CENOZOIC EXTENSION IN NORTHERN KENYA - A QUANTITATIVE MODEL OF RIFT BASIN DEVELOPMENT IN THE TURKANA REGION

Recent geological and geophysical studies in the Turkana region of the northern Kenya rift have highlighted the presence of a number of major sedimentary depocentres (Morley et al., 1992) overlying anomalously thin crust (Mechie et al., 1994). The region overlies a putative structure linking the Mesozoic-Palaeogene rifts of south Sudan and eastern Kenya (Ibrahim et al., 1991; Bosworth, 1992). The flexural cantilever model of continental extension (Kusznir et al., 1991) has been used to describe the rift evolution of the Turkana region, constrained by seismic reflection data and surface geology. Sections showing crustal structure, basin geometry and beta stretching factor profiles have been modelled for end Palaeogene (25 Ma), end Miocene (5 Ma) and end Pliocene (2 Ma). We also show that, whilst extension estimates giving a beta stretching factor estimate of 1.55-1.65 are consistent with geophysical data showing that the Moho has been shallowed to 20 km by extension, a significant portion of that extension may pre-date the Neogene. We propose that across the Turkana region during the Palaeogene a series of basins formed, linking known loci of extension in southern Sudan with extension in the Anza/Kaisut system and South Kerio basins in northern and central Kenya. These basins have no surface outcrop expression, but are defined by gravity anomalies and a number of pieces of indirect evidence. Continued extension on the later East African rift system resulted in reactivation of some of these structures in Neogene-Recent times, and the abandonment of others. Modelling of the rift basin geometry in the Turkana region using the flexural cantilever model gives a value df effective elastic thickness (T-e) of 3.5 km. Flexural bending of the lithosphere associated with extensional fault block rotations is shown (following Buck, 1988) to generate large bending stresses which give rise to substantial upper lithosphere brittle failure and plastic deformation, so producing low values of lithosphere flexural strength and T-e. The low values of T-e estimated locally for the rift region by the flexural cantilever model differ from the much larger values of T-e determined using gravity-topography coherence techniques (Ebinger et al., 1989) which sample regional lithosphere strength including that of strong continental shield outside the rift. The maximum beta stretching factor of 1.55-1.65 defined by the model is insufficient to generate the observed volcanics using published quantitative models of melt generation (McKenzie and Bickle, 1988). An anomalously hot mantle temperature is required to generate any melt at such low stretching factors. Whilst previous authors Karson and Curtis, 1989; Latin et al., 1993) have proposed that large volumes of melt have been intruded into and ponded within the lithosphere, we show that crustal underplating and intrusion do not significantly alter estimates of crustal thickness and extension across the rift system.