摘要
渭河作为黄河最大的支流,在黄河流域生态保护和高质量发展中处于重要地位.基于新型综合水质指数(WQI-DET),结合差异性分析方法对渭河水质状况和时空差异性进行综合评价分析,同时借助主成分分析法确定主要污染物种类,利用地理探测器量化流域水质可能的驱动因子影响力,并在此基础上运用机器学习算法和高频监测数据实现对WQI-DET的模拟计算和预测.结果表明:①渭河流域存在有机污染、富营养化营养盐污染、六价铬污染和氟化物污染,水体可生化性差,营养盐结构不平衡,氮污染严重;水质状况在过去3 a未能得到根本改善.②渭河流域水质指标时空差异显著,有机污染、氮污染以及浊度等指标呈现季节性特点;不同支流水质与干流上中下游间存在差异,污染物呈现出沿程累积的特点.③农业生产的面源污染和生活污水点源污染是渭河有机污染物、富营养化营养盐等的主要来源,同时渭河部分支流受到工业源污染影响,重金属污染较严重;地理探测结果显示流域水质受到人类活动和自然条件的共同作用.④机器学习模型可对渭河WQI-DET进行准确预测,利用日监测数据的6项指标通过COA+BP模型准确模拟计算了WQI-DET(R^(2)>0.92),依托高频次的日监测数据和COA+BP模型计算结果建立VMD+CNN-GUR-SE模型实现了未来WQI-DET的计算,从而实现了对渭河水质的预测计算;群智能优化算法、变分模态分解和注意力机制的引入大幅改善了模型性能.
As the largest tributary of the Yellow River,the Weihe River plays an important role in the ecological protection and high-quality development of the Yellow River Basin.Based on the new comprehensive water quality index(WQI-DET),a comprehensive evaluation and analysis of the water quality status and spatiotemporal differences of the Weihe River was conducted using differential analysis methods.Principal component analysis was used to determine the main types of pollutants,and geographic detectors were used to quantify the possible driving factors of water quality in the basin.On this basis,machine learning algorithms and high-frequency monitoring data were used to simulate and predict WQI-DET.The study produced some important results:①Organic pollution,eutrophication nutrient pollution,hexavalent chromium pollution,and fluoride pollution are all present in the Weihe River Basin,with poor water biodegradability,imbalanced nutrient structure,and severe nitrogen pollution.The water quality has not improved fundamentally in the past three years.The temporal and spatial differences in water quality indicators in the Weihe River Basin are significant,and there are seasonal characteristics of indicators such as organic pollution,nitrogen pollution,and turbidity.There are differences in water quality between different tributaries and the upstream,midstream,and downstream portions of the main stream,and pollutants show the characteristic of accumulating along the drainage.The non-point source pollution of agricultural production and the point source pollution of domestic wastewater are the main sources of organic pollutants and eutrophic nutrients in the Weihe River.At the same time,some tributaries of the Weihe River are affected by industrial source pollution,and heavy metal pollution is relatively serious.The geographical exploration results showed that the water quality of the watershed is influenced by both human activities and natural conditions.The machine learning model can accurately predict the WQI-DET of the Weihe River.Using six indicators from daily monitoring data,the WQI-DET(R^(2)>0.92)was accurately simulated and calculated through the COA+BP model.Based on high-frequency daily monitoring data and the calculation results of the COA+BP model,a VMD+CNN-GUR-SE model was established to achieve calculation of future WQI-DET,thus realizing prediction and calculation of Weihe River water quality.The introduction of swarm intelligence optimization algorithms,variable mode decomposition,and attention mechanisms significantly improved the performance of the model.
作者
张京新
蔡庆旺
康子怡
丛铭
韩玲
何皎洁
杨利伟
ZHANG Jing-xin;CAI Qing-wang;KANG Zi-yi;CONG Ming;HAN Ling;HE Jiao-jie;YANG Li-wei(School of Civil Engineering,Chang'an University,Xi'an 710061,China;College of Geology Engineering and Geomatics,Chang'an University,Xi'an 710054,China;School of Land Engineering,Chang'an University,Xi'an 710054,China)
出处
《环境科学》
北大核心
2025年第9期5619-5640,共22页
Environmental Science
基金
中央高校高新技术研究培育项目(300102281201)
中国博士后科学基金项目(2021M692510)
陕西省自然科学基础研究计划项目(2021JO-222)。
关键词
渭河流域
水质评价
主成分分析(PCA)
地理探测器
机器学习
Weihe River Basin
water quality assessment
principal component analysis(PCA)
geographic detector
machine learning
