| 朱海鹏,刘元元,杨超,焦乐,刘胜,冯峻林.基于机器学习和PLS-SEM的农田土壤重金属污染驱动机制——以湖南省湘潭县为例[J].农业环境科学学报,2025,44(11):2764-2782. |
| 基于机器学习和PLS-SEM的农田土壤重金属污染驱动机制——以湖南省湘潭县为例 |
| Spatial dynamics of heavy metal pollution and their driving factors using coupled machine learning and PLS-SEM: a case study in Xiangtan，China |
| 投稿时间：2025-03-06 |
| DOI：10.11654/jaes.2025-0223 |
| 中文关键词: 重金属 稻田土壤 随机森林 偏最小二乘法结构方程 地理加权回归 空间分布 |
| 英文关键词: heavy metal paddy soil random forest partial least squares structural equation modeling geographically weighted regression spatial distribution |
| 基金项目:国家自然科学基金青年基金项目（42307507）；中国电力建设股份有限公司科技项目成果 |
|
| 摘要点击次数: 1148 |
| 全文下载次数: 885 |
| 中文摘要: |
| 为了定量化评估自然因素和社会经济因素对稻田土壤重金属空间分布的影响，本研究以湖南省湘潭县为研究区，通过整合随机森林（RF）、地理加权回归模型（GWR）和偏最小二乘法结构方程模型（PLS-SEM），构建了重金属分布驱动机制耦合分析框架。结果表明：研究区稻田土壤重金属污染风险主要来自镉（Cd）。进一步对除镍（Ni）外含量超出背景值的7种重金属进行驱动机制分析，RF结果发现灌溉水用量对Cd含量的特征重要性值最高，而降水和PM10对汞（Hg）、砷（As）、铅（Pb）和锌（Zn）的特征重要性值最高。GWR模拟结果发现各因素对稻田土壤7种重金属分布的影响存在显著空间异质性。基于PLS-SEM的路径分析显示气候因素对土壤重金属Cd、Pb和Zn产生直接的积极作用（β=0.50~0.74），而土壤性质对Hg、As、Cr和Cu产生直接负面影响（β=-0.59~-0.47）。间接效应分析结果显示，土壤性质介导下社会经济因素对土壤Cd、Pb和Zn含量具有显著的正向效应（β=0.33~0.49），而地形和水文条件则对Hg、As、Cr和Cu产生显著的正向间接效应（β>0.30）。水文条件的中介作用缓解了气候因素对土壤属性的正向直接效应和地形对土壤的负向直接效应。本研究揭示了自然因素和社会经济因素在稻田土壤重金属分布中的复杂交互作用，为深入理解重金属分布驱动机制和制定有效治理策略提供了科学依据。 |
| 英文摘要: |
| To quantitatively assess the impact of natural and socioeconomic factors on heavy metal distribution in rice paddy soils, this study integrated random forest（RF）, geographically weighted regression（GWR）, and partial least squares structural equation modeling （PLS-SEM）to construct an integrated framework for analyzing the driving pathways of heavy metal distribution in Xiangtan County, Hunan Province. Results showed that the primary heavy metal risks in paddy soils in Xiangtan were associated with cadmium（Cd）. We further analyzed the driving mechanisms of 7 heavy metals（excluding Ni）with concentrations exceeding the background values. RF model showed the highest characteristic importance value of irrigation water usage on Cd distribution, while precipitation and PM 10 had the highest characteristic importance value on mercury（Hg）, arsenic（As）, lead（Pb）, and zinc（Zn）. GWR results demonstrated the significantly spatial heterogeneity in the effects of various factors on the distribution of seven heavy metals. Pathway analysis based on PLSSEM revealed a directly positive effect of climatic factors on Cd, Pb, and Zn（β=0.50~0.74）. In contrast, soil properties showed a directly negative effect on Hg, As, Cr, and Cu（β=- 0.59~- 0.47）. Indirect effect analysis indicated that socioeconomic factors, mediated by soil properties, had significant positive effects on Cd, Pb, and Zn distribution（β=0.33~0.49）. However, terrain and hydrological factors showed significantly positive effects on Hg, As, Cr and Cu（β>0.30）. The mediating role of hydrological conditions moderated the directly positive effect of climatic factors on soil properties and the directly negative effect of terrain on soil properties. This study reveals the complex interactions between natural and socioeconomic factors in heavy metal distribution in paddy soils, providing scientific evidence to identify the driving mechanism of heavy metal distribution and formulate of effective remediation strategies. |
| HTML
查看全文
查看/发表评论 下载PDF阅读器 |
|
|
|
