Damage zone heterogeneity on seismogenic faults in crystalline rock; a field study of the Borrego Fault, Baja California

Highlights

• 68 m² high-resolution fracture map of 11,114 fractures ranging from cm to m lengths.

• Power law decay in fracture density from fault gives a damage zone width of ~85 m.

• Heterogeneity in fracture density decreases away from the fault core to ~16 m.

• Inner and Outer damage zone defined by increasing heterogeneity from parent rock.

• Modified model of fault damage zone evolution explaining the heterogeneity trend.

Abstract

Complex fracture damage around large faults is often simplified to fit exponential or power law decay in fracture density with distance from the fault. Noise in these datasets is attributed to large subsidiary faults or random natural variation. Through a field study of the Borrego Fault (Baja California) damage zone, combining mm-resolution structural mapping and point sampling, we show that such variations are the expression of systematic damage heterogeneity. The oblique-slip Borrego Fault comprises NW and SE segments that ruptured during the Mw7.2 2010 El Mayor-Cucapah earthquake. Measurements of fracture density along eight linear fault-perpendicular transects and from a high-resolution 68 m² structural map display a power law decay and define footwall damage zone widths of ~85 m and ~120 m for the NW and SE segments respectively. Variance in fracture density decays with distance following an inverse exponential relationship, to background variance at ~16 m. Spatial analysis of the high-resolution fracture map reveals a patchy distribution of high- and low-intensity clusters at metre- and decimetre-scales. We attribute high-intensity clusters at these scales to local complexity caused by interactions between minor subsidiary faults (10¹ m length and 10⁻²-10⁻¹ m displacement). Fracture density differences between high- and low-intensity clusters decrease with distance from the fault, demonstrating a systematic change in outcrop-scale damage heterogeneity. Based on these observations we present a revised model for damage zone growth including growth of heterogeneity.

Introduction

In crustal fault zones, dynamic rupture and fault slip is typically hosted within a narrow fault core, surrounded by a fracture damage zone (Fig. 1) of up to 100s of metres in width (e.g. Ben-Zion and Sammis, 2003; Chester and Chester, 2000; Chester and Logan, 1986; Rowe et al., 2013; Savage and Brodsky, 2011; Scholz, 1987; Sibson, 1986, 2003). This damage is accrued by a combination of aseismic/quasi-static (e.g. Chester and Chester, 2000; Childs et al., 2009; Cowie and Scholz, 1992; Faulkner et al., 2011) and coseismic processes (e.g. Aben et al., 2016; Ben-Zion and Ampuero, 2009; Johri et al., 2014a; Okubo et al., 2019; Rempe et al., 2013; Rice et al., 2005; Sagy and Korngreen, 2012; Xu et al., 2012). Fractured rock can have significantly different mechanical and hydraulic properties to intact rock, and so the damage zone plays a fundamental role in crustal fluid flow and the mechanics of faulting and earthquakes. Firstly, damaged fault rocks are generally more permeable with higher surface area than intact rocks, and hence play a key role in the migration of fluids and precipitation of minerals in and around fault zones over the seismic cycle (e.g. Evans et al., 1997; Hennings et al., 2012; Lawther et al., 2016; Lockner et al., 2000; Seront et al., 1998; Sibson, 1994). Secondly, damaged rocks have reduced elastic moduli, cohesion and yield strength (e.g. Bruhn et al., 1994; Callahan et al., 2019; Faulkner et al., 2006; Griffith et al., 2012; Griffith et al., 2009; Walsh, 1965), resulting in reduced elastic wave velocity, which can cause attenuation and potentially non-linear wave propagation effects during ruptures (e.g. Wu et al., 2009). The amount and spatial variation of these reductions can directly modify rupture dynamics/style/shape (e.g. Cappa et al., 2014; Dunham et al., 2011; Huang and Ampuero, 2011; Okubo et al., 2019), and lead to the generation of slip pulses that can accelerate the transition to supershear rupture (e.g. Harris and Day, 1997; Huang and Ampuero, 2011). Significant velocity reductions within a fault zone results in the structures trapping seismic waves that can continuously perturb stresses on the fault during earthquakes. Finally, the dynamic generation of damage as the earthquake rupture propagates can itself influence the dynamics of rupture propagation. This can be done by increasing energy dissipation (e.g. Andrews, 2005), modulating the rupture velocity (Cappa, 2011; Huang et al., 2014; Thomas et al., 2017) and modifying the size of the earthquake, changing the efficiency of weakening mechanisms such as thermal pressurisation of pore fluids (e.g. Brantut and Mitchell, 2018; Noda and Lapusta, 2013), and even generating additional seismic waves (e.g. Ben-Zion and Ampuero, 2009).

With increasing displacement and fault maturity, fracture damage zones increase in both width and complexity (Fig. 1). This increased width and complexity is due to overprinting of incremental fracture damage, which leads to heterogeneity in off-fault damage structures. Furthermore, strong rock-type dependencies (Bistacchi et al., 2010; Loveless et al., 2011; O'Hara et al., 2017) and the influence of pre-existing structures (e.g. Brogi, 2011; Myers and Aydin, 2004) can also lead to spatial heterogeneities in damage formation. Heterogeneous damage patterns lead to heterogeneous mechanical and hydraulic properties of the same scale and distribution. Thus, quantifying damage heterogeneity is fundamental in understanding the complex effects and feedbacks on earthquake processes. To date, most observations of damage heterogeneity are limited to qualitative description only (Caine et al., 2010; Gudmundsson et al., 2002, 2010).

Most classical fundamental studies of fault zone damage were based on detailed qualitative structural geology techniques (e.g. Crider and Peacock, 2004; Price and Cosgrove, 1990). This approach identified three broad zones of damage, based on the type, intensity, and extent of fracturing; tip, wall, and interaction damage (Kim et al., 2000, 2003, 2004; Peacock et al., 2016) (Fig. 1a). Initially, interaction and tip zones show the most complex and intense damage, while wall zones develop more complexity as the fault grows through cumulative slip (Kim and Sanderson, 2008; Madariaga, 1983; Rousseau and Rosakis, 2003). More recent quantitative approaches of damage analysis have been developed in order to answer fundamental questions on the seismic cycle, such as fault strength, fluid flow properties and rupture dynamics. To do so, it was necessary to simplify the complex off-fault damage so that usable mathematical expressions describing the spatial and temporal distribution of damage could be derived (e.g. Chester et al., 2005; Choi et al., 2016; Savage and Brodsky, 2011; Shipton and Cowie, 2003) (e.g. Fig. 1d and e). For simplicity, we apply the following damage terminologies (adopted from Shipton and Cowie (2001)) for fault/fracture length scales relative to the main fault, where main fault length is >km: (1) Macro-damage, 1–3 orders of magnitude smaller [10¹–10³ m]; (2) Meso-damage, 3–5 orders of magnitude smaller [10⁻²-10¹ m]; and (3) Micro-damage, >5 orders of magnitude smaller [<10⁻² m]. Results from studies measuring micro- and meso-fracture densities on fault perpendicular transects show that across-fault 1-D damage profiles can be simplified to fit either an exponential decay model (log-normal linear regression) (Mitchell and Faulkner, 2009), or a power law decay model (log-log linear regression) (Johri et al., 2014b; O'Hara et al., 2017; Savage and Brodsky, 2011). These quantitative studies do not address the patterns in damage heterogeneity observed in many of the datasets, and although there are many field studies of off-fault damage, it is problematic to compare datasets due to a lack of consistency in the data sampling techniques, the scales at which damage is measured (micro, meso, and macro), terminology and nomenclature, lithological and tectonic differences, and variations in analytical approach (Choi et al., 2016).

Despite heterogeneous damage distributions within fault zones having been shown to theoretically have significant effects on earthquake ruptures (e.g. Cappa, 2011), our understanding of the distribution of off-fault damage heterogeneity and how it scales with increasing fault maturity is surprisingly poor. This is in part due to little being known about the relative contributions of quasi-static and dynamically induced fractures in seismic fault zones, and how this damage evolves cumulatively in time and space. To complicate matters, with increased pressure and temperature at depth, the structure, mechanical, and hydraulic characteristics of a fault zone are subject to constant change (e.g. healing and/or sealing) during the seismic cycle as the fault evolves (e.g. Eichhubl et al., 2009; Faulkner et al., 2010; Rempe et al., 2018; Williams et al., 2017).

In this study we aim to address the data gap between qualitative and quantitative descriptions of meso-scale (10⁻²-10¹ m long faults/fractures) fault damage heterogeneity by performing a comprehensive high-resolution field study and analysis of outcrop-scale fracture patterns along the km-scale active Borrego Fault, Baja California. While most existing studies are limited to measuring damage trends on one or two fault perpendicular transects, we collected an extensive along-strike dataset of eight transects and made a mm-scale resolution 2D fracture map from a damage zone outcrop on a river bed pavement with 100% exposure. This dataset allows us to quantify the distribution of heterogeneities at decimetre to decametre scales, providing a detailed characterisation of the distribution of meso-scale damage around large seismogenic faults. The 2D damage map presented here may offer improved insights into the cumulative growth of off-fault damage, and how this feeds back into the faulting and earthquake process. This dataset also allows the critical comparison of different fracture sampling techniques, and the impact of sampling resolution/density on quantifying fault damage.