Scholarlink

<p>With the intensification of extreme climate events, resource constraints, and price volatility, the factors affecting food security have become increasingly complex and dynamic. Understanding the relationship between agricultural risks and food security is therefore essential for identifying the vulnerabilities of the food system and exploring effective governance pathways. This study constructs a comprehensive agricultural risk assessment framework for 31 provinces in China, encompassing five dimensions—ecological, disaster, market, policy, and social risks. It reveals the spatiotemporal evolution of agricultural risks from 2000 to 2022 and employs an interpretable XGBoost model to explore the nonlinear relationships and interaction effects between agricultural risks and food security. The results show that overall agricultural risks in China have declined, with high-risk regions shifting from major grain-producing areas to frontier regions. Agricultural risk indicators exhibit nonlinear relationships with food security, where factors such as land governance, agrochemical use, and income distribution exert the most significant influence, each demonstrating threshold effects. Different types of risks negatively affect food security, and their interactions further amplify food insecurity. This research highlights the advantages of explainable machine learning in addressing agricultural risk-related issues and underscores the necessity of agricultural risk reduction management, providing crucial theoretical support for sustainable agricultural development and the resilience enhancement of global food systems.</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

Multidimensional framework of agricultural risk in China and its impact mechanism on food security using an explainable machine learning framework based on the XGBoost model

  • 期刊论文

  • 发表日期：

    2026年6月3日

查看全文
  • Dapeng Zhou,
  • Jingchun Wang,
  • Jing Zhang,
  • Zixuan Li,
  • Honghua Huan

摘要

With the intensification of extreme climate events, resource constraints, and price volatility, the factors affecting food security have become increasingly complex and dynamic. Understanding the relationship between agricultural risks and food security is therefore essential for identifying the vulnerabilities of the food system and exploring effective governance pathways. This study constructs a comprehensive agricultural risk assessment framework for 31 provinces in China, encompassing five dimensions—ecological, disaster, market, policy, and social risks. It reveals the spatiotemporal evolution of agricultural risks from 2000 to 2022 and employs an interpretable XGBoost model to explore the nonlinear relationships and interaction effects between agricultural risks and food security. The results show that overall agricultural risks in China have declined, with high-risk regions shifting from major grain-producing areas to frontier regions. Agricultural risk indicators exhibit nonlinear relationships with food security, where factors such as land governance, agrochemical use, and income distribution exert the most significant influence, each demonstrating threshold effects. Different types of risks negatively affect food security, and their interactions further amplify food insecurity. This research highlights the advantages of explainable machine learning in addressing agricultural risk-related issues and underscores the necessity of agricultural risk reduction management, providing crucial theoretical support for sustainable agricultural development and the resilience enhancement of global food systems.