Strasser, F. O.

The preparation of earthquake catalogues for seismic hazard analysis requires the use of uniform parameters, in particular for magnitudes, although the original data include a variety of formats, such as macroseismic intensities and various instrumental mag- nitude scales. In regions of low seismicity, such as South Africa, data are generally sparse and not always suffi- cient to develop locally calibrated conversion relations. They can nevertheless be used to test the applicability of imported conversion relations, as well as their consis- tency. The Koffiefontein region of South Africa pro- vides a good test case in view of its somewhat higher level of seismicity, and central geographic location with- in the country. This paper reviews determinations of location and magnitude parameters for a suite of moderate-to-large earthquakes that have occurred in this region between 1903 and 1985.

The seismicity of Grahamstown, in the Eastern Cape Province of South Africa, for the years between 1820 and 1936 is investigated with recourse to contemporaneous documentary sources, leading to the development of a seismic history incorporating consideration of the broader geo-political context. Individual studies of five regional events that were felt in Grahamstown during that period, and of one that was not, are presented. Each study includes the development of a full set of intensity data points (IDPs), which are used to determine reappraised epicentral locations and magnitudes, some of which differ significantly from previously listed parameters. The results thus obtained highlight the value of seeking out additional contemporary sources from different language groups when revisiting the source parameters of earthquakes for which no or only very limited instrumental information is available.

The influence of style-of-faulting on strong groundmotions has been the subject of debate for some time. Although some controversy persists, the general consensus is that ground motions produced by reverse faults are higher than those produced by normal faults, whereas motions from strike-slip faults are somewhere in between. In a recent article, Convertito and Herrero (2004) derived a correction factor for focal mechanism to be applied to predictive equations. This issue was previously addressed by Bommer et al. (2003). Although this article is cited by Convertito and Herrero, it seems that its aims and scope were not well understood, and we would therefore like to clarify what the method presented therein entails, especially because we feel that Convertito and Herrero’s approach of characterizing focal mechanisms based solely on the radiation pattern is difficult to justify. After presenting their correction scheme, Convertito and Herrero go on to present an implementation of probabilistic seismic hazard analysis (PSHA) explicitly accounting for focal mechanism. This represents a real innovation in terms of methodology because it allows propagation of the improvements in ground-motion prediction gained through the focal-mechanism adjustments to hazard estimation. Characterizing the dominant scenario in terms of focal mechanism furthermore has the advantage of providing constraints for numerical simulations that are derived directly from the hazard computation, rather than from arbitrary assumptions. However, in our opinion, the methodology presented by Convertito and Herrero has some serious shortcomings which would need to be addressed before it can lead to improvements of the PSHA methodology. Our discussion includes a comparison with the new Italian seismic hazard map, which was derived using the Bommer et al. (2003) adjustment methodology.

A database of intensity observations from instrumentally-recorded earthquakes in South Africa has been compiled as a contribution to the characterisation of seismic hazard. The database contains about 1,000 intensity data points (IDPs) that have been assigned from macroseismic observations retrieved from newspaper reports and questionnaires, and also digitised from previously published isoseismal maps. The database includes IDPs from 57 earthquakes with magnitudes in the range of Mw 2.2 to 6.4, for epicentral distances up to 1,000 km. Sixteen events have 20 or more IDPs, with half of these events having more than 80 IDPs. The database is dominated by relatively low intensity values, mostly determined from human perception of shaking rather than structural damage. However, 19 IDPs correspond to intensity values greater than VI MMI-56. Using geological maps of South Africa, the sites of 60% the IDPs were geologically classified as either ‘rock’ or ‘soil’, the uncertainty in locations precluding such a classification for the remaining data points. A few of the IDPs identified as being from soil sites appear to be strongly influenced by site effects and these were removed from the trimmed database created for exploring ground-motion levels. The trimmed database included 15 earthquakes which have a minimum of five useful IDPs, excluding those with intensity MMI=I and those based on a single observation. After removing such points, and those identified as clear ‘outliers’, a total of 436 useful IDPs were selected.