Item

Abstract

We constrain the source, depth, and rupture process of the Botswana earthquake of April 3, 2017, as well as its largest aftershock (5 April 2017, Mw 4.5). This earthquake is the largest recorded event (Mw 6.5) in the East African rift system since 1970, making one important case study to better understand source processes in stable continental regions. For the two events an automatic cepstrum analysis (Letort et al., 2015) is first applied on respectively 215 and 219 teleseismic records, in order to detect depth phase arrivals (pP, sP) in the P-coda. Coherent detections of depth phases for different azimuths allow us to estimate the hypocentral depths respectively at 28 and 23 km, suggesting that the events are located in the lower crust. A same cepstrum analysis is conducted on five other earthquakes with mb>4 in this area (from 2002 to 2017), and confirms a deep crustal seismicity cluster (around 20-30 km). The source mechanisms are then characterized using a joint inversion method by fitting both regional long-period surface-waves and teleseismic high-frequency body-waves. Combining regional and teleseismic data (as well as systematic comparisons between theoretical and observed regional surface-waves dispersion curves prior to the inversion) allows us to decrease epistemic uncertainties due to lack of regional data and poor knowledge about the local velocity structure. Focal mechanisms are both constrained as normal faulting with a northwest trending, and hypocentral depths are confirmed at 28 and 24 km. Finally, in order to study the mainshock rupture process, we originally apply a kymograph analysis method (an image processing method, commonly used in the field of cell biology for identifying motions of molecular motors, e.g. Mangeol et al., 2016). Here, the kymograph allows us to better identify high-frequency teleseismic P-arrivals inside the P-coda by tracking both reflected depth phase and direct P-wave arrivals radiated from secondary sources during the faulting process. Secondary P-arrivals are thus identified with a significant azimuthal variation of their arrival times (until 4s), allowing the localization of the source that generated these secondary waves. This analysis shows that the mainshock is probably a mix of at least two events, the second being 20-30 km further northwest along the fault.