The electrical conductivity of the continental lithospheric mantle: new insights from integrated geophysical and petrological modelling. Application to the Kaapvaal Craton and Rehoboth Terrane, southern Africa

The electrical conductivity of mantle minerals is sensitive to temperature and also, but far less so, to compositional variations, and hence can be used to characterize the lithospheric and sub-lithospheric mantle structure. In recent years significant effort has gone into designing and executing appropriate experiments for measuring the electrical conductivity of most upper mantle minerals under relevant temperature and pressure conditions. These experimental observations are essential for understanding the electrical behaviour of mantle rocks in situ. The magnetotelluric (MT) method makes use of the relationship between the temporal variations of the Earth’s electric and magnetic fields to infer the subsurface conductivity distribution. A caveat of all MT inversions is that even if they derive perfectly valid conductivity models (i.e., models that satisfy the MT observations), it is generally not clear whether these models correspond with physically plausible or meaningful petro-physical conditions inside the Earth, i.e., the chemical (e.g., iron and water content), mineralogical, temperature and pressure conditions. Nor is it clear in general whether these conductivity models are consistent with other geophysical observables, such as the geoid, gravity field and surface heat-flow and elevation. The link between the modelled subsurface conductivities and subsurface physical properties, calibrated by laboratory measurements, is therefore often missed. We propose a more petro-physically driven approach to modelling MT data based on the software package LitMod. This software combines petrological and geophysical modelling of the lithosphere and sub-lithospheric upper mantle within an internally consistent thermodynamic-geophysical framework, where all relevant properties are functions of temperature, pressure and composition. In particular, LitMod is used in this work to define realistic temperature and pressure distributions within the upper mantle, and to characterize the mineral assemblages given bulk chemical compositions as well as water contents. This allows us to firstly define a bulk conductivity/resistivity model of the upper mantle based on laboratory and xenolith data for the most relevant mantle minerals and secondly to produce corresponding predicted MT responses that are compared with observed MT responses. We apply this technique to two Precambrian regions in southern Africa: the Rehoboth Terrane and the Kaapvaal Craton. In contrast to the relatively well-studied (and diamondiferous) Kaapvaal Craton, the deep lithospheric structure and temperature distribution of the Rehoboth Terrane remains poorly known. In this work we generate self-consistent lithospheric/sub-lithospheric mantle models of the two regions that simultaneously fit geophysical and petrological observables: MT data, surface elevation, surface heat-flow and chemical compositions derived from xenolith data. Critically, we also assess the extent to which the lithospheric and sub-lithospheric mantle might be wet or dry within each terrane and the implications of the (potentially depth-variable) hydration state with respect to the lithospheric evolution of each terrane.