Calibration of an ML scale for South Africa using tectonic earthquake ...
Calibration of an ML scale for South Africa using tectonic earthquake ...
Calibration of an ML scale for South Africa using tectonic earthquake ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>Calibration</strong> <strong>of</strong> <strong>an</strong> M L <strong>scale</strong> <strong>for</strong> <strong>South</strong> <strong>Africa</strong><br />
<strong>using</strong> <strong>tectonic</strong> <strong>earthquake</strong> data recorded by<br />
the <strong>South</strong> Afric<strong>an</strong> National Seismograph<br />
Network: 2006 to 2009<br />
I<strong>an</strong> Saunders, Lars Ottemöller, Martin<br />
B. C. Br<strong>an</strong>dt & Christ<strong>of</strong>fel J. S. Fourie<br />
Journal <strong>of</strong> Seismology<br />
ISSN 1383-4649<br />
J Seismol<br />
DOI 10.1007/s10950-012-9329-0<br />
1 23
Your article is protected by copyright <strong>an</strong>d<br />
all rights are held exclusively by Springer<br />
Science+Business Media B.V.. This e-<strong>of</strong>fprint<br />
is <strong>for</strong> personal use only <strong>an</strong>d shall not be self-<br />
archived in electronic repositories. If you<br />
wish to self-archive your work, please use the<br />
accepted author’s version <strong>for</strong> posting to your<br />
own website or your institution’s repository.<br />
You may further deposit the accepted author’s<br />
version on a funder’s repository at a funder’s<br />
request, provided it is not made publicly<br />
available until 12 months after publication.<br />
1 23
J Seismol<br />
DOI 10.1007/s10950-012-9329-0<br />
ORIGINAL ARTICLE<br />
<strong>Calibration</strong> <strong>of</strong> <strong>an</strong> M L <strong>scale</strong> <strong>for</strong> <strong>South</strong> <strong>Africa</strong> <strong>using</strong> <strong>tectonic</strong><br />
<strong>earthquake</strong> data recorded by the <strong>South</strong> Afric<strong>an</strong> National<br />
Seismograph Network: 2006 to 2009<br />
I<strong>an</strong> Saunders & Lars Ottemöller &<br />
Martin B. C. Br<strong>an</strong>dt & Christ<strong>of</strong>fel J. S. Fourie<br />
Received: 18 April 2011 /Accepted: 21 August 2012<br />
# Springer Science+Business Media B.V. 2012<br />
Abstract A relation to determine local magnitude<br />
(<strong>ML</strong>) based on the original Richter definition is empirically<br />
derived from synthetic Wood–Anderson seismograms<br />
recorded by the <strong>South</strong> Afric<strong>an</strong> National<br />
Seismograph Network. In total, 263 <strong>earthquake</strong>s in<br />
the dist<strong>an</strong>ce r<strong>an</strong>ge 10 to 1,000 km, representing<br />
1,681 trace amplitudes measured in n<strong>an</strong>ometers from<br />
synthesized Wood–Anderson records on the vertical<br />
ch<strong>an</strong>nel were considered to derive <strong>an</strong> attenuation relation<br />
appropriate <strong>for</strong> <strong>South</strong> <strong>Africa</strong> through multiple<br />
regression <strong>an</strong>alysis. Additionally, station corrections<br />
were determined <strong>for</strong> 26 stations during the regression<br />
<strong>an</strong>alysis resulting in values r<strong>an</strong>ging between −0.31 <strong>an</strong>d<br />
I. Saunders (*) : M. B. C. Br<strong>an</strong>dt<br />
Council <strong>for</strong> Geoscience,<br />
Private Bag X112, Silverton,<br />
Pretoria, <strong>South</strong> <strong>Africa</strong> 0001<br />
e-mail: i<strong>an</strong>s@geoscience.org.za<br />
M. B. C. Br<strong>an</strong>dt<br />
email: martinb@geoscience.org.za<br />
L. Ottemöller<br />
Department <strong>of</strong> Earth Science, University <strong>of</strong> Bergen,<br />
Allegt 41, 5007, Norway<br />
e-mail: lars.ottmoller@geo.uib.no<br />
C. J. S. Fourie<br />
Environmental, Water <strong>an</strong>d Earth Science Department,<br />
Faculty <strong>of</strong> Science, Tshw<strong>an</strong>e University <strong>of</strong> Technology,<br />
Private Bag X460,<br />
Pretoria, <strong>South</strong> <strong>Africa</strong> 0001<br />
e-mail: fouriecjs@tut.ac.za<br />
Author's personal copy<br />
0.50. The most appropriate <strong>ML</strong> <strong>scale</strong> <strong>for</strong> <strong>South</strong> <strong>Africa</strong><br />
from this study satisfies the equation:<br />
<strong>ML</strong> ¼ log 10ðAÞþ1:149 log 10ðRÞþ0:00063R þ 2:04 S<br />
The <strong>an</strong>elastic attenuation term derived from this<br />
study indicates that ground motion attenuation is signific<strong>an</strong>tly<br />
different from <strong>South</strong>ern Cali<strong>for</strong>nia but comparable<br />
with stable continental regions.<br />
Keywords Local magnitude . <strong>South</strong> <strong>Africa</strong> .<br />
Attenuation relation . Stable continental region .<br />
Station corrections<br />
1 Introduction<br />
The use <strong>of</strong> <strong>earthquake</strong> magnitude <strong>scale</strong>s to express the<br />
physical size or energy released by <strong>earthquake</strong>s is<br />
routine practice at seismological observatories <strong>an</strong>d is<br />
based on qu<strong>an</strong>tities that are measured from wave<strong>for</strong>m<br />
recordings. In general, magnitude <strong>scale</strong>s are dist<strong>an</strong>cedependent<br />
<strong>an</strong>d are only valid <strong>for</strong> a specific wave type<br />
recorded in a narrow frequency b<strong>an</strong>d <strong>an</strong>d <strong>scale</strong> differently<br />
with seismic energy <strong>an</strong>d moment. The most<br />
popular <strong>an</strong>d widely used magnitude <strong>scale</strong> <strong>for</strong> expressing<br />
the size <strong>of</strong> <strong>an</strong> <strong>earthquake</strong> at local <strong>an</strong>d regional<br />
dist<strong>an</strong>ces remains the local magnitude (<strong>ML</strong>) <strong>scale</strong>,<br />
originally defined by Richter (1935).<br />
This Richter <strong>scale</strong> is <strong>of</strong> engineering interest, as emphasized<br />
by Borm<strong>an</strong>n (2002), where the import<strong>an</strong>ce <strong>of</strong> a
calibrated M L <strong>scale</strong> in seismic hazard studies was<br />
highlighted since the frequency b<strong>an</strong>d <strong>of</strong> the St<strong>an</strong>dard<br />
Wood–Anderson (SWA) seismograph (~0.8–10 Hz) is<br />
close to the reson<strong>an</strong>t frequencies <strong>of</strong> m<strong>an</strong>y engineering<br />
structures. The attenuation curve derived from the <strong>ML</strong><br />
<strong>scale</strong> c<strong>an</strong> be used in the risk assessment <strong>of</strong> these<br />
structures.<br />
The practice <strong>of</strong> determining <strong>ML</strong> in <strong>South</strong> <strong>Africa</strong><br />
during the period 1971 to 1997 is described by<br />
Fernández (1973, 1977, 1980, 1991, <strong>an</strong>d 1993).<br />
Maximum peak-to-peak amplitudes in millimeters<br />
<strong>of</strong> the S-Lg wavecomplexrecordedonthevertical<br />
ch<strong>an</strong>nel <strong>of</strong> short-period <strong>an</strong>alogue seismograms<br />
were measured <strong>an</strong>d then reduced to <strong>ML</strong> values<br />
<strong>using</strong> the magnitude–dist<strong>an</strong>ce attenuation curves<br />
derived <strong>for</strong> <strong>South</strong>ern Cali<strong>for</strong>nia (Richter 1935),<br />
modified <strong>for</strong> <strong>South</strong> Afric<strong>an</strong> conditions, through<br />
the relation:<br />
<strong>ML</strong> ¼ log10ðAÞ log10ðA0ÞþS ð1Þ<br />
where A is the maximum amplitude, –log10A0 is<br />
the dist<strong>an</strong>ce correction term, <strong>an</strong>d S is a station<br />
correction term.<br />
The <strong>an</strong>alogue instrumentation <strong>of</strong> the <strong>South</strong> Afric<strong>an</strong><br />
National Seismograph Network (SANSN) differed signific<strong>an</strong>tly<br />
from the SWA seismographs in Cali<strong>for</strong>nia,<br />
specifically the gain factors at 1 Hz, viz. 1,300 <strong>for</strong> the<br />
SWA (Anderson <strong>an</strong>d Wood 1925) <strong>an</strong>d 50,000 <strong>for</strong> the<br />
SANSN instrumentation, respectively (Fernández 1993).<br />
Fernández argued that the conversion <strong>of</strong> the<br />
SANSN seismograms to simulate SWA torsion seismographs<br />
was impractical since the SANSN consisted<br />
mostly <strong>of</strong> vertical records as opposed to the horizontal<br />
ch<strong>an</strong>nels <strong>of</strong> the SWA seismographs used to define the<br />
Richter magnitude. Hence, rather th<strong>an</strong> simulating<br />
SWA seismograms, Fernández used the ratio <strong>of</strong> magnification<br />
<strong>of</strong> the SANSN seismographs to that <strong>of</strong> the<br />
SWA seismograph to modify the correction factor (A0)<br />
<strong>for</strong> <strong>South</strong> <strong>Africa</strong>. In practice, readings from all observing<br />
stations were averaged after adjustment with<br />
station-specific corrections (Fernández 1993) to obtain<br />
the <strong>ML</strong> value.<br />
The question <strong>of</strong> <strong>an</strong> equivalent <strong>ML</strong> relation <strong>for</strong> <strong>South</strong><br />
<strong>Africa</strong> was considered by Br<strong>an</strong>dt (1997) withthe<br />
introduction <strong>of</strong> the SEISAN <strong>earthquake</strong> <strong>an</strong>alysis s<strong>of</strong>tware<br />
(Havskov <strong>an</strong>d Ottemöller 2005) during March<br />
1997 <strong>for</strong> routine data <strong>an</strong>alysis. It was concluded that<br />
the relation <strong>of</strong> Hutton <strong>an</strong>d Boore (1987) was the most<br />
relev<strong>an</strong>t since it compared well with the modified<br />
Author's personal copy<br />
J Seismol<br />
Richter (1935) relation introduced by Fernández<br />
(1993) to determine <strong>ML</strong> in <strong>South</strong> <strong>Africa</strong>.<br />
A preliminary study with the MAG program <strong>of</strong> the<br />
SEISAN Earthquake Analysis S<strong>of</strong>tware (Havskov<br />
<strong>an</strong>d Ottemöller 2005) to test coefficients <strong>for</strong> the<br />
<strong>ML</strong> <strong>scale</strong> in <strong>South</strong> <strong>Africa</strong> was undertaken during<br />
2008 (Saunders et al. 2008). The derived coefficients<br />
hinted that the geometrical spreading coefficient<br />
in <strong>South</strong> <strong>Africa</strong> is similar to other parts <strong>of</strong><br />
the world while the <strong>an</strong>elastic attenuation value is<br />
signific<strong>an</strong>tly lower th<strong>an</strong> that <strong>of</strong> the strongly attenuating<br />
crust <strong>of</strong> <strong>South</strong>ern Cali<strong>for</strong>nia (Hutton <strong>an</strong>d<br />
Boore 1987), indicating a high Q-value <strong>for</strong> the<br />
crystalline Archae<strong>an</strong> <strong>an</strong>d Mesozoic crust in <strong>South</strong><br />
<strong>Africa</strong>. The question <strong>of</strong> <strong>South</strong> <strong>Africa</strong> being located<br />
in a stable continental region (SCR) is debatable<br />
since the northeastern part <strong>of</strong> the country is in<br />
close proximity to the East Afric<strong>an</strong> Rift System.<br />
However, <strong>for</strong> practical purposes, we followed the<br />
criteria <strong>of</strong> the Electric Power Research Institute<br />
(EPRI) Report (1994) <strong>for</strong> delineating SCRs since<br />
<strong>Africa</strong> is included in the nine major SCRs identified<br />
in the report. Thus, the authors tentatively<br />
accept that <strong>South</strong> <strong>Africa</strong> meets the norms <strong>of</strong> SCRs<br />
being older, with lower crustal temperatures <strong>an</strong>d a<br />
low level <strong>of</strong> seismicity.<br />
Thus, since high Q values are expected in SCRs<br />
(Kvamme <strong>an</strong>d Havskov 1989), we expected to have<br />
geometrical spreading <strong>an</strong>d <strong>an</strong>elastic attenuation<br />
coefficients similar to other regions classified as<br />
being located in SCRs. Table 1 provides selected<br />
geometrical spreading <strong>an</strong>d <strong>an</strong>elastic attenuation<br />
coefficients values determined <strong>for</strong> other regions<br />
<strong>of</strong> the world.<br />
The primary reason <strong>for</strong> continuing with the <strong>ML</strong><br />
<strong>scale</strong> <strong>for</strong> qu<strong>an</strong>tifying <strong>earthquake</strong> sources in <strong>South</strong><br />
<strong>Africa</strong> is to maintain continuity <strong>of</strong> the magnitude <strong>scale</strong><br />
since the catalogue <strong>of</strong> instrumentally determined magnitudes<br />
dates back to 1971 based on maximumamplitude<br />
readings (Wright <strong>an</strong>d Fernández 2003).<br />
Secondly, <strong>South</strong> <strong>Africa</strong> measures no more that<br />
2,000 km in either the north–south or east–west direction.<br />
This is within the dist<strong>an</strong>ce limitations <strong>of</strong> the <strong>ML</strong><br />
<strong>scale</strong> while the country is additionally considered to be<br />
located in a SCR (EPRI 1994) with a low level <strong>of</strong><br />
seismic activity.<br />
On average, 11 <strong>earthquake</strong>s per <strong>an</strong>num are recorded<br />
in <strong>South</strong> <strong>Africa</strong> with M L≥4.0 (statistics based on the<br />
time period 1990 to 2009). It should be noted that this
J Seismol<br />
Table 1 Geometrical spreading<br />
<strong>an</strong>d <strong>an</strong>elastic attenuation values<br />
determined <strong>for</strong> other regions<br />
(from Havskov <strong>an</strong>d Ottemöller<br />
2010)<br />
Region Geometrical<br />
spreading<br />
number includes mining related tremors, which have<br />
decreased in recent years as production in the gold<br />
mining areas <strong>of</strong> <strong>South</strong> <strong>Africa</strong> had declined (see Fig. 1).<br />
The largest recorded <strong>tectonic</strong> <strong>earthquake</strong>s over the past<br />
century in southern <strong>Africa</strong> registered with magnitudes<br />
either equal or below the upper threshold <strong>of</strong> 7 on the<br />
M L <strong>scale</strong>. The last consideration is purely practical<br />
since the <strong>ML</strong> <strong>scale</strong> relies on relatively simple methodology<br />
in comparison to more computationally complicated<br />
magnitude <strong>scale</strong>s, e.g., moment or energy<br />
magnitude <strong>scale</strong>s.<br />
The aim <strong>of</strong> this study is to define coefficients <strong>for</strong><br />
the M L <strong>scale</strong> in <strong>South</strong> <strong>Africa</strong> through simult<strong>an</strong>eous<br />
inversion on a dataset <strong>of</strong> <strong>earthquake</strong>s with a good<br />
regional distribution within the borders <strong>of</strong> <strong>South</strong><br />
<strong>Africa</strong> <strong>an</strong>d to determine stations corrections.<br />
Fig. 1 Earthquakes recorded<br />
in <strong>South</strong> <strong>Africa</strong> with M L≥4.0<br />
during the period 1990 to<br />
2009. It is evident that more<br />
mining-related <strong>earthquake</strong>s<br />
(solid bars) were recorded by<br />
the SANSN during the period<br />
1990 to 2009. The figure<br />
further shows a general decline<br />
in the number <strong>of</strong><br />
mining-related <strong>earthquake</strong>s<br />
from 2001 (except <strong>for</strong> <strong>an</strong><br />
<strong>an</strong>omalous number during<br />
2005)<br />
Author's personal copy<br />
Anelastic<br />
attenuation<br />
Reference<br />
Alb<strong>an</strong>ia 1.66 0.00080 Muco <strong>an</strong>d Minga (1991)<br />
Central Cali<strong>for</strong>nia 1.00 0.00310 Bakun <strong>an</strong>d Joyner (1984)<br />
Ethiopia 1.20 0.00107 Keir et al. (2006)<br />
Northern Italy 1.00 0.00540 Bindi et al. (2005)<br />
Northwestern Turkey 1.00 0.00152 Baumbach et al. (2003)<br />
Norway 0.91 0.00087 Alsaker et al. (1991)<br />
<strong>South</strong> <strong>Africa</strong> 1.15 0.00063 This Study<br />
<strong>South</strong>ern Australia 1.10 0.00130 Greenhalgh <strong>an</strong>d Singh (1986)<br />
<strong>South</strong>ern Cali<strong>for</strong>nia 1.11 0.01890 Hutton <strong>an</strong>d Boore (1987)<br />
<strong>South</strong>western Germ<strong>an</strong>y 1.11 0.00095 St<strong>an</strong>ge (2006)<br />
T<strong>an</strong>z<strong>an</strong>ia 0.78 0.00090 L<strong>an</strong>gston et al. 1998<br />
The Great Basin 1.00 0.00690 Chávez <strong>an</strong>d Priestly (1985)<br />
Western USA 0.83 0.00260 Chávez <strong>an</strong>d Priestly (1985)<br />
2 Major geological provinces <strong>an</strong>d seismo<strong>tectonic</strong><br />
setting<br />
The basement geology in <strong>South</strong> <strong>Africa</strong> is <strong>of</strong> Arche<strong>an</strong><br />
to Mesozoic age hence attenuation <strong>of</strong> seismic waves<br />
with dist<strong>an</strong>ce is expected to be similar to that <strong>of</strong> other<br />
SCRs (Bakun <strong>an</strong>d McGarr 2002).<br />
The development <strong>of</strong> the basement geology <strong>of</strong> southern<br />
<strong>Africa</strong> (Fig. 2) sp<strong>an</strong>s a period <strong>of</strong> approximately<br />
2.9 Ga years, which commenced with the Kaapvaal<br />
craton becoming a stable 3.1 Ga assemblage (e.g., De<br />
Wit et al. 1992; McCourt <strong>an</strong>d Wilson 1992). The Zimbabwe<br />
craton (Wilson et al. 1990) fused with the Kaapvaal<br />
craton during a collision 2.7 Ga ago (e.g., Light<br />
1982; Treloar et al. 1992) to <strong>for</strong>m, what is collectively<br />
known as the Kalahari craton. Metamorphosed rocks <strong>of</strong>
Fig. 2 Simplified basement<br />
lithology <strong>of</strong> southern <strong>Africa</strong><br />
the Limpopo Belt define the suture between the cratons<br />
(e.g., Watkeys 1983; V<strong>an</strong> Reenen et al. 1990; Barton <strong>an</strong>d<br />
V<strong>an</strong> Reenen 1992).<br />
Continental shelf deposits <strong>an</strong>d sediments that accumulated<br />
along the western edge <strong>of</strong> the Kaapvaal <strong>an</strong>d<br />
Zimbabwe<strong>an</strong> cratons were subsequently folded <strong>an</strong>d<br />
thrusted eastwards (Kheis <strong>an</strong>d Magondi Belts)<br />
~1.8 Ga (e.g., Leyshon <strong>an</strong>d Tennick 1988; Treloar<br />
<strong>an</strong>d Kramers 1989), postulated as a collision between<br />
the Congo <strong>an</strong>d Kalahari craton. An extensive plat<strong>for</strong>m<br />
<strong>of</strong> continental crust was created during the collision <strong>of</strong><br />
the cratons (Congo <strong>an</strong>d Kalahari) to <strong>for</strong>m the Ubendi<strong>an</strong><br />
Belt (1.8 Ga).<br />
The assemblage <strong>of</strong> the palaeo-continent, Rodinia<br />
(1.1 Ga), through the amalgamation <strong>of</strong> Atl<strong>an</strong>tica (parts<br />
<strong>of</strong> present-day West <strong>Africa</strong> <strong>an</strong>d <strong>South</strong> America), Arctica<br />
(parts <strong>of</strong> present-day C<strong>an</strong>ada, northern Siberia,<br />
<strong>an</strong>d Greenl<strong>an</strong>d), <strong>an</strong>d Ur (parts <strong>of</strong> present-day <strong>South</strong><br />
<strong>Africa</strong>, Madagascar, India, <strong>an</strong>d Australia), resulted in<br />
the <strong>for</strong>mation <strong>of</strong> a global mountain r<strong>an</strong>ge that has<br />
since been eroded to expose extensive metamorphic<br />
belts (Kibaren/Grenville Belt) (Thomas et al. 1993)<br />
known locally as the Namaqua-Natal Mobile Belt<br />
fl<strong>an</strong>king the southern portion <strong>of</strong> the Kaapvaal craton.<br />
Sediments deposited in rifts that subsequently opened<br />
between the Congo <strong>an</strong>d Kalahari craton during the<br />
fragmentation <strong>an</strong>d break-up <strong>of</strong> Rodinia (700 Ma) were<br />
compressed, folded, <strong>an</strong>d metamorphosed (500 to<br />
550 Ma) with the assemblage <strong>of</strong> the supercontinent,<br />
P<strong>an</strong>gaea, to <strong>for</strong>m the P<strong>an</strong>-Afric<strong>an</strong> Belts, Damara Belts,<br />
<strong>an</strong>d the Cape Gr<strong>an</strong>ite Suite.<br />
Author's personal copy<br />
J Seismol<br />
Around 500 Ma, the Supercontinent Gondw<strong>an</strong>a<br />
consolidated from fragments <strong>of</strong> the Supercontinent<br />
P<strong>an</strong>gea (which had broken into the northern Supercontinent<br />
Laurasia <strong>an</strong>d southern Supercontinent Gondw<strong>an</strong>a).<br />
Gondw<strong>an</strong>a was comprised mainly <strong>of</strong> the<br />
continents <strong>of</strong> the southern hemisphere known today<br />
as <strong>Africa</strong>, Antarctica, Australia, <strong>South</strong> America,<br />
Madagascar, <strong>an</strong>d India (which today resides in the<br />
northern hemisphere). Thinning <strong>of</strong> the crust through<br />
rifting allowed the area presently known as the southern<br />
Cape to be inundated to <strong>for</strong>m the Agulhas Sea.<br />
Sedimentary rocks <strong>of</strong> the Cape Supergroup <strong>an</strong>d the<br />
Natal Group <strong>for</strong>med from sediments deposited in the<br />
Agulhas Sea (Johnson et al. 2006).<br />
Extensive folding <strong>of</strong> the Cape Supergroup (~310 Ma)<br />
created the Cape Mountains through compression <strong>of</strong> the<br />
interior <strong>of</strong> Gondw<strong>an</strong>a from a subduction zone that had<br />
<strong>for</strong>med along the southern edge <strong>of</strong> Gondw<strong>an</strong>a (e.g.,<br />
Hälbich 1983; De Beer 1990). A basin developed along<br />
the northern margin <strong>of</strong> the Cape Mountains in which<br />
sedimentary rocks <strong>of</strong> the Karoo Supergroup were deposited.<br />
Massive qu<strong>an</strong>tities <strong>of</strong> flood basalt lavas erupted<br />
onto the surface, disrupting sedimentation in the Karoo.<br />
The extensive outpouring <strong>of</strong> lavas is postulated to be<br />
related to a m<strong>an</strong>tle plume that had risen beneath, what is<br />
today known as southern <strong>Africa</strong>, <strong>an</strong>d may have initiated<br />
the break-up <strong>of</strong> Gondw<strong>an</strong>a at 182 Ma (e.g., Martin <strong>an</strong>d<br />
Hartnady 1986; Watkeys <strong>an</strong>d Sweeney 1988).<br />
By 140 Ma, oce<strong>an</strong>ic crust <strong>for</strong>med between West<br />
Gondw<strong>an</strong>a (<strong>Africa</strong> <strong>an</strong>d <strong>South</strong> America) <strong>an</strong>d East Gondw<strong>an</strong>a<br />
(Antarctica, Australia, India, <strong>an</strong>d Madagascar) to
J Seismol<br />
mark the beginning <strong>of</strong> the Indi<strong>an</strong> Oce<strong>an</strong>. West<br />
Gondw<strong>an</strong>a further fragmented around 120 Ma<br />
when oce<strong>an</strong>ic crust, possibly resulting from a separate<br />
<strong>an</strong>d distinct m<strong>an</strong>tle plume, separated <strong>Africa</strong><br />
from <strong>South</strong> America (proto-Atl<strong>an</strong>tic Oce<strong>an</strong>). Simult<strong>an</strong>eously,<br />
continental crust <strong>of</strong> the Falkl<strong>an</strong>d<br />
Plateau detached along a major fracture (Agulhas<br />
Falkl<strong>an</strong>d Fracture Zone) <strong>an</strong>d moved westwards<br />
with <strong>South</strong> America be<strong>for</strong>e completely separating from<br />
the Cape by 90 Ma (Von Veh <strong>an</strong>d Anderson 1990).<br />
The faulting regime in <strong>South</strong> <strong>Africa</strong> ch<strong>an</strong>ges from<br />
normal faulting in the northeast <strong>of</strong> the country, possibly<br />
related to the influence <strong>of</strong> the nearby East Afric<strong>an</strong><br />
Rift System to strike-slip along the southwestern part.<br />
The ch<strong>an</strong>ge to strike-slip faulting c<strong>an</strong> be explained<br />
through the dominating stresses becoming horizontal<br />
through ridge push from the Mid-Atl<strong>an</strong>tic Ridge<br />
against the Afric<strong>an</strong> plate (Singh et al. 2009).<br />
3 Earthquake data<br />
Earthquake data used in our study were obtained from the<br />
<strong>South</strong> Afric<strong>an</strong> National Seismograph Network (SANSN)<br />
Fig. 3 Distribution <strong>of</strong> mining-related <strong>earthquake</strong>s in the gold/<br />
platinum mining areas (gray polygons), collieries, <strong>an</strong>d iron/<br />
m<strong>an</strong>g<strong>an</strong>ese mines (black polygons) as well as smaller mining<br />
operations in the Western Cape aggregate mines (1), Lime Acres<br />
diamond mine (2) <strong>an</strong>d the Zeerust mines (3) during the period<br />
2006 to 2009. A large number <strong>of</strong> recorded seismic disturb<strong>an</strong>ces<br />
Author's personal copy<br />
operated by the Council <strong>for</strong> Geoscience (CGS) <strong>an</strong>d three<br />
primary stations <strong>of</strong> the International Monitoring System.<br />
A total <strong>of</strong> 263 <strong>tectonic</strong> events were selected that<br />
were recorded by at least five stations during the<br />
period 2006 to 2009 within <strong>an</strong> area bounded by approximate<br />
coordinates 22° to 36° <strong>South</strong> <strong>an</strong>d 15° to 35°<br />
East with amplitude readings in the dist<strong>an</strong>ce r<strong>an</strong>ge 10<br />
to 1,000 km <strong>an</strong>d magnitudes r<strong>an</strong>ging from 1.4 to 4.1.<br />
Stations with at least one clear phase reading <strong>an</strong>d<br />
where the S <strong>an</strong>d Lg phases were not truncated or<br />
clipped were used to measure the maximum amplitude.<br />
The displacement amplitudes were measured in<br />
n<strong>an</strong>ometers on the vertical (Z) component by deconvolving<br />
the wave<strong>for</strong>m with the instrument response,<br />
<strong>an</strong>d the resulting ground displacement wave<strong>for</strong>m convolved<br />
with the frequency response <strong>of</strong> the SWA instrument<br />
as given by Uhrhammer <strong>an</strong>d Collins (1990)<br />
(T 000.8 s <strong>an</strong>d damping factor H 000.7).<br />
The bulk <strong>of</strong> <strong>South</strong> Afric<strong>an</strong> seismicity originates in<br />
the gold/platinum mines <strong>of</strong> the country with <strong>an</strong>other<br />
signific<strong>an</strong>t number recorded from blasting activities in<br />
quarries (e.g., Saunders et al. 2010). The mining related<br />
events <strong>an</strong>d quarry blasts in collieries, the iron/<br />
m<strong>an</strong>g<strong>an</strong>ese mines, <strong>an</strong>d smaller quarries are identified<br />
by their spatial distribution relative to large-<strong>scale</strong><br />
(gray tri<strong>an</strong>gles) c<strong>an</strong> be attributed to explosions <strong>an</strong>d suspected<br />
explosions while <strong>an</strong> equal number <strong>of</strong> <strong>earthquake</strong>s c<strong>an</strong> spatially<br />
be related to mining operations (induced seismicity) in the gold/<br />
platinum mines <strong>of</strong> <strong>South</strong> <strong>Africa</strong>. These seismic events were not<br />
included in the dataset used during the inversion (<strong>for</strong> more<br />
detail, see text)
mining operations (see Fig. 3) <strong>an</strong>d are flagged in the<br />
database by fixing the depth to surface (explosions) or<br />
to 2-km depth (mining related <strong>earthquake</strong>s). Explosions<br />
are further identified by occurring in time between<br />
06:00 <strong>an</strong>d 18:00 (local time) which is the time<br />
when blasting is allowed through <strong>South</strong> Afric<strong>an</strong> Legislation.<br />
The depth <strong>of</strong> 2 km <strong>for</strong> mining-related <strong>earthquake</strong>s<br />
was chosen, since 2 km best represents the<br />
average depth <strong>of</strong> gold mines in <strong>South</strong> <strong>Africa</strong>.<br />
In <strong>South</strong> <strong>Africa</strong>, a simple one-dimensional velocity<br />
structure (Wright et al. 2002) is used <strong>for</strong> locating <strong>earthquake</strong>s<br />
with <strong>an</strong> average location error <strong>of</strong> less th<strong>an</strong> 10 km.<br />
Furthermore, the routine practice at the CGS is to measure<br />
maximum amplitude on the vertical component as<br />
opposed to Richter (1935) who defined the <strong>ML</strong> <strong>scale</strong><br />
from maximum amplitudes on the horizontal components.<br />
It is noted by Alsaker et al. (1991) thatthereis<br />
little difference between the maximum amplitudes measured<br />
on vertical <strong>an</strong>d horizontal components in Norway.<br />
Additionally, amplification on horizontal ch<strong>an</strong>nels is<br />
more pronounced at stations located on unconsolidated<br />
material whereas the vertical component is less affected.<br />
Using vertical component amplitudes thus leads to a<br />
more consistent M L estimation between stations located<br />
on different geological foundations (Havskov <strong>an</strong>d<br />
Fig. 4 Ratio <strong>of</strong> amplitudes<br />
(1,681) measured on the<br />
vertical <strong>an</strong>d horizontal components<br />
<strong>of</strong> <strong>earthquake</strong>s used<br />
in this study. The solid line<br />
delineates linear regression<br />
<strong>of</strong> data<br />
Author's personal copy<br />
LogA (Vertical)<br />
4.5<br />
4<br />
3.5<br />
3<br />
2.5<br />
2<br />
1.5<br />
1<br />
0.5<br />
0<br />
Ottemöller 2010). This assumption was tested <strong>for</strong><br />
<strong>South</strong> Afric<strong>an</strong> conditions (Fig. 4) by comparing the<br />
ratio <strong>of</strong> 1,681 maximum amplitudes measured on the<br />
vertical ch<strong>an</strong>nel (logA Vertical) to the arithmetic me<strong>an</strong> <strong>of</strong><br />
the maximum amplitudes measured on the horizontal<br />
ch<strong>an</strong>nels (logAHorizntal). The me<strong>an</strong> value <strong>of</strong> the vertical<br />
to horizontal ratio is 0.94 with a correlation coefficient<br />
<strong>of</strong> 0.96.<br />
Focal depths <strong>of</strong> <strong>earthquake</strong>s are not well constrained<br />
in <strong>South</strong> <strong>Africa</strong> due to the sparse station distribution <strong>of</strong><br />
the SANSN. We there<strong>for</strong>e accepted that <strong>South</strong> <strong>Africa</strong>,<br />
tentatively located in a SCR, would have <strong>earthquake</strong><br />
depths similar to other regions considered SCRs. V<strong>an</strong><br />
L<strong>an</strong>en <strong>an</strong>d Mooney (2004) noted that the overall majority<br />
<strong>of</strong> <strong>earthquake</strong>s occurring in SCRs have hypocentral<br />
depths equal or less th<strong>an</strong> 10 km with the maximum<br />
number occurring between 5–10 km. Thus, <strong>earthquake</strong>s<br />
resulting from <strong>tectonic</strong> activity in <strong>South</strong> <strong>Africa</strong> are restricted<br />
to depths <strong>of</strong> either 5 or 10 km depending on the<br />
best convergence calculated by the location s<strong>of</strong>tware<br />
(Saunders et al. 2010). Br<strong>an</strong>dt <strong>an</strong>d Saunders (2011)<br />
determined the <strong>earthquake</strong> depths <strong>of</strong> three <strong>tectonic</strong><br />
<strong>earthquake</strong>s in <strong>South</strong> <strong>Africa</strong> as shallow (between 9 <strong>an</strong>d<br />
11 km) during a study to determine moment tensor<br />
solutions, however, further studies are needed to qualify<br />
-0.5<br />
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
LogA (Horizontal)<br />
J Seismol
J Seismol<br />
Fig. 5 Spatial distribution<br />
<strong>of</strong> <strong>earthquake</strong>s (solid<br />
circles) with paths (gray<br />
lines) indicated to stations<br />
(solid tri<strong>an</strong>gles) used in this<br />
study. The number indicated<br />
at the station relates to the<br />
number assigned in Table 2<br />
the shallow hypocentral depth hypothesis <strong>for</strong> <strong>tectonic</strong><br />
<strong>earthquake</strong>s in <strong>South</strong> <strong>Africa</strong>.<br />
Figure 5 depicts the epicenters <strong>an</strong>d station locations<br />
together with the resulting ray pattern. It shows that<br />
the majority <strong>of</strong> <strong>South</strong> <strong>Africa</strong> is covered spatially, traversing<br />
all the major crustal units. Details <strong>of</strong> the stations<br />
are listed in Table 2 with station corrections<br />
resulting from the inversion.<br />
4 Methodology<br />
The concept <strong>of</strong> local magnitude was introduced by<br />
Richter (1935) <strong>an</strong>d is defined as the logarithm (log10A)<br />
<strong>of</strong> the peak average maximum displacement<br />
(millimeters) measured on the two horizontal components<br />
<strong>of</strong> a SWA (Anderson <strong>an</strong>d Wood 1925) torsion<br />
seismometer (see Eq. 1). The <strong>scale</strong> was defined such<br />
that <strong>an</strong> <strong>earthquake</strong> <strong>of</strong> <strong>ML</strong>03 situated at 100-km dist<strong>an</strong>ce<br />
from the recording stations resulted in 1 mm<br />
deflection on the seismogram. Hutton <strong>an</strong>d Boore<br />
(1987) proposed a reference dist<strong>an</strong>ce (Rref) <strong>of</strong>17km<br />
with a reference magnitude <strong>of</strong> 3 (resulting in a 10 mm<br />
deflection on a st<strong>an</strong>dard Wood–Anderson seismograph)<br />
in regions with dissimilar attenuation to <strong>South</strong>ern<br />
Cali<strong>for</strong>nia, as is the case in <strong>South</strong> <strong>Africa</strong>. However,<br />
Alsaker et al. (1991) cautioned that sufficient observations<br />
are required within the chosen reference dist<strong>an</strong>ce<br />
to justify deviating from the original reference<br />
dist<strong>an</strong>ce <strong>of</strong> 100 km. The hypocentral distribution <strong>of</strong> data<br />
used in this study (Fig. 6) indicates that 90 % <strong>of</strong> the<br />
Author's personal copy<br />
amplitude readings <strong>of</strong> our data occur in the dist<strong>an</strong>ce<br />
r<strong>an</strong>ge 100–800 km with only 3 % in the dist<strong>an</strong>ce r<strong>an</strong>ge<br />
0–100 km. We there<strong>for</strong>e used the same arbitrary constraint<br />
suggested by Richter (1935) during regression.<br />
The general <strong>for</strong>m <strong>of</strong> the local magnitude <strong>scale</strong> c<strong>an</strong><br />
be expressed as:<br />
<strong>ML</strong> ¼ logðAÞþalogðRÞþb ðRÞþc þ Si ð4Þ<br />
where a relates to the geometrical spreading, b the<br />
attenuation factor, Si is a station correction term <strong>for</strong><br />
station i, <strong>an</strong>d c a source region correction term (see,<br />
e.g., Borm<strong>an</strong>n (2002) <strong>an</strong>d Havskov <strong>an</strong>d Ottemöller<br />
(2010) <strong>for</strong> a more comprehensive discussion).<br />
We applied singular value decomposition (SVD)<br />
<strong>using</strong> the Numerical Recipe (Press et al. 2003) routines<br />
to invert our observations <strong>for</strong> a, b, <strong>an</strong>d Si.<br />
Attenuation characteristics between dissimilar<br />
regions vary resulting in different trace amplitudes<br />
measured <strong>for</strong> identical magnitude <strong>earthquake</strong>s observed<br />
at identical dist<strong>an</strong>ces. However, as mentioned<br />
previously, Richter (1935) defined the M L <strong>scale</strong> that<br />
the maximum trace amplitude on a SWA should measure<br />
1 mm from <strong>an</strong> <strong>earthquake</strong> registering 3 on the<br />
Richter <strong>scale</strong>, originating at a dist<strong>an</strong>ce <strong>of</strong> 100 km from<br />
<strong>an</strong> observation station in <strong>South</strong>ern Cali<strong>for</strong>nia. In order<br />
to compensate <strong>for</strong> the difference in attenuation along<br />
the path to source to obtain <strong>an</strong> equivalent M L in <strong>South</strong><br />
<strong>Africa</strong>, we set c such that 1 mm amplitude at 100 km<br />
on a SWA seismograph would also compute <strong>ML</strong>03.<br />
Since a least-squares inversion procedure was followed<br />
<strong>for</strong> solving Eq. 4, errors c<strong>an</strong> be expected during
Table 2 Details <strong>of</strong> stations used in the study with station corrections<br />
Number Station Code Sensor Period <strong>of</strong><br />
operation<br />
the inversion should the data not be normally distributed<br />
(St<strong>an</strong>ge 2006). There<strong>for</strong>e, the dependency on a<br />
specific station was applied to test whether the parameters<br />
determined during the SVD inversion was mutually<br />
free. The procedure was to create secondary<br />
datasets, omitting stations in succession, then<br />
Author's personal copy<br />
Site Lithology Station<br />
correction<br />
1 Belfast BFT CMG-40T 03/2006–present Vault Gabbro 0.06<br />
2 Bosh<strong>of</strong> a<br />
BOSA KS-54000 08/1999–present Borehole Gr<strong>an</strong>ite 0.50<br />
3 Buffelsbos BUFB L4C-3D 09/2008–present Vault S<strong>an</strong>dstone 0.22<br />
4 Ceres CER KS-2000 03/2002–present Vault Quartzite 0.15<br />
5 Calvinia CVNA KS-2000 04/2004–present Vault Dolerite −0.21<br />
6 East R<strong>an</strong>d<br />
propriety mines<br />
ERPM L4C-3D 05/2006–present Vault Quartzite 0.26<br />
7 Elim ELIM L4-3D 10/1996–03/2007 Vault Quartzitic −0.05<br />
CMG-40T 03/2007–present<br />
s<strong>an</strong>dstone<br />
8 Grahamstown GRM KS-2000 10/2002–present Vault Quartzite −0.08<br />
9 Gariep Dam HVD CMG-40T 02/2004–present Vault Dolerite −0.08<br />
10 Klo<strong>of</strong> KLOF L4-3D 08/2006–present Ground floor Dolomite −0.23<br />
11 Komaggas KOMG L4C-3D 06/2003–05/2007 Vault Gneiss −0.08<br />
CMG-40T 05/2007–present<br />
12 Kokstad KSD CMG-40T 07/2005–03/2007 Vault Dolerite 0.03<br />
KS-2000 03/2007–present<br />
13 Koster KSR S-13 04/2002–05/2006 Vault S<strong>an</strong>dstone −0.25<br />
CMG-40T 05/2006–present<br />
14 Lobatse a<br />
LBTB KS-54000 04/1993–present Borehole Gr<strong>an</strong>ite 0.37<br />
15 Mop<strong>an</strong>i MOPA CMG-40T 04/2004–present Vault Gr<strong>an</strong>ite 0.02<br />
16 Mussina MSNA CMG-40T 03/2003–11/2006 Vault Quartzite −0.11<br />
KS-2000 11/2006–present<br />
17 Observatory OBSV L4C-3D 10/2009–12/2009 Ground floor Quartzite 0.20<br />
18 Parys PRYS S-13 11/2004–09/2006 Vault Gr<strong>an</strong>ite −0.07<br />
CMG-40T 09/2006–present<br />
19 Prieska PKA CMG-40T 04/2004–present Vault Alluvium,<br />
calcrete<br />
0<br />
20 Pongola POGA KS-2000 03/2005–present Vault S<strong>an</strong>dstone −0.12<br />
21 Senekal SEK CMG-40T 12/2003–present Vault on hill S<strong>an</strong>dstone −0.30<br />
22 Somerset East SOE L4C-3D 03/2005–09/2008 Vault Mudstone, 0<br />
CMG-40T 09/2008–present<br />
s<strong>an</strong>dstone<br />
23 Sutherl<strong>an</strong>d b<br />
SUR STS-2 10/1990–present Vault Dolerite 0.20<br />
24 Schweizer-Reneke SWZ CMG-40T 04/2006–present Vault S<strong>an</strong>dstone −0.11<br />
25 Upington UPI CMG-40T 06/2005–05/2007 Vault Calcrete −0.03<br />
KS-2000 05/2007–present<br />
26 Western deep<br />
levels mine<br />
WDLM L4C-3D 12/2009–present Vault Quartzite −0.31<br />
a Station <strong>of</strong> the Global Telemetered Seismological Network (GTSN)<br />
b Station <strong>of</strong> the Incorporated Research Institutions <strong>for</strong> Seismology (IRIS)<br />
J Seismol<br />
executing the SVD inversion on each subset <strong>an</strong>d comparing<br />
the results. The deviation <strong>of</strong> the different solutions<br />
provides <strong>an</strong> indication <strong>of</strong> the error introduced<br />
during the inversion. Figure 7 indicates negligible<br />
deviation from the original results <strong>using</strong> the entire<br />
dataset <strong>an</strong>d subsets.
J Seismol<br />
Fig. 6 Dist<strong>an</strong>ce/magnitude<br />
distribution <strong>of</strong> events used<br />
in this study. The magnitudes<br />
shown were determined<br />
with the Hutton <strong>an</strong>d<br />
Boore (1987) relation (original<br />
dataset)<br />
5 Station corrections<br />
Station corrections were determined during the inversion<br />
<strong>for</strong> 26 stations used during the study, <strong>an</strong>d the<br />
results are presented in Table 2 <strong>an</strong>d Fig. 8.<br />
The IMS stations at Bosh<strong>of</strong> <strong>an</strong>d Lobatse have<br />
large corrections, viz. 0.50 <strong>an</strong>d 0.37, respectively.<br />
This observation indicates that the amplitude readings<br />
at these stations are lower th<strong>an</strong> expected which<br />
Dist<strong>an</strong>ce Corrrection (M L )<br />
5<br />
4<br />
3<br />
2<br />
1<br />
101 102 103 0<br />
Hypocentral dist<strong>an</strong>ce (km)<br />
Fig. 7 Graphical representation <strong>of</strong> Eq. 4 shown as a broken<br />
black line with the gray shaded area representing the st<strong>an</strong>dard<br />
deviation determined during the study to establish whether there<br />
is a dependency on <strong>an</strong>y particular station<br />
Author's personal copy<br />
may be related to the free surface effect (e.g., Lay<br />
<strong>an</strong>d Wallace 1995) since the sensors at both these<br />
stations are located in boreholes, as opposed to<br />
vaults located at the surface, as is the case <strong>for</strong> the<br />
other stations considered during this study. Furthermore,<br />
the stations at BUFB, CVNA, ERPM, KLOF,<br />
KSR, OBSV, SEK, SUR, <strong>an</strong>d WDLM have large<br />
station corrections (>±0.2) that are explained<br />
through poor seismic coupling <strong>for</strong> stations ERPM,<br />
KLOF, OBSV, <strong>an</strong>d WDLM since the sensors at<br />
these station are not located in ideal seismic vaults,<br />
while stations SEK <strong>an</strong>d CVNA are located on hillocks<br />
where wave interactions may lead to complex<br />
interference. The larger-th<strong>an</strong>-expected station residuals<br />
<strong>for</strong> stations BUFB, KSR, <strong>an</strong>d SUR are unclear<br />
but may be related to unknown geological structures<br />
at these sites or could be a result <strong>of</strong> poor<br />
seismic coupling. Figure 9a, b shows the magnitude<br />
residuals computed without station corrections <strong>using</strong><br />
the original dataset (Hutton <strong>an</strong>d Boore 1987)<br />
<strong>an</strong>d with station corrections (this study). The average<br />
residuals with station corrections considered are<br />
close to zero (0.05) with a st<strong>an</strong>dard deviation <strong>of</strong><br />
0.28 <strong>an</strong>d vari<strong>an</strong>ce σ 2 <strong>of</strong> 0.079 while residuals without<br />
station corrections have a st<strong>an</strong>dard deviation <strong>of</strong><br />
0.38 with a vari<strong>an</strong>ce σ 2 <strong>of</strong> 0.144. There<strong>for</strong>e, the<br />
vari<strong>an</strong>ce is reduced by 45.1 % when adopting the<br />
station corrections.
Fig. 8 Station corrections determined during the SVD shown in<br />
relation to the simplified surface geology <strong>of</strong> <strong>South</strong> <strong>Africa</strong>,<br />
Lesotho, <strong>an</strong>d Swazil<strong>an</strong>d. Positive station corrections are<br />
Fig. 9 Magnitude residuals<br />
versus dist<strong>an</strong>ce. a Magnitude<br />
residuals computed<br />
with A 0(R) (Hutton <strong>an</strong>d<br />
Boore 1987) without station<br />
corrections. The st<strong>an</strong>dard<br />
deviation is 0.38, <strong>an</strong>d the<br />
vari<strong>an</strong>ce is 0.144. b Magnitude<br />
residuals computed<br />
with A 0(R) (this study) with<br />
station corrections considered.<br />
The st<strong>an</strong>dard deviation<br />
is 0.28, <strong>an</strong>d the vari<strong>an</strong>ce is<br />
0.079. A vari<strong>an</strong>ce reduction<br />
<strong>of</strong> 45.1 % is achieved<br />
through applying station<br />
corrections<br />
Author's personal copy<br />
J Seismol<br />
indicated through tri<strong>an</strong>gles while inverted tri<strong>an</strong>gles indicate a<br />
negative station correction. The number indicated at the station<br />
relates to the number assigned in Table 2
J Seismol<br />
Fig. 10 Comparison <strong>of</strong> attenuation<br />
curves <strong>for</strong> <strong>South</strong>ern<br />
Cali<strong>for</strong>nia. The original<br />
relation (Richter 1935) is<br />
indicated as open squares,<br />
Hutton <strong>an</strong>d Boore (1987) as<br />
filled circles, Norway<br />
(Alsaker et al. 1991) asa<br />
solid line, <strong>an</strong>d <strong>South</strong> <strong>Africa</strong><br />
(this study) as crosses<br />
6 Results <strong>an</strong>d discussion<br />
The dist<strong>an</strong>ce correction term (−logA 0) determined during<br />
this study, <strong>using</strong> 100-km dist<strong>an</strong>ce normalization, is<br />
given by:<br />
logðA0Þ ¼ 1:149 logðRÞþ0:00063ðRÞþ2:04 ð5Þ<br />
Fig. 11 Magnitude determined<br />
during this study<br />
plotted against the magnitude<br />
determined with the<br />
Hutton <strong>an</strong>d Boore (1987)<br />
relation. The broken line<br />
indicates linear regression <strong>of</strong><br />
data. An overestimation <strong>of</strong><br />
magnitude <strong>using</strong> the Hutton<br />
<strong>an</strong>d Boore (1987) relation in<br />
<strong>South</strong> <strong>Africa</strong> is implied<br />
Author's personal copy<br />
<strong>ML</strong> (This study)<br />
5<br />
4.5<br />
4<br />
3.5<br />
3<br />
2.5<br />
2<br />
1.5<br />
The overall magnitude st<strong>an</strong>dard deviation obtained<br />
during the inversion (10–1,000 km) was 0.27 considering<br />
station residuals.<br />
A comparison <strong>of</strong> the attenuation curves obtained<br />
<strong>for</strong> <strong>South</strong>ern Cali<strong>for</strong>nia (Richter 1935; Hutton <strong>an</strong>d<br />
Boore 1987) <strong>an</strong>d <strong>South</strong> <strong>Africa</strong> (this study) indicates<br />
lower attenuation in <strong>South</strong> <strong>Africa</strong> with dist<strong>an</strong>ce<br />
1<br />
1 1.5 2 2.5 3 3.5 4 4.5 5<br />
<strong>ML</strong> (Hutton <strong>an</strong>d Boore (1987)
Fig. 12 Comparison <strong>of</strong><br />
magnitudes calculated at<br />
ten stations located at varying<br />
dist<strong>an</strong>ces (>100 to<br />
J Seismol<br />
<strong>of</strong> import<strong>an</strong>ce, since a modified Richter (1935) attenuation<br />
relation developed by Fernández (1993) <strong>an</strong>d<br />
that <strong>of</strong> Hutton <strong>an</strong>d Boore (1987) had been used to<br />
determine local magnitude in <strong>South</strong> <strong>Africa</strong> from the<br />
inception <strong>of</strong> the SANSN to present. A comparison<br />
with the attenuation relation obtained <strong>for</strong> Norway<br />
(Alsaker et al. 1991) indicates that the attenuation<br />
relation obtained during this study is similar to Norway<br />
which is located in a SCR, which cautiously<br />
suggests that <strong>South</strong> <strong>Africa</strong> displays attenuation characteristics<br />
like those <strong>of</strong> other SCRs. Thus, a<br />
frequency-dependent attenuation relation <strong>for</strong> <strong>South</strong><br />
<strong>Africa</strong> would assist in supporting results obtained<br />
during our study.<br />
Figure 12 indicates that the attenuation curve derived<br />
during our study more consistently estimates<br />
magnitude over different hypocentral dist<strong>an</strong>ces in<br />
comparison to previous M L <strong>scale</strong>s used in <strong>South</strong><br />
<strong>Africa</strong> which appear to have had a tendency to overestimate<br />
magnitude (Fernández 1993; Hutton <strong>an</strong>d<br />
Boore 1987).<br />
The me<strong>an</strong> magnitude residuals obtained by subtracting<br />
<strong>ML</strong>, independently calculated <strong>for</strong> each recording<br />
station, from the arithmetic me<strong>an</strong> <strong>of</strong> M L<br />
determined from all stations per <strong>earthquake</strong> is shown<br />
in Fig. 13. The me<strong>an</strong> values, averaged at 50-km dist<strong>an</strong>ce<br />
intervals, follow the zero baseline closely. This<br />
trend indicates that the relationship between attenuation<br />
<strong>an</strong>d hypocentral dist<strong>an</strong>ce in the region is accurately<br />
modeled with our attenuation curve.<br />
A calibrated local magnitude <strong>scale</strong> is a useful tool<br />
as a relative <strong>scale</strong> <strong>for</strong> general seismic studies but is not<br />
practical <strong>for</strong> studies <strong>of</strong> the seismic source. To this end,<br />
we pl<strong>an</strong> to develop a moment magnitude <strong>scale</strong> <strong>for</strong><br />
<strong>South</strong> <strong>Africa</strong> <strong>an</strong>d develop a relationship between the<br />
local <strong>an</strong>d moment magnitude <strong>scale</strong>s.<br />
7 Conclusion<br />
A local magnitude <strong>scale</strong>, based on the Richter (1935)<br />
definition, was developed <strong>for</strong> <strong>South</strong> <strong>Africa</strong> from 1,943<br />
maximum amplitude observations measured on vertical<br />
component synthetic Wood–Anderson seismograms<br />
in the dist<strong>an</strong>ce r<strong>an</strong>ge 10 to 1,000 km. Results<br />
from the inversion <strong>for</strong> event magnitudes resulted in a<br />
<strong>ML</strong> <strong>scale</strong> <strong>for</strong> <strong>South</strong> <strong>Africa</strong> expressed as Eq. 5.<br />
It was shown that the average difference between<br />
local magnitude calculated from the horizontal<br />
Author's personal copy<br />
components <strong>an</strong>d vertical components is 0.1 magnitude<br />
units. This confirms that the practice <strong>of</strong> computing<br />
local magnitude on the vertical ch<strong>an</strong>nel is valid <strong>for</strong><br />
<strong>South</strong> <strong>Africa</strong>.<br />
The attenuation curve up to 150 km is similar to<br />
<strong>South</strong>ern Cali<strong>for</strong>nia but deviates at dist<strong>an</strong>ces beyond<br />
150 km, indicating that ground motion attenuation is<br />
lower in <strong>South</strong> <strong>Africa</strong> in comparison to <strong>South</strong>ern Cali<strong>for</strong>nia.<br />
It is, however, consistent with observations<br />
from other regions located in SCRs.<br />
It was shown that the attenuation relation determined<br />
during this study provides a more consistent<br />
estimation <strong>of</strong> <strong>ML</strong> over a r<strong>an</strong>ge <strong>of</strong> hypocentral dist<strong>an</strong>ces<br />
when compared with previous <strong>ML</strong> relations used in<br />
<strong>South</strong> <strong>Africa</strong>.<br />
An <strong>an</strong>alysis <strong>of</strong> the me<strong>an</strong> magnitude residuals<br />
obtained with the attenuation curve from our study<br />
does not show pronounced complexities in the crustal<br />
structure <strong>of</strong> <strong>South</strong> <strong>Africa</strong>. The station corrections<br />
obtained during the study show that 13 <strong>of</strong> the 26<br />
stations considered during the study have correction<br />
factors equal or less th<strong>an</strong> ±0.1. This indicates that local<br />
site amplification does not have a signific<strong>an</strong>t influence<br />
on magnitude estimates at these stations. However, it<br />
was postulated that, in some inst<strong>an</strong>ces, less th<strong>an</strong> ideal<br />
vault conditions, topography, <strong>an</strong>d local site effects<br />
may lead to erroneous magnitude calculations.<br />
Acknowledgments This work was funded by the National<br />
Research Foundation under Gr<strong>an</strong>t Number TP2010061400012.<br />
We are grateful to the editors <strong>an</strong>d four <strong>an</strong>onymous reviewers <strong>for</strong><br />
their thoughtful suggestions <strong>an</strong>d positive criticism to improve<br />
the m<strong>an</strong>uscript <strong>an</strong>d a special word <strong>of</strong> gratitude to Pr<strong>of</strong>. Borm<strong>an</strong>n<br />
<strong>for</strong> contributing to the m<strong>an</strong>uscript through invaluable support<br />
<strong>an</strong>d insightful ideas. The authors further wish to express their<br />
gratitude to the copy editor, Sonja v<strong>an</strong> Eck, <strong>for</strong> her meticulous<br />
<strong>an</strong>d careful review <strong>of</strong> the m<strong>an</strong>uscript <strong>an</strong>d to Magda Roos <strong>for</strong> her<br />
assist<strong>an</strong>ce with the figures <strong>for</strong> the m<strong>an</strong>uscript.<br />
References<br />
Alsaker A, Kvamme LB, H<strong>an</strong>sen RA, Dahle A (1991) The M L<br />
<strong>scale</strong> in Norway. Bull Seismol Soc Am 81(2):379–398<br />
Anderson JA, Wood HO (1925) Description <strong>an</strong>d theory <strong>of</strong> the<br />
torsion seismometer. Bull Seismol Soc Am 15(1):1–72<br />
Bakun WH, Joyner WB (1984) The M L <strong>scale</strong> in Central Cali<strong>for</strong>nia.<br />
Bull Seismol Soc Am 74:1827–1843<br />
Bakun WH, McGarr A (2002) Differences in attenuation between<br />
the stable continental regions. Geophys Res Lett 29<br />
(23):36–1–36–4<br />
Barton JM, V<strong>an</strong> Reenen DD (1992) When was the Limpopo<br />
Orogeny? Precambri<strong>an</strong> Res 5:7–16
Baumbach M, Bindi D, Grosser H, Milkereit C, Paralai S, W<strong>an</strong>g<br />
R, Karakisa S, Zünbül S, Zschau J (2003) <strong>Calibration</strong> <strong>of</strong> a<br />
M L <strong>scale</strong> in Northwestern Turkey from 1999 Izmit aftershocks.<br />
Bull Seismol Soc Am 93:2289–2295<br />
Bindi D, Spallarossa D, Ev<strong>an</strong> C, Catt<strong>an</strong>eo M (2005) Local <strong>an</strong>d<br />
duration magnitudes in Northwestern Italy, <strong>an</strong>d seismic<br />
moment versus magnitude relationships. Bull Seismol<br />
Soc Am 95:592–604<br />
Borm<strong>an</strong>n P (ed) (2002) Magnitude <strong>of</strong> seismic events, in IASPEI<br />
New M<strong>an</strong>ual <strong>of</strong> Seismological Observatory Practice. Geo-<br />
ForschungsZentrum, Potsdam, vol. 1, chapter 3, 16–50<br />
Br<strong>an</strong>dt MBC (1997) Implementation <strong>of</strong> SEISAN <strong>earthquake</strong><br />
s<strong>of</strong>tware <strong>for</strong> the SUN to <strong>an</strong>alyze the data obtained through<br />
the <strong>South</strong> Afric<strong>an</strong> National Seismograph Network. Report<br />
no. 1970–0263, Council <strong>for</strong> Geoscience, 48 pp<br />
Br<strong>an</strong>dt MBC, Saunders I (2011) New regional moment tensors<br />
in <strong>South</strong> <strong>Africa</strong>. Seismol Res Lett 82(1):69–80<br />
Chávez DE, Priestly KF (1985) M L observations in the Great<br />
Basin <strong>an</strong>d M 0 versus M L relationship <strong>for</strong> the 1980 Mammoth<br />
Lakes, Cali<strong>for</strong>nia, <strong>earthquake</strong> sequence. Bull Seismol<br />
Soc Am 75:1583–1598<br />
De Beer CH (1990) Simult<strong>an</strong>eous folding in the western <strong>an</strong>d<br />
southern br<strong>an</strong>ches <strong>of</strong> the Cape Fold Belt. <strong>South</strong> Afr J Geol<br />
93:583–591<br />
De Wit MJ, Roering C, Hart RJ, Armstrong RA, De Ronde CEJ,<br />
Green RWE, Tredoux M, Peberdy E, Hart RA (1992)<br />
Formation <strong>of</strong> <strong>an</strong> Archae<strong>an</strong> continent. Nature 357:553–562<br />
Electric Power Research Institute (EPRI) (1994) The <strong>earthquake</strong>s<br />
<strong>of</strong> stable continental regions, EPRI vol. 1. In:<br />
Schneider JF. EPRI, Cali<strong>for</strong>nia, pp 1-1–6-21<br />
Fernández LM (1973) Program to evaluate epicentral dist<strong>an</strong>ce,<br />
origin time <strong>an</strong>d local magnitude <strong>of</strong> regional tremors. Report<br />
no. Gh 1968, Geological Survey <strong>of</strong> <strong>South</strong> <strong>Africa</strong>, 11 pp<br />
Fernández LM (1977) Evaluation <strong>of</strong> epicentral dist<strong>an</strong>ces, origin<br />
times <strong>an</strong>d local magnitudes <strong>of</strong> local seismic events. A<br />
computer program <strong>for</strong> the HP-9830. Report no. Gh 2256,<br />
Geological Survey <strong>of</strong> <strong>South</strong> <strong>Africa</strong>, 12 pp<br />
Fernández LM (1980) Adaptation <strong>of</strong> the Richter seismic local<br />
magnitude <strong>scale</strong> to <strong>South</strong> <strong>Africa</strong>. Report no. 1980-0175,<br />
Geological Survey <strong>of</strong> <strong>South</strong> <strong>Africa</strong>, 22 pp<br />
Fernández LM (1991) A UNIX operating s<strong>of</strong>tware package to<br />
process the data obtained by the <strong>South</strong> Afric<strong>an</strong> National<br />
digital Seismograph System. Report no. 1991-0009, Geological<br />
Survey <strong>of</strong> <strong>South</strong> <strong>Africa</strong>, 25 pp<br />
Fernández LM (1993) The practice <strong>of</strong> evaluating local Richter<br />
magnitude in <strong>South</strong> <strong>Africa</strong>. Report no. 1993–0038, Council<br />
<strong>for</strong> Geoscience, 16 pp<br />
Greenhalgh SA, Singh R (1986) A revised magnitude <strong>scale</strong> <strong>for</strong><br />
<strong>South</strong> Australi<strong>an</strong> <strong>earthquake</strong>s. Bull Seismol Soc Am 76<br />
(3):757–769<br />
Hälbich IW (1983) Intra<strong>for</strong>mational folding in the Cape Fold<br />
Belt. Spec Publ Geol Soc <strong>South</strong> Afr 12:125–130<br />
Havskov J, Ottemöller L (2005) SEISAN Earthquake Analysis<br />
S<strong>of</strong>tware <strong>for</strong> the Windows, Solaris, Linux <strong>an</strong>d MACOSX.<br />
University <strong>of</strong> Bergen, Bergen<br />
Havskov J, Ottemöller L (2010) Routine <strong>earthquake</strong> <strong>an</strong>alysis<br />
in <strong>earthquake</strong> seismology: with sample data, exercise<br />
<strong>an</strong>d s<strong>of</strong>tware. Chapter 6. Springer, ISBN 978-90-<br />
481-8696-9<br />
Hutton LK, Boore DM (1987) The M L <strong>scale</strong> in <strong>South</strong>ern Cali<strong>for</strong>nia.<br />
Bull Seismol Soc Am 77(6):2074–2094<br />
Author's personal copy<br />
J Seismol<br />
Johnson MR, Anhaeusser CR, Thomas RJ (2006) The geology<br />
<strong>of</strong> <strong>South</strong> <strong>Africa</strong>. Geological Society<strong>of</strong><strong>South</strong><strong>Africa</strong>,<br />
Joh<strong>an</strong>nesburg/Council <strong>for</strong> Geoscience, Pretoria<br />
Keir D, Stuart GW, Jackson A, Ayele A (2006) Local magnitude<br />
<strong>scale</strong> <strong>an</strong>d seismicity rate <strong>for</strong> the Ethiopi<strong>an</strong> Rift. Bull Seismol<br />
Soc Am 96:2221–2230<br />
Kvamme LB, Havskov J (1989) Q in <strong>South</strong>ern Norway. Bull<br />
Seismol Soc Am 79(5):1575–1599<br />
L<strong>an</strong>gston CA, Brazier R, Nyblade AA, Owens TJ (1998) Local<br />
magnitude <strong>scale</strong> <strong>an</strong>d seismicity rate <strong>for</strong> T<strong>an</strong>z<strong>an</strong>ia, East<br />
<strong>Africa</strong>. Bull Seismol Soc Am 88(3):712–721<br />
Lay T, Wallace TC (1995) Modern global seismology. Academic,<br />
S<strong>an</strong> Diego, ISBN: 978-0-12-732870-6<br />
Leyshon PR, Tennick FP (1988) The Proterozoic Magondi<br />
Mobile Belt in Zimbabwe—a review. <strong>South</strong> Afr J Geol<br />
91(1):114–131<br />
Light MPR (1982) The Limpopo Belt: a result <strong>of</strong> continental<br />
collision. Tectonics 1:325–342<br />
Martin AK, Hartnady CJH (1986) Plate <strong>tectonic</strong> development<br />
<strong>of</strong> the southwest Indi<strong>an</strong> Oce<strong>an</strong>: a revised reconstruction<br />
<strong>of</strong> East Antarctica <strong>an</strong>d <strong>Africa</strong>. J Geophys Res<br />
91:4767–4786<br />
McCourt S, Wilson JF (1992) Late Archae<strong>an</strong> <strong>an</strong>d Early Proterozoic<br />
<strong>tectonic</strong>s <strong>of</strong> the Limpopo <strong>an</strong>d Zimbabwe<strong>an</strong> provinces,<br />
southern <strong>Africa</strong>. In: Glover JE, Ho SE (eds) The<br />
Archae<strong>an</strong>: terrains, processes <strong>an</strong>d metallogeny pp 237–246<br />
Muco B, Minga P (1991) Magnitude determination <strong>of</strong> near<br />
<strong>earthquake</strong>s <strong>for</strong> the Alb<strong>an</strong>i<strong>an</strong> network. Bolletinno di Ge<strong>of</strong>isica<br />
Teorico ed Applicata 33(129):17–24<br />
Press WH, Tuekolsky SA, Vetterling WT, Fl<strong>an</strong>erry BP (2003)<br />
Numerical recipes in Fortr<strong>an</strong> 77, 2nd edn. Cambridge<br />
University Press, Cambridge<br />
Richter CF (1935) An instrumental <strong>earthquake</strong> <strong>scale</strong>. Bull Seismol<br />
Soc Am 25(1):1–32<br />
Saunders I, Br<strong>an</strong>dt MBC, Roblin DL (2008) Determination <strong>of</strong> a<br />
local magnitude <strong>scale</strong> <strong>for</strong> <strong>South</strong> <strong>Africa</strong>. Report No. 2008-<br />
0111, Council <strong>for</strong> Geoscience, 18 pp<br />
Saunders I, Br<strong>an</strong>dt MBC, Molea TT, Akromah L, Sutherl<strong>an</strong>d<br />
BE (2010) Seismicity <strong>of</strong> southern <strong>Africa</strong> during 2006 with<br />
special reference to the MW 7 Machaze <strong>earthquake</strong>. <strong>South</strong><br />
Afric<strong>an</strong> Journal <strong>of</strong> Geology 113(4):369–380<br />
Singh M, Kijko A, Durfheim R (2009) Seismo<strong>tectonic</strong> models<br />
<strong>for</strong> <strong>South</strong> <strong>Africa</strong>: synthesis <strong>of</strong> geoscientific in<strong>for</strong>mation,<br />
problems, <strong>an</strong>d the way <strong>for</strong>ward. Seismol Res Lett 80<br />
(1):71–80<br />
St<strong>an</strong>ge S (2006) M L determinations <strong>for</strong> local <strong>an</strong>d regional events<br />
<strong>using</strong> a sparse network in southwestern Germ<strong>an</strong>y. J Seismol<br />
10:247–257<br />
Thomas RJ, von Veh MW, McCourt S (1993) The <strong>tectonic</strong><br />
evolution <strong>of</strong> southern <strong>Africa</strong>: <strong>an</strong> overview. J Afr Earth Sci<br />
16:5–24<br />
Treloar PJ, Kramers JD (1989) Metamorphism <strong>an</strong>d geochronology<br />
<strong>of</strong> gr<strong>an</strong>ulites <strong>an</strong>d migmatitic gr<strong>an</strong>ulites from the<br />
Magondi Mobile Belt, Zimbabwe. Precambri<strong>an</strong> Res<br />
46:277–289<br />
Treloar PJ, Coward MP, Harris NBW (1992) Himalay<strong>an</strong>–Tibet<strong>an</strong><br />
<strong>an</strong>alogies <strong>for</strong> the evolution <strong>of</strong> the Zimbabwe craton <strong>an</strong>d<br />
Limpopo Belt. Precambri<strong>an</strong> Res 55:571–587<br />
Uhrhammer RA, Collins ER (1990) Synthesis <strong>of</strong> Wood–Anderson<br />
seismograms from broadb<strong>an</strong>d digital records. Bull<br />
Seismol Soc Am 89(8):702–716
J Seismol<br />
V<strong>an</strong> L<strong>an</strong>en X, Mooney WD (2004) Integrated geologic <strong>an</strong>d<br />
geophysical studies <strong>of</strong> North Americ<strong>an</strong> intraplate seismicity.<br />
In: Stein S, Mazotti S (eds) Continental intraplate<br />
<strong>earthquake</strong>s: science, hazard <strong>an</strong>d policy issues. Geological<br />
Society <strong>of</strong> America Special Paper, p 425, doi: 10.1130/<br />
2007.2425(19)<br />
V<strong>an</strong> Reenen DD, Roering C, Br<strong>an</strong>dl G, Smit CA, Barton<br />
JM (1990) The gr<strong>an</strong>ulite facies rocks <strong>of</strong> the Limpopo<br />
belt in southern <strong>Africa</strong>. In: Vielzeuf D, Vidal P (eds)<br />
Gr<strong>an</strong>ulites <strong>an</strong>d crustal evolution. Kluwer, Dordrecht, pp<br />
257–289<br />
Von Veh MW, Anderson NJB (1990) Normal-slip faulting in the<br />
coastal areas <strong>of</strong> northern Natal <strong>an</strong>d Zulul<strong>an</strong>d, <strong>South</strong> <strong>Africa</strong>.<br />
<strong>South</strong> Afr J Geol 93(4):574–582<br />
Watkeys MK (1983) Brief expl<strong>an</strong>atory notes on the provisional<br />
geological map <strong>of</strong> the Limpopo Belt <strong>an</strong>d environs. Spec<br />
Publ Geol Soc <strong>South</strong> Afr 8:5–8<br />
Author's personal copy<br />
Watkeys MK, Sweeney RJ (1988) Tuli-Lebombo volc<strong>an</strong>ism <strong>an</strong>d<br />
Gondw<strong>an</strong>a rifting. Extended abstracts, Geocongres 1988,<br />
Geological Society <strong>of</strong> <strong>South</strong> <strong>Africa</strong>, Durb<strong>an</strong>, pp 725–228<br />
Wilson JF, Baglow N, Orpen JL, <strong>an</strong>d Tsomondo JM (1990) A<br />
reassessment <strong>of</strong> some regional correlations <strong>of</strong> greenstonebelt<br />
rocks in Zimbabwe <strong>an</strong>d their signific<strong>an</strong>ce in the development<br />
<strong>of</strong> the Archae<strong>an</strong> craton. In: Glover JE <strong>an</strong>d Ho SE<br />
(Comps) Third International Archae<strong>an</strong> Symposium, extended<br />
volume, pp 43–44<br />
Wright C, Fernández LM (2003) International h<strong>an</strong>dbook <strong>of</strong><br />
<strong>earthquake</strong> <strong>an</strong>d engineering seismology (Hung K<strong>an</strong> Lee<br />
W) chapter 79. Elsevier, ISBN 0124406580<br />
Wright C, Kwadiba MTO, Kgasw<strong>an</strong>e EM, Simon RE (2002)<br />
The structure <strong>of</strong> the crust <strong>an</strong>d upper m<strong>an</strong>tle to depths <strong>of</strong> 320<br />
km beneath the Kaapvaal craton, from P-arrivals generated<br />
by regional <strong>earthquake</strong>s <strong>an</strong>d mining-induced tremors. J Afr<br />
Earth Sci 35(4):477–488