Detecting sea ice freeze-onset from Sentinel-1 synthetic aperture radar

The Canadian Arctic Archipelago is undergoing rapid climate-driven changes, including prolonged melt seasons and delayed freeze-up that accelerate the loss of thick multi-year ice. Existing freeze-onset datasets rely on localized in situ observations or coarse-resolution NASA passive microwave (PMW) products at km scale. However, in situ measurements lack systematic spatial coverage, while PMW data suffer from severe land contamination in the narrow channels and inlets of the Archipelago, limiting their reliability in this complex coastal environment. Detecting freeze-onset using high-resolution synthetic aperture radar (SAR) have historically been challenging due to sea ice motion that introduce uncertainty by disrupting pixel-level correspondence and backscatter continuity across image sequences. This study introduces a novel approach that integrates temporal backscatter slope analysis with environmental constraints, such as surface air temperature thresholds to overcome motion-related challenges. The method enables robust detection of freeze-onset events across both multi-year ice and open-water areas at 40 m pixel resolution, producing spatially comprehensive maps of freeze-up timing across the Canadian Arctic. Analysis during 2015–2025 will reveal fine-scale spatial and temporal variability in freeze-up patterns within the Last Ice Area. This work will deliver the first high-resolution freeze-up dataset for the Canadian Arctic, providing a scalable foundation for pan-Arctic freeze-onset estimation.