Development and Application of Remote Sensing-based Land-use Classification System for the Zhoushan Archipelago

Current land use classification systems using remote sensing are primarily oriented to terrestrial areas, without accounting for the complex marine environment and periodic tidal influences, leading to poor applicability and low information extraction accuracy in island areas. As the largest island group in China, Zhoushan Archipelago serves as a vital platform for safeguarding oceanic ecosystems and expanding the scope of the blue economy. Focused on the complexity of the geographical units of Zhoushan Archipelago and proceeding from the perspective of land-sea coupling and the needs of long-term research, the remote sensing classification system for Zhoushan Archipelago was constructed while fully considered the particularity of the geographical environment and the characte-ristics of land use patterns. First, a land use remote sensing classification system was constructed based on 30 m resolution Landsat remote sensing images. Second, multiple feature parameters were integrated, and a random forest algorithm was introduced to extract land use information. Finally, the spatiotemporal evolution patterns of land use in the Zhoushan Archipelago were analyzed. The research results showed that: (1) By incorporating the characte-ristics of periodic tidal inundation and land-water interactions, the land-use remote-sensing classification system developed for the Zhoushan Archipelago can more accurately represent the natural resource endowments and land-use patterns of the region. (2) Integrating multi-feature parameters, the random forest algorithm(RFA) demonstrated strong performance in land use information retrieval from remote sensing data of the Zhoushan Archipelago between 1985 and 2024, achieving an average overall accuracy of 93.56% and a Kappa coefficient of 0.93. (3) Du-ring 1985-2024, the cumulative land area of the Zhoushan Archipelago increased from 1 274.42 km² to 1 456.57 km², with an annual increment of approximately 4.67 km². Urbanization triggered dramatic expansion of construction land, increasing from 47.62 km² to 426.50 km², expanding to approximately 8.96 times its 1985 level. (4) The area of salt pans was comparable to that of construction land in 1985. However, due to industrial transformation, the area of salt pans declined by up to 94.88%, resulting in near total loss. Driven by the fishery economy, aquaculture ponds areas experienced substantial expansion, from 8.08 km² to 35.09 km², expanding to approximately 4.34 times its 1985 level. Coastal reclamation, salt marsh vegetation, and bare intertidal flats have been significantly affected by human activities, exhibiting pronounced stage-specific variations. Coastal seawater and bare intertidal flats were the primary sources for coastal reclamation. The newly reclaimed land was predominantly used for construction land, cropland/grassland, and aquaculture ponds, with cumulative increases of 119.12 km², 29.04 km², and 17.66 km², respectively. In conclusion, the classification system developed in this study can effectively characterize the natural and anthropogenic attributes of the land-use pattern in the Zhoushan Archipelago, thereby supporting the sustainable utilization and management of regional land resources, and providing a scientific refe-rence for more comprehensive and in-depth island land use research.