Interannual Glacier Variability and Accelerated Albedo Decline in Northeastern Tibetan Plateau: Multidecadal Remote Sensing Insights (1986–2024)

Abstract

Mountain glaciers are key indicators of climate change, with their retreat and surface albedo variations exerting significant impacts on regional water resources and surface energy balance. This study introduces an innovative machine-learning framework that combines high-resolution Landsat imagery with moderate-resolution imaging spectroradiometer-derived albedo products to enable precise annual glacier boundary extraction and comprehensive assessment of long-term glacier surface albedo dynamics. Applying this approach to the Qilian Mountains National Park (QMNP), we achieve a boundary delineation accuracy of 98.65%, with a Kappa coefficient of 0.98, outperforming the conventional methods and exhibiting consistency exceeding 90% when benchmarked against the Randolph Glacier Inventory V7.0 dataset. Since 1986, QMNP glaciers have undergone significant retreat, losing 686.27 km2 (40.03%) of their total area, with the mean glacier size decreasing from 0.81 to 0.65 km2, and over 500 small- to medium-sized glaciers (≤1 km2) disappearing entirely. Notably, a pronounced decline in glacier surface albedo has been observed since 2019, coinciding with the rapid expansion of debris-covered glaciers, which have surged by more than 12-fold since 1986. Our analysis reveals a strong negative correlation between air temperature and both glacier extent and surface albedo, highlighting air temperature as the dominant driver of glacier retreat and albedo degradation. In contrast, the influence of precipitation is weaker and more variable, suggesting a more complex role in glacier mass balance. These findings provide critical insights into glacier–climate interactions and their cascading effects on hydrological systems, offering a transferable framework for global assessments of glacier change and climate responses.