V. Poggi, R. Durrheim, G.M. Tuluka, G.A. Weatherill, R. Gee, M. Pagani, A. Nyblade, D. Delvaux
The Sub-Saharan Africa (SSA) Earthquake Model was developed by GEM in collaboration with AfricaArray within the USAID-supported SSAHARA project. The original model is extensively described in Poggi et al. (2017), while an extended and improved version was developed in 2018 by introducing a procedure of earthquake-driven redistribution of activity rates (smoothed seismicity approach) on the previously defined source zones. Moreover, the current model includes now Madagascar, for which an ad-hoc seismicity analysis was carried out.
Information about the OQ model versions and input files can be found on the Results and Dissemination page.
The viewer below depicts the seismic sources and hazard results in terms of PGA for a return period of 475 years. Click on the menu in the upper right corner to select the layer.
The East African Rift System (EARS) is an example of an active continental rift system. This divergent plate boundary runs roughly north-south through eastern Africa, separating the Nubian and Somalian plates. It intersects the Afar depression in northern Ethiopia, where a triple junction connects it north-west to the Red Sea rift and north-east to the Gulf of Aden rift, which extends as far as the Indian Ocean Ridge. Towards the south, the EARS splits into two branches—the eastern and western rifts—that bracket the Tanzanian craton. The eastern rift extends along the coast of Mozambique into the Indian Ocean, and eventually joins the Southwest Indian Ocean Ridge (SWIR). The western rift continues through Lake Malawi into central Mozambique, with several splays that extend into continental Africa. The EARS was likely initiated in the region from the present-day Turkana Rift during the mid-Tertiary. The Western branch of the EARS formed subsequently around 25 Myr, simultaneously with the Eastern branch, within a spreading process that is still on going and is responsible for the largest seismicity experienced in the Africa continent.
An earthquake catalogue for Sub-Saharan Africa with homogenous magnitude representation (Mw) was obtained by merging available global catalogues (e.g. ISC-Reviewed, ISC-GEM, GCMT, GEM Historical Catalogue) with information from local agencies and regional projects, particularly from AfricaArray temporary deployments (e.g. The Tanzanian Broadband Seismic Experiment, The Ethiopian Plateau Catalogue, The AfricaArray Eastern Africa Seismic experiment).
The homogeneous catalogue (hereinafter SSA-GEM) was then declustered by removing fore- and aftershock sequences and seismic swarms, using the algorithm introduced by Gardner and Knopoff (1974). The declustered SSA-GEM catalogue consists of 7,259 events out of the original 29,803 in the magnitude range 3 ≤ Mw ≤ 7.53 (Figure 1).
Figure 1 - Distribution of earthquake events (Mw > 3) from the homogenised SSA-GEM earthquake catalogue. Names of the major rift systems associated with seismicity are indicated on the map with blue labels.
Seismic Source Characterisation
The study area was initially discretised into 21 independent source zones (Figure 2), following the guidelines proposed by Villanova et al. (2014) that provide a set of objective criteria to delineate regions of supposedly homogenous seismic potential. The main constraint for the development of the source model came from the analysis of the earthquake catalogue (stationarity of the completeness periods, evaluation of the mean activity rate, distribution of seismogenic depths) and from a set of geological and seismotectonic considerations, such as style, geometry, and distribution of existing faulting systems and their relation to the local stress and deformation regimes.
The source zones were gathered into six main tectonic domains, assumed to have comparable rheological and mechanical behaviour with respect to the underlying crustal geology under the regional stress regime. Source grouping is particularly useful for earthquake occurrence analysis in low seismicity regions, where the limited earthquake record might be insufficient for the proper calibration of poorly constrained seismicity parameters, such as the maximum magnitude or the slope (b-value) of the assumed frequency-magnitude occurrence model. Tectonic grouping was also used for the regional characterization of main faulting style and hypocentral depth distribution of the seismic source model.
Figure 2 - Source zonation model used in this study. Area sources belonging to the same tectonic group are represented with the same colour. The outermost red dashed line marks the PSHA calculation area. In the background is the SSA-GEM homogenised catalogue (non-declustered, Mw > 3) and the faults from the database of Macgregor (2015).
Seismicity in each area source is assumed to follow a double truncated Gutenberg-Richter magnitude occurrence relation (or magnitude-frequency distribution, MFD). Lower truncation is arbitrarily assigned to Mw 4.5.
Gutenberg-Richter b-values were calibrated for the whole catalogue and independently for each source group. Conversely, occurrence rates (a-values) were calculated separately for each source zone by imposing the previously calibrated b-values.
A different maximum magnitude (Mw-Max) estimate is derived independently for each source group as the largest observed event plus an arbitrary - although quite conservative - increment of 0.5 magnitude units. Seismiciy parameters are summarised in Table 1.
Table 1 - Seismicity parameters used in the SSA model. Mmax and b-values are consistent within source groups.
In a second step, to better represent the spatial variability of seismicity across the study area, the annual occurrence rates previously obtained for the homogeneous source zones were redistributed within each polygon using a procedure that accounts for the irregular spatial pattern of the observed events. The procedure shares some similarity with the popular smoothed seismicity approach (e.g. Frankel, 1995), but is more convenient in that a unique fit of the MFD is required for each zone, while the corresponding total earthquake occurrence is a-posteriori spatially reorganised as a function of the epicentral distance to all neighbouring events. Moreover, the combined use of zones gives the possibility to account for different modelling parameters (b-value, depth distribution, rupture mechanism) in separate regions.
Ground Motion Characterisation
Table 2 shows the ground motion logic tree, which distinguishes between two main tectonic domains: active shallow crust (ASC) for areas involving plate boundary segmentation, and and stable crust (SCC) for intra-plate areas. These correspond to Tectonic_Type_A and Tectonic_Type_E, respectively. Other tectonic regions, Tectonic_Type_B, Tectonic_Type_C and Tectonic_Type_D, are prescribed for transition zones of intermediate characteristics between SCC and ASC, in order to avoid abrupt variations of ground motion predicted by GMPEs calibrated for different tectonic settings.
For every tectonic region, epistemic uncertainty is considered by using multiple GMPEs, each with an associated logic tree weight. Four GMPEs were selected for this study, two models for ASC (Chiou and Youngs 2014; Akkar et al., 2014) and two models for SCC (Atkinson and Boore, 2006; Pezeshk et al., 2011). All GMPEs are assigned to each source zone, with the corresponding logic-tree weight varying with the likelihood for each specific tectonic type. Assignment of weights was agreed on the basis of the direct judgement of local seismotectonic conditions by a pool of experts from the region.
Table 2 - GMPEs used in the SSA model.
Hazard curves were computed with the OQ engine for peak ground acceleration (PGA) and spectral acceleration (SA) at 0.2s, 0.5s, 1.0s, and 2s. The computation was performed on a grid of 123924 sites (spaced at approximately 10 km) with reference soil conditions corresponding to a shear wave velocity in the upper 30 meters (Vs30) of 760-800 m/s.
The hazard map for PGA corresponding to a 10% probability of exceedance in 50 years (475 year return period), can be seen using the interactive viewer. For a more comprehensive set of hazard and risk results, please see the GEM Visualization Tools.
Poggi, V., Durrheim, R., Mavonga Tuluka, G., Weatherill, G., Gee, R., Pagani, M., Nyblade, A., Delvaux, D., 2017. Assessing Seismic Hazard of the East African Rift: a pilot study from GEM and AfricaArray. Bulletin of Earthquake Engineering. Volume 15, Issue 11, 4499–4529, DOI: 10.1007/s10518-017-0152-4
Vilanova SP, Nemser ES, Besana-Ostman GM, Bezzeghoud M, Borges JF, Da Silveira AB, Cabral J, Carvalho J, Cunha PP, Dias RP, Madeira J, Lopes FC, Oliveira CS, Perea H, García-Mayordomo J, Wong I, Arvidsson R, Fonseca JFBD (2014) Incorporating descriptive metadata into seismic source zone models for seismic-hazard assessment: a case study of the Azores-West Iberian region. Bull Seismol Soc Am 104:1212–1229