A Hierarchical Framework for Arctic Shoreline Type Mapping Using SAR and Optical Satellite Imagery

Although intertidal-zone delineation and shoreline or coastal geomorphological classification have been extensively investigated, relatively few studies have integrated these complementary information layers within a unified and operational remote-sensing framework for oil-spill emergency response. With increasing development and marine transportation activities in the Arctic, detailed and spatially explicit shoreline information is essential for oil-spill response planning and coastal hazard assessment along the Canadian Arctic coast. A key challenge is to reliably characterize the spatial extent and geomorphological variability of Arctic intertidal shorelines under substantial temporal variability in tidal inundation and environmental conditions. This study presents a hierarchical shoreline classification framework designed to support oil-spill emergency preparedness and response planning. The framework integrates Sentinel-1 synthetic aperture radar (SAR) backscatter, temporal-change and structural descriptors, and Sentinel-2 water-sensitive indices to derive a composite shore-zone proxy that captures spatial and temporal information on tidal inundation and coastal surface characteristics. This proxy supports Level 1 shore-zone mapping, while high-resolution optical imagery enables subsequent Level 2 shoreline-type classification. The framework was evaluated in four Canadian Arctic coastal areas representing diverse coastal landscapes and environmental conditions, including differences in tidal regimes, water turbidity, floating ice, and geomorphology. Results indicated that mapping accuracy remains constrained for tidal zones and substrates smaller than or comparable to the sensor’s spatial resolution. For Level 1 shore-zone mapping, unsupervised classification of the temporal and composite indices achieved overall mapping accuracies of 56–73% for the Upper Shore Zone across three study areas. For Level 2 shoreline-type mapping, Random Forest classification of high-resolution satellite imagery, including WorldView and GeoEye imagery, achieved overall accuracies of 81–85% across the four study areas. These results demonstrate the potential of the composite shore-zone proxy to integrate complementary SAR and optical information for consistent characterization of Arctic shore zones, while the subsequent shoreline-type classification provides additional geomorphological detail. Together, the hierarchical products provide spatially explicit shoreline information that can support the identification and prioritization of shoreline areas for oil-spill response planning and resource allocation.