Earthquake triggering in southeast Africa following the 2012 Indian Ocean earthquake

Earthquake triggering in southeast Africa following the 2012 Indian Ocean earthquake Summary In this paper we present evidence of earthquake dynamic triggering in southeast Africa. We analysed seismic waveforms recorded at 53 broad-band and short-period stations in order to identify possible increases in the rate of microearthquakes and tremor due to the passage of teleseismic waves generated by the Mw8.6 2012 Indian Ocean earthquake. We found evidence of triggered local earthquakes and no evidence of triggered tremor in the region. We assessed the statistical significance of the increase in the number of local earthquakes using β-statistics. Statistically significant dynamic triggering of local earthquakes was observed at 7 out of the 53 analysed stations. Two of these stations are located in the northeast coast of Madagascar and the other five stations are located in the Kaapvaal Craton, southern Africa. We found no evidence of dynamically triggered seismic activity in stations located near the structures of the East African Rift System. Hydrothermal activity exists close to the stations that recorded dynamic triggering, however, it also exists near the East African Rift System structures where no triggering was observed. Our results suggest that factors other than solely tectonic regime and geothermalism are needed to explain the mechanisms that underlie earthquake triggering. Africa, Earthquake interaction, forecasting and prediction, Seismicity and tectonics 1 INTRODUCTION Earthquake–earthquake interactions provide critical insights on the transfer of stress in the lithosphere and on earthquake nucleation processes (Hill & Prejean 2015; King & Devès 2015). Such interactions, of which aftershock sequences are a prime example, have been documented for long (Omori 1894; Utsu et al. 1995; Kanamori & Brodsky 2004). In this paper we focus on dynamic triggering—a significant increase of local seismic activity, including earthquakes and tremor, by the dynamic stresses imposed by the passing seismic waves generated by remote large earthquakes, which temporarily alter the local stress field (Hill & Prejean 2015). Dynamic triggering was first convincingly documented in 1992, when the seismicity rate in the western United States conspicuously increased on the wake of the Mw7.3 Landers, California, earthquake (Hill et al. 1993). Remote triggering has since been extensively documented elsewhere around the world (Brodsky & van der Elst 2014; Hill & Prejean 2015). Velasco et al. (2008) analysed seismic data recorded worldwide following 15 earthquakes of magnitude greater than 7.0 and concluded that dynamic triggering was a widespread phenomenon that took place in various tectonic environments. Although dynamic triggering has been documented in various tectonic settings, most triggered microearthquakes have been observed in extensional regimes, geothermal areas and volcanic settings (e.g. Aiken & Peng 2014; Hill & Prejean 2015). Recent studies suggest that areas of induced anthropogenic seismicity are also susceptible to remote dynamic triggering (van der Elst et al. 2013; Wang et al. 2015; Han et al. 2017). In comparison, triggered deep tectonic tremor is most commonly observed in compressive and transpressive settings: subduction zones, strike-slip faults and arc-continental collisions (e.g. Peng & Chao 2008; Peng et al. 2009; Peng & Gomberg 2010; Aiken et al. 2016). Triggered tremor typically shows a sustained increase in high-frequency seismic energy with no clear impulsive phases and is often modulated by the passing surface waves of the teleseismic event (Rubinstein et al. 2007; Peng et al. 2009). In dynamic triggering, the onset of the increase in the rate of local seismic activity typically coincides with the passage of the surface wave train, in which case it is referred to as instantaneous dynamic triggering (Hill & Prejean 2015). When the rate of local seismic activity remains abnormally high in the hours to days following the passage of teleseismic waves, it is referred to as delayed dynamic triggering (Parsons 2005; Brodsky 2006; Shelly et al. 2011; Castro et al. 2015). Although the mechanisms underlying dynamic triggering are not yet completely understood (Brodsky & van der Elst 2014), proposed mechanisms include triggering by frictional failure (Gonzalez-Huizar & Velasco 2011) and triggering through the excitation of crustal fluids (Hill & Prejean 2015). The April 11, 2012, Mw8.6 Indian Ocean earthquake (08:38:37 UTC)—the largest instrumentally recorded strike-slip earthquake—ruptured a complex sequence of 3–4 orthogonal fault segments (McGuire & Beroza 2012; Meng et al. 2012; Yue et al. 2012; Wei et al. 2013). It was followed by an Mw8.2 aftershock that occurred approximately 2 hr after the main shock. The waves generated both by the main shock and by its aftershocks triggered deep tremor and earthquakes globally (e.g. Pollitz et al. 2012; Wu et al. 2012; Tape et al. 2013; Wang et al. 2015; Bansal et al. 2016; Chao & Obara 2016). In this paper, we present the results of the first systematic search for remote triggering in southeast (SE) Africa, including the southern section of the East African Rift System (EARS). In the past, the region has been poorly monitored by seismic networks. We take advantage of the occurrence of the 2012 Indian Ocean earthquake at a time when SE Africa was well monitored by permanent and temporary seismic networks to study the effect of dynamic stresses imposed by teleseismic waves on active fault structures of SE Africa. 2 SOUTHEAST AFRICA The African continent (Fig. 1) was assembled during the Late Precambrian by the amalgamation of several Archean cratons (Craig et al. 2011). Proterozoic mobile belts, which resulted from the Pan-African tectono-thermal event, mark the boundaries between Archean cratons (Craig et al. 2011). During the Phanerozoic, Africa remained mostly a stable continent. Currently, east Africa is undergoing active continental rifting along the EARS, which separates the Nubia Plate to the west, from the Somalia Plate to the east. The EARS is characterized at the surface by a succession of adjacent basins controlled by faults, which form subsiding grabens or troughs (Chorowicz 2005). Most earthquakes in southeast Africa are associated with the EARS (Foster & Jackson 1998). Figure 1. View largeDownload slide Map of central and south Africa. The small ‘+’ mark the seismicity from 1901 to 2016 (International Seismological Centre 2016). The seismicity clearly defines the structures of the EARS. Clusters of earthquakes in South Africa result both from natural and anthropogenic causes. Blue triangles represent the stations where no significant increase in activity was found. Red triangles represent the stations where remotely triggered earthquakes were observed: two stations in the northeast coast of Madagascar and five stations around one of the South Africa most active clusters. Circles around these stations mark the possible locations of triggered earthquakes computed from S–P time differences; green and red circles mark the shortest and most distant circles for each station. The location of the Mw8.6 2012 Indian Ocean earthquake (red star) relative to the study area (black square) is shown in the top right inset map. Figure 1. View largeDownload slide Map of central and south Africa. The small ‘+’ mark the seismicity from 1901 to 2016 (International Seismological Centre 2016). The seismicity clearly defines the structures of the EARS. Clusters of earthquakes in South Africa result both from natural and anthropogenic causes. Blue triangles represent the stations where no significant increase in activity was found. Red triangles represent the stations where remotely triggered earthquakes were observed: two stations in the northeast coast of Madagascar and five stations around one of the South Africa most active clusters. Circles around these stations mark the possible locations of triggered earthquakes computed from S–P time differences; green and red circles mark the shortest and most distant circles for each station. The location of the Mw8.6 2012 Indian Ocean earthquake (red star) relative to the study area (black square) is shown in the top right inset map. The EARS can be divided in two main branches—the Eastern Branch and the Western Branch (Fig. 1). The Eastern Branch is more active in the north and spreads from the Afar region southwards until the north of Tanzania. The Western Branch separates from the Eastern Branch between Afar and Tanzania, in the Aswa shear zone. It extends from Lake Albert, Uganda, down to the Malawi rift, from where its propagation further south becomes unclear. Different hypothesis for its southward continuation have been proposed: (1) through the Urema Rift in Central Mozambique (Ebinger et al. 1987), supported by recent seismological investigations (Fonseca et al. 2014; Domingues et al. 2016); (2) a bifurcation south of Lake Tanganyika that continues through the Okavango Rift (Scholz et al. 1976); (3) a third southeastern branch that spreads offshore through the seismically active Davie Ridge in the Mozambique Channel (Mougenot et al. 1986); and (4) a diffuse plate boundary and the existence of several microplates between the Nubia and Somalia plates (Stamps et al. 2008), which considers a continuation through Madagascar, possibly through the central Madagascar Alaotra-Ankay rift valley (Kusky et al. 2010). The Eastern branch is currently in a more mature stage of the rifting process than the Western branch (Craig et al. 2011). It is also shorter and seismic activity along its extent is less common than in the Western Branch, which is seismically active along its entire length (Craig et al. 2011). In opposition, Cenozoic volcanism along the EARS is widespread in the north, especially along the eastern branch, but sparse in the south (Chorowicz 2005). Natural and anthropogenic earthquakes unrelated to the EARS are also documented in southeast Africa, particularly—but not exclusively—in South Africa. A noteworthy natural M6.5 earthquake occurred in Botswana on 2017 April 3. Anthropogenic events have been related to mining and to the impoundment of water in dams (Singh et al. 2009). Clusters of moderate magnitude seismicity in central and southwest South Africa have also been associated with geothermal activity (Singh et al. 2009). 3 DATA AND METHODS We inspected data recorded at 53 broad-band and short-period seismic stations in southeast Africa (Fig. 1), including 16 stations from a temporary deployment in Mozambique (Fonseca et al. 2014) and 37 openly available seismic stations (networks AF, GE, GT, II, IU, XV, YA). For each station, we analysed the waveforms recorded 6 hr before and after the origin time of the 2012 Indian Ocean earthquake. The detailed data analysis procedure is shown in Fig. 2, and is briefly summarized here. Figure 2. View largeDownload slide Schematic description of the data processing steps followed in this study. GCP: Great Circle Path. Figure 2. View largeDownload slide Schematic description of the data processing steps followed in this study. GCP: Great Circle Path. We first deconvolved the instrumental response and carried out all the analysis using velocity waveforms. When all three-components of ground motion recordings were available, we rotated the horizontal components to transverse and radial (Figs 3b and 4b). In order to suppress the long period energy of teleseismic waves and rather focus on the search for high-frequency local events, we applied a high-pass (≥5 Hz) filter (Figs 3a and c). When cultural noise affected the waveforms in the frequency bands relevant for the analysis, we used band-pass filters to isolate the signals of interest (Figs 4a and c). In these cases, the frequency pass-bands were chosen by visual inspection of spectrograms (Fig. 4d). Figure 3. View largeDownload slide Remote dynamic triggering at station SBV in Madagascar. (a) Envelope of the high-pass filtered (≥5 Hz) vertical velocity recorded in the 6 hr before and after the 2012 Mw8.6 Indian Ocean main shock. Blue squares mark picked events before the instantaneous triggering window and red circles picked events in the triggering windows. (b) Transverse, radial and vertical seismograms during the passage of the main shock and Mw8.2 aftershock. (c) High-pass filtered ≥5 Hz velocity waveform (same time window as in (b). (d) Spectrogram of the vertical velocity waveform (same time window as in (b). (e) Zoom of the band-passed velocity waveform (shown in c) around the time of the main shock surface waves. Figure 3. View largeDownload slide Remote dynamic triggering at station SBV in Madagascar. (a) Envelope of the high-pass filtered (≥5 Hz) vertical velocity recorded in the 6 hr before and after the 2012 Mw8.6 Indian Ocean main shock. Blue squares mark picked events before the instantaneous triggering window and red circles picked events in the triggering windows. (b) Transverse, radial and vertical seismograms during the passage of the main shock and Mw8.2 aftershock. (c) High-pass filtered ≥5 Hz velocity waveform (same time window as in (b). (d) Spectrogram of the vertical velocity waveform (same time window as in (b). (e) Zoom of the band-passed velocity waveform (shown in c) around the time of the main shock surface waves. Figure 4. View largeDownload slide Remote dynamic triggering at station LBTB in Southern Botswana. Same figure caption as for Fig. 3, except now the waveforms were band-pass filtered (2–8 Hz) due to the strong cultural noise above 8 Hz. Figure 4. View largeDownload slide Remote dynamic triggering at station LBTB in Southern Botswana. Same figure caption as for Fig. 3, except now the waveforms were band-pass filtered (2–8 Hz) due to the strong cultural noise above 8 Hz. We started by visually identifying stations that recorded possible triggered seismic activity during and after the passage of teleseismic waves. We looked both for triggered earthquakes and triggered tremor. We defined triggered earthquakes as those generating signals with clearly distinct P and S phases above the background noise level. We only searched for signals with S–P times <40 s, according to the interstation distances between analysed stations. We defined triggered tremor as generating signals above the background noise level, with longer durations, no clear phase arrivals and modulated by long-period surface waves. In the cases where we visually identified possible triggered activity, we assessed the significance of the rate of change of seismic activity using β-statistics (Aron & Hardebeck 2009). We manually identified and picked earthquakes in the 12 hr of analysed data. We then computed the β-value to compare the number of events that occurred in a time window after the arrival of teleseismic waves with the number of events that would be expected in the same time window using the background seismicity rate estimated based on the 6 hr of data recorded before the arrival of the teleseismic P-wave arrival. The β-value is calculated as follows:   \begin{equation} \beta = \frac{N_a - N(T_a/T)}{\sqrt{N(T_a/T)(1-(T_a/T))}} \end{equation} (1)where Ta is the time length of the triggering window, T is the time length of the background seismicity window plus the triggering window, and Na and N are the number of events detected in Ta and T, respectively. A statistically significant increase in the rate of seismic activity is found when β ≥ 2 (Hill & Prejean 2015). We defined two consecutive triggering windows to search for both instantaneous and delayed triggering. The instantaneous triggering window encompasses the passage of surface waves, estimated as the time between the arrival of seismic energy travelling at 5 and 2 km s−1 (Peng et al. 2009). The delayed triggering window extends from the end of the instantaneous triggering window up until 6 hr after the main shock’s origin time (Aiken & Peng 2014). The background seismicity rate was computed considering the time window extending from 6 hr before the main shock until the teleseismic P-wave arrival. For those stations where the β-statistics analysis revealed a significant (β ≥ 2) increase in seismic activity, we further investigated the robustness of triggering. To this end, we retrieved waveforms recorded 24 hr before and after the origin time of the main shock. We then identified and picked earthquakes in the 48 hr of data and performed three additional tests with the new catalogues: Significance of β > 2. We defined a moving triggering window with the length of the instantaneous triggering window sweeping along the 48 hr of data. For each second of data we compared the triggering window to the background window and calculated the respective β-value. We then assessed which sections of our catalogue presented β ≥ 2 and whether the increase in seismic activity during the passage of the surface waves was significant. The background seismicity rate was estimated from a fixed time window extending from 24 hr before the main shock until the teleseismic P-wave arrival. Dependence on background window. We randomly chose 1000 background window lengths, varying between 1 to 24 hr, and for each computed the β-value with the window starting at 1000 different times in the 24 hr before the main shock. We considered the same instantaneous triggering window as in the original β-statistics analysis (time between the arrival of seismic energy traveling at 5 and 2 km s−1). We then analysed the distribution of β-values to assess whether the increase in seismic activity was independent of the chosen background window. Dependence on triggering window. We also calculated the β-value for an increasing length of the considered triggering window. Starting with the instantaneous triggering window, we calculated β-values for triggering windows with successive additional 1 s until 24 hr after the main shock. We defined the background window starting at 24 hr before the main shock and ending at the teleseismic P-wave arrival. 4 RESULTS The visual inspection of data recorded at the 53 stations resulted in the identification of 10 stations with enhanced microearthquake activity during or after the passage of teleseismic waves. We found no evidence of triggered tremor. Most stations affected by strong cultural noise recorded high-amplitude seismic energy at frequencies above 10 Hz. In these cases, we applied a bandpass filter of 2–8 Hz in order to eliminate the cultural noise. Fig. 4(d) shows an example of a station (LBTB) affected by cultural noise that is particularly intense at 10–15 Hz. In those cases where noise was very strong in the frequency band of interest to our analysis (2–8 Hz), we did not use the station for further analysis. Table 1 shows the β-values computed for the instantaneous and delayed triggering windows. Seven out of the ten stations presented a statistically significant increase (β ≥ 2) in the rate of seismic activity (Fig. 5). Two are located in north Madagascar and the remaining five are located in the Kaapvaal craton, South Africa (Fig. 1). Figure 5. View largeDownload slide Mw8.6 2012 Indian Ocean main shock surface wave train (light blue) recorded in the seven stations that presented a statistically significant increase in the rate of seismic activity. The envelopes of the corresponding filtered waveforms are shown in black. The dynamically triggered local microearthquakes that were manually identified are marked as red dots. For stations 3434, 3901 and LBTB, we applied a band-pass filter of 2–8 Hz. For the other 4 stations we applied a high-pass filter of ≥5 Hz. Figure 5. View largeDownload slide Mw8.6 2012 Indian Ocean main shock surface wave train (light blue) recorded in the seven stations that presented a statistically significant increase in the rate of seismic activity. The envelopes of the corresponding filtered waveforms are shown in black. The dynamically triggered local microearthquakes that were manually identified are marked as red dots. For stations 3434, 3901 and LBTB, we applied a band-pass filter of 2–8 Hz. For the other 4 stations we applied a high-pass filter of ≥5 Hz. Table 1. β-values computed for the 10 stations with a visually detected increase in earthquake rate during or after the passage of the Indian Ocean earthquake teleseismic waves. Tb indicates the length of the background window, 6 or 24 hr before the main shock until the teleseismic P-wave arrival. Underlined values identify β ≥ 2, suggesting statistically significant triggering. Tb (hr)    3434  3901  3A16  BOSA  KIBK  LBTB  LAHA  MBAR  SBV  UPI  6  βi  5.03  4.52  3.55  4.01  −0.50  6.85  4.88  0.91  6.02  0.99  6  βd  3.60  1.86  3.62  1.48  1.88  2.70  2.14  0.29  2.14  0.71  24  βi  8.97  6.91  3.00  5.67    7.87  5.92    10.51    Tb (hr)    3434  3901  3A16  BOSA  KIBK  LBTB  LAHA  MBAR  SBV  UPI  6  βi  5.03  4.52  3.55  4.01  −0.50  6.85  4.88  0.91  6.02  0.99  6  βd  3.60  1.86  3.62  1.48  1.88  2.70  2.14  0.29  2.14  0.71  24  βi  8.97  6.91  3.00  5.67    7.87  5.92    10.51    Note. βi, instantaneous triggering window; βd, delayed triggering window. View Large 4.1 Triggered seismic activity 4.1.1 Madagascar We found evidence of locally triggered earthquakes in the northeast coast of Madagascar, at stations LAHA and SBV (Fig. 5). Fig. 3 shows the triggered earthquakes recorded at station SBV. Local earthquakes are recorded during the passage of Rayleigh waves of the Mw8.6 Indian Ocean earthquake and continue to be recorded even after the end of the surface wave train. The β-statistics analysis confirms the significance of the increase in seismic activity both during and after the passage of surface waves at station LAHA (Table 1). Both at station SBV and LAHA, the recorded activity is stronger during the passage of the main shock and of the Mw8.2 aftershock surface waves. At station SBV, we detected 4 microearthquakes during the passage of the main shock Rayleigh wave and 3 during the passage of the aftershock Rayleigh wave. At LAHA, we detected three microearthquakes during the passage of the main shock Rayleigh wave, two microearthquakes during the passage of the aftershock Love wave and three microearthquakes during the passage of the aftershock Rayleigh wave. The limited data set precludes a robust location of the detected microearthquakes. The S–P time differences of triggered earthquakes range between 0.5 s and 36 s. If we assume nominal crustal P- and S-wave velocities of 6 km s−1 and 3.45 km s−1, respectively, the corresponding epicentral distances range between 4 km and 265 km (Fig. 1). 4.1.2 Southern Africa We found a significant increase in the rate of recorded local earthquakes at five stations located in the Kaapvaal Craton: LBTB, southeast Botswana; BOSA, central South Africa; and 3A16, 3434 and 3901, in the border between South Africa and Mozambique, near the Lebombo monocline. The increase in seismic activity at stations LBTB (Fig. 4), 3434 and 3A16 (Fig. 6) is statistically significant both in the instantaneous and delayed triggering windows (Table 1). At stations BOSA and 3901 the increase is only significant during the passage of surface waves (Table 1). The onset of the increase in the rate of seismic activity occurs during the passage of the main shock surface waves at all stations except 3434 (Fig. 5). At this station, the increase of seismic activity in the instantaneous triggering window is due to a microearthquake recorded before the actual arrival of the surface wave train. Fig. 4 shows an example of the triggered seismic activity at station LBTB. The onset of seismic activity is detected during the passage of the Rayleigh wave at stations LBTB, 3901 and 3A16 and during the passage of the Love wave at station BOSA. The S–P times of the local events range between 1 s and 40 s, indicating epicentral distances between 9 and 325 km (Fig. 1). Figure 6. View largeDownload slide Remote dynamic triggering at station 3A16 in southern Africa. Same figure caption as for Fig. 3. Figure 6. View largeDownload slide Remote dynamic triggering at station 3A16 in southern Africa. Same figure caption as for Fig. 3. 4.2 Additional tests for triggering The additional tests performed for these stations support a significant increase in seismic activity during the passage of teleseismic surface waves: (1) When considering a longer background window of 24 hr, the β-values increased (Table 1). (2) The significance of β ≥ 2 test (Figs 7b and 8b) showed that, for most stations, parts of the catalogue in the background 24 hr also generate β-values≥2, for example, stations 3434 (maximum β of 4.52), 3A16 (3.00), BOSA (3.29), LAHA (4.00), LBTB (5.37) and SBV (3.14). Nonetheless, these β-values are smaller than those determined for the instantaneous triggering window—by a difference of at least 2—and for the window that encompasses the waves of the Mw8.2 aftershock, except for station 3A16. These results confirm a statistically significant increase in seismic activity during the passage of teleseismic waves. It further suggests that we cannot define a single β threshold for the whole study region. (3) The dependence on background window test (Figs 9a and c) showed that most determined β-values remain higher than 2 independent of background window. (4) Finally, we observed a decay of β with an increasing triggering window (Figs 7c and 8c). At most stations, a relative maximum is also observed around the time of the passage of the Mw8.2 aftershock surface waves. These results indicate that the observed increase in the rate of seismic activity is connected to the dynamic stresses imposed by the passing seismic waves of the Indian Ocean main shock and aftershock. Figure 7. View largeDownload slide Additional tests performed to validate the significance of triggering for station 3901 in southern Africa. (a) Hand-picked earthquakes in the 24 hr before and after the main shock with the respective amplitudes (P-wave PGV) and cumulative number of events (blue line). The vertical dashed grey line marks the origin time of the main shock. The two shaded windows mark the approximate surface wave window (5-2 km s−1) of the 2012 Mw8.6 Indian Ocean main shock and its Mw8.2 aftershock. (b) Results of the significance of β ≥ 2 test. Only the time periods around the passage of the main shock and aftershock result in β ≥2 values. The horizontal dashed red line marks β = 2. (c) Dependence on triggering window test, showing a decay of β-values as the triggering window length increases. The results show also a relative maximum around the time of passage of the seismic waves of the aftershock. These results suggest that the observed increase in the rate of seismic activity results from the dynamic stresses imposed by passing main shock and aftershock seismic waves. Figure 7. View largeDownload slide Additional tests performed to validate the significance of triggering for station 3901 in southern Africa. (a) Hand-picked earthquakes in the 24 hr before and after the main shock with the respective amplitudes (P-wave PGV) and cumulative number of events (blue line). The vertical dashed grey line marks the origin time of the main shock. The two shaded windows mark the approximate surface wave window (5-2 km s−1) of the 2012 Mw8.6 Indian Ocean main shock and its Mw8.2 aftershock. (b) Results of the significance of β ≥ 2 test. Only the time periods around the passage of the main shock and aftershock result in β ≥2 values. The horizontal dashed red line marks β = 2. (c) Dependence on triggering window test, showing a decay of β-values as the triggering window length increases. The results show also a relative maximum around the time of passage of the seismic waves of the aftershock. These results suggest that the observed increase in the rate of seismic activity results from the dynamic stresses imposed by passing main shock and aftershock seismic waves. Figure 8. View largeDownload slide Same caption as for Fig. 7, now for station SBV in Madagascar. (b) A β ≥2 is observed at times before the arrival of the main-shock seismic waves. However, these β-values are smaller than the ones observed during the instantaneous triggering window and during the passage of the aftershock seismic waves, confirming that the observed increase in seismic activity with the passage of teleseismic waves is robust. Figure 8. View largeDownload slide Same caption as for Fig. 7, now for station SBV in Madagascar. (b) A β ≥2 is observed at times before the arrival of the main-shock seismic waves. However, these β-values are smaller than the ones observed during the instantaneous triggering window and during the passage of the aftershock seismic waves, confirming that the observed increase in seismic activity with the passage of teleseismic waves is robust. Figure 9. View largeDownload slide Distribution of β-values determined by dependence on background window test, for station: (a) 3901 in southern Africa; (b) SBV in Madagascar. Most β-values are higher than 2 (vertical dashed red line), indicating that the observed increase in seismic activity is statistically significant. Figure 9. View largeDownload slide Distribution of β-values determined by dependence on background window test, for station: (a) 3901 in southern Africa; (b) SBV in Madagascar. Most β-values are higher than 2 (vertical dashed red line), indicating that the observed increase in seismic activity is statistically significant. 4.3 Absence of triggering along the southern East African Rift System Twenty-three of the 53 analysed stations are located near active structures of the EARS. Surprisingly, we found no evidence of triggered tremor or local earthquake activity at these stations during the passage of seismic waves generated by the Mw8.6 Indian Ocean earthquake and its Mw8.2 aftershock (Fig. S1). The visual inspection of waveforms recorded at stations KIBK and MBAR (Fig. S2) revealed a possible increase in seismic activity after the passage of surface waves, but the β-statistics analysis showed that the increase was not statistically significant (Table 1). 5 DISCUSSION AND CONCLUSION We systematically searched for dynamically triggered seismic activity in southeast Africa following the Mw8.6 2012 Indian Ocean earthquake. We found evidence of local earthquakes triggered in two different regions: the northeast coast of Madagascar and the Kaapvaal craton, central-east South Africa. No evidence was found for the triggering of seismic activity in structures associated with the East African Rift System. Two of the eleven stations in Madagascar—SBV and LAHA—recorded a statistically significant increase in earthquake rate. The two neighbouring stations are located in the northeast coast of Madagascar, in a region where some seismic activity has been previously reported (Bertil & Regnoult 1998; International Seismological Centre 2016). In particular, an mb5.0 earthquake was recorded in this region in 1992. Bertil & Regnoult (1998) identify a major reactivated fault, the Nosy-Be Antongil fracture, which is associated with current seismic activity. This fault is located approximately 100 km SW of station LAHA and 200 km S of station SBV. The region is also known for its geothermal activity (Bertil & Regnoult 1998). Because we were not able to locate the detected earthquakes, we cannot clearly associate the triggered earthquakes with structures in the region. It is intriguing to note that we found evidences of triggering in the northeast coast of Madagascar and not in other regions of Madagascar considered more seismically active, such as in the central Madagascar Alaotra-Ankai rift system (Kusky et al. 2010). This observation suggests that faults along the northeast coast of Madagascar may be more easily perturbed by transient dynamic stresses. Five stations located in the Kaapvaal Craton, central-east South Africa, recorded a significant increase in earthquake rate. Earthquake catalogues show active seismicity in this region, which has been attributed to a combination of natural and anthropogenic factors (Singh et al. 2009). Stations 3434, 3901 and 3A16 are located near the Lebombo monocline, which marks the eastern border of the Kaapvaal Craton. The region of the Lebombo monocline hosts low-level background seismic activity (Singh et al. 2009). The detected local earthquakes were observed at most in two stations and therefore we could not locate them robustly. We verified that the observed increase in the rate of seismic activity was not related with mining working hours by assessing the level of seismic activity in the same period of time in the 10 days before the main shock. We did not find a similar pattern of increase in the rate of seismic activity in the days before the main shock. We also did not find other time periods with a similar rate of seismic activity as that observed during the passage of teleseismic waves. This result may corroborate previous indications that areas of anthropogenic seismic activity are more prone to remote dynamic triggering (van der Elst et al. 2013; Wang et al. 2015; Han et al. 2017). It is interesting to note the existence of hydrothermalism close to all stations that recorded dynamic triggering. Hot springs can be found in Madagascar (Bertil & Regnoult 1998), close to the stations that detected triggered activity (Fig. 1). Singh et al. (2009) reported hydrothermalism associated with the Zebediela fault, located in the region of the 5 stations that detected dynamic triggering in the Kaapvaal craton. However, it should be noted that hydrothermalism is also common in other regions of southeast Africa, including close to structures of the EARS, where no triggering was observed. The absence of triggered earthquakes near the extensive structures of the EARS confirms that dynamic triggering is not ubiquitous along major plate-boundary fault-systems (e.g. Bansal et al. 2016). It also suggests that tectonic regime and geothermalism cannot solely explain remote dynamic triggering mechanisms. Other factors, such as local crustal stress field or pore pressure, should also be considered (Zhang et al. 2017). Fig. 10(a) shows the peak ground velocity (PGV) observed at each station during the passage of teleseismic waves. A clear relationship between surface wave PGV and dynamic triggering does not exist. However, stations where triggering was observed present a high vertical PGV in comparison to the horizontal PGV (Fig. 10b). The high vertical to horizontal PGV ratio suggests a local distortion of the wavefield, possibly due to topography, structure or site effects, which may facilitate triggering. Figure 10. View largeDownload slide (a) PGV and corresponding dynamic stress recorded on the vertical and transverse components of velocity waveforms in the study region. (b) Ratio between vertical and transverse PGV. A clear relationship between observed PGV and dynamic triggering does not exist. However, stations where triggered earthquakes were recorded appear to present a higher vertical to horizontal ratio PGV. Stations not shown in the graph present anomalous peak ground velocity values and were not considered in the analysis. The dynamic stress was estimated assuming a shear rigidity of 30 GPa and a phase velocity of 3.5 km s−1, following Aiken & Peng (2014). Figure 10. View largeDownload slide (a) PGV and corresponding dynamic stress recorded on the vertical and transverse components of velocity waveforms in the study region. (b) Ratio between vertical and transverse PGV. A clear relationship between observed PGV and dynamic triggering does not exist. However, stations where triggered earthquakes were recorded appear to present a higher vertical to horizontal ratio PGV. Stations not shown in the graph present anomalous peak ground velocity values and were not considered in the analysis. The dynamic stress was estimated assuming a shear rigidity of 30 GPa and a phase velocity of 3.5 km s−1, following Aiken & Peng (2014). Finally, our extensive β-statistic tests suggest that a β-value larger than 2 should not be taken as a universal value for determining a significant increase in the rate of seismic activity. The results also indicate that a single β threshold cannot be determined for the study region and that individual thresholds should be considered for different stations. Acknowledgements We thank the MOZART (MOZAmbique Rift Tomography) project for making available the data from the deployment in Mozambique. Seismic data used in this study are available from the Incorporated Research Institutes for Seismology (IRIS) Data Management Center (AF, doi:10.7914/SN/AF; GT, doi:10.7914/SN/GT; II, doi:10.7914/SN/II; IU, doi:10.7914/SN/IU; XV, doi:10.7914/SN/XV_2011; YA, doi:10.7914/SN/YA_2012) and the GEOFON Data Centre (GE, doi:10.14470/TR560404). The seismic catalogue used in this work was retrieved from the International Seismological Centre (2016). We thank Chenyu Li for providing the MATLAB scripts used to create Figs 3, 4 and 6. The following software packages were used in this study: Seismic Analysis Code (SAC; Goldstein et al.1998), ObsPy (Beyreuther et al. 2010) and Generic Mapping Tools (GMT; Wessel & Smith 1998). This work was supported by the Portuguese Science and Research Foundation—Fundação Portuguesa para a Ciência e Tecnologia (FCT) through project SPIDER (PTDC/GEO-FIQ/2590/2014). Zhigang Peng is partially supported by NSF grant EAR-1543399. REFERENCES Aiken C., Peng Z., 2014. Dynamic triggering of microearthquakes in three geothermal/volcanic regions of California, J. geophys. Res. , 119( 9), 6992– 7009. Google Scholar CrossRef Search ADS   Aiken C., Chao K., Gonzalez-Huizar H., Douilly R., Peng Z., Deschamps A., Calais E., Haase J.S., 2016. Exploration of remote triggering: A survey of multiple fault structures in Haiti, Earth planet. Sci. Lett. , 455, 14– 24. Google Scholar CrossRef Search ADS   Aron A., Hardebeck J.L., 2009. Seismicity rate changes along the Central California Coast due to stress changes from the 2003 M 6.5 San Simeon and 2004 M 6.0 Parkfield Earthquakes, Bull. seism. Soc. Am. , 99( 4), 2280– 2292. Google Scholar CrossRef Search ADS   Bansal A.R. Yao D. Peng Z. Sianipar D., 2016. Isolated regions of remote triggering in South/Southeast Asia following the 2012 Mw8.6 Indian Ocean earthquake, Geophys. Res. Lett. , 43( 20), 10 654–10 662. Bertil D., Regnoult J.M., 1998. Seismotectonics of Madagascar, Tectonophysics , 294( 1–2), 57– 74. Google Scholar CrossRef Search ADS   Beyreuther M., Barsch R., Krischer L., Megies T., Behr Y., Wassermann J., 2010. ObsPy: a Python toolbox for seismology, Seismol. Res. Lett. , 81( 3), 530– 533. Google Scholar CrossRef Search ADS   Brodsky E.E., 2006. Long-range triggered earthquakes that continue after the wave train passes, Geophys. Res. Lett. , 33, L15313, doi:10.1029/2006GL026605. Brodsky E.E., van der Elst N.J., 2014. The uses of dynamic earthquake triggering, Annu. Rev. Earth Planet. Sci. , 42( 1), 317– 339. Google Scholar CrossRef Search ADS   Castro R.R., González-Huízar H., Ramón Zúñiga F., Wong V.M., Velasco A.A., 2015. Delayed dynamic triggered seismicity in Northern Baja California, México caused by large and remote earthquakes, Bull. seism. Soc. Am. , 105( 4), 1825– 1835. Google Scholar CrossRef Search ADS   Chao K., Obara K., 2016. Triggered tectonic tremor in various types of fault systems of Japan following the 2012 Mw8.6 Sumatra earthquake, J. geophys. Res. , 121( 1), 170– 187. Google Scholar CrossRef Search ADS   Chorowicz J., 2005. The East African rift system, J. Afr. Earth Sci. , 43( 1–3), 379– 410. Google Scholar CrossRef Search ADS   Craig T.J., Jackson J.A., Priestley K., McKenzie D., 2011. Earthquake distribution patterns in Africa: their relationship to variations in lithospheric and geological structure, and their rheological implications, Geophys. J Int. , 185( 1), 403– 434. Google Scholar CrossRef Search ADS   Domingues A., Silveira G., Ferreira A.M., Chang S.-J., Custódio S., Fonseca J.F., 2016. Ambient noise tomography of the East African Rift in Mozambique, Geophys. J Int. , 204( 3), 1565– 1578. Google Scholar CrossRef Search ADS   Ebinger C., Rosendahl B., Reynolds D., 1987. Tectonic model of the Malaŵi rift, Africa, Tectonophysics , 141( 1), 215– 235. Google Scholar CrossRef Search ADS   Fonseca J.F.B.D.et al.  , 2014. MOZART: A seismological investigation of the east African rift in central Mozambique, Seismol. Res. Lett. , 85( 1), 108– 116. Google Scholar CrossRef Search ADS   Foster A.N., Jackson J.A., 1998. Source parameters of large African earthquakes: implications for crustal rheology and regional kinematics, Geophys. J. Int. , 134( 2), 422– 448. Google Scholar CrossRef Search ADS   Goldstein P., Dodge D., Firpo M., Ruppert S., 1998. What’s new in SAC2000? Enhanced processing and database access, Seismol. Res. Lett. , 69( 3), 202– 205. Google Scholar CrossRef Search ADS   Gonzalez-Huizar H., Velasco A.A., 2011. Dynamic triggering: Stress modeling and a case study, J. geophys. Res. , 116, B02304, doi:10.1029/2009JB007000. Google Scholar CrossRef Search ADS   Han L. Peng Z. Johnson C.W., Pollitz F.F. Li L. Wang B. Wu J. Li Q. , 2017. Shallow microearthquakes near Chongqing, China triggered by the Rayleigh waves of the 2015 M7.8 Gorkha, Nepal earthquake, Earth, Planets Sci. Lett. , 479, 231– 240. Google Scholar CrossRef Search ADS   Hill D. Prejean S., 2015. Dynamic Triggering, in Treatise on Geophysics , 2nd edn, pp. 273– 304, ed. Schubert G., Elsevier. Hill D.P.et al.  , 1993. Seismicity Remotely Triggered by the Magnitude 7.3 Landers, California, Earthquake, Science , 260( 5114), 1617– 1623. Google Scholar CrossRef Search ADS PubMed  International Seismological Centre 2016. On-line Bulletin , Int. Seismol. Cent., Thatcham, United Kingdom, Available at: http://www.isc.ac.uk. Kanamori H. Brodsky E.E., 2004. The physics of earthquakes, Rep. Prog. Phys. , 67( 8), 1429– 1496. Google Scholar CrossRef Search ADS   King G., Devès M.. 2015. Fault Interaction, Earthquake Stress Changes, and the Evolution of Seismicity, in Treatise on Geophysics , 2nd edn, pp. 243– 271, ed. Schubert G., Elsevier. Kusky T.M., Toraman E., Raharimahefa T., Rasoazanamparany C., 2010. Active tectonics of the Alaotra–Ankay Graben System, Madagascar: Possible extension of Somalian–African diffusive plate boundary?, Gondwana Res. , 18( 2–3), 274– 294. Google Scholar CrossRef Search ADS   McGuire J.J., Beroza G.C., 2012. A Rogue Earthquake Off Sumatra, Science , 336( 6085), 1118– 1119. Google Scholar CrossRef Search ADS PubMed  Meng L., Ampuero J.-P., Stock J., Duputel Z., Luo Y., Tsai V.C., 2012. Earthquake in a maze: compressional rupture branching during the 2012 Mw8.6 Sumatra earthquake, Science , 337, 724– 726. Google Scholar CrossRef Search ADS PubMed  Mougenot D., Recq M., Virlogeux P., Lepvrier C., 1986. Seaward extension of the east African rift, Nature , 321( 6070), 599– 603. Google Scholar CrossRef Search ADS   Omori F., 1894. On the aftershocks of earthquakes, J. Colloid Sci. Imp. Univ. Tokyo , 7, 111– 200. Parsons T., 2005. A hypothesis for delayed dynamic earthquake triggering, Geophys. Res. Lett. , 32, L04302, doi:10.1029/2004GL021811. Peng Z., Chao K., 2008. Non-volcanic tremor beneath the Central Range in Taiwan triggered by the 2001 Mw 7.8 Kunlun earthquake, Geophys. J Int. , 175( 2), 825– 829. Google Scholar CrossRef Search ADS   Peng Z., Gomberg J., 2010. An integrated perspective of the continuum between earthquakes and slow-slip phenomena, Nat. Geosci. , 3( 9), 599– 607. Google Scholar CrossRef Search ADS   Peng Z. Vidale J.E. Wech A.G. Nadeau R.M. Creager K.C., 2009. Remote triggering of tremor along the San Andreas Fault in central California, J. geophys. Res. , 114, B00A06, doi:10.1029/2008JB006049. Google Scholar CrossRef Search ADS   Pollitz F.F., Stein R.S., Sevilgen V., Burgmann R., 2012. The 11 April 2012 east Indian Ocean earthquake triggered large aftershocks worldwide, Nature , 490( 7419), 250– 253. Google Scholar CrossRef Search ADS PubMed  Rubinstein J.L., Vidale J.E., Gomberg J., Bodin P., Creager K.C., Malone S.D., 2007. Non-volcanic tremor driven by large transient shear stresses, Nature , 448( 7153), 579– 582. Google Scholar CrossRef Search ADS PubMed  Scholz C.H., Koczynski T.A., Hutchins D.G., 1976. Evidence for incipient rifting in southern Africa, Geophys. J. R. astr. Soc. , 44( 1), 135– 144. Google Scholar CrossRef Search ADS   Shelly D.R., Peng Z., Hill D.P., Aiken C., 2011. Triggered creep as a possible mechanism for delayed dynamic triggering of tremor and earthquakes, Nat. Geosci , 4( 6), 384– 388. Google Scholar CrossRef Search ADS   Singh M., Kijko A., Durrheim R., 2009. Seismotectonic models for South Africa: synthesis of geoscientific information, problems, and the way forward, Seismol. Res. Lett. , 80( 1), 71– 80. Google Scholar CrossRef Search ADS   Stamps D.S. Calais E. Saria E. Hartnady C. Nocquet J.-M. Ebinger C.J. Fernandes R.M., 2008. A kinematic model for the East African Rift, Geophys. Res. Lett. , 35, L05304, doi:10.1029/2007GL032781. Google Scholar CrossRef Search ADS   Tape C., West M., Silwal V., Ruppert N., 2013. Earthquake nucleation and triggering on an optimally oriented fault, Earth planet. Sci. Lett. , 363, 231– 241. Google Scholar CrossRef Search ADS   Utsu T., Ogata Y., Matsu’ura R.S., 1995. The centenary of the Omori formula for a decay law of aftershock activity, J. Phys. Earth , 43( 1), 1– 33. Google Scholar CrossRef Search ADS   van der Elst N.J., Savage H.M., Keranen K.M., Abers G.A., 2013. Enhanced remote earthquake triggering at fluid-injection sites in the midwestern United States, Science , 341( 6142), 164– 167. Google Scholar CrossRef Search ADS PubMed  Velasco A.A., Hernandez S., Parsons T., Pankow K., 2008. Global ubiquity of dynamic earthquake triggering, Nat. Geosci. , 1( 6), 375– 379. Google Scholar CrossRef Search ADS   Wang W., Meng X., Peng Z., Chen Q.-F., Liu N., 2015. Increasing background seismicity and dynamic triggering behaviors with nearby mining activities around Fangshan Pluton in Beijing, China, J. geophys. Res. , 120( 8), 5624– 5638. Google Scholar CrossRef Search ADS   Wei S., Helmberger D., Avouac J.-P., 2013. Modeling the 2012 Wharton basin earthquakes off-Sumatra: complete lithospheric failure, J. geophys. Res. , 118( 7), 3592– 3609. Google Scholar CrossRef Search ADS   Wessel P., Smith W. H.F., 1998. New, improved version of generic mapping tools released, EOS, Trans. Am. geophys. Un. , 79( 47), 579. Google Scholar CrossRef Search ADS   Wu J. Peng Z. Wang W. Gong X. Chen Q. Wu C., 2012. Comparisons of dynamic triggering near Beijing, China following recent large earthquakes in Sumatra, Geophys. Res. Lett. , 39, L21310, doi:10.1029/2012GL053515. Yue H., Lay T., Koper K.D., 2012. En echelon and orthogonal fault ruptures of the 11 April 2012 great intraplate earthquakes, Nature , 490( 7419), 245– 249. Google Scholar CrossRef Search ADS PubMed  Zhang Q., Lin G., Zhan Z., Chen X., Qin Y., Wdowinski S., 2017. Absence of remote earthquake triggering within the Coso and Salton Sea geothermal production fields, Geophys. Res. Lett. , 44( 2), 726– 733. Google Scholar CrossRef Search ADS   SUPPORTING INFORMATION Supplementary data are available at GJI online. Figure S1. The absence of triggering in 23 stations located near structures of the EARS during the passage of seismic waves generated by the Mw8.6 2012 Indian Ocean main shock. Envelope of the high-pass filtered (≥5 Hz) waveforms (black) and raw three-component waveform at TETE station (light blue). Figure S2. The absence of remote dynamic triggering at station MBAR in Uganda, EARS Western Branch. Same caption as for Fig. 3. At this station waveforms were high-pass filtered (≥5 Hz). The detected increase in activity is not statistically significant. Figure S3. The absence of remote dynamic triggering at station LODK in Kenya, EARS Eastern Branch. Same caption as for Fig. 3. At this station waveforms were high-pass filtered (≥5 Hz). Although some activity was detected during and after the surface wave train, it is similar to the activity detected before. Table S1. Details of the 53 analysed stations. Stations with strong noise in the frequency band of interest are marked with *. Stations of the MOZART temporary deployment are marked with network code (Moz). Please note: Oxford University Press is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the paper. © The Author(s) 2017. Published by Oxford University Press on behalf of The Royal Astronomical Society. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Geophysical Journal International Oxford University Press

Earthquake triggering in southeast Africa following the 2012 Indian Ocean earthquake

Loading next page...
 
/lp/ou_press/earthquake-triggering-in-southeast-africa-following-the-2012-indian-5T9uO0prvA
Publisher
The Royal Astronomical Society
Copyright
© The Author(s) 2017. Published by Oxford University Press on behalf of The Royal Astronomical Society.
ISSN
0956-540X
eISSN
1365-246X
D.O.I.
10.1093/gji/ggx462
Publisher site
See Article on Publisher Site

Abstract

Summary In this paper we present evidence of earthquake dynamic triggering in southeast Africa. We analysed seismic waveforms recorded at 53 broad-band and short-period stations in order to identify possible increases in the rate of microearthquakes and tremor due to the passage of teleseismic waves generated by the Mw8.6 2012 Indian Ocean earthquake. We found evidence of triggered local earthquakes and no evidence of triggered tremor in the region. We assessed the statistical significance of the increase in the number of local earthquakes using β-statistics. Statistically significant dynamic triggering of local earthquakes was observed at 7 out of the 53 analysed stations. Two of these stations are located in the northeast coast of Madagascar and the other five stations are located in the Kaapvaal Craton, southern Africa. We found no evidence of dynamically triggered seismic activity in stations located near the structures of the East African Rift System. Hydrothermal activity exists close to the stations that recorded dynamic triggering, however, it also exists near the East African Rift System structures where no triggering was observed. Our results suggest that factors other than solely tectonic regime and geothermalism are needed to explain the mechanisms that underlie earthquake triggering. Africa, Earthquake interaction, forecasting and prediction, Seismicity and tectonics 1 INTRODUCTION Earthquake–earthquake interactions provide critical insights on the transfer of stress in the lithosphere and on earthquake nucleation processes (Hill & Prejean 2015; King & Devès 2015). Such interactions, of which aftershock sequences are a prime example, have been documented for long (Omori 1894; Utsu et al. 1995; Kanamori & Brodsky 2004). In this paper we focus on dynamic triggering—a significant increase of local seismic activity, including earthquakes and tremor, by the dynamic stresses imposed by the passing seismic waves generated by remote large earthquakes, which temporarily alter the local stress field (Hill & Prejean 2015). Dynamic triggering was first convincingly documented in 1992, when the seismicity rate in the western United States conspicuously increased on the wake of the Mw7.3 Landers, California, earthquake (Hill et al. 1993). Remote triggering has since been extensively documented elsewhere around the world (Brodsky & van der Elst 2014; Hill & Prejean 2015). Velasco et al. (2008) analysed seismic data recorded worldwide following 15 earthquakes of magnitude greater than 7.0 and concluded that dynamic triggering was a widespread phenomenon that took place in various tectonic environments. Although dynamic triggering has been documented in various tectonic settings, most triggered microearthquakes have been observed in extensional regimes, geothermal areas and volcanic settings (e.g. Aiken & Peng 2014; Hill & Prejean 2015). Recent studies suggest that areas of induced anthropogenic seismicity are also susceptible to remote dynamic triggering (van der Elst et al. 2013; Wang et al. 2015; Han et al. 2017). In comparison, triggered deep tectonic tremor is most commonly observed in compressive and transpressive settings: subduction zones, strike-slip faults and arc-continental collisions (e.g. Peng & Chao 2008; Peng et al. 2009; Peng & Gomberg 2010; Aiken et al. 2016). Triggered tremor typically shows a sustained increase in high-frequency seismic energy with no clear impulsive phases and is often modulated by the passing surface waves of the teleseismic event (Rubinstein et al. 2007; Peng et al. 2009). In dynamic triggering, the onset of the increase in the rate of local seismic activity typically coincides with the passage of the surface wave train, in which case it is referred to as instantaneous dynamic triggering (Hill & Prejean 2015). When the rate of local seismic activity remains abnormally high in the hours to days following the passage of teleseismic waves, it is referred to as delayed dynamic triggering (Parsons 2005; Brodsky 2006; Shelly et al. 2011; Castro et al. 2015). Although the mechanisms underlying dynamic triggering are not yet completely understood (Brodsky & van der Elst 2014), proposed mechanisms include triggering by frictional failure (Gonzalez-Huizar & Velasco 2011) and triggering through the excitation of crustal fluids (Hill & Prejean 2015). The April 11, 2012, Mw8.6 Indian Ocean earthquake (08:38:37 UTC)—the largest instrumentally recorded strike-slip earthquake—ruptured a complex sequence of 3–4 orthogonal fault segments (McGuire & Beroza 2012; Meng et al. 2012; Yue et al. 2012; Wei et al. 2013). It was followed by an Mw8.2 aftershock that occurred approximately 2 hr after the main shock. The waves generated both by the main shock and by its aftershocks triggered deep tremor and earthquakes globally (e.g. Pollitz et al. 2012; Wu et al. 2012; Tape et al. 2013; Wang et al. 2015; Bansal et al. 2016; Chao & Obara 2016). In this paper, we present the results of the first systematic search for remote triggering in southeast (SE) Africa, including the southern section of the East African Rift System (EARS). In the past, the region has been poorly monitored by seismic networks. We take advantage of the occurrence of the 2012 Indian Ocean earthquake at a time when SE Africa was well monitored by permanent and temporary seismic networks to study the effect of dynamic stresses imposed by teleseismic waves on active fault structures of SE Africa. 2 SOUTHEAST AFRICA The African continent (Fig. 1) was assembled during the Late Precambrian by the amalgamation of several Archean cratons (Craig et al. 2011). Proterozoic mobile belts, which resulted from the Pan-African tectono-thermal event, mark the boundaries between Archean cratons (Craig et al. 2011). During the Phanerozoic, Africa remained mostly a stable continent. Currently, east Africa is undergoing active continental rifting along the EARS, which separates the Nubia Plate to the west, from the Somalia Plate to the east. The EARS is characterized at the surface by a succession of adjacent basins controlled by faults, which form subsiding grabens or troughs (Chorowicz 2005). Most earthquakes in southeast Africa are associated with the EARS (Foster & Jackson 1998). Figure 1. View largeDownload slide Map of central and south Africa. The small ‘+’ mark the seismicity from 1901 to 2016 (International Seismological Centre 2016). The seismicity clearly defines the structures of the EARS. Clusters of earthquakes in South Africa result both from natural and anthropogenic causes. Blue triangles represent the stations where no significant increase in activity was found. Red triangles represent the stations where remotely triggered earthquakes were observed: two stations in the northeast coast of Madagascar and five stations around one of the South Africa most active clusters. Circles around these stations mark the possible locations of triggered earthquakes computed from S–P time differences; green and red circles mark the shortest and most distant circles for each station. The location of the Mw8.6 2012 Indian Ocean earthquake (red star) relative to the study area (black square) is shown in the top right inset map. Figure 1. View largeDownload slide Map of central and south Africa. The small ‘+’ mark the seismicity from 1901 to 2016 (International Seismological Centre 2016). The seismicity clearly defines the structures of the EARS. Clusters of earthquakes in South Africa result both from natural and anthropogenic causes. Blue triangles represent the stations where no significant increase in activity was found. Red triangles represent the stations where remotely triggered earthquakes were observed: two stations in the northeast coast of Madagascar and five stations around one of the South Africa most active clusters. Circles around these stations mark the possible locations of triggered earthquakes computed from S–P time differences; green and red circles mark the shortest and most distant circles for each station. The location of the Mw8.6 2012 Indian Ocean earthquake (red star) relative to the study area (black square) is shown in the top right inset map. The EARS can be divided in two main branches—the Eastern Branch and the Western Branch (Fig. 1). The Eastern Branch is more active in the north and spreads from the Afar region southwards until the north of Tanzania. The Western Branch separates from the Eastern Branch between Afar and Tanzania, in the Aswa shear zone. It extends from Lake Albert, Uganda, down to the Malawi rift, from where its propagation further south becomes unclear. Different hypothesis for its southward continuation have been proposed: (1) through the Urema Rift in Central Mozambique (Ebinger et al. 1987), supported by recent seismological investigations (Fonseca et al. 2014; Domingues et al. 2016); (2) a bifurcation south of Lake Tanganyika that continues through the Okavango Rift (Scholz et al. 1976); (3) a third southeastern branch that spreads offshore through the seismically active Davie Ridge in the Mozambique Channel (Mougenot et al. 1986); and (4) a diffuse plate boundary and the existence of several microplates between the Nubia and Somalia plates (Stamps et al. 2008), which considers a continuation through Madagascar, possibly through the central Madagascar Alaotra-Ankay rift valley (Kusky et al. 2010). The Eastern branch is currently in a more mature stage of the rifting process than the Western branch (Craig et al. 2011). It is also shorter and seismic activity along its extent is less common than in the Western Branch, which is seismically active along its entire length (Craig et al. 2011). In opposition, Cenozoic volcanism along the EARS is widespread in the north, especially along the eastern branch, but sparse in the south (Chorowicz 2005). Natural and anthropogenic earthquakes unrelated to the EARS are also documented in southeast Africa, particularly—but not exclusively—in South Africa. A noteworthy natural M6.5 earthquake occurred in Botswana on 2017 April 3. Anthropogenic events have been related to mining and to the impoundment of water in dams (Singh et al. 2009). Clusters of moderate magnitude seismicity in central and southwest South Africa have also been associated with geothermal activity (Singh et al. 2009). 3 DATA AND METHODS We inspected data recorded at 53 broad-band and short-period seismic stations in southeast Africa (Fig. 1), including 16 stations from a temporary deployment in Mozambique (Fonseca et al. 2014) and 37 openly available seismic stations (networks AF, GE, GT, II, IU, XV, YA). For each station, we analysed the waveforms recorded 6 hr before and after the origin time of the 2012 Indian Ocean earthquake. The detailed data analysis procedure is shown in Fig. 2, and is briefly summarized here. Figure 2. View largeDownload slide Schematic description of the data processing steps followed in this study. GCP: Great Circle Path. Figure 2. View largeDownload slide Schematic description of the data processing steps followed in this study. GCP: Great Circle Path. We first deconvolved the instrumental response and carried out all the analysis using velocity waveforms. When all three-components of ground motion recordings were available, we rotated the horizontal components to transverse and radial (Figs 3b and 4b). In order to suppress the long period energy of teleseismic waves and rather focus on the search for high-frequency local events, we applied a high-pass (≥5 Hz) filter (Figs 3a and c). When cultural noise affected the waveforms in the frequency bands relevant for the analysis, we used band-pass filters to isolate the signals of interest (Figs 4a and c). In these cases, the frequency pass-bands were chosen by visual inspection of spectrograms (Fig. 4d). Figure 3. View largeDownload slide Remote dynamic triggering at station SBV in Madagascar. (a) Envelope of the high-pass filtered (≥5 Hz) vertical velocity recorded in the 6 hr before and after the 2012 Mw8.6 Indian Ocean main shock. Blue squares mark picked events before the instantaneous triggering window and red circles picked events in the triggering windows. (b) Transverse, radial and vertical seismograms during the passage of the main shock and Mw8.2 aftershock. (c) High-pass filtered ≥5 Hz velocity waveform (same time window as in (b). (d) Spectrogram of the vertical velocity waveform (same time window as in (b). (e) Zoom of the band-passed velocity waveform (shown in c) around the time of the main shock surface waves. Figure 3. View largeDownload slide Remote dynamic triggering at station SBV in Madagascar. (a) Envelope of the high-pass filtered (≥5 Hz) vertical velocity recorded in the 6 hr before and after the 2012 Mw8.6 Indian Ocean main shock. Blue squares mark picked events before the instantaneous triggering window and red circles picked events in the triggering windows. (b) Transverse, radial and vertical seismograms during the passage of the main shock and Mw8.2 aftershock. (c) High-pass filtered ≥5 Hz velocity waveform (same time window as in (b). (d) Spectrogram of the vertical velocity waveform (same time window as in (b). (e) Zoom of the band-passed velocity waveform (shown in c) around the time of the main shock surface waves. Figure 4. View largeDownload slide Remote dynamic triggering at station LBTB in Southern Botswana. Same figure caption as for Fig. 3, except now the waveforms were band-pass filtered (2–8 Hz) due to the strong cultural noise above 8 Hz. Figure 4. View largeDownload slide Remote dynamic triggering at station LBTB in Southern Botswana. Same figure caption as for Fig. 3, except now the waveforms were band-pass filtered (2–8 Hz) due to the strong cultural noise above 8 Hz. We started by visually identifying stations that recorded possible triggered seismic activity during and after the passage of teleseismic waves. We looked both for triggered earthquakes and triggered tremor. We defined triggered earthquakes as those generating signals with clearly distinct P and S phases above the background noise level. We only searched for signals with S–P times <40 s, according to the interstation distances between analysed stations. We defined triggered tremor as generating signals above the background noise level, with longer durations, no clear phase arrivals and modulated by long-period surface waves. In the cases where we visually identified possible triggered activity, we assessed the significance of the rate of change of seismic activity using β-statistics (Aron & Hardebeck 2009). We manually identified and picked earthquakes in the 12 hr of analysed data. We then computed the β-value to compare the number of events that occurred in a time window after the arrival of teleseismic waves with the number of events that would be expected in the same time window using the background seismicity rate estimated based on the 6 hr of data recorded before the arrival of the teleseismic P-wave arrival. The β-value is calculated as follows:   \begin{equation} \beta = \frac{N_a - N(T_a/T)}{\sqrt{N(T_a/T)(1-(T_a/T))}} \end{equation} (1)where Ta is the time length of the triggering window, T is the time length of the background seismicity window plus the triggering window, and Na and N are the number of events detected in Ta and T, respectively. A statistically significant increase in the rate of seismic activity is found when β ≥ 2 (Hill & Prejean 2015). We defined two consecutive triggering windows to search for both instantaneous and delayed triggering. The instantaneous triggering window encompasses the passage of surface waves, estimated as the time between the arrival of seismic energy travelling at 5 and 2 km s−1 (Peng et al. 2009). The delayed triggering window extends from the end of the instantaneous triggering window up until 6 hr after the main shock’s origin time (Aiken & Peng 2014). The background seismicity rate was computed considering the time window extending from 6 hr before the main shock until the teleseismic P-wave arrival. For those stations where the β-statistics analysis revealed a significant (β ≥ 2) increase in seismic activity, we further investigated the robustness of triggering. To this end, we retrieved waveforms recorded 24 hr before and after the origin time of the main shock. We then identified and picked earthquakes in the 48 hr of data and performed three additional tests with the new catalogues: Significance of β > 2. We defined a moving triggering window with the length of the instantaneous triggering window sweeping along the 48 hr of data. For each second of data we compared the triggering window to the background window and calculated the respective β-value. We then assessed which sections of our catalogue presented β ≥ 2 and whether the increase in seismic activity during the passage of the surface waves was significant. The background seismicity rate was estimated from a fixed time window extending from 24 hr before the main shock until the teleseismic P-wave arrival. Dependence on background window. We randomly chose 1000 background window lengths, varying between 1 to 24 hr, and for each computed the β-value with the window starting at 1000 different times in the 24 hr before the main shock. We considered the same instantaneous triggering window as in the original β-statistics analysis (time between the arrival of seismic energy traveling at 5 and 2 km s−1). We then analysed the distribution of β-values to assess whether the increase in seismic activity was independent of the chosen background window. Dependence on triggering window. We also calculated the β-value for an increasing length of the considered triggering window. Starting with the instantaneous triggering window, we calculated β-values for triggering windows with successive additional 1 s until 24 hr after the main shock. We defined the background window starting at 24 hr before the main shock and ending at the teleseismic P-wave arrival. 4 RESULTS The visual inspection of data recorded at the 53 stations resulted in the identification of 10 stations with enhanced microearthquake activity during or after the passage of teleseismic waves. We found no evidence of triggered tremor. Most stations affected by strong cultural noise recorded high-amplitude seismic energy at frequencies above 10 Hz. In these cases, we applied a bandpass filter of 2–8 Hz in order to eliminate the cultural noise. Fig. 4(d) shows an example of a station (LBTB) affected by cultural noise that is particularly intense at 10–15 Hz. In those cases where noise was very strong in the frequency band of interest to our analysis (2–8 Hz), we did not use the station for further analysis. Table 1 shows the β-values computed for the instantaneous and delayed triggering windows. Seven out of the ten stations presented a statistically significant increase (β ≥ 2) in the rate of seismic activity (Fig. 5). Two are located in north Madagascar and the remaining five are located in the Kaapvaal craton, South Africa (Fig. 1). Figure 5. View largeDownload slide Mw8.6 2012 Indian Ocean main shock surface wave train (light blue) recorded in the seven stations that presented a statistically significant increase in the rate of seismic activity. The envelopes of the corresponding filtered waveforms are shown in black. The dynamically triggered local microearthquakes that were manually identified are marked as red dots. For stations 3434, 3901 and LBTB, we applied a band-pass filter of 2–8 Hz. For the other 4 stations we applied a high-pass filter of ≥5 Hz. Figure 5. View largeDownload slide Mw8.6 2012 Indian Ocean main shock surface wave train (light blue) recorded in the seven stations that presented a statistically significant increase in the rate of seismic activity. The envelopes of the corresponding filtered waveforms are shown in black. The dynamically triggered local microearthquakes that were manually identified are marked as red dots. For stations 3434, 3901 and LBTB, we applied a band-pass filter of 2–8 Hz. For the other 4 stations we applied a high-pass filter of ≥5 Hz. Table 1. β-values computed for the 10 stations with a visually detected increase in earthquake rate during or after the passage of the Indian Ocean earthquake teleseismic waves. Tb indicates the length of the background window, 6 or 24 hr before the main shock until the teleseismic P-wave arrival. Underlined values identify β ≥ 2, suggesting statistically significant triggering. Tb (hr)    3434  3901  3A16  BOSA  KIBK  LBTB  LAHA  MBAR  SBV  UPI  6  βi  5.03  4.52  3.55  4.01  −0.50  6.85  4.88  0.91  6.02  0.99  6  βd  3.60  1.86  3.62  1.48  1.88  2.70  2.14  0.29  2.14  0.71  24  βi  8.97  6.91  3.00  5.67    7.87  5.92    10.51    Tb (hr)    3434  3901  3A16  BOSA  KIBK  LBTB  LAHA  MBAR  SBV  UPI  6  βi  5.03  4.52  3.55  4.01  −0.50  6.85  4.88  0.91  6.02  0.99  6  βd  3.60  1.86  3.62  1.48  1.88  2.70  2.14  0.29  2.14  0.71  24  βi  8.97  6.91  3.00  5.67    7.87  5.92    10.51    Note. βi, instantaneous triggering window; βd, delayed triggering window. View Large 4.1 Triggered seismic activity 4.1.1 Madagascar We found evidence of locally triggered earthquakes in the northeast coast of Madagascar, at stations LAHA and SBV (Fig. 5). Fig. 3 shows the triggered earthquakes recorded at station SBV. Local earthquakes are recorded during the passage of Rayleigh waves of the Mw8.6 Indian Ocean earthquake and continue to be recorded even after the end of the surface wave train. The β-statistics analysis confirms the significance of the increase in seismic activity both during and after the passage of surface waves at station LAHA (Table 1). Both at station SBV and LAHA, the recorded activity is stronger during the passage of the main shock and of the Mw8.2 aftershock surface waves. At station SBV, we detected 4 microearthquakes during the passage of the main shock Rayleigh wave and 3 during the passage of the aftershock Rayleigh wave. At LAHA, we detected three microearthquakes during the passage of the main shock Rayleigh wave, two microearthquakes during the passage of the aftershock Love wave and three microearthquakes during the passage of the aftershock Rayleigh wave. The limited data set precludes a robust location of the detected microearthquakes. The S–P time differences of triggered earthquakes range between 0.5 s and 36 s. If we assume nominal crustal P- and S-wave velocities of 6 km s−1 and 3.45 km s−1, respectively, the corresponding epicentral distances range between 4 km and 265 km (Fig. 1). 4.1.2 Southern Africa We found a significant increase in the rate of recorded local earthquakes at five stations located in the Kaapvaal Craton: LBTB, southeast Botswana; BOSA, central South Africa; and 3A16, 3434 and 3901, in the border between South Africa and Mozambique, near the Lebombo monocline. The increase in seismic activity at stations LBTB (Fig. 4), 3434 and 3A16 (Fig. 6) is statistically significant both in the instantaneous and delayed triggering windows (Table 1). At stations BOSA and 3901 the increase is only significant during the passage of surface waves (Table 1). The onset of the increase in the rate of seismic activity occurs during the passage of the main shock surface waves at all stations except 3434 (Fig. 5). At this station, the increase of seismic activity in the instantaneous triggering window is due to a microearthquake recorded before the actual arrival of the surface wave train. Fig. 4 shows an example of the triggered seismic activity at station LBTB. The onset of seismic activity is detected during the passage of the Rayleigh wave at stations LBTB, 3901 and 3A16 and during the passage of the Love wave at station BOSA. The S–P times of the local events range between 1 s and 40 s, indicating epicentral distances between 9 and 325 km (Fig. 1). Figure 6. View largeDownload slide Remote dynamic triggering at station 3A16 in southern Africa. Same figure caption as for Fig. 3. Figure 6. View largeDownload slide Remote dynamic triggering at station 3A16 in southern Africa. Same figure caption as for Fig. 3. 4.2 Additional tests for triggering The additional tests performed for these stations support a significant increase in seismic activity during the passage of teleseismic surface waves: (1) When considering a longer background window of 24 hr, the β-values increased (Table 1). (2) The significance of β ≥ 2 test (Figs 7b and 8b) showed that, for most stations, parts of the catalogue in the background 24 hr also generate β-values≥2, for example, stations 3434 (maximum β of 4.52), 3A16 (3.00), BOSA (3.29), LAHA (4.00), LBTB (5.37) and SBV (3.14). Nonetheless, these β-values are smaller than those determined for the instantaneous triggering window—by a difference of at least 2—and for the window that encompasses the waves of the Mw8.2 aftershock, except for station 3A16. These results confirm a statistically significant increase in seismic activity during the passage of teleseismic waves. It further suggests that we cannot define a single β threshold for the whole study region. (3) The dependence on background window test (Figs 9a and c) showed that most determined β-values remain higher than 2 independent of background window. (4) Finally, we observed a decay of β with an increasing triggering window (Figs 7c and 8c). At most stations, a relative maximum is also observed around the time of the passage of the Mw8.2 aftershock surface waves. These results indicate that the observed increase in the rate of seismic activity is connected to the dynamic stresses imposed by the passing seismic waves of the Indian Ocean main shock and aftershock. Figure 7. View largeDownload slide Additional tests performed to validate the significance of triggering for station 3901 in southern Africa. (a) Hand-picked earthquakes in the 24 hr before and after the main shock with the respective amplitudes (P-wave PGV) and cumulative number of events (blue line). The vertical dashed grey line marks the origin time of the main shock. The two shaded windows mark the approximate surface wave window (5-2 km s−1) of the 2012 Mw8.6 Indian Ocean main shock and its Mw8.2 aftershock. (b) Results of the significance of β ≥ 2 test. Only the time periods around the passage of the main shock and aftershock result in β ≥2 values. The horizontal dashed red line marks β = 2. (c) Dependence on triggering window test, showing a decay of β-values as the triggering window length increases. The results show also a relative maximum around the time of passage of the seismic waves of the aftershock. These results suggest that the observed increase in the rate of seismic activity results from the dynamic stresses imposed by passing main shock and aftershock seismic waves. Figure 7. View largeDownload slide Additional tests performed to validate the significance of triggering for station 3901 in southern Africa. (a) Hand-picked earthquakes in the 24 hr before and after the main shock with the respective amplitudes (P-wave PGV) and cumulative number of events (blue line). The vertical dashed grey line marks the origin time of the main shock. The two shaded windows mark the approximate surface wave window (5-2 km s−1) of the 2012 Mw8.6 Indian Ocean main shock and its Mw8.2 aftershock. (b) Results of the significance of β ≥ 2 test. Only the time periods around the passage of the main shock and aftershock result in β ≥2 values. The horizontal dashed red line marks β = 2. (c) Dependence on triggering window test, showing a decay of β-values as the triggering window length increases. The results show also a relative maximum around the time of passage of the seismic waves of the aftershock. These results suggest that the observed increase in the rate of seismic activity results from the dynamic stresses imposed by passing main shock and aftershock seismic waves. Figure 8. View largeDownload slide Same caption as for Fig. 7, now for station SBV in Madagascar. (b) A β ≥2 is observed at times before the arrival of the main-shock seismic waves. However, these β-values are smaller than the ones observed during the instantaneous triggering window and during the passage of the aftershock seismic waves, confirming that the observed increase in seismic activity with the passage of teleseismic waves is robust. Figure 8. View largeDownload slide Same caption as for Fig. 7, now for station SBV in Madagascar. (b) A β ≥2 is observed at times before the arrival of the main-shock seismic waves. However, these β-values are smaller than the ones observed during the instantaneous triggering window and during the passage of the aftershock seismic waves, confirming that the observed increase in seismic activity with the passage of teleseismic waves is robust. Figure 9. View largeDownload slide Distribution of β-values determined by dependence on background window test, for station: (a) 3901 in southern Africa; (b) SBV in Madagascar. Most β-values are higher than 2 (vertical dashed red line), indicating that the observed increase in seismic activity is statistically significant. Figure 9. View largeDownload slide Distribution of β-values determined by dependence on background window test, for station: (a) 3901 in southern Africa; (b) SBV in Madagascar. Most β-values are higher than 2 (vertical dashed red line), indicating that the observed increase in seismic activity is statistically significant. 4.3 Absence of triggering along the southern East African Rift System Twenty-three of the 53 analysed stations are located near active structures of the EARS. Surprisingly, we found no evidence of triggered tremor or local earthquake activity at these stations during the passage of seismic waves generated by the Mw8.6 Indian Ocean earthquake and its Mw8.2 aftershock (Fig. S1). The visual inspection of waveforms recorded at stations KIBK and MBAR (Fig. S2) revealed a possible increase in seismic activity after the passage of surface waves, but the β-statistics analysis showed that the increase was not statistically significant (Table 1). 5 DISCUSSION AND CONCLUSION We systematically searched for dynamically triggered seismic activity in southeast Africa following the Mw8.6 2012 Indian Ocean earthquake. We found evidence of local earthquakes triggered in two different regions: the northeast coast of Madagascar and the Kaapvaal craton, central-east South Africa. No evidence was found for the triggering of seismic activity in structures associated with the East African Rift System. Two of the eleven stations in Madagascar—SBV and LAHA—recorded a statistically significant increase in earthquake rate. The two neighbouring stations are located in the northeast coast of Madagascar, in a region where some seismic activity has been previously reported (Bertil & Regnoult 1998; International Seismological Centre 2016). In particular, an mb5.0 earthquake was recorded in this region in 1992. Bertil & Regnoult (1998) identify a major reactivated fault, the Nosy-Be Antongil fracture, which is associated with current seismic activity. This fault is located approximately 100 km SW of station LAHA and 200 km S of station SBV. The region is also known for its geothermal activity (Bertil & Regnoult 1998). Because we were not able to locate the detected earthquakes, we cannot clearly associate the triggered earthquakes with structures in the region. It is intriguing to note that we found evidences of triggering in the northeast coast of Madagascar and not in other regions of Madagascar considered more seismically active, such as in the central Madagascar Alaotra-Ankai rift system (Kusky et al. 2010). This observation suggests that faults along the northeast coast of Madagascar may be more easily perturbed by transient dynamic stresses. Five stations located in the Kaapvaal Craton, central-east South Africa, recorded a significant increase in earthquake rate. Earthquake catalogues show active seismicity in this region, which has been attributed to a combination of natural and anthropogenic factors (Singh et al. 2009). Stations 3434, 3901 and 3A16 are located near the Lebombo monocline, which marks the eastern border of the Kaapvaal Craton. The region of the Lebombo monocline hosts low-level background seismic activity (Singh et al. 2009). The detected local earthquakes were observed at most in two stations and therefore we could not locate them robustly. We verified that the observed increase in the rate of seismic activity was not related with mining working hours by assessing the level of seismic activity in the same period of time in the 10 days before the main shock. We did not find a similar pattern of increase in the rate of seismic activity in the days before the main shock. We also did not find other time periods with a similar rate of seismic activity as that observed during the passage of teleseismic waves. This result may corroborate previous indications that areas of anthropogenic seismic activity are more prone to remote dynamic triggering (van der Elst et al. 2013; Wang et al. 2015; Han et al. 2017). It is interesting to note the existence of hydrothermalism close to all stations that recorded dynamic triggering. Hot springs can be found in Madagascar (Bertil & Regnoult 1998), close to the stations that detected triggered activity (Fig. 1). Singh et al. (2009) reported hydrothermalism associated with the Zebediela fault, located in the region of the 5 stations that detected dynamic triggering in the Kaapvaal craton. However, it should be noted that hydrothermalism is also common in other regions of southeast Africa, including close to structures of the EARS, where no triggering was observed. The absence of triggered earthquakes near the extensive structures of the EARS confirms that dynamic triggering is not ubiquitous along major plate-boundary fault-systems (e.g. Bansal et al. 2016). It also suggests that tectonic regime and geothermalism cannot solely explain remote dynamic triggering mechanisms. Other factors, such as local crustal stress field or pore pressure, should also be considered (Zhang et al. 2017). Fig. 10(a) shows the peak ground velocity (PGV) observed at each station during the passage of teleseismic waves. A clear relationship between surface wave PGV and dynamic triggering does not exist. However, stations where triggering was observed present a high vertical PGV in comparison to the horizontal PGV (Fig. 10b). The high vertical to horizontal PGV ratio suggests a local distortion of the wavefield, possibly due to topography, structure or site effects, which may facilitate triggering. Figure 10. View largeDownload slide (a) PGV and corresponding dynamic stress recorded on the vertical and transverse components of velocity waveforms in the study region. (b) Ratio between vertical and transverse PGV. A clear relationship between observed PGV and dynamic triggering does not exist. However, stations where triggered earthquakes were recorded appear to present a higher vertical to horizontal ratio PGV. Stations not shown in the graph present anomalous peak ground velocity values and were not considered in the analysis. The dynamic stress was estimated assuming a shear rigidity of 30 GPa and a phase velocity of 3.5 km s−1, following Aiken & Peng (2014). Figure 10. View largeDownload slide (a) PGV and corresponding dynamic stress recorded on the vertical and transverse components of velocity waveforms in the study region. (b) Ratio between vertical and transverse PGV. A clear relationship between observed PGV and dynamic triggering does not exist. However, stations where triggered earthquakes were recorded appear to present a higher vertical to horizontal ratio PGV. Stations not shown in the graph present anomalous peak ground velocity values and were not considered in the analysis. The dynamic stress was estimated assuming a shear rigidity of 30 GPa and a phase velocity of 3.5 km s−1, following Aiken & Peng (2014). Finally, our extensive β-statistic tests suggest that a β-value larger than 2 should not be taken as a universal value for determining a significant increase in the rate of seismic activity. The results also indicate that a single β threshold cannot be determined for the study region and that individual thresholds should be considered for different stations. Acknowledgements We thank the MOZART (MOZAmbique Rift Tomography) project for making available the data from the deployment in Mozambique. Seismic data used in this study are available from the Incorporated Research Institutes for Seismology (IRIS) Data Management Center (AF, doi:10.7914/SN/AF; GT, doi:10.7914/SN/GT; II, doi:10.7914/SN/II; IU, doi:10.7914/SN/IU; XV, doi:10.7914/SN/XV_2011; YA, doi:10.7914/SN/YA_2012) and the GEOFON Data Centre (GE, doi:10.14470/TR560404). The seismic catalogue used in this work was retrieved from the International Seismological Centre (2016). We thank Chenyu Li for providing the MATLAB scripts used to create Figs 3, 4 and 6. The following software packages were used in this study: Seismic Analysis Code (SAC; Goldstein et al.1998), ObsPy (Beyreuther et al. 2010) and Generic Mapping Tools (GMT; Wessel & Smith 1998). This work was supported by the Portuguese Science and Research Foundation—Fundação Portuguesa para a Ciência e Tecnologia (FCT) through project SPIDER (PTDC/GEO-FIQ/2590/2014). Zhigang Peng is partially supported by NSF grant EAR-1543399. REFERENCES Aiken C., Peng Z., 2014. Dynamic triggering of microearthquakes in three geothermal/volcanic regions of California, J. geophys. Res. , 119( 9), 6992– 7009. Google Scholar CrossRef Search ADS   Aiken C., Chao K., Gonzalez-Huizar H., Douilly R., Peng Z., Deschamps A., Calais E., Haase J.S., 2016. Exploration of remote triggering: A survey of multiple fault structures in Haiti, Earth planet. Sci. Lett. , 455, 14– 24. Google Scholar CrossRef Search ADS   Aron A., Hardebeck J.L., 2009. Seismicity rate changes along the Central California Coast due to stress changes from the 2003 M 6.5 San Simeon and 2004 M 6.0 Parkfield Earthquakes, Bull. seism. Soc. Am. , 99( 4), 2280– 2292. Google Scholar CrossRef Search ADS   Bansal A.R. Yao D. Peng Z. Sianipar D., 2016. Isolated regions of remote triggering in South/Southeast Asia following the 2012 Mw8.6 Indian Ocean earthquake, Geophys. Res. Lett. , 43( 20), 10 654–10 662. Bertil D., Regnoult J.M., 1998. Seismotectonics of Madagascar, Tectonophysics , 294( 1–2), 57– 74. Google Scholar CrossRef Search ADS   Beyreuther M., Barsch R., Krischer L., Megies T., Behr Y., Wassermann J., 2010. ObsPy: a Python toolbox for seismology, Seismol. Res. Lett. , 81( 3), 530– 533. Google Scholar CrossRef Search ADS   Brodsky E.E., 2006. Long-range triggered earthquakes that continue after the wave train passes, Geophys. Res. Lett. , 33, L15313, doi:10.1029/2006GL026605. Brodsky E.E., van der Elst N.J., 2014. The uses of dynamic earthquake triggering, Annu. Rev. Earth Planet. Sci. , 42( 1), 317– 339. Google Scholar CrossRef Search ADS   Castro R.R., González-Huízar H., Ramón Zúñiga F., Wong V.M., Velasco A.A., 2015. Delayed dynamic triggered seismicity in Northern Baja California, México caused by large and remote earthquakes, Bull. seism. Soc. Am. , 105( 4), 1825– 1835. Google Scholar CrossRef Search ADS   Chao K., Obara K., 2016. Triggered tectonic tremor in various types of fault systems of Japan following the 2012 Mw8.6 Sumatra earthquake, J. geophys. Res. , 121( 1), 170– 187. Google Scholar CrossRef Search ADS   Chorowicz J., 2005. The East African rift system, J. Afr. Earth Sci. , 43( 1–3), 379– 410. Google Scholar CrossRef Search ADS   Craig T.J., Jackson J.A., Priestley K., McKenzie D., 2011. Earthquake distribution patterns in Africa: their relationship to variations in lithospheric and geological structure, and their rheological implications, Geophys. J Int. , 185( 1), 403– 434. Google Scholar CrossRef Search ADS   Domingues A., Silveira G., Ferreira A.M., Chang S.-J., Custódio S., Fonseca J.F., 2016. Ambient noise tomography of the East African Rift in Mozambique, Geophys. J Int. , 204( 3), 1565– 1578. Google Scholar CrossRef Search ADS   Ebinger C., Rosendahl B., Reynolds D., 1987. Tectonic model of the Malaŵi rift, Africa, Tectonophysics , 141( 1), 215– 235. Google Scholar CrossRef Search ADS   Fonseca J.F.B.D.et al.  , 2014. MOZART: A seismological investigation of the east African rift in central Mozambique, Seismol. Res. Lett. , 85( 1), 108– 116. Google Scholar CrossRef Search ADS   Foster A.N., Jackson J.A., 1998. Source parameters of large African earthquakes: implications for crustal rheology and regional kinematics, Geophys. J. Int. , 134( 2), 422– 448. Google Scholar CrossRef Search ADS   Goldstein P., Dodge D., Firpo M., Ruppert S., 1998. What’s new in SAC2000? Enhanced processing and database access, Seismol. Res. Lett. , 69( 3), 202– 205. Google Scholar CrossRef Search ADS   Gonzalez-Huizar H., Velasco A.A., 2011. Dynamic triggering: Stress modeling and a case study, J. geophys. Res. , 116, B02304, doi:10.1029/2009JB007000. Google Scholar CrossRef Search ADS   Han L. Peng Z. Johnson C.W., Pollitz F.F. Li L. Wang B. Wu J. Li Q. , 2017. Shallow microearthquakes near Chongqing, China triggered by the Rayleigh waves of the 2015 M7.8 Gorkha, Nepal earthquake, Earth, Planets Sci. Lett. , 479, 231– 240. Google Scholar CrossRef Search ADS   Hill D. Prejean S., 2015. Dynamic Triggering, in Treatise on Geophysics , 2nd edn, pp. 273– 304, ed. Schubert G., Elsevier. Hill D.P.et al.  , 1993. Seismicity Remotely Triggered by the Magnitude 7.3 Landers, California, Earthquake, Science , 260( 5114), 1617– 1623. Google Scholar CrossRef Search ADS PubMed  International Seismological Centre 2016. On-line Bulletin , Int. Seismol. Cent., Thatcham, United Kingdom, Available at: http://www.isc.ac.uk. Kanamori H. Brodsky E.E., 2004. The physics of earthquakes, Rep. Prog. Phys. , 67( 8), 1429– 1496. Google Scholar CrossRef Search ADS   King G., Devès M.. 2015. Fault Interaction, Earthquake Stress Changes, and the Evolution of Seismicity, in Treatise on Geophysics , 2nd edn, pp. 243– 271, ed. Schubert G., Elsevier. Kusky T.M., Toraman E., Raharimahefa T., Rasoazanamparany C., 2010. Active tectonics of the Alaotra–Ankay Graben System, Madagascar: Possible extension of Somalian–African diffusive plate boundary?, Gondwana Res. , 18( 2–3), 274– 294. Google Scholar CrossRef Search ADS   McGuire J.J., Beroza G.C., 2012. A Rogue Earthquake Off Sumatra, Science , 336( 6085), 1118– 1119. Google Scholar CrossRef Search ADS PubMed  Meng L., Ampuero J.-P., Stock J., Duputel Z., Luo Y., Tsai V.C., 2012. Earthquake in a maze: compressional rupture branching during the 2012 Mw8.6 Sumatra earthquake, Science , 337, 724– 726. Google Scholar CrossRef Search ADS PubMed  Mougenot D., Recq M., Virlogeux P., Lepvrier C., 1986. Seaward extension of the east African rift, Nature , 321( 6070), 599– 603. Google Scholar CrossRef Search ADS   Omori F., 1894. On the aftershocks of earthquakes, J. Colloid Sci. Imp. Univ. Tokyo , 7, 111– 200. Parsons T., 2005. A hypothesis for delayed dynamic earthquake triggering, Geophys. Res. Lett. , 32, L04302, doi:10.1029/2004GL021811. Peng Z., Chao K., 2008. Non-volcanic tremor beneath the Central Range in Taiwan triggered by the 2001 Mw 7.8 Kunlun earthquake, Geophys. J Int. , 175( 2), 825– 829. Google Scholar CrossRef Search ADS   Peng Z., Gomberg J., 2010. An integrated perspective of the continuum between earthquakes and slow-slip phenomena, Nat. Geosci. , 3( 9), 599– 607. Google Scholar CrossRef Search ADS   Peng Z. Vidale J.E. Wech A.G. Nadeau R.M. Creager K.C., 2009. Remote triggering of tremor along the San Andreas Fault in central California, J. geophys. Res. , 114, B00A06, doi:10.1029/2008JB006049. Google Scholar CrossRef Search ADS   Pollitz F.F., Stein R.S., Sevilgen V., Burgmann R., 2012. The 11 April 2012 east Indian Ocean earthquake triggered large aftershocks worldwide, Nature , 490( 7419), 250– 253. Google Scholar CrossRef Search ADS PubMed  Rubinstein J.L., Vidale J.E., Gomberg J., Bodin P., Creager K.C., Malone S.D., 2007. Non-volcanic tremor driven by large transient shear stresses, Nature , 448( 7153), 579– 582. Google Scholar CrossRef Search ADS PubMed  Scholz C.H., Koczynski T.A., Hutchins D.G., 1976. Evidence for incipient rifting in southern Africa, Geophys. J. R. astr. Soc. , 44( 1), 135– 144. Google Scholar CrossRef Search ADS   Shelly D.R., Peng Z., Hill D.P., Aiken C., 2011. Triggered creep as a possible mechanism for delayed dynamic triggering of tremor and earthquakes, Nat. Geosci , 4( 6), 384– 388. Google Scholar CrossRef Search ADS   Singh M., Kijko A., Durrheim R., 2009. Seismotectonic models for South Africa: synthesis of geoscientific information, problems, and the way forward, Seismol. Res. Lett. , 80( 1), 71– 80. Google Scholar CrossRef Search ADS   Stamps D.S. Calais E. Saria E. Hartnady C. Nocquet J.-M. Ebinger C.J. Fernandes R.M., 2008. A kinematic model for the East African Rift, Geophys. Res. Lett. , 35, L05304, doi:10.1029/2007GL032781. Google Scholar CrossRef Search ADS   Tape C., West M., Silwal V., Ruppert N., 2013. Earthquake nucleation and triggering on an optimally oriented fault, Earth planet. Sci. Lett. , 363, 231– 241. Google Scholar CrossRef Search ADS   Utsu T., Ogata Y., Matsu’ura R.S., 1995. The centenary of the Omori formula for a decay law of aftershock activity, J. Phys. Earth , 43( 1), 1– 33. Google Scholar CrossRef Search ADS   van der Elst N.J., Savage H.M., Keranen K.M., Abers G.A., 2013. Enhanced remote earthquake triggering at fluid-injection sites in the midwestern United States, Science , 341( 6142), 164– 167. Google Scholar CrossRef Search ADS PubMed  Velasco A.A., Hernandez S., Parsons T., Pankow K., 2008. Global ubiquity of dynamic earthquake triggering, Nat. Geosci. , 1( 6), 375– 379. Google Scholar CrossRef Search ADS   Wang W., Meng X., Peng Z., Chen Q.-F., Liu N., 2015. Increasing background seismicity and dynamic triggering behaviors with nearby mining activities around Fangshan Pluton in Beijing, China, J. geophys. Res. , 120( 8), 5624– 5638. Google Scholar CrossRef Search ADS   Wei S., Helmberger D., Avouac J.-P., 2013. Modeling the 2012 Wharton basin earthquakes off-Sumatra: complete lithospheric failure, J. geophys. Res. , 118( 7), 3592– 3609. Google Scholar CrossRef Search ADS   Wessel P., Smith W. H.F., 1998. New, improved version of generic mapping tools released, EOS, Trans. Am. geophys. Un. , 79( 47), 579. Google Scholar CrossRef Search ADS   Wu J. Peng Z. Wang W. Gong X. Chen Q. Wu C., 2012. Comparisons of dynamic triggering near Beijing, China following recent large earthquakes in Sumatra, Geophys. Res. Lett. , 39, L21310, doi:10.1029/2012GL053515. Yue H., Lay T., Koper K.D., 2012. En echelon and orthogonal fault ruptures of the 11 April 2012 great intraplate earthquakes, Nature , 490( 7419), 245– 249. Google Scholar CrossRef Search ADS PubMed  Zhang Q., Lin G., Zhan Z., Chen X., Qin Y., Wdowinski S., 2017. Absence of remote earthquake triggering within the Coso and Salton Sea geothermal production fields, Geophys. Res. Lett. , 44( 2), 726– 733. Google Scholar CrossRef Search ADS   SUPPORTING INFORMATION Supplementary data are available at GJI online. Figure S1. The absence of triggering in 23 stations located near structures of the EARS during the passage of seismic waves generated by the Mw8.6 2012 Indian Ocean main shock. Envelope of the high-pass filtered (≥5 Hz) waveforms (black) and raw three-component waveform at TETE station (light blue). Figure S2. The absence of remote dynamic triggering at station MBAR in Uganda, EARS Western Branch. Same caption as for Fig. 3. At this station waveforms were high-pass filtered (≥5 Hz). The detected increase in activity is not statistically significant. Figure S3. The absence of remote dynamic triggering at station LODK in Kenya, EARS Eastern Branch. Same caption as for Fig. 3. At this station waveforms were high-pass filtered (≥5 Hz). Although some activity was detected during and after the surface wave train, it is similar to the activity detected before. Table S1. Details of the 53 analysed stations. Stations with strong noise in the frequency band of interest are marked with *. Stations of the MOZART temporary deployment are marked with network code (Moz). Please note: Oxford University Press is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the paper. © The Author(s) 2017. Published by Oxford University Press on behalf of The Royal Astronomical Society.

Journal

Geophysical Journal InternationalOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial