Going Beyond Rate Changes as the Sole Indicator for Dynamic Triggering of Earthquakes

Abstract

Remote earthquake triggering is a well-established phenomenon. Triggering is commonly identified from statistically significant increases in earthquake rate coincident with the passage of seismic energy. in establishing rate changes, short duration earthquake catalogs are commonly used, and triggered sequences are not typically analyzed within the context of background seismic activity. Using 500 mainshocks and four western USA 33-yearlong earthquake catalogs, we compare the ability of three different statistical methods to identify remote earthquake triggering. Counter to many prior studies, we find remote dynamic triggering is rare (conservatively, <2% of the time). For the mainshocks associated with remote rate increases, the spatial and temporal signatures of triggering differ. We find that a rate increase coincident in time with mainshock energy alone is insufficient to conclude that dynamic triggering occurred. to classify dynamically triggered sequences, we suggest moving away from strict statistical measurements and instead use a compatibility assessment that includes multiple factors, like spatial and temporal indicators.

Characteristics of Frequent Dynamic Triggering of Microearthquakes in Southern California

“…Intriguingly, our observed triggering rate (over 20%) is significantly larger than the triggering rate documented in Pankow and Kilb (2020), which also examined the possible triggering cases in Anza, California (SJFZ‐I). They found that remote dynamic triggering is rare (<2%) by examining 500 triggering earthquakes and four 33‐year‐long earthquake catalogs in the Western US.…”

“…One possible explanation for the higher triggering rate in this study is the magnitude of completeness (M c ) of the QTM catalog. The nominal M c of the QTM catalog is 0.3, which is almost one magnitude smaller than the M c of the catalogs analyzed in Pankow and Kilb (2020). For example, our statistical procedure would suggest that the 2012 Haida Gwaii earthquake did not dynamically trigger microearthquakes in Southern California within 24 h of its occurrence, when only using earth-FAN ET AL.…”

“…Thus, the key to identifying triggering cases is a robust characterization of seismicity rate changes. A few statistical approaches developed over the years seek to quantify the changes of seismicity rate by requiring that their statistic exceeds some threshold value(s) to be classified as a significant rate change (Aiken et al., 2018; Habermann, 1987; Matthews & Reasenberg, 1988; Pankow & Kilb, 2020). The β ‐statistic is the most widely used statistic to identify dynamic triggering cases (Delbridge et al., 2017; Marsan & Nalbant, 2005): where N a is the number of earthquakes during the time period of interest ( δt a ), and Λ is the expected number of events, which is computed from with N b as the number of events during a reference time period δt b .…”

“…Despite its popularity, using the β ‐statistic to identify dynamic triggering cases comes with a few limitations (Marsan & Nalbant, 2005; Pankow & Kilb, 2020). The statistic might not be robust when evaluating short time windows, so there can be a wide range of β for both triggering and nontriggering cases, with β ‐thresholds often larger than 2 (Cattania et al., 2017; Prejean & Hill, 2018).…”

“…In other words, with the detected seismicity, we have no evidence of dynamic triggering of earthquakes with M ≥ 0.3 in SJFZ‐I from this mainshock during these 2 h, and, if delayed triggering did occur, it must have been at least 2 h after the seismic wave passage. As we mentioned, using the 95% confidence interval as a threshold value ( β 0 ≥ β 95% ) is merely a subjective convention; it can and sometimes should be higher significance levels (Pankow & Kilb, 2020).…”