The intraplate Mw 7 Machaze earthquake in Mozambique: Improved point source model, stress drop, and geodynamic implications
Graphical abstract
Introduction
Large continental intraplate earthquakes are infrequent and the physics governing these events is not fully understood yet (Stein, 2007). Given the significant seismic risk they pose (e.g. Mw 7.7 Bhuj and Mw 7.9 Sichuan events in 2001 and 2008 respectively) and their implications to conventional plate tectonics theory, it is imperative that we gain a greater understanding of these seismic sources. The February 22nd 2006 Mw 7 earthquake in Machaze is probably the largest event ever instrumentally recorded within continental Africa in the past century (Table 1). Due to the large magnitude and the timing of this particular event, 176 globally distributed broadband seismic stations recorded very high quality seismic waveforms from it at teleseismic distances (Fig. 1). Thus, it provides a rare and unique dataset to investigate the physics of a continental intraplate earthquake, which in turn will improve our understanding of slow deformation taking place within the East African Rift System (EARS) and in analog continental intraplate environments and potential of seismic hazard.
Seismic body waveforms can be used to extract spatiotemporal fault slip history and constrain fault-averaged macroscopic parameters describing fault mechanics and dynamics. From forward modeling experiments, Yang and Chen (2008) constrained the focal mechanism of the Machaze event and showed that its P and SH waveforms are consistent with a west dipping normal fault with a strike 175° whose dip angle is anomalously steep (76°), making it the steepest normal fault reported hitherto. This anomalous dip angle broadly agrees with surface observations (Fenton and Bommer, 2006) and results of body waveform inversion (Craig et al., 2011). A slip distribution map for this particular event has also been constructed from a joint inversion of seismic and geodetic data (Copley et al., 2012), which indicates a peak displacement of ∼4.5 m at depths of 10–15 km and a slip deficit towards the surface from about a 10 km-depth that was partially recovered in post-seismic after-slip. Observation of accelerated post-seismic deformation of about 3.5 cm/yr in the source region is consistent with this slip model (Raucoules et al., 2010). The 16 MPa stress drop estimated by Copley et al. (2012) is significantly greater than the global average of ∼6 MPa estimated for intraplate events (Allmann and Shearer, 2009).
On the one hand, some uncertainties still remain about the source parameters of the Machaze event partly because the full seismic dataset has not been exploited in some previous studies. For instance, in constructing the slip distribution map, Copley et al. (2012) used seismic waveforms from only 21 stations in the 30°–80° epicentral distance range. On the other hand, methodological improvements can be made to place stricter constrains on source parameters. For example, the duration of the Source Time Function (STF) in Yang and Chen (7.5s, [2008]) and Copley et al. (∼10.5s, [2012]) differs by up to 3s. Also, best-fitting values of the strike of the fault in point source models inverted from body waveforms vary by up to 15° (Yang and Chen, 2008, Craig et al., 2011). In addition, there is some uncertainty in the moment magnitude estimated for the Machaze earthquake. The long period GCMT (Ekström et al., 2012) estimate puts the magnitude at 7.0, whereas results of body waveform inversion suggest a magnitude of 6.9 (Craig et al., 2011).
Accurate estimates of source parameters along with stress drop, particularly of larger events, have important implications in understanding source processes (Kanamori and Allen, 1986, Houston, 1990), to interpretations of tectonics (Liu and Kanamori, 1980, Allmann and Shearer, 2009), and to strong ground motion prediction (Cotton et al., 2013). From the perspective of seismic hazard, improving the precision with which source parameters can be measured is important. In particular, Machaze event is the likely Mmax event based on historical seismicity of this region and in analog intraplate environments via space-for-time substitution. Because characteristics of the Mmax event strongly correlates with the predicted hazard level (Ida, 1973, Mueller, 2010, Anderson, 2015), further independent work is required to improve the confidence with which source parameters can be estimated.
The purpose of the current paper is two fold. First, we inverted the complete high quality azimuthally distributed teleseismic P waveform dataset in the 30°–90° epicentral distance range for a best fitting point source model to obtain stricter constraints on the focal mechanism solution, focal depth, scalar moment, and the STF. Unlike in previous work, we took in to account the effects of higher order reverberations around the source region whose influence on the amplitude and shape of observed waveforms can be significant (Hong and Fujita, 1981). For the Machaze event, it can be inferred that these effects are important because the seismic structure in the source region, as indicated by past tectonics (Salman and Abdula, 1995), might not be simple. We also explicitly compute cross correlation coefficients to assess waveform misfit between observations and synthetics, which allows for a direct comparison between our best-fitting model and previously published ones. Secondly, we use source spectra with necessary corrections to estimate the stress drop to independently verify if the previously determined estimate of 16 MPa (Copley et al., 2012) using a slip distribution map, which is significantly higher than the global intraplate average, is robust. Note that stress drop and its variability are key parameters in strong ground motion prediction (Cotton et al., 2013). Thus, accurate estimates of these parameters are particularly important for continental earthquakes, where the seismic risk can be extreme both in-situ and in analog environments. In section 2, we describe our data and methods followed by results in section 3. We discuss our results in section 4, where we consider implications of the Machaze event along with a newly discovered active structure to geodynamics in a broad sense.
Section snippets
Data and methods
We collected a total of 279 broadband vertical displacement P waveforms in the 30°–90° epicentral distance range from the Incorporated Research Institutions for Seismology Data Management Center (IRIS-DMC). This distance range is selected to minimize the interference of waves interacting with complex geologic structure in the crust and the uppermost mantle at shorter distances and near the core mantle boundary at longer distances. Our dataset was assembled from recordings of stations in both
Point source solution
Our best fitting point source solution confirms steep normal faulting during the Machaze earthquake (Fig. 6). The best-fitting fault parameters are as follows with values of auxiliary plane given in parentheses; strike = 173° (304°), dip = 73° (24°), rake = −72° (−132°). The best-fitting focal depth is 14.8 km and is consistent with previous estimates of 13–15 km (Yang and Chen, 2008, Copley et al., 2012). Our inverted scalar moment is 3.98 × 1019 N m. This value inverted from the higher
Discussion
Spatiotemporal properties of the earthquake process in plate interiors differ from those along plate boundaries. Fault networks at plate boundaries are loaded systematically from rather predictable relative plate motion, consistent with a quasi-periodic recurrence model at least in some regions (Boettcher and McGuire, 2009). On the other hand, significant earthquakes in continental interiors result from slowly deforming fault systems and tend to cluster and migrate over space and time with
Conclusions
We inverted 176 very high quality displacement P waveforms of the 2006 February 22nd Mw 7 Machaze earthquake in Mozambique for a best-fitting point source model. The focal mechanism is consistent with normal faulting. The inverted parameters were strike (173°), dip (73°), slip (−72°), scalar moment (3.98x1019 Nm), and the focal depth (14.8 km). We also inverted these waveforms for the Source Time Function with a best-fitting duration of 9.5 s and a peak moment rate of 8.29 × 1018 N m/s at
Acknowledgement
The data were downloaded from IRIS-DMC. The Figures were made with GMT (Wessel et al., 2013) and initial data processing was done with Seismic Analysis Code (Goldstein and Snoke, 2005). Earthquake catalogue data for Fig. 1 and Table 1 were obtained from the ISC catalog (ISC, 2013). We also appreciate L.V. Pinto for providing us earthquake locations given in Fig. 9. We thank V. F. Cormier, P. M. Shearer, and Y. Kaneko for their helpful comments. Finally, we also thank two anonymous reviewers and
References (57)
Physical processes that control strong ground motion
The East African Rift System
J. Afr. Earth Sci.
(2005)- et al.
The global CMT project 2004-2010: Centroid-moment tensor for 13,017 earthquakes
Phys. Earth Planet. Inter
(2012) - et al.
Modeling of depth phases and source processes of some central Aleutian earthquakes
Earth Planet. Sci. Lett.
(1981) - et al.
Mode of the strain release along the Gibbs fracture zone, Mid Atlantic Ridge
Phys. Earth Planet. Inter
(1976) - et al.
Surface displacement of the Mw 7 Mozambique earthquake (Mozambique):Complementary use of multiband InSAR and radar amplitude image correlation with elastic modeling
Remote Sens. Environ.
(2010) - et al.
Development of the Mozambique and Ruvuma sedimentary basins, offshore Mozambique
Sediment. Geol.
(1995) Slip on shallow-dipping normal faults
Geology
(2009)- et al.
Global variations of stress drop for moderate to large earthquakes
J. Geophys. Res.
(2009) A spectral theory for circular seismic sources; simple estimates of source dimension, dynamic stress drop, and radiated seismic energy
Bull. Seismol. Soc. Am.
(1980)