Landslides pose significant geohazards in both mining and natural slopes, necessitating effective temporal prediction of failure for slope-scale mitigation. The conventional understanding of creep behavior, derived from controlled laboratory testing, often falls short in explaining the dynamic in-situ creep characteristics of heterogeneous soil slopes. This study presents in-situ creep assessments through continuous displacement monitoring of an inactive coal mine waste dump, employing Ground-Based Interferometric Synthetic Aperture Radar (GBInSAR). This study explores the applicability of the inverse velocity (IV) method for landslide failure prediction. Additionally, it evaluates the performance of two widely adopted filters, namely, moving average and exponential smoothing. In contrast to prior findings, the study reveals a distinct pattern in the IV vs. time curve, transitioning from an initial linear trend to a new steady-state. Notably, the research highlights instances of false predictions, underscoring the importance of considering potential stick-slip behavior, particularly in dump slopes. Furthermore, the findings on tension crack formation and propagation offer insights crucial for selecting an appropriate constitutive model tailored to the characteristics of the waste dump. The findings from this study contribute valuable knowledge to the field of slope stability assessment and aid in refining the methodologies for accurate landslide failure prediction in complex, real-world scenarios.

来源平台：MINING METALLURGY & EXPLORATION