AI 中文总结
该研究扩展了White（1980）侵蚀公式，结合ROV剖面数据建立模型，发现其预测的芬布尔冰架前缘水位线侵蚀量仅为卫星观测退缩量的约1/2.4，剩余差异或源于破波湍流、崩冰等因素。
AI 中文摘要
海浪会侵蚀南极冰架前缘的水位线并刻蚀出热侵蚀凹痕。该凹痕无法被卫星观测到，但它会为前缘崩塌和游离式崩冰创造条件，因此凹痕生长与可观测前缘退缩之间的关联，对受海浪影响的冰架质量损失研究至关重要。我们于2024年1-2月在东南极洲的芬布尔冰架研究了这一关联。对广泛使用的White（1980）侵蚀公式从两方面进行了扩展：一是对不规则海浪进行分谱处理，二是采用考虑破碎的波浪剖面，该剖面兼顾了水下冰脚处的波浪浅水效应，而水下冰脚的形态由遥控潜水器（ROV）剖面测量揭示。采用逐小时ERA5海浪数据作为驱动，扩展模型在匹配的1月7日至2月27日窗口内预测出约110米的累计水位线侵蚀量。卫星观测显示的退缩量明显更大：基于S1引导的Sentinel-2平台断裂方法测得约264米，通过手动数字化Sentinel-1前缘在稍长窗口内测得约266米。在基线假设（α=1，ΔT_wi=1℃）下，模型预测的退缩量比观测值小约2.4倍。部分剩余差异可能源于水动力因素，因为模型未包含破波后的湍流；部分可能源于力学因素，因为崩塌和游离式崩冰可将凹痕侵蚀转化为更大的可观测退缩。未测量的近冰热驱动仍是首要不确定性，仅前缘位置观测无法区分这些贡献。
英文摘要
Ocean waves erode the waterline of Antarctic ice-shelf fronts and carve a thermo-erosional notch. The notch is hidden from satellites, yet it preconditions front collapse and footloose calving, so the link between notch growth and observable front retreat matters for how wave-exposed ice shelves lose mass. We study this link at the Fimbul Ice Shelf, East Antarctica, in January-February 2024. The widely used White (1980) erosion formulation is extended in two directions: a component-wise spectral treatment of irregular seas, and a breaking-aware wave profile that accounts for shoaling over the submerged ice foot revealed by remotely operated vehicle (ROV) profiling. Forced with hourly ERA5 waves, the extended model predicts about 110 m of cumulative waterline erosion over the matched 7 January-27 February window. Satellite observations show considerably more retreat: about 264 m from an S1-guided Sentinel-2 plateau-break method, and 266 m from manually digitised Sentinel-1 fronts over a slightly longer window. Under the baseline assumptions (alpha = 1, Delta T_wi = 1 degree C), the model therefore falls short of the observed retreat by a factor of about 2.4. Part of this residual may be hydrodynamic, since post-breaking turbulence is not represented; part may be mechanical, because collapse and footloose calving can convert notch erosion into larger observable retreat. Unmeasured near-ice thermal driving remains a first-order uncertainty, and front-position observations alone cannot separate these contributions.
Comments34 pages, 11 figures, 2 tables. Code and data: https://doi.org/10.5281/zenodo.21622437
