• 关键词：营养盐  氮、磷  水质预测  机器学习  自动监测站  高时间分辨率  辽河

• 基金项目：广东省基础与应用基础研究基金面上项目（No.2022A1515010789）；“珠江人才计划”引进创新创业团队资助项目（No.2019ZT08L213）

• 杨宇锋
• 广东工业大学生态环境与资源学院，广东省流域水环境治理与水生态修复重点实验室，广州 510006

• 武暕
• 辽宁省生态环境监测中心，沈阳 110161

• 王璐
• 辽宁省生态环境监测中心，沈阳 110161

• 郭杨
• 辽宁省生态环境监测中心，沈阳 110161

• 惠婷婷
• 辽宁省生态环境监测中心，沈阳 110161

• 高伟
• 广东工业大学生态环境与资源学院，广东省流域水环境治理与水生态修复重点实验室，广州 510006

• 张远
• 广东工业大学生态环境与资源学院，广东省流域水环境治理与水生态修复重点实验室，广州 510006

• 摘要：随着环境信息获取技术的飞速发展，水质监测数据逐渐呈现出高时间分辨率的特点.针对传统方法在模拟高频检测数据时的不足，本文基于随机森林建立了河流氮、磷逐日浓度预测模型，识别了影响辽河干流马虎山断面氮、磷浓度变化的主要因素，并预测了未来不同情景下氮、磷浓度变化特征.结果表明：①马虎山断面氮、磷浓度变化的主要影响因素是上游珠尔山断面的来水水质和本断面的流量；②随机森林模型在高时间分辨率的水质指标模拟中具有误差低和拟合优度高的特点，其中，TN浓度预测模型的RMSE为0.40 mg·L^-1，R²为0.95，TP浓度预测模型的RMSE为0.01 mg·L^-1，R²为0.96；③在不同水文、污染控制和来水水质的变化情景下，未来马虎山断面氮、磷浓度变化主要取决于上游 来水水质，加强全流域营养盐控制是确保断面水质稳定达标的重要基础.

• Abstract：With the rapid development of environmental information acquisition technology， the resolution of water quality monitoring data becomes much higher. To overcome shortcomings of traditional methods in simulating high frequency monitoring data， we developed a random forest model to predict daily nitrogen and phosphorus concentrations in rivers. Then， we identified key factors impacting nitrogen and phosphorus concentrations of Mahushan section of the main stream of the Liaohe River. Next， we predicted dynamics of nitrogen and phosphorus concentrations under different future scenarios. The results shown that key factors impacting nitrogen and phosphorus concentrations of Mahushan section of the main stream of the Liaohe River were water quality of the upstream Zhuershan section and the flow of this section. We also found the that random forest model had low errors and a high goodness of fit in the simulation of high temporal resolution water quality indicators. The RMSE and R² for total nitrogen were 0.40 mg·L^-1 and 0.95， while those for total phosphorus were 0.01 mg·L^-1 and 0.96， respectively. We thereby concluded that under different future scenarios of hydrology， pollution control， and upstream water quality， the dynamics of nitrogen and phosphorus concentrations at the Mahushan section will mainly depend on the upstream water quality. Strengthening nutrient control in the entire watershed is the basis to ensure the stable compliance of water quality of this section.

• Key words：nutrient  nitrogen and phosphorus  water quality prediction  machine learning  auto-monitoring station  high time resolution  Liao River

• 
  摘要点击次数： 177 全文下载次数： 253