基于随机森林法的弥河-潍河流域地下水质量评价研究
DOI:
作者:
作者单位:

1.中国地质环境监测院地质灾害技术指导中心;2.山东省第六地质矿产勘查院;3.济南大学水利与环境学院;4.山东省地质测绘院;5.山东省国土空间生态修复中心

作者简介:

通讯作者:

中图分类号:

P641.2

基金项目:

国家自然科学基金(41772257);山东省自然科学基金资助项目(ZR2019MD029);山东省高校院所创新团队项目(2018GXRC012);院科研基金(801KY202004)


Research on Groundwater Quality Evaluation of Mihe-Weihe River Basin Based on Random Forest Method
Author:
Affiliation:

1.China Geological Environment Monitoring Institute Geological Disaster Technical Guidance Center;2.No Institution of Geology and Mineral Resources Exploration of Shandong Province;3.School of Water Conservancy and Environment,University of Jinan;4.Shandong Geo-surveying Mapping Institute;5.Shandong Provincial Space Ecological Restoration Center

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    准确掌握地下水的环境质量状况是合理确定地下水资源开发策略和有效进行地下水资源保护的重要前提。通过随机森林(random forest)法构建了地下水质量评价模型,利用弥河-潍河流域的152组地下水水质数据进行模型训练,并利用剩余的65组水质数据进行验证,结果表明:(1)随机森林法在进行地下水水质分类时具有分类精度高、泛化能力强及收敛速度快等特点,且在进行超参数优化后,其分类精度和运算效率均会进一步提高,证明将随机森林法应用于地下水质量评价是可行的,并且其综合性能要优于逻辑回归模型;(2)研究区地下水水样均为Ⅳ类和Ⅴ类水,说明地下水水质状况整体较差;(3)通过分类指标重要性评价可以看出,研究区地下水水质的主要影响指标为硝酸盐、总硬度和溶解性总固体,而此类指标的主要来源是蔬菜种植化肥的不合理使用及河流污染入渗,因此要进一步加强对蔬菜种植污染排放及河流水质的监测和控制。

    Abstract:

    Accurately understanding the environmental quality of groundwater is an important prerequisite for rational and effective development and protection of groundwater resources. A groundwater quality evaluation model was constructed through the random forest algorithm, 152 sets of groundwater quality sample in the Mihe-Weihe River Basin were used for model training, and the remaining 65 sets of water quality sample were used for verification. The results showed: (1) random forest algorithm has the characteristics of high classification accuracy, strong generalization ability and fast convergence speed when classifying groundwater quality; moreover, after hyperparameter optimization, its classification accuracy and operation efficiency will be further improved, which proves that it’s feasible to apply random forest algorithm for groundwater quality evaluation, and its comprehensive performance is better than the logistic regression model. (2) the groundwater samples are all Class IV and V, indicating that the groundwater quality is generally poor; (3) it can be seen from the evaluation of the importance of classification indicators that the main influencing indicators of groundwater quality in the study area are nitrate, total hardness and total dissolved solids, and the main sources of such indicators are the unreasonable use of fertilizers and the infiltration of river pollution; therefore, it is necessary to further strengthen the monitoring and control of pollution emissions from vegetable planting and river water quality

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2021-12-08
  • 最后修改日期:2021-12-08
  • 录用日期:2022-04-20
  • 在线发布日期:
  • 出版日期: