In defence of a resistive oceanic upper mantle: reply to a Comment by Tarits, Chave and Schultz

Abstract

In a previous paper (Heinson & Constable 1992) we discussed the effect of coastlines on sea-floor magnetotelluric (MT) data collected deep within the ocean basins. In order to quantify the geomagnetic coast effect we developed a model for oceanic upper mantle electrical conductivity independent of previous sea-floor MT interpretations, based on sea-floor controlled-source electromagnetic (CSEM) soundings, laboratory studies of mantle materials, and global geomagnetic-sounding estimates for lower mantle conductivity. We demonstrated that the coast effect of our model was large, was not manifest as severe anisotropy in the MT response if the ocean basins were bounded by coastlines on three or more sides, and that a qualitative agreement existed between the MT response of our simple coastline model and published sea-floor MT data. The implication of the latter is that by including the effect of the coastlines additional mechanisms for enhanced upper mantle conductivity, such as volatiles, carbon, melt, hydrogen and other factors, may not be required by the MT data. Tarits, Chave & Schultz (1993) question the validity of a large number of aspects of our work. Their principal conclusions are that (a) horizontal or vertical leakage paths between the ocean and other oceans or the mantle will substantially reduce the magnitude of the ocean-wide coast effect; (b) the principle of extrapolating laboratory electrical-conductivity measurements to mantle conditions is unreliable from uncertainties in the understanding of pressure, temperature and petrological states of the upper mantle; (c) published sea-floor MT data sets examined by us were of such vintage that any conclusions drawn were unreliable or incorrect; and (d) the conductivity profile compiled by Chave, Flosadottir & Cox (1990), based on 1-D interpretations of sea-floor MT and CSEM data that took no account of the coast effect, is consequently a more reliable estimator of mantle-conductivity structure than our model. While all aspects of Tarits et d’s comments were addressed to some extent in our original paper, we did not investigate all details fully either in the interests of conciseness, or because the impact on our conclusions was minimal. We will take this opportunity to delve further into some of these issues, but before considering the complications it will be useful to reiterate the motivation for our work. First, Oldenburg’s (1981) and Oldenburg, Whittal & Parker’s (1984) 1-D interpretations of three sea-floor MT data sets included a high-conductivity zone (HCZ) in the oceanic upper mantle that decreased in magnitude and deepened with increasing age. The existence of a HCZ has become a paradigm not only amongst the electromagnetic community but also with seismologists, petrologists and mineral physicists. Many entire papers are written on the subject, attempting to explain the disparities between laboratory measurements of olivine conductivity at upper mantle temperatures and the HCZ, correlating the HCZ with seismic low-velocity zones (LVZ), examining the temporal evolution of the HCZ, etc. The impact of Oldenburg’s work has been so large that it seemed appropriate to re-examine the basis of the HCZ and question whether it is indeed required by the data. Note that