachella Valley has impinged farther onto the upstream portion of Super Creek. This last interval of slip also displaces the incised edge of the Whitewater Canyon, resulting in a preserved terrace with a width of up to ~200 m. Increased incision into the canyon of Whitewater River leads to tributary growth and drainage captures on the west flank of the river, on the uplifting flank of the Kitching Peak block (Figs. 4 and 13E). Tributaries to the east of Whitewater River further disrupted and captured sections of Super Creek, leading to the discontinuous preservations of the drainage system observed today.

The discussion above suggests that the San Bernardino and Banning strands have experienced a punctuated history of slip, with very old deposits displaced on these strands from the source rock types by ~5–6 km. The long history of slip is also indicated by the very different morphologies of the Kitching Peak and Pisgah Peak blocks. A younger history is also present on these strands, as evidenced by the paleoseismic record and the displacement of younger landforms, discussed above. Indications that this slip history is punctuated instead of continuous include the discontinuity of the mapped active fault trace, the lack of intermediate offsets, both temporally and spatially, and the lack of long-term evidence of faulting on secondary strands and sections that are now active (i.e., Gandy Ranch strand and Burro Flats section of the San Bernardino strand).

These observations are not the only indication that there may be punctuated activity and strand switching in the SGPr. Kendrick et al. (2015) proposed that the Mill Creek strand of the SAF has been displaced by 1–1.25 km of left-lateral slip along the Pinto Mountain fault. Matti et al. (1985, 1992a) proposed that the 15 km left bend in the Mission Creek strand was caused by left-lateral slip on the Pinto Mountain fault, though we would modify that hypothesis and suggest, as discussed above, that it was left-lateral slip along the Morongo Valley fault, specifically, that offset the Mission Creek strand. Summed, these offsets approximate the cumulative slip of 16–19 km reported for the Pinto Mountain fault system (Dibblee, 1975; Bacheller, 1978) and the 16 km offset of geophysical features at the western end of the Pinto Mountain fault system (Langenheim and Powell, 2009). These displacements of SAF strands during two separate intervals along faults of the Pinto Mountain fault zone suggest alternating intervals of northwest dextral slip with northeast-oriented sinistral slip, likely associated with increased compression across the SGPr as a whole.

Alternating periods of activity between the eastern California shear zone and sections of the SAF system on a multi-millennial scale were noted by Dolan et al. (2007) and attributed to the development of the eastern California shear zone as an active part of the plate boundary. Similarly, the evolution of a complex fault system may be the driving force to the alternating intervals of slip we observe in the SGPr, between right-lateral slip along NW-trending faults and N-S compression and left-lateral slip on NE-trending faults. Ingersoll and Coffey (2017) propose that the double left restraining bend within the southern SAF, with the SGPr as the southeastern of the pair, is a function of the interaction between the rigid Peninsular Range and Sierra Nevada batholiths, and the response of less rigid blocks moving through these restraining bends. Based on the geologic and geomorphic evidence presented in this study, we propose that the SGPr is currently in a stage of N-S compression, with much of the plate motion stepping off of the SAF onto secondary fault systems, including the San Jacinto fault and eastern California shear zone. This model has implications for establishing regional hazard based on proposed scenario earthquakes.

We propose a model to explain the geologic and geomorphic observations we have made of the SGPr. We use as criteria the cumulative slip along various faults and SAF strands and the youthfulness of tectonic features along faults. In this model, we seek to explain the presence of low-density units present below the Kitching Peak block, based on gravity measurements (Langenheim et al., 2005), and to reconcile recent findings by Fosdick and Blisniuk (2018) that demonstrate provenance for the oldest deposits in the Mission Creek alluvial complex included sources in the Little San Bernardino Mountains.

In the presentation of this model, we hold the San Gorgonio block, including San Gorgonio Mountain, fixed, with the other blocks allowed to move (Fig. 14). By necessity, we make further subdivisions in our block configuration. We include the Morongo Valley fault, Pinto Mountain fault, and Lion Canyon lineament zone as block boundaries.

Our first snapshot of the model reconstructs the landscape at the end of activity on the Mission Creek strand. The Mission Creek strand has a long history with ~90 km of cumulative right-lateral slip (Matti et al., 1992a; Matti and Morton, 1993). At this point in time, we assume that no other faults are active or have significant offset, and the segments of the Mission Creek strand are aligned in a NW-striking system with a gentle convex to the NE curve. Based on our reconstruction, the overall strike of the Mission Creek strand through the SGPr is ~305°. The curve to the fault trace in our reconstruction likely indicates the initiation of disruption to through-going slip. At this time, the drainage systems Hell For Sure and North Fork Whitewater are combining and flowing down Big Morongo Canyon, and the headwaters of Mission Creek are flowing down Little Morongo Canyon. The headwaters of San Gorgonio River flow down Potato Canyon (current path of Oak Glen Creek) to San Timoteo Canyon, ultimately joining the Santa Ana River. Flow from the Wood Canyon headwater region is along Banning Canyon, the current path of San Gorgonio River. Although we lack age control, previous studies have suggested that this phase of activity occurred prior to initiation of slip along the San Jacinto fault. Studies have suggested that the San Jacinto fault inception was ca. 1.1 Ma (Janecke et al., 2011), 1.2 Ma (Matti and Morton, 1993), and less than 1.5 Ma (Morton and Matti, 1993). This latter estimate used several approaches, including age control of the uppermost San Timoteo Beds. Subsequent findings by Albright (1999) refine that age to less than 0.78 Ma. We agree with the model that the San Jacinto fault formed in response to development of the left bend along the Mission Creek strand of the SAF in SGPr; these age estimates provide a reasonable estimate of the time frame for our model at this stage and the subsequent stage.

During this interval, the Morongo Valley fault, part of the Pinto Mountain fault system, becomes active. Left-lateral displacement along this fault disrupts the trace of the Mission Creek strand. The faults are responding to an episode of compression; we envision this as the impingement of the rigid Peninsular Range batholith on the Little San Bernardino Mountains and San Bernardino Mountains, sequentially, after the model proposed by Ingersoll and Coffey (2017). Fattaruso et al. (2016) alternatively propose that this disruption of the Mission Creek strand is created by E-W extension on the West Salton detachment fault. In both models, the Pinto Mountain fault system, including the Morongo Valley fault are passive. This displacement of the Mission Creek strand is also manifest as uplift along the strand west of the fault intersection; uplift of the Yucaipa Ridge and San Gorgonio blocks is occurring. The Galena Peak and Ford Canyon faults might also be active at this time, accommodating some of the uplift generated from this left-lateral movement. We also propose that compressional movement along the SGPFZ is initiated at this time.

After ~7 km of left-lateral motion along the Morongo Valley fault, Hell For Sure and North Fork Whitewater flow southward down Boundary Canyon and either Stubbe Canyon or Cottonwood Creek; Mission Creek is aligned to flow down Big Morongo Canyon; and drainage from the headwaters of Big Morongo Canyon is flowing down Little Morongo Canyon (depicted in Fig. 14C).

After continued left-lateral displacement on the Morongo Valley fault, compression along the SGPFZ, and uplift of the Kitching Peak block, flow from the Hell For Sure and North Fork Whitewater might abandon Boundary Canyon in favor of a drainage system farther east; evidence of this old drainage remains in a series of perched landscape fragments, wind gaps, and knickpoints (Fig. 4).

At this step, the Mission Creek strand has been offset left laterally ~15 km by the Morongo Valley fault. With completion of slip along the Morongo Valley fault, and completion of the initial compressional phase in the SGPr, there has been enough uplift of the Kitching Peak block and wedging of low-density material below the SGPFZ (Langenheim et al., 2005) to further divert flow from Hell For Sure and North Fork Whitewater drainages.

We interpret that tilting initiates at this interval in time, in response to the uplift described, and continues through much of the history of the region. This tilting of the Kitching Peak block and the adjacent terrain to the east is indicated by the uplift of the oldest surfaces in the Mission Creek alluvial complex (Kendrick et al., 2015) and east-flowing drainages east of the modern Whitewater River, discussed above. Within the bounds of the Kitching Peak block, three samples were collected and apatite (U/Th)/He ages were measured by Spotila and others (Spotila et al., 2001, 2002). Spotila et al. (2002) attributed the resulting pattern of exhumation to doming, as per Yule and Sieh (2003), but an alternate explanation for the apparent eastward decrease in exhumation is east-directed tilting of the Kitching Peak block.

At this time, flow from the headwater region of Whitewater drainage system, Hell For Sure and North Fork Whitewater, follows the existing Mission Creek strand fault zone, either because it contains easily erodible material, or because the Kitching Peak block is actually uplifted along this bounding fault resulting in differential movement between Kitching Peak and Yucaipa Ridge blocks. The drainage then follows the route along the southern edge of the hill associated with Wathier Landing before flowing down a drainage system to the east of Boundary Canyon (Fig. 14D). At some point, this drainage links with the (current) headwaters of Super Creek, and deposition begins at the location of the western edge of deposits associated with Whitewater Hill (Figs. 13A and 14E).

At this time, Mission Creek drainage is now flowing down the axis of the Mission Creek alluvial fan complex and depositing the older two units (Kendrick et al., 2015). In their analysis of detrital zircon U/Pb geochronology and clast compositions, Fosdick and Blisniuk (2018) propose that these older deposits were sourced from both Mission Creek headwaters and the Little San Bernardino Mountains, and not from the headwaters of modern Whitewater River. Allowing the Hell For Sure and North Fork Whitewater rivers to flow down Boundary Canyon and other ancestral drainages to the east of Boundary Canyon results in a reconstruction consistent with this finding. Both Big Morongo and Little Morongo creeks were positioned to have contributed clasts to the oldest units in the Mission Creek alluvial complex, providing the source from the Little San Bernardino Mountains. During the interval represented by Figures 14C–14E, we envision that the dextral slip along the SAF in Coachella Valley is stepping around the SGPr, either onto the San Jacinto fault or onto the eastern California shear zone. The degree of soil development associated with the deposits in Mission Creek alluvial complex indicates that this deposition occurred before ca. 0.5 Ma.

At this point in time, ~2 km right-lateral cumulative slip has occurred along the Banning and San Bernardino strands. We propose that these faults were linked by a structure that was located along the Lion Canyon lineament zone, defined by the morphologic boundary, as described and proposed above. This is consistent with our Model 1 (Fig. 12B), although the reconstruction could be accomplished instead with motion along the Burro Flats section of the San Bernardino strand.

During this interval, Mission Creek continues to deposit older units in the Mission Creek alluvial complex. Hell For Sure and North Fork Whitewater drainages continue to flow down the same route as in the previous developmental interval; the central portion of Whitewater Hill is adjacent to Super Creek and receives deposits from this ancient drainage (also shown in Fig. 13B).

The Pine Bench landslide is emplaced then bisected by the San Bernardino strand and begins to be displaced. Falls Creek deposits are beginning to be separated from the headwater source and canyon mouth. San Gorgonio River headwaters are positioned to flow down Little San Gorgonio Creek; before the next interval, it will subsequently occupy Noble Creek. Drainage from Wood Canyon flows along Hathaway Creek; subsequent motion causes beheading of Hathaway Creek in favor of succeeding tributaries to the east of the main channel. The degree of soil development associated with the deposits at Whitewater Hill and the Falls Creek deposits indicates an age estimate for this step at greater than 0.5 Ma.

After ~4.5–5 km of slip has occurred on the San Bernardino strand, Lion Canyon lineament zone, and Banning strand, through-going slip seems to again be interrupted by a compressional episode. This results in further uplift of the Kitching Peak block and continued emplacement of the low-density material beneath Kitching Peak (Langenheim et al., 2005).

We propose that compressional motion along the Garnet Hill strand occurs at this time, uplifting the Whitewater Hill. Compression also occurs along the SGPFZ. Alignment at this point favors the establishment of the Whitewater drainage along the modern trace, following the Whitewater fault gouge zone. This shift, combined with uplift along the Garnet Hill strand, causes the Whitewater River to incise into the Super Creek fan deposits.

Flow from the headwaters of Big Morongo is coursed through Dry Morongo Canyon, and aligned to incise into the Mission Creek alluvial complex, along the future path of the Mission Creek drainage. Flow from the headwaters of the San Gorgonio River previously flowed sequentially down the Oak Glen Creek, Little San Gorgonio River, and Noble Creek and is now deflected by the Pine Bench landslide deposit and flows down Banning Canyon, depositing the units associated with the Banning Bench.

Right-lateral slip initiates on the Mill Creek strand. After ~4 km of right-lateral motion has occurred along this strand, Mission Creek occupies its modern flow path (Fig. 14H), Whitewater River is located along the current path, and incision of Whitewater Canyon is continuing. This incision leads to stream captures to the west of the modern drainage, on the flank of the uplifting Kitching Peak block. The slope of the oldest surfaces in the Mission Creek alluvial complex (Kendrick et al., 2015) as well as the establishment of east-trending drainages to the south of the alluvial complex indicate tilting at this time. These east-trending drainages, also discussed in the previous section, have captured some of the catchment of the ancestral Super Creek drainage system. Tributaries of the Whitewater River have also disrupted Super Creek drainage by capture.

This interval marks the completion of cumulative slip of 7.1–8.7 km along the Mill Creek strand, indicated by offset drainages (Kendrick et al., 2015). Based on luminescence dating, Kendrick et al. (2015) suggest this slip concluded at ca. 100 ka. Blisniuk et al. (2021) concluded that the high rate of slip they propose for the Coachella Valley continues into the SGPr, with the combined Mill Creek and Mission Creek strands carrying the majority of the recent SAF slip. We uphold the analyses and conclusion of Kendrick et al. (2015) that there is little evidence for recent activity and no evidence for a rapid rate of slip for these strands of the fault. Slip observed along the Mission Creek and Mill Creek strands to the SE of SGPr could be stepping off onto other faults such as the eastern California shear zone.

Subsequent to this right-lateral slip on the Mill Creek strand, and represented in this interval, the Mill Creek strand is left-laterally offset 1–1.25 km by the Pinto Mountain fault. We propose that this left-lateral slip feeds into the existing NW-striking faults to the west and is resolved in further uplift of the Yucaipa Ridge and San Gorgonio blocks; slip might also step off onto the Ford Canyon and Galena Peak faults.

As with the previous interval in which the Pinto Mountain fault system was active, we propose that this interval is also accompanied by regional N-S compression, including along the eastern portion of the SGPFZ and the Garnet Hill strand. This compression would result in further uplift of the Kitching Peak block and would also contribute to obscuring the fault connection between the Lion Canyon lineament zone and the Banning strand. In addition, the uplift of the Kitching Peak block has resulted in numerous landslides; several of these obscure the trace of the Lion Canyon lineament zone.

This interval represents our knowledge of the currently active faults and fault sections, based on mapping of displaced late Holocene units and results from paleoseismic and slip-rate investigations (summarized above). The Burro Flats section of the SBS is initiated; right-lateral slip continues on the San Bernardino and Banning strands, and likely on the Garnet Hill strand. We interpret this phase as re-establishing a pathway for strike-slip motion, following disruption by a compressive interval. An unresolved question is why the active faulting would shift from a structure aligned with the far-field strike of the SAF to the northwest and southeast (the Lion Canyon lineament zone) to a fault segment (the Burro Flats section) that strikes more southerly. We note, however, that the orientation of the Burro Flats section is subparallel with the mapped tear faults of the SGPFZ. This stage, as well as the most recent left-lateral slip along the Pinto Mountain fault, would have occurred in the past 100 k.y., based on the timing of completion of slip along the Mill Creek strand.

We consider the implications to our proposed model of fault activity if the lineaments along our Lion Canyon lineament zone instead represent the surface expression of a blind, high-angle structure in the footwall of the SGPFZ, as discussed previously. In this alternate configuration, this structure would link the San Bernardino and Banning strands through the SGPr at depth, beneath the hanging wall of the SGPFZ. The position of the Lion Canyon lineament zone is in agreement with the geophysical data (Langenheim et al., 2005) and is approximately aligned with the San Bernardino and Banning strands. In our proposed model, the Lion Canyon lineament zone is active during the interval represented by Figures 14F and 14G. If, during this time, slip was occurring at depth along this structure, then we would expect that the surface manifestation of this slip would be quite distributed between oblique slip on the SGPFZ and on the numerous discontinuous faults in the SGPr, (e.g., Gandy Ranch strand). This dispersal of surface slip would be occurring between the northern edge of Burro Flats, to the northwest, and the mouth of Cottonwood Creek, to the southeast (indicated by black triangles on Fig.1). This would not impact the displacements and reconstructions we have discussed at the Falls Creek deposits, Pine Bench Landslide, or Whitewater Hill, as all of these sites are beyond the bounds of this obscured portion of the fault system.

One of the primary questions facing the southern California earthquake hazards community is whether or not slip is able to transfer through the SGPr (e.g., Jones et al., 2008). This is critical to evaluate the magnitude and impact of large earthquakes on the SAF zone. Research indicates that in the SGPr, the San Bernardino, Banning, and Garnet Hill strands of the SAF, and the SGPFZ, sometimes referred to as the southern faults in the SGPr, are the active faults that transfer slip through the SGPr (e.g., Yule and Sieh, 2003). Geologic observations, however, reveal a discontinuity between active faults at the surface. The San Bernardino strand does not unequivocally continue as far south as the SGPFZ; at best, there is very subtle evidence, not suggestive of large or frequent ground rupture.

In this study, we have addressed the difficulty in mapping a through-going, active fault system within the SGPr, the southeastern component of the double restraining bend of the southern SAF. We demonstrate that the geomorphology of the SGPr indicates that the two blocks separated by the San Bernardino strand have morphologies distinct from each other. We interpret these differences to indicate that the two blocks, the Kitching Peak and Pisgah Peak blocks, have experienced different uplift histories. From this interpretation, we propose that a bounding structure, either the San Bernardino strand along the Burro Flats section or our hypothesized Lion Canyon lineament zone, has been actively transferring slip during a prior interval of uplift. We have used mapping to document the areal extent of landslides and late Quaternary deposits that show the differences in surface processes operating in each terrain.

We describe two duripans located within our study area. These soil profiles, along with the soil associated on the oldest deposit in the Mission Creek alluvial complex (Kendrick et al., 2015), represent the oldest preserved remnants of the Quaternary landscape in the SGPr and help us to constrain fault activity. We define offsets along the San Bernardino and Banning strands and use these displacements, clast provenance, and age estimates based on the degree of soil development to propose a sequence of slip along the faults in the SGPr during the late Quaternary. We propose that the Morongo Valley fault, part of the Pinto Mountain fault system, rather than the Pinto Mountain fault, was responsible for the displacement of the Mission Creek strand.

We propose that flow from the headwaters of Whitewater River and Mission Creek once coursed along Big Morongo and Little Morongo canyons, which became incised with uplift of the San Bernardino Mountains along the Morongo Valley fault. After some motion along the left-lateral Morongo Valley fault, flow from the headwaters of Whitewater River shifted. We propose an alternate model of this flow, modified slightly from previous work (Kendrick et al., 2015) to explain the presence of Boundary Canyon, its alignment with Cottonwood and Stubbe canyons, and ancestral drainage remnants between Boundary Canyon and the modern Whitewater River. We propose that one of these ancestral drainages, Super Creek, was responsible for the emplacement of deposits associated with Whitewater Hill.

We conclude that there have been two phases of activity along the southern set of faults through the SGPr, including San Bernardino strand, Lion Canyon lineament zone, and Banning and Garnet Hill strands. We define two segments of the SGPFZ that differ in tectonic geomorphology and propose that this reflects differing fault history. We consider the possible interpretation that the Lion Canyon lineament zone is instead the surface manifestation of a high-angle strand of the SAF in the footwall of the SGPFZ or an abandoned range-front thrust fault. Our reconstruction model suggests that there have been alternating intervals of compression and strike-slip plate boundary motion. We interpret that this is an outgrowth of the continued tightening of the double restraining bend along the southern SAF.

We appreciate helpful reviews of earlier drafts of this manuscript by Peter Sadler, Victoria Langenheim, Keith Knudsen, Nate Onderdonk, Lisa Grant Ludwig, and Michele Cooke. Support for this study was provided by the Earthquake Hazards and the National Cooperative Geologic Mapping Programs of the U.S. Geological Survey. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.