Dou et al. (2026) High-resolution annual desertification mapping in northern China using a novel comprehensive desertification index and unsupervised algorithm

Identification

• Journal: Remote Sensing of Environment
• Year: 2026
• Date: 2026-01-08
• Authors: Yaqing Dou, Huaiqing Zhang, Hua Sun, Hui Lin, Yang Liu, Meng Zhang
• DOI: 10.1016/j.rse.2025.115230

Research Groups

• Hunan Provincial Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security, Central South University of Forestry and Technology, Changsha 410004, China
• Research Center of Forestry Remote Sensing & Information Engineering, Changsha 410004, China
• Research Institute of Forest Resources Information Techniques, Chinese Academy of Forestry, Beijing 100091, China

Short Summary

This study developed a novel Comprehensive Desertification Index (CDI) using Sentinel-1/2 data and applied a Gaussian Mixture Model (GMM) to create the first 10 m-resolution annual desertification dataset for northern China (2016–2024), achieving high accuracy and robust spatio-temporal reliability.

Objective

• To develop a robust, high-resolution method for annual desertification mapping in northern China and generate a corresponding dataset (NCDMD) by constructing a Comprehensive Desertification Index (CDI) and applying an unsupervised algorithm.

Study Configuration

• Spatial Scale: Northern China, 10 meter resolution.
• Temporal Scale: Annual, 2016–2024.

Methodology and Data

• Models used: Comprehensive Desertification Index (CDI), Gaussian Mixture Model (GMM), Pixel Dichotomy Model (FVC-based), DDI feature space method, K-Means, MiniBatch K-Means, Jenks natural breaks, Weka LVQ algorithms (for comparison).
• Data sources: Sentinel-1 satellite data, Sentinel-2 satellite data, field survey data (for validation).

Main Results

• The CDI-GMM method achieved an overall accuracy of 93.5 % for desertification scope and 86.4 % for desertification degree in northern China in 2019, validated with field survey data.
• Traditional approaches showed significantly lower accuracy: the pixel dichotomy model (FVC-based) achieved 82.2 % for scope and 50.3 % for degree, while the DDI feature space method reached 86.1 % for scope and 64.2 % for degree.
• Comparative experiments with five unsupervised classification methods indicated that CDI combined with the GMM clustering algorithm optimized desertification extraction with an overall classification accuracy exceeding 93 %.
• The NCDMD dataset maintained consistent desertification mapping accuracies above 83 % throughout the 2016–2024 period, demonstrating robust spatio-temporal reliability.

Contributions

• Developed a novel Comprehensive Desertification Index (CDI) by integrating multisource remote sensing data (Sentinel-1/2) based on the multidimensional driving mechanisms of desertification.
• Generated the first 10 m-resolution annual desertification dataset for northern China (NCDMD, 2016–2024), filling a gap in high-precision desertification monitoring in China.
• Provided a robust and high-precision method for large-scale desertification monitoring, outperforming traditional approaches in accuracy and stability.
• Offers a nationally significant high-resolution base dataset to support ecological restoration assessment and land management policy-making.

Funding

• Not specified in the provided text.

Citation

@article{Dou2026Highresolution,
  author = {Dou, Yaqing and Zhang, Huaiqing and Sun, Hua and Lin, Hui and Liu, Yang and Zhang, Meng},
  title = {High-resolution annual desertification mapping in northern China using a novel comprehensive desertification index and unsupervised algorithm},
  journal = {Remote Sensing of Environment},
  year = {2026},
  doi = {10.1016/j.rse.2025.115230},
  url = {https://doi.org/10.1016/j.rse.2025.115230}
}