Scholarlink

<p>Gully erosion severely threatens agricultural landscapes, causing arable land loss and compromising environmental and socio-economic stability in arid/semi-arid regions. This study aims to delineate gully-affected areas and assess their impacts by evaluating the Modified Normalized Difference Water Index (MNDWI) against other remote sensing indices. Advanced machine learning techniques, including unsupervised Fuzzy C-Means (FCM) and K-means clustering, were employed for accurate gully extraction. Feature-selection algorithms, such as Principal Component Analysis (PCA) and the Boruta algorithm, identified key drivers of gully expansion. The Fuzzy-Best-Worst Method (Fuzzy BWM) was used to map areas susceptible to gully erosion and quantify agricultural land loss. Results showed that the humidity index outperformed other indices in detecting gully-prone zones, with FCM achieving 88% classification accuracy compared to 69% for K-means. Fuzzy BWM analysis indicated that 49% of the study area is at risk of gully erosion. Longitudinal data from 2000 to 2020 revealed a 20% loss of cultivated land, particularly in southern regions, exacerbating economic losses and threatening local livelihoods. PCA highlighted soil texture as a primary factor in gully formation, outperforming the Boruta algorithm. Receiver Operating Characteristic (ROC) analysis validated the robustness of the Fuzzy BWM approach, supporting the reliability of the predictive models. This study highlights gully erosion’s multifaceted environmental impact through soil degradation, its economic burden from diminished agricultural productivity, and its social ramifications for farming communities. It establishes a comprehensive framework for land managers to implement effective control strategies, fostering sustainable land use and bolstering regional socio-economic resilience.</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

Enhancing gully erosion susceptibility prediction and agricultural land loss assessment through machine learning and remote sensing techniques

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Marzieh Mokarram,
  • Hamid Reza Pourghasemi,
  • Foroogh Golkar,
  • Tam Minh Pham

摘要

Gully erosion severely threatens agricultural landscapes, causing arable land loss and compromising environmental and socio-economic stability in arid/semi-arid regions. This study aims to delineate gully-affected areas and assess their impacts by evaluating the Modified Normalized Difference Water Index (MNDWI) against other remote sensing indices. Advanced machine learning techniques, including unsupervised Fuzzy C-Means (FCM) and K-means clustering, were employed for accurate gully extraction. Feature-selection algorithms, such as Principal Component Analysis (PCA) and the Boruta algorithm, identified key drivers of gully expansion. The Fuzzy-Best-Worst Method (Fuzzy BWM) was used to map areas susceptible to gully erosion and quantify agricultural land loss. Results showed that the humidity index outperformed other indices in detecting gully-prone zones, with FCM achieving 88% classification accuracy compared to 69% for K-means. Fuzzy BWM analysis indicated that 49% of the study area is at risk of gully erosion. Longitudinal data from 2000 to 2020 revealed a 20% loss of cultivated land, particularly in southern regions, exacerbating economic losses and threatening local livelihoods. PCA highlighted soil texture as a primary factor in gully formation, outperforming the Boruta algorithm. Receiver Operating Characteristic (ROC) analysis validated the robustness of the Fuzzy BWM approach, supporting the reliability of the predictive models. This study highlights gully erosion’s multifaceted environmental impact through soil degradation, its economic burden from diminished agricultural productivity, and its social ramifications for farming communities. It establishes a comprehensive framework for land managers to implement effective control strategies, fostering sustainable land use and bolstering regional socio-economic resilience.