Geologic and structural controls on rupture zone fabric: A field-based study of the 2010 M_w 7.2 El Mayor–Cucapah earthquake surface rupture

For more than a century, geologists have recognized that surface ruptures related to large earthquakes contain important structural relations that provide insight into the mechanics of earthquake rupture (e.g., Lawson and Reid, 1908; Reid, 1910; Gilbert, 1928). Surface ruptures display key structural relationships that give important insight into the origin of fractures (e.g., Tchalenko, 1970; Johnson et al., 1994), parameters that control rock failure (e.g., Bray et al., 1994; Johnson et al., 2002), and the nature of rupture propagation through the crust (e.g., Tchalenko, 1970; Sibson, 2003). However, all geologic mapping requires simplification of structural relationships, and the challenge of mapping any surface rupture is that they typically extend tens to hundreds of kilometers across regional structural domains, yet are composed of fracture arrays that require the finest levels of detail in order to be adequately characterized and documented. Johnson et al. (1997) pointed out that surface ruptures typically occur as belts of shear zones that are commonly hundreds of meters in width and can be composed of distinct sets of fractures that accommodate different components of the coseismic slip, and they argued that “reality has been grossly misrepresented” (p. 20) when the mapping of such belts is simplified to a single line. Another challenge is that erosional degradation of fault scarps begins immediately after the earthquake, and, within a period as short as a few months, many important structural relationships of the surface rupture may have become completely erased and cannot be documented at any scale. Successive improvements in consumer level technology, such as computers, geographical information system (GIS) mapping software, digital single lens reflex cameras, and global positioning system devices, allowed small groups of geologists working in the short window of time following the 2010 El Mayor–Cucapah (EMC) earthquake (Baja California, northwestern Mexico) to document structural relationships in unprecedented detail over large regions (Fletcher et al., 2014; this study).

In this study we identify and systematically measure the key parameters that compose the rupture zone fabric, which we here define as the distribution and internal configuration of shearing in the rupture zone. Among the many parameters that define rupture zone fabric, the map-view width of the rupture zone is one of the most important and easiest to systematically document. Rupture zone width can be converted to thickness if the inclination of the master fault is known. Rupture zone fabric is also characterized by the way in which fracture sets are arranged relative to each other, their degree of interconnectivity, and patterns of splaying. In addition, the distribution and partitioning of coseismic slip among the different fracture sets is another parameter used to define rupture zone fabric. In this study we systematically classify each fault scarp based on the amount of coseismic slip that it accommodates relative to the total measured slip in fault-perpendicular transects. This readily distinguishes fault sections with slip that is focused onto a single principal scarp from those with more broadly distributed slip on multiple overlapping scarps. It also is key for defining the symmetry of coseismic slip distribution relative to important structural elements such as the principal fault scarp, the tectonic contact separating distinct fault blocks, and the boundaries of the long-lived master fault zone. Although this list of parameters may not include all measurable aspects of rupture zone fabric, we demonstrate that together, they provide an observational basis for testing hypotheses concerning the mechanics of earthquake rupture and how it propagates from seismogenic depths to the surface, as well as for siting and designing facilities to be constructed in regions near active faults.

As summarized in Table 1, analogue and theoretical modeling studies together with field mapping surveys have been performed throughout the past century to better understand the main parameters that control both rock failure and rupture zone fabric. Important differences in the nature of surface failure were documented by some of the earliest surveys of earthquake ruptures (e.g., Reid, 1910). These surveys recognized that the physical properties of ruptured materials strongly control the localization of shearing, which changes markedly along strike as rupture associated with a single earthquake passes from bedrock to unconsolidated and/or water-saturated sediments (Reid, 1910; Tchalenko and Ambraseys, 1970; Tchalenko and Berberian, 1975; Irvine and Hill, 1993; Bray et al., 1994; Lazarte et al., 1994; Johnson et al., 1997). In addition, as rupture propagates through previously undeformed alluvial cover, it converts to an array of secondary fractures, and other factors such as overburden thickness, magnitude of coseismic slip and slip kinematics have been proposed to explain variations in many of the defining parameters of rupture zone fabric, including its overall thickness, distribution and partitioning of coseismic slip, and diversity of individual fracture sets (Tchalenko, 1970; Horsfield, 1977; Naylor et al., 1986; Bray et al., 1994; Schlische et al., 2002; Quigley et al., 2012).

Rock rheology is also widely recognized as a control on the architecture of brittle faults located entirely within bedrock, and narrow fault zones composed of a single zone of high shear strain are typically associated with quartzofeldspathic protoliths, whereas wide zones with multiple zones of high shear strain are commonly associated with phyllosilicate-rich protoliths (Chester and Logan, 1986; Faulkner et al., 2003, 2008). Despite these well-documented differences in fault zone architecture, there are virtually no systematic studies that demonstrate their effects on rupture zone fabric, and thus there is little understanding of how fault zone architecture evolves with each increment of coseismic slip.

In terms of geometric controls, fault orientation is known to have a significant effect on kinematics (Wallace, 1951; Bott, 1959), and numerous studies have shown that off-fault coseismic strain is asymmetrically localized in the hanging walls of inclined faults (Axen et al., 1999; Fletcher and Spelz, 2009; Ma, 2009; Huang and Johnson, 2010). Other geometric factors that control the complexity and extent of surface ruptures include fault bends (Aydin and Du, 1995), stepovers (Lazarte et al., 1994; Aydin and Du, 1995; Johnson et al., 1997; Oskin et al., 2012; Quigley et al., 2012), and macroscopic slip partitioning between distinct faults (King et al., 2005; Liu-Zeng et al., 2009; Barth et al., 2012). Although bends and stepovers are observed along the 2010 EMC earthquake surface rupture, we restrict the scope of this study to the documentation of detailed structural relations along individual faults, and such macroscopic interactions between multiple faults will be further considered elsewhere.

This study presents detailed mapping of the 2010 EMC surface rupture through the sparsely vegetated Sierra Cucapah (Baja California, northwestern Mexico). We demonstrate that the EMC rupture produced an unparalleled diversity of rupture zone fabric that reflects significant variations in most of the imposed physical conditions identified in Table 1. In the Sierra Cucapah, the 2010 EMC rupture activated slip on four distinct master faults that show a diverse range of geometries and kinematics and have orders of magnitude differences in the finite geologic slip that they accommodate. Fault zone architecture also varies along the preexisting faults that accommodated the 2010 EMC rupture. The mechanical properties and heterogeneities of the faulted host rocks vary greatly throughout the Sierra Cucapah. Systematic mapping along the diverse fault system in the Sierra Cucapah allows us to compare faults with similar structural and lithologic characteristics and improve our understanding of how each affects the observed fabric of the surface rupture.

As a general organizational note, this is the second paper that contains data collected by a team of geologists working on the 2010 EMC rupture. The first (Fletcher et al., 2014) contains the full data sets of fault orientations and coseismic displacements across discrete scarps. This paper presents a set of georectified aerial photographs and detailed map traces of coseismic fractures and fault scarps in the Sierra Cucapah. All of these data sets are published in their entirety as files in Google Earth (KML format) and can be downloaded as Geosphere supplemental files. A fault-by-fault description of the 2010 EMC rupture was presented in Fletcher et al. (2014). In this paper we present a global synthesis of structural patterns of the surface rupture, and we describe only the interesting portions of individual faults that provide exemplary systems for understanding the factors that control rupture zone fabric.

The 4 April 2010 EMC M_w (moment magnitude) 7.2 earthquake produced one of the most complex surface ruptures ever documented on the Pacific–North American plate boundary (Fletcher et al., 2014). On average, ∼2 m of oblique dextral-normal displacement was identified on at least 7 major faults within a zone that extends ∼120 km from the Gulf of California to the U.S.-Mexico border along a trend of ∼315° (Fig. 1; Wei et al., 2011; Rymer et al., 2011; Oskin et al., 2012; Fletcher et al., 2014). The rupture sequence initiated near the center of the surface rupture zone and propagated bilaterally to the northwest and southeast (Hauksson et al., 2010; Wei et al., 2011; Uchide et al., 2013; Fletcher et al., 2014). Southeast of the epicenter, primary slip occurred on a southwest-dipping fault beneath the fluid-saturated sediments of the Colorado River delta, and surface faulting was largely masked by coseismically induced liquefaction and lateral spreading features (Fig. 1; Wei et al., 2011; Oskin et al., 2012; Fletcher et al., 2014). In contrast, to the northwest of the epicenter, surface rupture is well expressed through the ∼55-km-long Sierra Cucapah, and here the sense of coseismic vertical slip changes polarity and becomes dominantly northeast side down (Fletcher et al., 2014). North of the Sierra Cucapah, the rupture zone is characterized by small-displacement (<10 cm) cross-faulting widely distributed throughout the Yuha Desert (Fig. 1; Rymer et al., 2011). Most of the detailed structural analyses of the rupture zone fabrics presented in this study are associated with the series of faults in the Sierra Cucapah, which are both exceptionally well exposed and accommodated large magnitudes of coseismic slip (as much as 4 m; Fletcher et al., 2014).

In Fletcher et al. (2014), the 2010 EMC surface rupture in the Sierra Cucapah was divided into four main fault sections separated by one branching intersection and two accommodation zones. Nearest to the epicenter at the southeastern end of the Sierra Cucapah, primary coseismic slip splayed off of the Laguna Salada fault and onto the Pescadores fault. In the central Sierra Cucapah, coseismic rupture was observed along the Borrego fault over a length of ∼12 km, but slip was rarely concentrated on to this single master fault. Instead, activated sections of the Borrego fault extend several kilometers into both the Puerta and Paso Inferior accommodation zones where coseismic slip is distributed among multiple overlapping master faults (Fig. 2; Fletcher et al., 2014). In the northwestern end of the Sierra Cucapah, coseismic slip becomes consolidated onto a single master fault called the Paso Superior detachment, which is observed to dip as shallowly as 20° along some sections (Fig. 2; Fletcher et al., 2014).

With the exception of the Laguna Salada fault, faults activated by the 2010 EMC earthquake generally do not strongly control local topography (Fletcher et al., 2014) and have not accommodated slip in a historical rupture (Fig. 1). Therefore, one of the main challenges of mapping this particular rupture was to identify and locate the primary rupture path through a complex network of faults that cut the uplifted massif of Cretaceous crystalline basement in the Sierras Cucapah and El Mayor. Rupture mapping based on helicopter and ground-based reconnaissance was performed for 1 week after the main event, but this left a ∼24-km-long gap between known locations of primary rupture (Fig. 3A).

High elevations and very limited access within the gap region made for a daunting task to check each of the overlapping series of known faults for primary rupture. However, within days of the main shock, Wei et al. (2011) used the COSI-Corr (co-registration of optically sensed images and correlation) technique of subpixel correlation of preearthquake and postearthquake SPOT satellite images to eliminate many of the possibilities and identify the exact traces of primary rupture through the gap. Thus, the COSI-Corr results were key for directing field geologists to reach the primary rupture in this remote region (Fig. 3B). Surface breaks defined by COSI-Corr coincided extremely well with primary rupture (Fig. 3B), but simplified and masked the rich structural detail of the fault scarp arrays, especially where primary rupture became distributed across multiple faults in the stepover region of the Puerta accommodation zone (Fig. 3C). Mapping in the Puerta accommodation zone proved challenging, as much of the surface evidence was erased by seismically induced mass wasting on the steep mountainous slopes. Another remote sensing technique using pre-event and postevent lidar (light detection and ranging) (Oskin et al., 2012) proved the most successful in mapping surface offsets in this small but complex section of the rupture (Fig. 3C).

Surface rupture in the Sierra Cucapah typically occurs as multiple overlapping scarps that can be divided into kinematic sets that occur throughout the width of the preexisting fault zones. In order to document detailed structural relationships, we systematically mapped surface rupture at scales finer than 1:500, and documented the distribution of all fractures with >5 cm of vertical offset, which totaled ∼6500 individual scarps. This mapping database has been made available as KMZ and ArcGIS shape files that can be viewed in Google Earth and other GIS applications (see Supplemental File 1¹). The mapping was carried out using observations from the field, optical imagery, and lidar-derived topographic data. Using a Nikon D5000 camera and small fixed-wing aircraft with a hole in the floor, low-elevation, subvertical aerial photographs were acquired during hours with low sun angles, which, depending on the facing direction of the scarp, either highlighted or shadowed its free face (Fig. 4A). These subvertical aerial photos were georectified with manually picked ground-control points from postevent hillshaded lidar topography. On average, ∼100 ground control points were used for photos that have an average footprint of ∼0.36 km². As one of the products of this study, we have made available the entire set of georectified aerial photographs, which have an average pixel resolution of ∼20 cm (see Supplemental File 2²). Numerous oblique air photographs of the rupture, taken during various helicopter surveys, were invaluable in the documentation of the structural details of the surface rupture. The first stage of systematic mapping was performed by visually locating ground fractures seen in oblique photos on a base map of pre-event Google Earth imagery (Figs. 4C, 4D). Within ∼4 months of the main event, a lidar-based high-resolution (50 cm/pixel) digital elevation model (DEM) was available for the entire rupture (Oskin et al., 2012). Mapping using only the lidar data was a rapid way to compile the overall macroscopic distribution of surface rupture, but was not sufficient for documentation of the detailed rupture zone fabric (Figs. 4E and 4F). In particular, ruptures with scarp heights less than 3 cm cannot be discriminated with airborne lidar. Nonetheless, the lidar DEM was critical for georectifying vertical air photos acquired within one month of the main event, which gave the most detailed view of rupture zone fabric (Fig. 4B).

We report variations in readily observable parameters that define rupture zone fabric. These include (1) rupture zone thickness, (2) existence of a principal displacement scarp, (3) distribution and partitioning of coseismic slip within the rupture zone, and (4) geometric arrangement of different fracture sets, patterns of splaying, and degree of interconnectivity. The specific measurements and classifications used for each parameter are defined in the following, and our main findings for the factors that control different aspects of rupture zone fabric are summarized in Table 1.

Most faults in the Sierra Cucapah are either completely exposed at the surface or only shallowly buried by sedimentary cover (Fig. 2A). Nonetheless, rupture zone width in map view varies by more than an order of magnitude, from meters to hundreds of meters (Figs. 2B, 2C, and 5). In general, the widest rupture zones are associated with the Paso Superior detachment in the north (Fig. 2B), and the narrowest rupture zones are found along the Pescadores and Laguna Salada faults in the southern Sierra Cucapah (Figs. 2C and 5A). In order to systematically study these variations, the rupture zones in the Sierra Cucapah were divided into 77 sections based on along-strike changes in master fault orientation and/or rupture zone width in map view (see Supplemental File 3³ for pdf of tabulated fault section geometry, location, and rupture zone dimensions). Using the fault dips reported in Fletcher et al. (2014), the thickness of the rupture zone for each fault section was calculated from simple trigonometric relationships using the direct measurements of the rupture width measured in map view together with the average orientation of the ground surface and master fault over any given rupture section.

All the faults with primary rupture were classified into three main categories based on the rock types that they juxtapose: basement-basement, basement-sediment, and sediment-sediment faults. Basement-basement faults, such as the Pescadores fault, cut across the rugged topography of the high Sierras that are composed of Cretaceous plutonic rocks (Fig. 2C; Barnard, 1968; Chora-Salvador, 2003). Basement-sediment faults control sedimentary basins, and include all sections of the Borrego fault and the central and northern Paso Superior detachment (Fig. 2B). Sediment-sediment faults are those that juxtapose sediments and are typically buried in the near surface, such as along the margin of the Laguna Salada basin, where tectonic subsidence is enhanced by the superposition of east- and west-directed oblique normal faulting that characterizes the Paso Inferior accommodation zone (Fig. 2B; Fletcher et al., 2014). The southern Laguna Salada fault controls an alluviated fault-line valley, and thus also belongs to the sediment-sediment fault class (Fig. 2C).

Of the 77 sections defined in the Sierra Cucapah, 8 were filtered from the analysis due to (1) uncertainty about the dip of the master fault (sections 50 and 51; Fig. 2B), (2) stepover within the southern Borrego fault and Paso Superior detachment (sections 47 and 5, respectively; Fig. 2B), and (3) rupture zones associated with closely spaced or intersecting master faults (sections 31, 35, 41, 42, 66; Figs. 2B, 2C; see also Supplemental File 3 [see ^{footnote 3}]). The multiple fault strands observed in the macroscopic accommodation zones were treated as individual master faults, which provides a useful convention for comparing thickness versus fault dip, yet these stepovers actually represent the widest parts of the rupture zone in the Sierra Cucapah.

One of the most important parameters that is correlated with rupture zone thickness is the distribution of coseismic slip, which we measured as the number of fault scarps across a given rupture zone as well as the relative magnitude of slip that each accommodated. In order to systematically characterize the distribution of coseismic slip, all scarps were classified into four categories based on relative magnitude of total coseismic slip as reported by Fletcher et al. (2014; Supplemental File 1 [see ^{footnote 1}]): these include >90%, 60%–90%, 30%–60%, <30%.

The maximum number of scarps that exist in any given transect varies from 2 to 18, and the plot of Figure 6A shows that this fabric parameter increases systematically with the width of the rupture zone. It is surprising that a rupture zone as thick as 200 m only has 4 scarps (Fig. 6A), but this occurs along a buried section of the Paso Superior detachment (fault section 16; Fig. 2B), where much of the coseismic surface deformation is accommodated by penetrative shear and warping of the surface instead of discrete scarps (discussed herein). In general, fault sections of the basement-basement fault class are clustered at the low end of the range of observed values for the number of scarps in a given transect (Fig. 6A), and with few exceptions basement-basement faults typically do not exceed 50 m in width. Although the fields of basement-sediment and sediment-sediment rupture zones overlap significantly, there is a crude separation such that for the same number of scarps, sediment-sediment rupture zones are thicker than those found in basement-sediment fault sections (Fig. 6A).

The expected tendency of focused coseismic slip occurring in narrow rupture zones is clearly documented by the fact that single principal scarps that accommodate >90% of coseismic slip are only observed in the narrowest rupture zones (∼25 m wide; Fig. 6B). These rupture zones would be considered by most to have an extremely localized slip distribution (Heermance et al., 2003; Rockwell and Ben-Zion, 2007). As the rupture zone widens the largest scarps systematically accommodate a relatively smaller amount of the total coseismic slip measured in any given transect (Fig. 6B).

Our data clearly demonstrate that rupture zone thickness systematically increases with decreasing dip of the master fault, and faults with shallower dips have more variation in rupture zone thickness (Fig. 7A). Both fault dip and rupture zone thickness vary systematically with lithologic class of the master faults in the Sierra Cucapah. In general, basement-basement faults are steeply dipping (>70°), whereas basement-sediment faults generally dip <60° (Fig. 7A). Faults from both of these lithologic classes become buried beneath a sedimentary cover that generally does not exceed 300 m, and thus faults of the sediment-sediment class span the entire range of dips (Fig. 7A). Fault sections of the basement-basement fault class, however, do not show significant variations in thickness over their limited ranges in dip (Fig. 7A), and they are generally narrow with a weighted average rupture zone thickness of 27 m. In contrast, rupture zone thickness reaches ∼250 m on sections of both the sediment-sediment and basement-sediment classes, which show systematic increases of rupture zone thickness with decreasing dip of the master fault (Fig. 7A).

It is important to note that the structural relationships established here are actually magnified in the map-view changes in width of the surface rupture. For example, rupture zone thickness related to a vertical fault does not change from the map view, but the surface rupture width above a 20° dipping fault is three times as wide in map view (Fig. 7B), and thus a simple examination of map-view width can be very useful in characterizing this essential aspect of the master fault.

In the Sierra Cucapah, the 2010 EMC earthquake activated slip on numerous master faults that are well exposed at the surface and display a great diversity of fault zone architectural characteristics. Fault zone architecture is defined in terms of three main rock units: unfractured protolith, damage zone, and fault core (e.g., Chester and Logan, 1986; Caine et al., 1996). Most slip is thought to occur in the fault core, which is composed of discrete slip surfaces, unconsolidated clay-rich gouge zones, brecciated and geochemically altered zones, and/or zones of highly indurated cataclasite (Caine et al., 1996, and references therein). The damage zone can include relatively small displacement slip surfaces (e.g., Chester and Logan, 1986), but is more generally characterized by subsidiary structures that diminish in intensity and grade into unfractured protolith with distance from the fault core (Chester and Logan, 1986; Morton et al., 2012).

Systematic mapping of fault zone width throughout the Sierra Cucapah was enabled by a strong contrast in outcrop weathering resistance at the outer contact between the damage zone and host rock (Fig. 5; see Supplemental File 4⁴ for a KML file containing the mapped extent of fault zones). We find that as fault zone architecture increases in complexity and width, so does the expression of coseismic surface rupture (Figs. 5 and 8A). Master faults like the Laguna Salada and Pescadores faults commonly have narrow fault zones where coseiesmic slip is strongly concentrated onto a single well-defined principal scarp of the >90% or 60%–90% classes (Figs. 5A, 6B, and 8A). Master faults like the Paso Superior fault and sections of the Borrego fault have wide, complex fault zones composed of multiple zones of high shear strain, and these faults also typically have more broadly distributed coseismic slip through wider rupture zones (Figs. 5B, 5C, and 8A). Nonetheless, our data demonstrate that the thickness of the fault zone in any given section is systematically greater than the thickness of the surface rupture zone (Fig. 8A). Owing to the strong correlation of fault zone and rupture zone thickness, we find that both of these parameters are inversely correlated with the dip of the master fault (Figs. 7A and 8B).

The structural complexity and thickness of fault zones are commonly thought to be related to fault rock rheology as well as the amount of finite displacement the fault has accommodated (Chester and Logan, 1986; Scholz, 1987; Ben-Zion and Sammis, 2003; Faulkner et al., 2003, 2008). Our data show that there are also systematic variations in fault zone architecture with fault orientation. Even if all faults were subjected to the same regional stress state, individually they would have undergone tractions that would have been very different in both overall magnitude as well as the key ratio of shear stress to normal stress, which is known to control seismogenic failure (e.g., Wallace, 1951; Bott, 1959). It is beyond the scope of this paper to document the detailed structural relations required to evaluate the relative influence of rheology, finite displacement, and tectonic loading for each fault section, but the data set compiled for this study clearly establishes the strong correlation between fault zone thickness and rupture zone thickness.

The most significant along-strike variation in the map-view width of rupture zones coincides with the boundary between the Sierra Cucapah and Colorado River delta (Fig. 1). In the Sierra Cucapah, rupture zones are generally confined to single master faults and have finite widths <135 m (Fig. 8A). In two accommodation zones formed between major left steps in the surface rupture, coseismic slip is distributed along faults in a zone that reaches the width of ∼4 km (Fig. 2A). However, even these stepover zones are narrow compared to the penetrative fracturing and liquefaction that spread across zones as wide as 25 km in the delta (Fig. 1; Fletcher et al., 2014). Vertical coseismic slip of ∼2 m detected by remote sensing techniques in the delta (Wei et al., 2011; Oskin et al., 2012) is thought to be similar in magnitude to that observed in the Sierra Cucapah (Fletcher et al., 2014). However, due to the thick overburden sequence, coseismic slip did not reach the surface by discrete faulting in the delta (Oskin et al., 2012). Most ground failure in the Colorado River delta is located on the hanging-wall side of buried master faults, demonstrating asymmetrically developed intensity of ground motion on different fault blocks (Fig. 1; Fletcher et al., 2014). Liquefaction and fracturing are noticeably absent along the section of the rupture that coincides with a dry reach of the Colorado river (Fig. 1; Fletcher et al., 2014), which is several meters higher in elevation than the adjacent regions of the delta due to aggradation of sediment between its (artificial) levees. Therefore, although contrasts in the thickness of overburden explains well the large increase in map-view width of the rupture zone in the delta, the distribution of water-saturated sediments also controls the extent of ground fracturing.

The effects of both magnitude and sense of fault slip on the fabric related to fault-tip propagation are best observed along fault sections that are buried by previously unruptured sediments. Because the rocks on both side of the fault at the surface are the same, the rupture zone is mechanically homogeneous. Therefore, fractures formed at the tips of propagating faults are less likely to be guided by preexisting weaknesses and are more likely to adopt mechanically optimal orientations reflecting differences in slip kinematics (e.g., McGrath and Davison, 1995). To further isolate kinematic effects on fracture patterns, we compare fault sections that are uniformly steeply dipping (≥60°). Faults that meet these lithologic and geometric criteria include normal and oblique normal-dextral faults in the Paso Inferior accommodation zone, two sections (63 and 64) of the Pescadores fault, and the subvertical dominantly strike slip southern Laguna Salada fault (Figs. 2B, 2C).

Gradients in the magnitude of slip demonstrate how individual sets of secondary fractures coalesce into throughgoing fault zones. Incipient slip observed in mesoscopic fractures commonly shows the development of multiple sets of secondary fractures (i.e., synthetic P shear fractures and T extensional fractures), and P-T arrays are commonly associated with slip magnitudes <5 cm (Fig. 9). As slip increases, early-formed T fractures become transected by multiple strands of P fractures, which likely go on to form the throughgoing fault at higher slip magnitudes.

An evolutionary sequence of fracture development is also observed at a macroscopic scale (Fig. 10). Macroscopic fracture arrays with oblique normal-dextral slip of <30 cm are generally characterized by isolated fractures that do not intersect except at the very smallest scales, and fractures with >10 cm of slip were not observed to overlap (Fig. 10A). Instead of forming relay arrays with more or less constant along-strike slip, steps between dominant fractures were characterized by significant decreases in slip on discrete faults due to their nonoverlapping character (Fig. 10A). However, this does not preclude the accommodation of deformation by other mechanisms such as rotation, penetrative shearing, and/or granular flow in the area between discrete faults. As oblique displacement increases to ∼45 cm, isolated fractures are much less common and the array forms an anastomosing pattern with complex branching of fault scarps (Fig. 10B). Slip is generally distributed and relayed among multiple overlapping strands, and in this manner the fault array accommodates relatively constant slip along strike. The clear development of a principal scarp is only observed on fault sections with >60 cm of oblique slip (Fig. 10C). In such sections, secondary fractures with oblique Riedel geometries emanate from, but do not crosscut, the principal scarp and commonly have asymmetric displacement gradients with the greatest slip magnitude near the intersection with the principal scarp (Fig. 10C). These systematic variations in rupture zone fabric with finite slip can be considered an evolutionary sequence, which is consistent with laboratory experiments that document the growth of individual fractures into coalescing networks and throughgoing fault planes (e.g., Tchalenko, 1970; Naylor et al., 1986; Schlische et al., 2002).

An important difference between the development of field and laboratory rupture zone fabrics is the scale of faulting. The same sequence observed with slip magnitudes between 30 and 60 cm in the field are observed with magnitudes of only ∼9–36 mm in the lab, which reflects the differences in rheology and thickness of unfaulted, sand-dominated sediments as compared to clay-cake substrate. The evolution of the distribution of coseismic slip within field rupture zones may also be fundamentally different from laboratory experiments. Tchalenko (1970) reported that the accumulation of slip on secondary fractures ceased immediately following the linking of isolated fault strands to form a throughgoing fault. However, as detected by lidar difference models (Oskin et al., 2012), many buried sections of the EMC rupture accommodate >2 m of oblique coseismic slip and as much as half is partitioned onto secondary fault scarps and zones of penetrative off-fault shearing that are distinct from the throughgoing principal scarp (Fig. 11). We observed that regardless of the stage of formation of a principal scarp, secondary faults generally accommodate 40%–60% of the total coseismic slip (Fig. 11). This strongly suggests that secondary faults must continue to be active after the formation of a throughgoing principal scarp. In this example the principal scarps are dominated by dip-slip displacement, and thus it is possible that secondary faults remained active because they were required to accommodate the overall oblique shear sense. In contrast, in the laboratory experiments of Tchalenko (1970), secondary and principal scarps have nearly identical kinematics and only one set is required to accommodate the net slip.

In addition to the changes observed with increasing slip magnitude, rupture zone fabric shows marked variations with overall kinematics of seismogenic slip. Riedel shears are the most common secondary fractures and they take on different orientations relative to the principal scarps depending on overall kinematics (Fig. 12). If finite slip has a significant component of strike slip, secondary fractures dominantly have an en echelon configuration and the strike is oblique to the principal scarp (Fig. 12A). In contrast, if finite slip is dominated by normal dip-slip, then the Riedel shears generally strike parallel to the principal scarp, but dip more steeply (Fig. 12B).

Among the subset of steeply dipping buried faults, an order-of-magnitude of increase in rupture zone thickness (12–135 m) is strongly correlated with variations in strike (Figs. 13A, 13B, and 14A). In Fletcher et al. (2010), it was demonstrated that the rake of slip changes systematically with fault orientation, indicating that rupture is controlled by regional stress with high phi values (>0.8) such that σ1 is close in magnitude to σ2, the axes of which are oriented in a vertical plane perpendicular to a subhorizontal σ3 that trends S80W (Fig. 14B). Combining these results with the analysis of rupture zone fabric demonstrates that the rupture zone thickness is narrowest in fault sections dominated by either pure dip-slip or pure strike-slip kinematics, and the fivefold increase in thickness is associated with oblique-slip kinematics characterized by subequal amounts of dip slip and strike slip (Fig. 14).

One possible explanation for the correlation of rupture zone thickness with kinematics involves the partitioning of slip into different sets of secondary fractures. Secondary fracture sets associated with steeply dipping buried faults can be classified into either subparallel and/or anastomosing, or en echelon arrays (Figs. 13C–13E). Several studies have documented that the same families of secondary fractures form on both normal faults and strike-slip faults, but are rotated 90° from each other (e.g., Petit, 1987; Davis et al., 2000). Therefore, it is likely that the en echelon rupture fabric develops in fault sections dominated by strike slip (Figs. 13E and 15A), whereas the rupture fabric of a normal-slip dominant fault appears subparallel and/or anastomosing in map view (Figs. 13C and 15C). We observed that only one set of secondary fractures is present along faults with orientations that coincide with either of these two end-member kinematic cases. In contrast, both sets of en echelon and subparallel and/or anastomosing secondary fractures are present along faults with orientations consistent with oblique slip (Figs. 13D and 15B). Therefore, the partitioning of slip in multiple sets of kinematically discrete secondary fractures may simply affect a greater volume of rock above oblique-slip master faults as compared to those that accommodate pure strike slip or pure dip slip.

One of the most complex classes of rupture zone fabric in the EMC surface rupture is associated with the buried sections of the Paso Superior detachment (Fig. 2B, sections 10–16). This is largely due to the structural adjustments related to the transfer of slip from the low-angle master fault to an array of more steeply dipping faults cutting sediment. Furthermore, this section of the Paso Superior detachment is located near the active margin of the Laguna Salada basin, where it is buried together with the crosscutting Laguna Salada fault beneath as much as several hundred meters of sedimentary cover. Thus, rupture zone fabric is affected by the propagation of coseismic slip through not only a wide complex fault zone, but also through a relatively thick section of sediments.

Along the buried section of the Paso Superior detachment, fault scarps generated in the EMC rupture follow the base of a series of low-lying hills of uplifted basin fill (Fig. 16), which forms the southward continuation of the narrow horst block of crystalline basement that exists between well-exposed sections of the Laguna Salada fault and Paso Superior detachment (Fig. 2B). Coseismic displacement averaged ∼2.5 m along the buried section of the Paso Superior detachment and was dominated by dip slip (Fletcher et al., 2014). The amount of lateral coseismic slip is generally low in magnitude and its sense changes along strike from predominately dextral to locally sinistral (Fletcher et al., 2014). In general, the lateral component of slip was accommodated on strands with the greatest vertical offset.

The EMC rupture zones developed above the buried Paso Superior detachment are consistently wide (97–262 m thick), and most coseismic slip was accommodated by an array of moderately dipping (generally <45°) synthetic faults within the hanging-wall side of the buried detachment (Fig. 16; Fletcher et al., 2014). Principal scarps that accommodate 60%–90% of the coseismic slip are generally only along very short segments of the rupture zone (50–75 m long; Supplemental File 1 [see ^{footnote 1}]). More typically, the rupture zone contains multiple scarps that accommodate 30%–60% of coseismic slip (Supplemental File 1 [see ^{footnote 1}]). These are configured crudely into en echelon arrays relay slip in regions of overlap (Fig. 16B).

The consistently wide EMC rupture zones along the buried Paso Superior detachment reflect the transfer of coseismic slip from the low-angle master fault at depth to more steeply dipping faults that cut the sedimentary cover. In general, synthetic east-down scarps are spatially partitioned from west-down antithetic scarps, which are respectively located on the southwest and northeast margins of the rupture zone (Fig. 16A). Very few scarps exist in the middle of the rupture zone between the two bands of kinematically partitioned scarps along the margins. In analogue laboratory experiments, Horsfield (1977) demonstrated that an upward-steepening normal fault requires slip on an antithetic set of normal faults in order to accommodate the same heave:throw ratio as that of the master fault at depth. The existence of antithetic scarps that define the eastern margin of the rupture zone strongly suggests that the buried master fault dips more shallowly than the synthetic scarps at the surface (Fig. 16A).

The pre-event and postevent lidar difference model of Oskin et al. (2012) defines well the vertical component of coseismic displacement and demonstrates that antithetic slip is also accommodated by gentle warping of the surface, which we interpret to be analogous to a rollover anticline above a listric normal fault (Fig. 17). The net heave:throw ratio of coseismic slip at the surface is readily calculated using the orientation of scarps at the surface and the ratio of synthetic to antithetic vertical offsets defined by the lidar differencing (Figs. 17B, 17D). We assume that the antithetic rollover warping of the surface occurs by penetrative shearing on optimally oriented surfaces that dip 65° (Fig. 17E). Thus the calculated net heave:throw ratios indicate that scarps at the surface are controlled by a master fault that dips ∼30°, which is consistent with direct measurements of the Paso Superior detachment exposed at the surface farther north along the strike of the rupture zone. Projecting antithetic and synthetic scarps indicates that they should intersect at depths of ∼100 to >300 m, which likely is the depth of burial of the Paso Superior detachment along these sections (Figs. 17C, 17E).

Paleoscarps are poorly preserved in the fine-grained surficial deposits above the buried Paso Superior detachment. However, we have identified both synthetic and antithetic paleoscarps that coincide well with the traces of EMC fault scarps (Fig. 16A), permitting the hypothesis that patterns of rupture observed in the EMC earthquake had been replicated in previous events.

The documentation of rupture zone fabric presented in this study could have important applications for engineers responsible for siting and designing facilities to be constructed in regions near active faults. For example, the Alquist-Priolo Act of California defines building setback distances from active faults, set to a minimum of 50 ft (∼15.25 m; Bryant and Hart, 2007). Similar setback distances were defined for the Wasatch fault based on 40 trenches evaluated by McCalpin (1987). The results from this study show that this minimum setback distance is only adequate for steep to subvertical faults with a simple architecture and simple rupture zone fabric, and does not apply to faults that dip <60° and/or those buried by thick sedimentary covers. The Borrego fault dips ∼40° and exhibits rupture zones that reach ∼260 m in map-view width (Fig. 2B); this is 17 times wider than the minimum setback defined by the Alquist-Priolo Act. Moreover, along many sections of the low-angle Paso Superior detachment, the surface rupture does not even coincide with the actual fault trace and is >400 m wide in map view (Fig. 2B). Therefore, the results of this study demonstrate that the definition of nonarbitrary setback distances requires consideration of master fault characteristics on a case by case basis. The diverse array of faults activated in the 2010 EMC event shows that rupture zones vary markedly with differences in parameters such as master fault dip, physical properties of ruptured materials, overburden thickness, magnitude of coseismic slip, sense of relative slip, and fault zone architecture.

Rupture zone fabric is characterized by the thickness, distribution, and internal configuration of shearing in a rupture zone. We developed a list of easily measurable parameters to help understand what controls the nature of rupture zone fabric at the surface. These parameters include rupture zone thickness, presence of a principal displacement scarp, number and type of secondary fractures, their interconnectivity and patterns of splaying, and the distribution and partitioning of coseismic slip.

Rupture zone thickness, which is the extent of surface faulting measured perpendicular to the master fault plane, increases with decreasing dip of the master fault and increasing thickness of sedimentary cover. An additional fabric parameter is the distribution of coseismic slip, which was measured by the number of fault scarps in a given rupture zone as well as the relative magnitude of slip that each accommodated. The focusing of coseismic slip onto a single principal scarp is only associated with rupture zones that are generally <50 m wide. Rupture zones >75 m wide generally lack a single principal scarp and may contain as many as 18 individual scarps. We find that rupture zone width and distribution of coseismic slip is strongly correlated with the width of the long-lived fault zone.

Shallowly buried faults with steep dips demonstrate how rupture fabric develops at the tips of propagating faults and how it is affected by the magnitude of coseismic slip. As the magnitude of coseismic slip increases from 0 to 60 cm, arrays of isolated fractures transition into coalescing networks that in turn develop into a single throughgoing principal scarp surrounded by secondary fractures. However, in contrast to the results of laboratory experiments, our data suggest that secondary fractures continue to accumulate slip even after the formation of the principal scarp.

Another parameter of rupture zone fabric is the number and types of secondary fractures, and we show that these change with master fault orientation and kinematics. The partitioning of oblique slip into different sets of secondary fractures affects a greater volume of rock and generates rupture zones that are as much as five times wider than those that accommodate pure strike slip or pure dip slip. This study demonstrates the importance of characterizing rupture zone fabric, which can be widely applied by scientists studying processes of rock failure and dynamic rupture propagation as well as mitigating seismic hazards associated with earthquake surface ruptures (i.e., fault zoning and building setback limits).

This work was financed by CONACYT (Consejo Nacional de Ciencia y Tecnología) grants CB-2007-81463 and CB-2009-133042. The GEER Association (Geotechnical Extreme Events Reconnaissance) provided funding for initial field work. This research was also supported by the Southern California Earthquake Center (SCEC), funded by National Science Foundation Cooperative Agreement EAR-1033462 and U.S. Geological Survey Cooperative Agreement G12AC20038. This is SCEC contribution 1963. We are grateful for the excellent technical support provided by Jose Mojarro, Sergio Arregui, and Luis Gradilla. We also thank Nicholas C. Barth, Colin Amos, and an anonymous reviewer for extensive and insightful reviews.