Performance of the South African National Seismograph Network from October 2012 to February 2017: Spatially varying magnitude completeness

The spatially varying magnitude completeness, Mc, of the earthquake catalogue recorded by the South African National Seismograph Network (SANSN) from October 2012 to February 2017 is presented. Two methods were utilised to estimate locally observed magnitude completeness:

• Goodness-of-fit test (GFT) of the frequency-magnitude distribution (FMD) and

• mode of the non-cumulative magnitude histogram.

The GFT method only estimated a small number of locally observed magnitude completeness, Mc^obs, for areas of high seismicity and was influenced by event spatial heterogeneity. The mode method was unaffected by spatial heterogeneity and estimated more locally observed magnitude completeness, Mm^obs, but is less reliable and underestimated the magnitude completeness by Mc^obs ~ Mm^obs + 0.5 ± 0.2. Mc^obs and Mm^obs were hence combined to derive the magnitude completeness model. The predicted completeness Mc^pred map was derived with the default Bayesian magnitude completeness (BMC) prior model. The posterior completeness Mc^post map (BMC estimate) was derived by merging prior information Mc^pred with observations Mm^obs by means of Bayes theorem and correcting afterwards with 0.5 units on the magnitude scale for the difference between Mc^obs and Mm^obs. Magnitude completeness varies between ~1.7 for the interior of South Africa near the gold mines and ~2.9 at the east coast and north-eastern border. New stations are being installed since March 2017 to address these gaps. The SANSN performance is an estimated Mc ~0.3 units of magnitude higher than Mc^pred derived with the default BMC prior model.