摘要
过量氮输入导致的水体富营养化问题严重威胁水质与生态健康.本研究以白洋淀流域为对象,创新性地耦合土壤与水评估工具(SWAT)与长短期记忆网络-多层感知器混合模型(LSTM-MLP),旨在揭示流域氮通量的时空变化特征、提升预测精度并提出管理对策.SWAT模型分析表明,流域氮通量存在显著的季节性差异,汛期负荷远高于非汛期,且主要受降水、径流和泥沙影响.为提升预测精度,所构建的LSTM-MLP混合模型在氮通量峰值预测方面表现优越(NSE>0.8),显著优于传统LSTM模型.基于此,情景模拟显示,通过削减泥沙(例如,削减60%泥沙可使特定区域氮负荷降低超过40%)可有效控制氮负荷.本研究首次将SWAT与LSTM-MLP模型结合应用于流域氮通量研究,深化了对白洋淀流域氮循环时空特征的理解,并为流域水质改善和生态修复提供了基于泥沙控制的有效管理策略与技术支撑,具有重要的科学与实践价值.
Excessive nitrogen input can lead to water eutrophication,which seriously threatens water quality and ecological health.In this study,Baiyangdian basin was selected as the study area,and the soil and water assessment tool(SWAT)was coupled with a long short-term memory-multilayer perceptron(LSTM-MLP)to reveal the spatial-temporal variation of nitrogen flux in the Basin.The SWAT model results showed that the nitrogen flux in the basin appeared significantly seasonal difference,and the nitrogen load was higher in the flood season than that in the non-flood season,which was impacted by the precipitation,runoff,and sediment.In order to improve the predication accuracy,the LSTM-MLP hybrid model performed superiority in the predication of nitrogen flux peak(NSE>0.8).It was significantly superior to the traditional LSTM model.Based on this,scenario simulation showed that the nitrogen load can be effectively controlled by reducing sediment(e.g.reducing sediment by 60%can lower the nitrogen load by more than 40%in a specific area).This study first combined the SWAT and LSTM-MLP models and applied them to the study of nitrogen fluxes in basins.The understanding of the spatio-temporal variation of the nitrogen cycle in the Baiyangdian basin was deepened,and it can provide effective management strategies and technical support based on sediment control for the improvement of water quality and ecological restoration in the basin.It has important scientific and practical value.
作者
郑子帅
徐童
高宇昂
王浩加
李子琼
李晓宁
杜兆楠
崔建升
张璐璐
ZHENG Zishuai;XU Tong;GAO Yu′ang;WANG Haojia;LI Ziqiong;LI Xiaoning;DU Zhaonan;CUI Jiansheng;ZHANG Lulu(College of Environmental Science and Engineering,Hebei University of Science and Technology,Shijiazhuang 050018;Hebei Key Laboratory of Pollution Prevention Biotechnology,Shijiazhuang 050018)
出处
《环境科学学报》
北大核心
2025年第12期219-232,共14页
Acta Scientiae Circumstantiae
基金
河北省自然科学基金(No.D2019208152)
河北省教育厅重点项目(No.ZD2021046)。
