A finite-difference approach for simulating ground responses in sedimentary basins: Quantitative modelling of the Nile Valley, Egypt

Abstract

On 1992 October 12, Egypt was struck by an earthquake with Mb = 5.9 or equivalently Ms = 5.2. It occurred southwest of Greater Cairo (Cairo and Giza) and was widely felt from Alexandria in the north to Aswan in the south. This earthquake also caused widespread destruction and damage, mostly concentrated in greater Cairo, along the Nile Valley and in the Nile Delta area. The extent of damage caused by this earthquake motivated this study. We computed a 2-D finite-difference (FD) synthetic wavefield for a prototype model of the Nile Valley with a maximum sedimentary basin thickness of 5 km and assumed uniform crustal thickness of 35 km. A variant of the latter model tested is a thin crust (Moho bump) beneath the basin. We used a 2.5 Hz Ricker wavelet source that excites both P and/or S waves, initiated as point sources. Focal depths vary from 5 to 20 km to simulate different scenarios. We investigated how various model features such as source location (focal depth and epicentral distance), deep sedimentary basin, near-surface attenuation and crustal thinning beneath the sedimentary basin may influence the wavefield across the Nile Valley. The outstanding result of the wavefield simulations is that the Nile Valley sedimentary basin and the associated strong impedance contrast produce significant focusing and amplification effects. It was found that the peak amplitudes might vary by one order of magnitude in the 3-8 Hz frequency band, across the Nile Valley. The high vulnerability of the buildings in Greater Cairo in combination with focusing/amplification in the Nile Valley explain qualitatively the extensive damage in the city itself even for moderate size earthquakes. In fact, the predicted area of high amplification coincides closely with that of greatest housing damage. The effects of near-surface attenuation in the uppermost 500 m within the basin and/or crustal thinning beneath the basin is less significant. These results are interesting on two accounts: (1) the extent of variation in the synthetic wavefield will always be less than that observed, but unfortunately there are no Egyptian records to corroborate these results and (2) wavefield variations are significantly larger than those deduced from ranges of Q-variations and seismotectonic modelling as used in most seismic risk analysis. These results imply that the wavefield simulations can be very informative and provide realistic results in particular when accurate structural model information is available and even more so if the synthetics can be corroborated with dense seismic network recordings.