Monitoring the river ice phenology along the Inner Mongolia reach of the Yellow River using time-series images from landsat and Sentinel-2

Inner Mongolia section of the Yellow River in northern China. This study developed a segment-based area-ratio framework for the automated extraction of river-ice phenology utilizing multi-source optical remote sensing data from Landsat-8/9 and Sentinel-2A/B. Random Forest models were employed to differentiate between ice and water, and time-series variations in the ice-cover ratio facilitated the identification of freeze-up, break-up, and ice-covered duration along standardized river segments. Six complete ice seasons (2018–2023) were analyzed to characterize the spatiotemporal dynamics of river-ice formation and decay under the combined effects of temperature, flow, and reservoir regulation effects. The results demonstrated a distinct spatial regularity in river-ice evolution: freeze-up was initiated earlier downstream and progressed upstream, whereas break-up proceeded in the reverse direction. The duration of ice cover increased progressively downstream, reflecting the combined influence of thermal conditions and hydrodynamic regulation. River-ice processes are primarily controlled by temperature, with flow dynamics and reservoir operations exerting secondary effects that amplify spatial heterogeneity. The proposed segment-based phenology extraction framework provides a transferable methodology for large-scale river-ice monitoring, supporting hydrological research, climate assessment, and ice-flood hazard management in seasonally frozen basins. • Developed a remote sensing method for river ice phenology on the Yellow River. • Mapped freeze-up and break-up patterns using segment-based analysis. • River ice duration varies between reservoir-influenced and natural river sections. • Provided transferable methods for river ice monitoring in cold regions.