Scholarlink

<p>The Northeast region serves as a vital grain production base in China, with an increasingly prominent role in national food security. This study integrated existing crop classification products and land cover data to extract samples of rice, maize, soybean, and other land cover types within Northeast China covering the period 1985-2024. An inter-annual sample migration method based on Spectral Angle Mapper (SAM) was adopted to construct a comprehensive sample database, to be used as input for crop type classification, covering the entire study period. By leveraging Landsat imagery and applying the Random Forest algorithm, we generated 30-meter-resolution crop maps for rice, maize, and soybean from 1985 to 2024. The resulting maps achieved high accuracy, with crop area estimates highly consistent with municipal-level statistical data. Among the three provinces, the maximum overall accuracy and Kappa coefficient could reach 0.99. Liaoning Province achieved the best average classification accuracy, with mean overall accuracy and Kappa values of 0.87 and 0.81, respectively. These findings provide a robust data foundation for ensuring food security and informing agricultural management strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Long-term crop mapping in the three provinces of Northeast China based on multi-source sample fusion

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月21日

查看全文
  • Zhaoyang Bai,
  • Maofang Gao,
  • Fukang Feng,
  • Shilei Li,
  • Qiang Li,
  • Guofei Shang,
  • Jianxi Huang,
  • Zhao-Liang Li,
  • Xia Zhang,
  • Gianfausto Bottini,
  • Francesco N. Tubiello

摘要

The Northeast region serves as a vital grain production base in China, with an increasingly prominent role in national food security. This study integrated existing crop classification products and land cover data to extract samples of rice, maize, soybean, and other land cover types within Northeast China covering the period 1985-2024. An inter-annual sample migration method based on Spectral Angle Mapper (SAM) was adopted to construct a comprehensive sample database, to be used as input for crop type classification, covering the entire study period. By leveraging Landsat imagery and applying the Random Forest algorithm, we generated 30-meter-resolution crop maps for rice, maize, and soybean from 1985 to 2024. The resulting maps achieved high accuracy, with crop area estimates highly consistent with municipal-level statistical data. Among the three provinces, the maximum overall accuracy and Kappa coefficient could reach 0.99. Liaoning Province achieved the best average classification accuracy, with mean overall accuracy and Kappa values of 0.87 and 0.81, respectively. These findings provide a robust data foundation for ensuring food security and informing agricultural management strategies.