Moussaid et al. (2026) K-means clustering applied to vegetation indices for mapping cultivated areas using high-resolution Moroccan Mohammed VI satellite imagery

Identification

• Journal: Scientific Reports
• Year: 2026
• Date: 2026-02-25
• Authors: Abdellatif Moussaid, Mohamed Bayad, Yousra Gamoussi, Yassir El Hamdouni, Hamza Briak
• DOI: 10.1038/s41598-026-41167-1

Research Groups

• Center for Sustainable Soil Sciences (C3S), College of Agriculture and Environmental Sciences (CAES), University Mohammed VI Polytechnic (UM6P), Ben Guerir, Morocco
• Center of Remote Sensing Applications (CRSA), College of Agriculture and Environmental Sciences (CAES), Mohammed VI Polytechnic University (UM6P), Ben Guerir, Morocco
• Department of Earth Sciences, Faculty of Sciences and Techniques of Tangier (FST), Abdelmalek Essaadi University (UAE), Tangier, Morocco

Short Summary

This study developed a pixel-based unsupervised classification method combining K-means clustering with vegetation indices (NDVI, MNDWI) and the Near-Infrared band to accurately map cultivated areas using high-resolution Moroccan Mohammed VI satellite imagery. The method achieved a low relative error of 1.41%, demonstrating its superior performance compared to traditional approaches for field-scale agricultural mapping.

Objective

• To develop and evaluate a pixel-based unsupervised classification approach for mapping cultivated land using high-resolution Moroccan Mohammed VI satellite imagery, integrating K-means clustering with spectral features derived from vegetation indices (NDVI, MNDWI) and the Near-Infrared band.

Study Configuration

• Spatial Scale: A 175-hectare agricultural region in northern Morocco.
• Temporal Scale: Not explicitly defined for imagery acquisition, but uses high-resolution satellite imagery.

Methodology and Data

• Models used: K-means clustering algorithm.
• Data sources: High-resolution imagery from the Moroccan Mohammed VI satellite, processed to derive Normalized Difference Vegetation Index (NDVI), Modified Normalized Difference Water Index (MNDWI), and Near-Infrared (NIR) band data.

Main Results

• The proposed method achieved a relative error of 1.41% in mapping cultivated areas.
• This performance significantly outperformed other classification methods:
  • NIR threshold-based classification: 7.2% error.
  • NDVI-based classification: 6.95% error.
  • Standard K-means classification using spectral bands only: 5.47% error.
• The output classified map successfully distinguished three classes: background, bare soil, and crop-dominated areas.

Contributions

• Presents an effective unsupervised classification approach for high-resolution satellite imagery by integrating K-means clustering with a combination of vegetation indices (NDVI, MNDWI) and the NIR band.
• Demonstrates the superior accuracy of this combined approach for cultivated land mapping compared to single-index or spectral-band-only methods.
• Highlights the potential of high-resolution Moroccan Mohammed VI satellite imagery for detailed field-scale agricultural mapping, supporting precision irrigation, and sustainable land management practices.

Funding

• This research received no external funding.

Citation

@article{Moussaid2026Kmeans,
  author = {Moussaid, Abdellatif and Bayad, Mohamed and Gamoussi, Yousra and Hamdouni, Yassir El and Briak, Hamza},
  title = {K-means clustering applied to vegetation indices for mapping cultivated areas using high-resolution Moroccan Mohammed VI satellite imagery},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-026-41167-1},
  url = {https://doi.org/10.1038/s41598-026-41167-1}
}

Generated by BiblioAssistant using gemini-2.5-flash (Google API)