Intense foreshock swarm preceding the 2019 M_W 6.5 Ambon (Seram, Indonesia) earthquake and its implication for the earthquake nucleation process

Highlights

• Template matching detects the foreshock swarm prior to the mainshock.

• The foreshocks occurred near the initial break of the mainshock.

• The foreshock has a high b-value.

• No temporal variation of b-value before the mainshock.

Abstract

We apply a template matching detection to report a swarm-like foreshock sequence concentrated near the northern edge of the strike-slip fault rupture of the 25 September 2019 M_W 6.5 Ambon earthquake in Seram, Indonesia. The template matching helps to identify small earthquakes that the routine earthquake catalog has not listed. The magnitude of completeness decreases from 3.1 (Badan Meteorologi Klimatologi dan Geofisika – BMKG catalog) to 1.7 in the improved catalog. We detect 3201 foreshocks which are tens fold the number of foreshocks in the routine earthquake catalog. We did not find a significant temporal variation in the b-value of the frequency–magnitude distribution (FMD) of the foreshock seismicity prior to initiating the dynamic rupture of the mainshock. The foreshocks around the mainshock hypocenter start to be busy since the beginning of July 2019 and continuously accelerates, with a minor variation in the seismicity rates and b-values. Although possible, we did not suggest that the stochastic cascade effect plays a role in the mainshock nucleation process. We discuss and highlight the possibility of fluid triggering to explain the high b-values of foreshock seismicity.

Introduction

On 25 September 2019 at 23:46:45 UTC (08:46:45 local time), a damaging shallow earthquake occurred near Ambon Island, Indonesia (Fig. 1). The population felt the mainshock with a maximum intensity of VI-VII MMI (Modified Mercalli Intensity) caused at least 30 casualties and destroyed hundreds of houses and infrastructures (www.bnpb.go.id). Badan Meteorologi Klimatologi dan Geofisika (BMKG) of Indonesia determined the epicenter of the earthquake at 3.43° S, 128.44° E, a depth of 16.0 km (± 2.0 km) (Fig. 1). In addition, BMKG determined the local magnitude of the mainshock of M_Lv 6.50 (± 0.24). The moment tensor product of BMKG (http://repogempa.bmkg.go.id) suggests a right-lateral strike-slip faulting mechanism with a nearly vertical dip and a moment magnitude of M_W 6.46.

The Global–Centroid-Moment-Tensor (Global CMT; www.globalcmt.org; Ekström et al., 2012) provides a focal mechanism indicating a right-lateral strike-slip mechanism (Fig. 1), with the nodal plane 1 (NP1) with strike₁ = 345°, dip₁ = 78°, and rake₁ = −174°; and nodal plane 2 (NP2) with strike₂ = 253°, dip₂ = 84°, and rake₂ = −12°. Global CMT placed the centroid of the mainshock at a depth of 12.7 km with a moment magnitude of M_W 6.5. The SCARDEC Geoscope (http://geoscope.ipgp.fr) obtained the centroid of the mainshock at a depth of 21 km, with a slightly larger magnitude (M_W 6.7). The U. S. Geological Survey (USGS; earthquake.usgs.gov) determined the hypocenter at a shallower depth, i.e., 12.3 km. The USGS also indicates that the mainshock has a strike-slip faulting mechanism. The body-wave moment tensor of USGS suggests the centroid of the mainshock positioned at a depth of 5.0 km, with a significant non-double couple (NDC) component of up to ~70%. This significant NDC percentage may indicate that the mainshock consists of complex multiple fault ruptures (e.g., Yue and Lay, 2020). Meilano et al. (2021) suggest that this earthquake rupture might dominantly consist of strike-slip motion but with thrust movement at the northern part of the fault rupture (Fig. 2).

The mainshock was followed by rich aftershocks that formed a ~ 35–km-long north-south trending cluster, with the mainshock epicenter located north of the seismicity cluster (Fig. 2, Fig. 3). The completeness magnitude of the BMKG earthquake catalog related to the Ambon mainshock is ~3.1. The aftershocks are primarily located offshore between Ambon and Haruku Islands, extended to the north in the south of Kairatu in Seram Island (Fig. 2). The largest aftershock was an M_W 5.5 event that occurred on 26 September 2019 at 00:40:00 UTC (marked as A1 in Fig. 3), near Haruku Island. Then there were two large aftershocks located in Ambon Island, an M_W 5.2 event that occurred on 10 October 2019 at 04:39:44 UTC (see A2 in Fig. 3), and an M_W 5.2 event that occurred on 12 November 2019 at 10:10:41 UTC (mark as A3 in Fig. 3). Sahara et al. (2021) suggest that this M_W 5.2 event and its nearby seismicity occurred on the NE-SW-trending reactivated fault in Ambon Island, triggered by the Ambon mainshock.

The Ambon mainshock was preceded by series of foreshocks. We define foreshocks retrospectively as the smaller-magnitudes seismicity before the mainshock that spatially and temporally in and close to the possible impending rupture of the mainshock. The magnitude of these foreshocks in the BMKG catalog ranges from ~2 to 3.5 (Fig. 4a). The spatiotemporal distribution of the seismicity surrounding the mainshock (latitude vs. time) reveals there seem to be two locations of the foreshock (Fig. 4b). The first foreshock series was located in the north near the mainshock epicenter, and the second foreshock series was situated in the south of the seismicity. The north cluster of foreshock occurred in January–February 2019, and they were off from March to June 2019. They activate again in July 2019 and accelerates toward the mainshock rupture (Fig. 4b). The foreshock in the south cluster was active only during August 2019 and did not accelerate toward the mainshock rupture (Fig. 4b). The occurrence of these two foreshock series is intriguing as they may be manifestations of the nucleation process of the damaging Ambon earthquake. However, the role of these foreshocks in the mainshock nucleation process and initiation is still not investigated and needs more information from smaller foreshocks. Understanding the preparatory phase of an intraplate strike-slip earthquake is essential for quantifying the prognostic of foreshock seismicity (e.g., Mignan, 2014).

This study performs a template matching detection by using broadband waveform retrieved from BMKG regional seismic network. This template matching algorithm utilizing BMKG broadband data has been used for some earthquake cases in Indonesia's tectonics (e.g., Sianipar et al., 2020; Passarelli et al., 2018). We scan through three months of continuous waveforms from July to September 2019 (Fig. 4). By applying waveform-based detection like this study, we enhance the temporal evolution of the foreshock series prior to the mainshock rupture. We analyze the frequency–magnitude distribution (FMD) of the new, improved catalog. We provide insights into the possible role of these foreshocks in the mainshock triggering.