Scholarlink

<p>Crop rotations play a crucial role in shaping the nature, timing, and intensity of agricultural management strategies, as well as their environmental impacts. While certain rotations can offer environmental benefits, others may pose risks. By mapping, characterizing, and accurately predicting crop rotations, it becomes possible to promote sustainable agricultural practices. Crop rotation modelling studies often overlook spatiotemporal non-stationarity, fail to integrate real-time updates, lack uncertainty quantification, and underutilize high-resolution spatial data, limiting their predictive accuracy and ability to incorporate stochasticity. To address these gaps, we introduce a stochastic crop rotation modeling approach called DCTR-PCA-PF, which integrates dynamic candidate transition rules (DCTRs), partitioned cellular automata (PCA), and the particle filter (PF) algorithm. We applied the proposed model to a detailed case study in the Mojen Agricultural Area, Iran, to predict crop rotations and uncover the underlying factors driving these changes. Using remote sensing techniques, we obtained high-resolution crop type maps from Sentinel-2 and Landsat-8 satellite images. Additionally, a field survey was conducted to identify potential factors influencing crop rotations. The results show that the DCTR-PCA-PF model improves overall accuracy (to 80%) by at least 12% and up to 25% compared to alternative models. The proposed model also consistently enhances the figure of merit (FOM) metric across all crop classes by 10–25%, underscoring its robustness. Beyond predictive performance, the model highlights precipitation and market prices as key drivers of crop rotation. These insights provide a data-driven foundation for more effective and sustainable agricultural policies.</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

High-resolution satellite data assimilation for stochastic crop rotation modeling and identifying spatiotemporal key drivers

  • 期刊论文

  • 发表日期：

    2025年8月8日

查看全文
  • Mostafa Jalilehvand,
  • Barat Mojaradi,
  • Hosein Alizadeh

摘要

Crop rotations play a crucial role in shaping the nature, timing, and intensity of agricultural management strategies, as well as their environmental impacts. While certain rotations can offer environmental benefits, others may pose risks. By mapping, characterizing, and accurately predicting crop rotations, it becomes possible to promote sustainable agricultural practices. Crop rotation modelling studies often overlook spatiotemporal non-stationarity, fail to integrate real-time updates, lack uncertainty quantification, and underutilize high-resolution spatial data, limiting their predictive accuracy and ability to incorporate stochasticity. To address these gaps, we introduce a stochastic crop rotation modeling approach called DCTR-PCA-PF, which integrates dynamic candidate transition rules (DCTRs), partitioned cellular automata (PCA), and the particle filter (PF) algorithm. We applied the proposed model to a detailed case study in the Mojen Agricultural Area, Iran, to predict crop rotations and uncover the underlying factors driving these changes. Using remote sensing techniques, we obtained high-resolution crop type maps from Sentinel-2 and Landsat-8 satellite images. Additionally, a field survey was conducted to identify potential factors influencing crop rotations. The results show that the DCTR-PCA-PF model improves overall accuracy (to 80%) by at least 12% and up to 25% compared to alternative models. The proposed model also consistently enhances the figure of merit (FOM) metric across all crop classes by 10–25%, underscoring its robustness. Beyond predictive performance, the model highlights precipitation and market prices as key drivers of crop rotation. These insights provide a data-driven foundation for more effective and sustainable agricultural policies.