| 基于INLA-SPDE模型的区域土壤硒元素空间预测及富硒区优选 |
| 摘要点击 597 全文点击 75 投稿时间：2024-05-11 修订日期：2024-08-01 |
| 查看HTML全文
查看全文 查看/发表评论 下载PDF阅读器 |
| 中文关键词 富硒土地 集成嵌套拉普拉斯逼近-随机偏微分方程 土壤重金属 低重金属污染富硒区 淄博市 |
| 英文关键词 selenium-enriched land integrated nested Laplace approximation-stochastic partial differential equation soil heavy metals selenium-enriched land with low heavy metals pollution Zibo City |
| DOI 10.13227/j.hjkx.20250650 |
|
| 中文摘要 |
| 土壤是动植物和人类获取生长发育所需硒元素（Se）的主要介质，但经济的快速发展造成了严重的土壤重金属污染，土壤环境和人类身体健康受到威胁. 因此，预测土壤Se空间分布，划定低重金属污染风险的“清洁富硒土地”，可为工农业健康发展提供参考. 以淄博市部分地区为研究区，系统采集212个土壤样品进行Se及重金属元素测试；采用集成嵌套拉普拉斯逼近-随机偏微分方程（INLA-SPDE），整合距道路距离和距工厂距离、土壤类型、降雨量和土壤粒度这5种环境协变量，对土壤Se元素空间分布进行预测，并结合有限混合分布模型（FMDM）确定的污染阈值，进行富硒区优选. 结果表明：① 研究区土壤ω（Se）平均值为0.34 mg·kg-1，高于山东省土壤Se背景值和全国土壤Se含量平均值. ② INLA-SPDE方法预测结果显示，土壤Se含量空间变异与自然因素和人类活动相关；富硒土地集中分布研究区中部，面积为423.49 km2. 空间预测标准差分布表明，采样点密度大，预测不确定性小. ③ 依据FMDM模型提供的污染临界值，划定低于中污染阈值和高污染阈值的富硒土地面积分别是2.22 km2和291.81 km2. 富硒土地的优选可为持续开发、土壤环境管控和人类健康保护提供参考. |
| 英文摘要 |
| Soil is the main medium for plants， animals， and humans to obtain the selenium element required for growth and development. Rapid economic development has caused serious soil heavy metal pollution， threatening soil environment and human health. Therefore， predicting the spatial distribution of soil selenium elements and delineating “clean and selenium-rich land” with low heavy metal pollution risk can provide a reference for the healthy development of agriculture and industry. A total of 212 soil samples were collected in part of Zibo City for selenium and heavy metals testing. The integrated nested Laplace approximation-stochastic partial differential equation （INLA-SPDE） method was used to integrate five environmental covariates， including distance to roads and factories， soil type， rainfall， and soil particle size， to predict the spatial distribution of soil selenium elements. Combined with the pollution thresholds determined by the finite mixture distribution model （FMDM）， the selenium-rich areas were optimized. The results show that： ① The average selenium content in the study area was 0.34 mg·kg-1， higher than the background value of selenium in Shandong soil and the national average. ② The INLA-SPDE method prediction results showed that the spatial variation of soil selenium content was related to natural factors and human activities； the selenium-rich land was concentrated in the middle of the study area， with an area of 423.49 km2. The distribution of spatial prediction standard deviation showed that the sampling density was high， and the prediction uncertainty was small. ③ According to the pollution threshold provided by the FMDM model， the areas with selenium-rich land lower than the medium pollution threshold and high pollution threshold were 2.22 km2 and 291.81 km2， respectively. The optimization of selenium-rich land can provide a reference for sustainable development， soil environment control， and human health protection. |
