近日，陕西师范大学段克勤团队研究了基于物理过程的冰川演化模型在祁连山老虎沟12号高山冰川中的应用。相关论文于2025年4月8日发表在《中国科学：地球科学》杂志上。

在过去的几十年里，青藏高原冰川的消融速度加快。了解冰川演变背后的机制，并根据物理过程预测未来的变化，对于管理冰川的变化及其影响至关重要。然而，目前的模型依赖于经验关系，这过于简化了冰川的热液过程，忽视了地形、反照率和辐射引起的冰川变异的空间异质性，导致模拟存在重大不确定性。

为了更好地了解未观测地区的冰川变化，研究组建立了一个分布式冰川质量能量平衡模型，该模型充分考虑了地形对太阳辐射的影响，并通过深度学习确定了冰川表面反照率。此外，它还与冰川动力学相结合，创建了一个三维准物理过程冰川演化模型（QPGM）。

以祁连山老虎沟12号冰川为例，研究组证明了QPGM在模拟高山冰川变化方面的强适用性。该模型的预测表明，根据SSP2-4.5，到本世纪末，冰川的质量可能会减少60%，而根据SSP5-8.5，冰川可能会完全消失。因此，QPGM代表了一种新的改进的冰川模拟和预测方法。

附：英文原文

Title: Glacier evolution model based on physical processes: Application to alpine glacier Laohugou No. 12, Qilian Mountains

Author: Keqin DUAN, Qiong WANG, Tandong YAO, Ninglian WANG, Jinping HE, Wei SHANG, Jiajia JIANG

Issue&Volume: 2025/04/08

Abstract: The ablation of glaciers on the Tibetan Plateau has accelerated over the past decades. Understanding the mechanism behind glacier evolution and projecting future variations based on physical processes is crucial for managing changes in glaciers and their impacts. However, current models rely on empirical relationships, which oversimplify the hydrothermal processes of glaciers and neglect the spatial heterogeneity of glacier variability caused by topography, albedo, and radiation, leading to significant uncertainties in simulation. To better understand glacier variations in unobserved regions, we established a distributed glacier mass-energy balance model that fully considers the impact of topography on solar radiation and determines glacial surface albedo through deep learning. Additionally, it is coupled with glacier dynamics to create a 3D Quasi-physical Process Glacier evolution Model (QPGM). With the Laohugou No. 12 glacier in the Qilian Mountains as an example, the strong applicability of the QPGM in simulating alpine glacier variations is demonstrated. The model’s projections suggest that the glacier could lose 60% of its mass by the end of this century under SSP2-4.5 and could completely disappear under SSP5-8.5. Thus, the QPGM represents a new and improved approach to glacier simulation and projection.

DOI: 10.1007/s11430-024-1524-8

Source: https://www.sciengine.com/SCES/doi/10.1007/s11430-024-1524-8