Annual Layering of Gas Ratios in Ice Cores

J P Severinghaus (Scripps Institution of Oceanography, University of California, San Diego, CA 92093; ph. 619 534-0000; fax 619 534-7889; Internet: jeffs@ucsd.edu)

Air trapped in bubbles in glacial ice has provided our primary source of information about past atmospheres, and more recently, has placed

constraints on the timing, speed, and origin of millenial-scale climate change. Unlike the well-known annual layering in the ice matrix, studies

of gases in ice cores have generally assumed that annual variations in gases are smoothed beyond recognition due to air mixing in the firn and

the gradual bubble close-off process. Likewise, samples for gas ratio measurements are generally taken with the assumption of homogeneity in gases on a hand sample scale. Here I propose that annual layering of some gas ratios is indeed present in glacial ice.

Air withdrawn from the firn at a variety of sites in Greenland and Antarctica shows sharp increases in oxygen and argon with depth in the firn-ice transition region. The simplest interpretation of these data are that summer layers have lower density and retain permeability longer than winter layers, and thus permit air to be sampled in the transition region. Under this scenario, oxygen and argon are preferentially excluded from the bubbles as they close off, and accumulate in the residual firn air in the summer layers. When the summer layers ultimately close off, bubbles in the summer layers should be enriched in oxygen and argon. Samples of firn air from Siple Dome, Antarctica, and bubble air taken from ice samples from the same drill hole, lend support to this

hypothesis. In the ice, oxygen is enriched by up to 0.3% in summer layers and typically depleted by 1% in winter layers. Argon shows similar patterns with smaller magnitudes. The covariation of O2/N2

with Ar/Kr in the firn air suggests that Ar/Kr is fractionated 37% as much as O2/N2 by the bubble close-off process. The mechanism of preferential exclusion is unknown but is probably dependent on

molecular size, as O2 is smaller than N2, and Ar is intermediate.