This study integrates multi-temporal Landsat imagery (1993–2023), hydrological records (1990–2022), and UAV surveys (2023) to quantify riverbank line migration and channel evolution processes in the Tarim River of China. Utilising imagery for visual interpretation and Pearson Type III distribution for hydrological frequency analysis, three distinct channel reaches were identified: wandering (R1–R2), characterised by high erosion (mean migration rate: 41.9 m/a) and lateral instability; transitional (R3), exhibiting moderate migration (28.6 m/a); and meandering (R4–R5), displaying low migration (12.6 m/a) and a predominance of cutoff processes. Key findings reveal a nonlinear relationship between channel-forming discharge and bank migration rate, modelled using the GaussAmp function (R2 = 0.769): the migration rate peaks at 54.41 m/a when the channel-forming discharge reaches 184.38 m3/s, beyond which erosion and deposition tend to reach equilibrium. Human interventions since 2000, have reduced the average bank migration rate from 28.4 m/a to 22.3 m/a by enhancing vegetation cover and implementing flow-regulating structures, thereby mitigating local erosion by up to 160 m in engineered reaches. Spatially, the upstream wandering reaches exhibited the highest sediment transport capacity, resulting in frequent channel shifts, whereas the downstream meandering reaches demonstrated greater stability owing to reduced flow energy. Temporally, 2013–2018 experienced the most severe erosion (125.4 km2), corresponding to elevated channel-forming discharge (244.79 m3/s in R3), whereas 2018–2023 was characterised by dominant deposition (152.5 km2), attributed to improved ecological conditions.
