Land use mapping for the forest-steppe ecotone in the Greater Khingan Mountains, 2019 to 2021
Generating accurate thematic land use maps is importance in ecologically vulnerable regions, especially considering the challenges associated with extracting the forest-steppe ecotone and its associated uncertainties and high error rates. By employing the Principal Component Analysis (PCA) method to integrate Sentinel-1 and Sentinel-2 imagery, high-resolution (10 meters) land use cover products were generated for the forest-steppe ecotone of the Greater Khingan Mountains from 2019 to 2021. The classification process utilized prior knowledge and an object-oriented classification-based approach. The main objective was to evaluate the accuracy improvement achieved through the integration of multi-source remote sensing data, while highlighting the advantages of the object-based classification method and comparing it with existing products. The results demonstrated a significant enhancement in classification accuracy, surpassing the accuracy obtained from individual Sentinel-1 or Sentinel-2 images. Furthermore, the object-oriented analysis approach yielded classification results that more accurately represented real-world land cover conditions while reducing salt-and-pepper noise. The research also showcased superior accuracy in delineating complex riverine wetlands, outperforming other existing land use/land cover (LULC) datasets. The generated 10m land use products provide valuable information for supporting sustainable development, effective management, and ecological assessment and conservation efforts in the Greater Khingan Mountains.