Ice at depth in ice-stream shear margins is thought to commonly be temperate, with interstitial meltwater that softens ice. Models that include this softening extrapolate results of a single experimental study in which ice effective viscosity decreased by a factor of ~3 over water contents of ~0.01 to 0.8%. Modeling indicates this softening by water localizes strain in shear margins and through shear heating increases meltwater at the bed, enhancing basal slip. To extend data to higher water contents, we shear lab-made ice in confined compression with a large ring-shear device. Ice rings with initial mean grain sizes of 2-4 mm are kept at the pressure-melting temperature and sheared at controlled rates with peak stresses of ~0.06-0.20 MPa, spanning most of the estimated shear-stress range in West Antarctic shear margins. Final mean grain sizes are 8-13 mm. Water content is measured by inducing a freezing front at the ice-ring edges, tracking its movement inward with thermistors, and fitting the data with solutions of the relevant Stefan problem. Results indicate two creep regimes, below and above a water content of ~0.6%. Comparison of effective viscosity values in secondary creep with those of tertiary creep from the earlier experimental study indicate that for water contents of 0.2%-0.6%, viscosity in secondary creep is about twice as sensitive to water content than for ice sheared to tertiary creep. Above water contents of 0.6%, viscosity values in secondary creep are within 25% of those of tertiary creep, suggesting a stress-limiting mechanism at water contents greater than 0.6% that is insensitive to ice fabric development in tertiary creep. At water contents of ~0.6-1.7%, effective viscosity is independent of water content, and ice is nearly linear-viscous. Minimization of intercrystalline stress heterogeneity by grain-scale melting and refreezing at rates that approach an upper bound as grain-boundary water films thicken might account for the two regimes.

中文翻译：

---

间隙水软化温带冰

冰流切变边缘深处的冰通常被认为是温带的，间质融水可以软化冰。包含这种软化的模型推断出单项实验研究的结果，其中冰的有效粘度降低了约 3 倍的水含量，约为 0.01 至 0.8%。建模表明，这种水软化使剪切边缘的应变局部化，并且通过剪切加热增加了床层的融水，增强了基础滑移。为了将数据扩展到更高的水含量，我们使用大型环形剪切装置在有限压缩中剪切实验室制造的冰。初始平均粒度为 2-4 毫米的冰环保持在压力熔化温度下并以受控速率剪切，峰值应力约为 0.06-0.20 兆帕，跨越南极西部剪切边缘的大部分估计剪切应力范围。最终平均晶粒尺寸为 8-13 毫米。通过在冰环边缘诱导冻结锋，用热敏电阻跟踪其向内移动，并将数据与相关 Stefan 问题的解决方案拟合，来测量水含量。结果表明两种蠕变状态，低于和高于约 0.6% 的水含量。二次蠕变中的有效粘度值与早期实验研究中三次蠕变的有效粘度值的比较表明，对于 0.2%-0.6% 的水含量，二次蠕变中的粘度对水含量的敏感度约为冰剪切至三次蠕变的两倍。高于 0.6% 的水含量，二次蠕变的粘度值在三次蠕变的 25% 以内，表明水含量大于 0.6% 时存在应力限制机制，对三次蠕变中的冰结构发展不敏感。在 ~0.6-1.7% 的水含量下，有效粘度与水含量无关，冰几乎是线性粘性的。当晶界水膜增厚时，通过晶粒尺度熔化和再冻结以接近上限的速率使晶间应力不均匀性最小化可能是这两种状态的原因。