Areal distribution of the oxygen-isotope ratio in Antarctica: an assessment based on multivariate models

Abstract

Mean oxygen-isotope ratio values relative to standard mean ocean water (δ18 O, in ‰) reported for 406 sites in Antarctica are compiled together with data on mean annual surface temperature, latitude, surface elevation, and mean annual shortest distance to open ocean denoted by the 20% sea-ice concentration boundary. Stepwise regression analyses with δ18O as the dependent variable are used as a model-building procedure based on statistical, rather than physical, criteria. Multivariate models sensitive to covariation between independent variables are defined using the whole dataset (N406 where N denotes the number of sites), as well as sub-sets for areas of conterminous grounded ice (N206) and ice shell (N110). The models show improvement over bivariate regression models. Distance to the open ocean enters all models at the second step. Inversions of the set and sub-set models applied to a database for 1351 gridpoint locations 100 km apart (it excludes the regions of Graham Land and eastern Palmer Land) are used to produce contoured distributions of δ18O. These may be used to assess the effects of atmospheric advection, as well as derive ice-flow adjustments for δ18O series obtained from deep-core or ablation-zone samples. Suggestions are made to improve model reliability.