Reply to Comment on “On the recent bimodal magmatic processes and their rates in the Torfajökull–Veidivötn area, Iceland” by K.M. Cooper

Cooper [Cooper, K.M., 2009. Comment on “On the recent bimodal magmatic processes and their rates in the Torfajökull–Veidivötn area, Iceland” by K.M. Cooper. Earth Planet. Sci. Lett.] argues that pre-eruptive crustal magma residence times of 1.4–3.2 ka determined from 226Ra–230Th disequilibrium in co-eruptive Icelandic rhyolites and basalts are erroneous in Zellmer et al. [Zellmer, G.F., Rubin, K.H., Grönvold, K., Jurado-Chichay, Z., 2008. On the recent bimodal magmatic processes and their rates in the Torfajökull–Veidivötn area, Iceland. Earth and Planetary Science Letters, 269: 388–398]. The core of Cooperʹs comment is that our not having applied model parameterizations that predict differences in mineral-melt partitioning of Ra (DRa) and Ba (DBa) introduce non-equilibrium conditions yielding inaccurate isochron ages. Yet the actual correction arising from the DRa ≠ DBa model is trivial relative to the DRa = DBa model we employed for the two mineral phases upon which most of our age interpretations are based: magnetite (23 years) and clinopyroxene (2 years). Cooperʹs correction scheme might have a larger effect on our lone plagioclase separate were it possible to accurately predict what DRa/DBa to employ (correction with DRa/DBa = 0.15 proposed by Cooper [op. cit.], results in an impossible age). Yet even using an ad hoc 0.15 < DRa/DBa < 1 that could generate an allowable age would still not change our interpretation of when mineral formation began in these magmas. 238U, 232Th and Ba abundances in our mineral separates are within the range of literature values for those phases. But Cooper [Ibid.] also argues that these phase separates are highly impure, using non-realistic Th, U and Ba distribution coefficients (particularly for rhyolite phases and basaltic titanomagnetites) to estimate equilibrium concentrations relative to their host magmas. The Cooper comment thus also faults us for not correcting our data for the perceived impurities, using the flawed assumption that generic, theoretical Di values are apparently applicable to any magma composition. Such a correction also introduces ~ 10× greater uncertainty to age estimates (errors that are not adequately characterized in the comment or prior work using this correction scheme). Cooper subsequently asserts that our extremely well correlated data (R2 > 0.99) on Ba-normalized 230Th–226Ra activity diagrams are mixing arrays without age significance, but this assertion is not supported by careful analysis of the data. Zellmer et al. [Ibid.] discussed the effects of potential impurities and disagree that they disallow or dominate age interpretations of our data. The significant problems associated with the DRa ≠ DBa-and-phase-impurity correction scheme advocated by Cooper result in age estimates that are generally less precise and less accurate when compared to Ra–Th ages determined by the traditional 230Th–226Ra–Ba mineral isochron method, which generates superior results and should be incorporated into future studies, with consideration of DRa ≠ DBa when minerals are very young (generally < 50–300 years, depending on the phase) or when DRa >> DTh.