Extended Data Fig. 3: One-dimensional steady-state models of the depth–age relationship of the ice column at the Prudhoe drill site.

For each model parameter in each model/panel, a range of five values is considered (see legend), and the resulting depth-age relationship shown is darker for increasing values. For each model/panel, the best estimate of the modern value is shown as a thicker dashed line. A): Nye (sandwich) model depends only on accumulation rate (bdot), the ice column deforms uniformly by pure shear and is not melting at the bed. B): Nye+melt model is the same as Nye (A), except that basal melting (mdot) is included. C): Dansgaard-Johnsen model, where the ice column of thickness H deforms by pure shear above a height h above the bed, and by simple shear below it. The first two models (Nye and Nye+melt) tend to produce younger ice columns, and regardless of parameter selection, they consistently indicate a completely Holocene ice column. The latter model (Dansgaard–Johnsen) indicates that a non-negligible basal layer of Pleistocene ice is possible at lower accumulation rates and higher basal shear layer thicknesses. From this initial modeling, we conclude that no past major basal shear or melting is required to reproduce the measured the δ¹⁸O drill ice chip record, and that simplest explanation for that record is an ice cap that regrew under a mean ice accumulation rate slightly lower than present.