Stress, Distance, Magnitude, and Clustering Influences on the Success or Failure of an Aftershock Forecast: The 2013 M 6.6 Lushan Earthquake and Other Examples

Abstract

Immediately after the 12 May 2008 M 7.9 Wenchuan earthquake, we began calculating the static stress change on major faults surrounding the rupture zone (Parsons et al. , 2008), and Toda et al. (2008) mapped out regional Coulomb stress changes (Fig. 1). The purpose was twofold: (1) to identify the most likely locations (stress increases) of dangerous aftershocks, and (2) to conduct a prospective test of stress mapping as a rapid‐response forecast tool. The occurrence of the 20 April 2013 M 6.6 Lushan earthquake in the Longmen fault zone near Ya’an was consistent with the prospective static stress forecast; for example, press coverage in 2008 (Kuang, 2008) stated According to their results, published in Nature on 6 July, Ya’an Thrust, Xiong Po Thrust, and Xianshuihe Fault, all near Longmen Fault where the earthquake occurred, show a notable increase in static stress changes, and therefore face a high risk of aftershocks. “The 12 May earthquake … caused grievous losses, yet its legacy includes possible large aftershocks in the near future because it increased failure stress on important faults within and around the Sichuan basin,” write the authors. Figure 1. Earthquake activity ( M ≥2) following the 12 May 2008 M 7.9 Wenchuan mainshock in Sichuan Province, China. Epicenters are superimposed on static stress‐change calculations made shortly after the mainshock in 2008 on mapped faults (red and blue lines) and optimally oriented faults (Parsons et al. , 2008). It is evident that many aftershocks occurred in regions and faults with calculated stress decreases, which led Parsons et al. (2012) to conclude that the forecast was a failure. However, a formal evaluation of the post‐Wenchuan forecast performance by Parsons et al. (2012) was not favorable, as …we show two fully prospective static stress forecasts that have failed to reproduce spatial patterns of microseismicity … over a span of 1 to 4 years we …