Scholarlink

<p>Monitoring land use and land cover is essential for understanding agricultural dynamics, especially in data-scarce regions such as the Tadla Plain (Morocco), where agriculture plays a significant economic role. This study investigates the potential of remote sensing and machine-learning techniques for crop classification using Normalized Difference Vegetation Index (NDVI) time-series data derived from Landsat 8, combined with a Support Vector Machine (SVM). A series of 15 Landsat 8 images, spanning September 2018 to July 2019, was processed with radiometric and atmospheric corrections and subsequently fused to produce phenological profiles from NDVI. These profiles enabled identification of the main crop types: sugar beet, alfalfa, cereals, arboriculture, and other crops. Classification using machine-learning algorithms, such as the Support Vector Machine (SVM), showed high class-separability (1.9-2.0) and an overall accuracy exceeding 97%, as validated against existing crop maps and field observations. Results indicate that sugar beet (3500&#xa0;ha) predominates, followed by cereals (2900&#xa0;ha), arboriculture (2400&#xa0;ha), and alfalfa (1200&#xa0;ha), consistent with local agro-industrial demand. This study demonstrates the effectiveness of integrating machine-learning algorithms with NDVI time-series for crop mapping in semi-arid areas. It proposes a replicable methodological framework for agricultural monitoring in data-scarce environments.</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

Remote sensing-based crop monitoring in data-scarce environments: lessons from the Tadla Plain (Morocco)

  • 期刊论文

  • 发表日期：

    2026年3月9日

查看全文
  • Driss Elhamdouni,
  • Abdessamad Hilali,
  • Ismail Karaoui,
  • Abdenbi Elaloui

摘要

Monitoring land use and land cover is essential for understanding agricultural dynamics, especially in data-scarce regions such as the Tadla Plain (Morocco), where agriculture plays a significant economic role. This study investigates the potential of remote sensing and machine-learning techniques for crop classification using Normalized Difference Vegetation Index (NDVI) time-series data derived from Landsat 8, combined with a Support Vector Machine (SVM). A series of 15 Landsat 8 images, spanning September 2018 to July 2019, was processed with radiometric and atmospheric corrections and subsequently fused to produce phenological profiles from NDVI. These profiles enabled identification of the main crop types: sugar beet, alfalfa, cereals, arboriculture, and other crops. Classification using machine-learning algorithms, such as the Support Vector Machine (SVM), showed high class-separability (1.9-2.0) and an overall accuracy exceeding 97%, as validated against existing crop maps and field observations. Results indicate that sugar beet (3500 ha) predominates, followed by cereals (2900 ha), arboriculture (2400 ha), and alfalfa (1200 ha), consistent with local agro-industrial demand. This study demonstrates the effectiveness of integrating machine-learning algorithms with NDVI time-series for crop mapping in semi-arid areas. It proposes a replicable methodological framework for agricultural monitoring in data-scarce environments.