Tsunami simulations along the Eastern African coast from mega-earthquake sources in the Indian Ocean

The catastrophic 2004 Indonesian tsunami reached the shores of Eastern Africa, where it affected at least 12 countries and caused several hundred casualties, principally in Somalia. Significant variations in run-up were documented by various post-tsunami surveys (note that the latter remain incomplete, especially in Southern Tanzania and Mozambique). In a previous study, Okal et al. (S Afr J Geology 112:343–358, 2009) suggested that these variations could depend on the precise location of the tsunami sources, as a result of the combined effect of source directivity and refraction by irregular bathymetry. In this context, we present the results of a significantly enhanced study, which considers a total of twelve potential sites of mega-earthquakes, along both the Sunda Arc, and the Makran subduction zone. Numerical simulations are carried out at a total of 35 virtual gauges, spanning the East African coast from Socotra in the North to Port Elizabeth in the South, as well as adjoining islands (Madagascar, Comoros, Mascarenes), the Gulf of Aden, and distant islands in the far Southern Indian Ocean, from Amsterdam to Prince Edward. We conclude that at many locations, in particular in Mozambique, South Africa, Madagascar, and the Mascarenes, the 2004 tsunami (which to a large extent awakened the awareness of the continent to tsunami danger) may not have represented a worst case scenario. For tsunamis originating in the Makran, amplitudes larger than in 1945 require the simultaneous rupture of several blocks of the plate boundary, a scenario which, although not impossible, remains at this stage speculative.