Pervasive diffusion of climate signals recorded in ice-vein ionic impurities

Abstract. A theory of vein impurity transport conceived two decades ago predicts that signals in the bulk concentration of soluble ions in ice migrate under a temperature gradient. If valid, it would mean that some palaeoclimatic signals deep in ice cores (signals from vein impurities as opposed to matrix/grain-boundary impurities) suffer displacements that upset their dating and alignment with other proxies. We revisit the vein physical interactions to show that a strong diffusion prevents such signals from surviving into deep ice. It arises because the Gibbs–Thomson effect, which the original theory had neglected, perturbs the impurity concentration of the vein water wherever the bulk impurity concentration carries a signal. Thus no distinct vein signals will reach a depth where their displacement matters; accordingly, the palaeoclimatic concern posed by the original theory no longer stands. Simulations with signal peaks introduced in shallow ice at the GRIP and EPICA Dome C ice-core sites confirm that rapid damping and broadening eradicates their form by two-thirds way down the ice column; artificially reducing the solute diffusivity in water (to mimic partially-connected veins) by 10³ times or more is necessary for signals to penetrate into the lowest several hundred metres with minimal loss of amplitude. The deep solute peaks observed in ice cores can only be explained by widespread vein disconnection or a dominance of matrix/grain-boundary impurities at depth (including their recent transfer to veins); in either case, the deep peaks would not have displaced far. Decomposing the vein and matrix impurity contributions will aid robust reconstruction from ion records.