On 3 April 2017, an Mw 6.5 earthquake occurred in central Botswana, situated inside the southern African plate. However, in a region with sparse station coverage such as Botswana, the earthquake catalog is usually largely incomplete at smaller magnitudes. Here, we develop a modified back-projection method using a correlation characteristic function, which represents energy similarity and mitigates interference due to waveform complexity in regional seismic networks, aiming to enhance the detection of small seismic events. The energy-related characteristic function exhibits robust detections for low signal-to-noise ratio (S/N) data, a capability not achievable by traditional arrival-time picking methods. Detection reliability is ensured through a dual search constraint, optimizing the effective signal count, considering nearby epicentral station signals, and accounting for seismic events clustering, which involves discarding seismic events from regions with infrequent earthquake occurrences. A total of 365 events (magnitude completeness of approximately M 2), approximately four times more than the catalog obtained by other methods, have been detected within seven days after the Botswana mainshock. Most are low S/N events that are not localized by traditional methods. The resulting earthquake sequence occurred on at least three northwest–southeast fault strands between the Kaapvaal Craton and the Limpopo Belt. Our results, when combined with other geophysical data, suggest that the Botswana earthquake mainly results from the difference in intraplate basal dragging forces at various tectonic regions driven by clockwise plate motions, which would generate extensional strains along the weak edges of rigid blocks.
Read full abstract