Scholarlink

<p>High-resolution, accurate mapping of crops is critical to enhance food security, resource efficiency, and policy effectiveness in agriculture across Africa, where maize remains a crucial staple crop. However, mosaic landscapes, common cloud cover, and scarce ground information have hindered large-area&#xa0;and field-level maize monitoring. This study presents a novel continent-wide framework for mapping maize cultivation across Africa for the 2023–2024 growing season at 10-m resolution using multi-temporal and multi-sensor remote sensing data. Our approach integrates Sentinel-1 SAR and Sentinel-2 optical imagery with the support of expert-validated pseudo-ground truth samples, region-based spectral-temporal signature analysis, and object-oriented segmentation through the Simple Non-Iterative Clustering (SNIC) algorithm. Maize classification was performed using a random forest model, which achieved an overall accuracy of 87.8% (kappa = 0.81), with regional performance of above 91% in Southern Africa. The harvested maize area in Africa was estimated to be 44.1 million hectares, with the largest share belonging to the West African region (31.4%). Model estimates showed strong alignment with national agricultural statistics (Pearson’s <i>r</i> = 0.88 when compared to FAOSTAT-reported areas). The resulting maps capture spatial variability in yield, cropping intensity, and field size. This study delivers one of the most detailed, cross-validated maize maps across Africa and proposes a method suitable for operational crop monitoring and food system planning amid climate variability.</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

Continental maize mapping and distribution in Africa by integrating radar and optical imagery

  • 期刊论文

  • 发表日期：

    2025年9月1日

查看全文
  • Nasser A. M. Abdelrahim,
  • Shuanggen Jin

摘要

High-resolution, accurate mapping of crops is critical to enhance food security, resource efficiency, and policy effectiveness in agriculture across Africa, where maize remains a crucial staple crop. However, mosaic landscapes, common cloud cover, and scarce ground information have hindered large-area and field-level maize monitoring. This study presents a novel continent-wide framework for mapping maize cultivation across Africa for the 2023–2024 growing season at 10-m resolution using multi-temporal and multi-sensor remote sensing data. Our approach integrates Sentinel-1 SAR and Sentinel-2 optical imagery with the support of expert-validated pseudo-ground truth samples, region-based spectral-temporal signature analysis, and object-oriented segmentation through the Simple Non-Iterative Clustering (SNIC) algorithm. Maize classification was performed using a random forest model, which achieved an overall accuracy of 87.8% (kappa = 0.81), with regional performance of above 91% in Southern Africa. The harvested maize area in Africa was estimated to be 44.1 million hectares, with the largest share belonging to the West African region (31.4%). Model estimates showed strong alignment with national agricultural statistics (Pearson’s r = 0.88 when compared to FAOSTAT-reported areas). The resulting maps capture spatial variability in yield, cropping intensity, and field size. This study delivers one of the most detailed, cross-validated maize maps across Africa and proposes a method suitable for operational crop monitoring and food system planning amid climate variability.