Coseismic slip gradient at the western terminus of the 1920 Haiyuan Mw 7.9 earthquake

Abstract

Stepovers in strike-slip fault systems play important roles in controlling the propagation of earthquake ruptures, depending partially on the competition between the stepover width and the coseismic slip gradient approaching the step. The 1920 Mw 7.9 Haiyuan earthquake is the most recent major earthquake that has occurred along the Haiyuan fault. The earthquake rupture broke through multiple stepovers along the fault and finally ended at the 4 km-wide Jingtai pull-apart basin (releasing stepover) at the western end. To understand the process of this termination of the Haiyuan earthquake rupture, we conducted detailed mapping of the surface rupture geometry, coseismic slip measurement and slip gradient calculation in the vicinity of the endpoint based on the 0.2 m-resolution SfM-derived DEM along the 20 km section east of the Jingtai pull-apart basin. Combining coseismic slip measurements from this study and published slip data from fieldwork in the 1980s, we calculated slip gradients of 84–160 cm/km as the rupture approached the Jingtai releasing stepover. These values are high for the 4 km-wide Jingtai releasing stepover compared to those from a dataset of worldwide historical earthquakes compiled by Elliott et al. (2009). The high values imply that the rupture could have a relatively high likelihood of breaching through the Jingtai releasing stepover. Alternatively, the slip gradient may be overestimated. Detailed mapping and field investigation, however, show that the rupture may extend ~1.86 km farther west of the end location indicated in previous studies. We acquired the new slip gradient with lower value of 50–82 cm/km, which is still considerably high compared to dataset of Elliott et al. (2009). We speculate that the slip gradient could be larger variation along the stepover boundary to stop fault slip in different geologic setting. Another factor control slip termination is related to fault properties. Creep along the southern boundary fault of the Jingtai stepover, a velocity-strengthening region, may also have played a role in stopping the rupture at this location. Our observations further indicate that the increasing LOS velocity of InSAR data within the Jingtai stepover is probably related to fault behavior rather than a nontectonic signal of subsidence.

Introduction

Fault geometrical complexities such as bends, branches and stepovers are common in strike-slip fault systems, separating the faults into multiple segments (King and Nabelek, 1985; Sibson, 1985). Faults comprising several disjointed segments can behave as distinct faults rupturing separately in individual earthquakes (e.g., Schwartz and Coppersmith, 1984; Wesnousky, 1994; Zhang et al., 1999) or coalesce to rupture simultaneously in large, multifault events (Elliott et al., 2015；; Hamling et al., 2017; Liu-Zeng et al., 2009; Sieh et al., 1993; Wesnousky, 1988, 2006；; Zhang et al., 1999). Earthquake ruptures often stop at geometrical complexities on faults (Wesnousky, 1988, Wesnousky, 2006). Wesnousky’s (2006) compilation of worldwide historical strike-slip earthquake rupture maps indicates that stepovers play a crucial role in controlling rupture length, constituting 2/3 of historical rupture endpoints, and there is a limiting dimension of fault stepover (3–4 km). An analysis of mapped surface ruptures for relationships between geometrical discontinuities in fault traces and earthquake rupture extents by Biasi and Wesnousky (2016) shows that ~90% of ruptures have at least one end at a mappable discontinuity, either a fault end or a step of 1 km or greater, especially along strike-slip faults. Harris and Day's (1993) numerical simulation also shows that a strike-slip rupture is unlikely to jump a stepover wider than 5 km. The basic controls on whether a rupture will stop at a step, however, are also related to the slip-behavior as it approaches the stepover (Elliott et al., 2009； Liu and Duan, 2016；; Oglesby, 2008). It is suggested that earthquake ruptures can jump through stepovers when their slip near the fault segment tip decreases rapidly and cannot propagate through fault stepovers when their slip decreases gradually, according to the compilation and comparative analysis of slip distributions from Elliott et al. (2009).

The 1920 Mw 7.9 Haiyuan earthquake was the latest major earthquake on the Haiyuan fault, forming a ~237-km-long surface rupture (Deng et al., 1986; IGCEA and NBCEA, 1990). This rupture broke through multiple stepovers along the fault while stopping at the ~4 km-wide Jingtai pull-apart basin at the western endpoint (Deng et al., 1986; IGCEA and NBCEA, 1990). However, why the surface rupture could break through multiple stepovers along the fault while stopping at the special structure of the left-stepping releasing Jingtai basin remains unclear. Therefore, a study on the mechanism of rupture cessation at the western stepover during the 1920 Haiyuan earthquake could provide important and valuable information for our understanding of the dynamic controls on rupture propagation and arrest.

To understand the interactions of earthquake ruptures and fault complexity, we need a detailed map of rupture geometry, coseismic slip, and its gradient along strike in the vicinity of the rupture endpoint (Wesnousky, 2006; Elliott et al., 2009). The quantification of slip data is usually based on reconstructing offset landforms along a fault trace, such as deflected stream channels, alluvial fans, and offset ridgelines (Sieh, 1978; McGill and Sieh, 1991). For the 1920 Haiyuan earthquake, such data were collected through field investigation in the 1980s (Deng et al., 1986; IGCEA and NBCEA, 1990). This early work provides valuable data but also poses some difficulties. The investigation was conducted 60 years after the earthquake and before the era of high-resolution satellite images and topographic maps (IGCEA and NBCEA, 1990). The surface rupture map was simplified, and offsets were measured with an old-fashioned simple measuring tape; thus, the numbers lack precise coordinates for coseismic slip measurements to be checked with reliability and uncertainty (IGCEA and NBCEA, 1990). Recent advances in remotely sensed high-resolution topography and optical imaging have improved upon this approach; for example, light detection and ranging (LiDAR) systems can acquire high-resolution topographic data, enabling the detailed characterization of landforms and measurements of small displacements (e.g., Arrowsmith and Zielke, 2009; Behr et al., 2010; Chen et al., 2018; Frankel and Dolan, 2007; Liu et al., 2013; Oskin et al., 2012; Ren et al., 2016; Salisbury et al., 2012; Zielke et al., 2010; 2012; Zielke and Arrowsmith, 2012).

In this study, we present slip measurements based on a high-resolution digital elevation model (DEM) near the western end of the 1920 Haiyuan earthquake rupture. We calculate the coseismic slip gradient as the rupture approached the Jingtai releasing stepover. This value is compared with the dataset of worldwide historical earthquakes compiled by Elliott et al. (2009). We also investigate the possible westward extension section of the surface rupture of the 1920 Mw 7.9 Haiyuan earthquake. In addition, we compare the section with discontinuities in the line-of-sight (LOS) velocity field revealed by synthetic aperture radar interferometry (InSAR) on the northern boundary of the Jingtai basin with the extension segment of the Haiyuan fault to discuss additional factors that stopped the earthquake rupture as it approached the releasing stepover.