Scholarlink

<p>Arsenic (As), a harmful metalloid, presents serious risks to both the environment and human health because of its high toxicity and widespread presence. Human exposure to As primarily occurs through polluted water consumption and the ingestion of food with high As content. The As levels in crops are determined not only by the As content in soil but also by the interactions of As with other soil elements. In this research, the Qinghai-Tibet Plateau was chosen as the research area. Using the random forest (RF) algorithm, Sr and leachable Pb were identified from 29 soil indicators to predict As levels in rapeseed, wheat, potato, grass, and chicory crops. The results show that, compared to traditional multiple linear regression methods, the RF model offers higher accuracy and precision in predicting crop As content. In particular, in cross-species prediction, RF models have demonstrated excellent predictive performance. This study marks the first successful attempt at cross-species research using this model. This approach avoids the requirement for redundant evaluations of various crop types within the same region, signifying significant innovation. Moreover, the unique environment of the Qinghai‒Tibet Plateau increases the value of this research. The findings provide valuable insights for effective farmland planning and management, facilitating better organization of crop cultivation areas.</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

Using a random forest model for cross-species prediction of crop arsenic contamination

  • 期刊论文

  • 发表日期：

    2025年4月4日

查看全文
  • Qiaohui Zhu,
  • Jie Luo,
  • Min Cao,
  • Siyao Feng,
  • Heran Jia

摘要

Arsenic (As), a harmful metalloid, presents serious risks to both the environment and human health because of its high toxicity and widespread presence. Human exposure to As primarily occurs through polluted water consumption and the ingestion of food with high As content. The As levels in crops are determined not only by the As content in soil but also by the interactions of As with other soil elements. In this research, the Qinghai-Tibet Plateau was chosen as the research area. Using the random forest (RF) algorithm, Sr and leachable Pb were identified from 29 soil indicators to predict As levels in rapeseed, wheat, potato, grass, and chicory crops. The results show that, compared to traditional multiple linear regression methods, the RF model offers higher accuracy and precision in predicting crop As content. In particular, in cross-species prediction, RF models have demonstrated excellent predictive performance. This study marks the first successful attempt at cross-species research using this model. This approach avoids the requirement for redundant evaluations of various crop types within the same region, signifying significant innovation. Moreover, the unique environment of the Qinghai‒Tibet Plateau increases the value of this research. The findings provide valuable insights for effective farmland planning and management, facilitating better organization of crop cultivation areas.