基于多源数据融合的典型丘陵区水源地非点源氮素迁移机制研究

The drinking water sources in hilly areas of China show a trend of water quality deterioration and eutrophication. However, the studies on diffuse nitrogen leaching regimes at this type of catchments are not sufficient. In this study, Huashan experimental catchment, which is located at the upstream of the drinking water sources of Chuzhou city, Anhui province is selected as study area. Artificially-controlled rainfall-runoff experiments are conducted under three dominant land uses (forest, arable land and paddy field) to investigate nitrogen losses via different hydrological pathways. Main diffuse sources contributed to nitrate in river, denitrification process and their responses to variations in meteorological and hydrological conditions are analyzed and quantified using stable isotope (oxygen and nitrogen in nitrate) tracing approach. The developed catchment spatially-distributed hydrological and nitrogen transport model is then set up at Huashan catchment to simulate nitrogen losses. In order to improve model performance on simulation of nitrogen loads and stream water nitrogen concentrations as well as decrease prediction uncertainty, parameter calibration, uncertainty analysis, and model validation are carried out using Bayesian inference based multi-data diffusion with combination of daily mean discharge and stream water nitrogen concentrations observed at catchment outlet, findings from artificially-controlled rainfall-runoff experiments, as well as diffuse sources apportionment and denitrification analysis. The outcomes of this study will help to improve understandings of diffuse nitrogen leaching regimes at hilly region, provide guidelines for taking measures to prevent diffuse nitrogen losses and protect surface water quality.

我国丘陵区水源地呈水质恶化和富营养化的趋势，而这类流域非点源氮素淋失机制的研究尚显不足。本研究拟选择安徽省滁州市饮用水源地的上游区—花山实验流域作为研究区，开展主要土地利用（林地、旱地、水田）人工控制降雨条件下的坡面小区径流实验，揭示氮素随不同径流路径的迁移规律；基于δ15N-NO3-和δ18O-NO3-示踪技术，解析河流氮素来源与反硝化及其对气象水文条件的响应；进而在流域上搭建改进的分布式水文与氮素迁移耦合模型模拟氮素径流输出，结合流域出口断面日均流量和氮素浓度监测数据、小区径流实验和同位素源解析与反硝化信息，采用基于贝叶斯理论的多源数据融合技术，开展模型参数率定、不确定性分析和模型验证，提高流域氮素输出通量和河流水体氮素浓度的模拟精度，减小模型预测的不确定性。研究结果有助于加深对丘陵区非点源氮素淋失的认识，为流域非点源氮素流失防治和水质保护提供科学参考。

非点源氮素污染输移流失规律、负荷定量计算、氮污染源解析、分布式流域非点源污染模拟研究是控制流域内氮素污染的重要技术问题。基于不同降雨条件下氮素浓度的动态变化、时空分布特征差异性，亟需开展基于多源数据融合的典型丘陵区水源地非点源氮素迁移机制研究，改进模型参数化以提高流域非点源氮素输移模拟的精度与可信度。项目选择典型丘陵区水源地花山实验流域作为研究区，通过资料收集与实地查勘相结合的技术手段，分析了流域内地形地貌、土壤植被、水文气象等自然地理特征，选取胡庄（三）作为关键监测断面，以高频监测的技术手段，研究分析了次降雨条件下与人工模拟降雨条件下花山实验流域降雨径流时空特征、不同形态氮素（溶解态氮、硝态氮、铵态氮、亚硝态氮）输移流失时空分布规律。结果表明，林地中总氮、硝态氮浓度较高，农田中氨氮浓度高，氮素流失的高峰期一般集中在5月-7月。影响氮素输出的主控因子为降雨径流量。通过氮素平衡收支法研究分析了流域内行政人口、农业种植（化肥、有机肥）、畜禽养殖（家畜粪便）的量值，指出区域内化肥氮素为主要氮素输入源，约占贡献率的50%以上，而作物的收获为主要输出端，约占贡献率的50%以上，研究表明，花山流域总氮湿沉降取决于降水量的大小。通过流域氮同位素分析的技术手段，于花山实验流域布点监测分析，研究分析氮同位素分布特征，指出汛期降水前，河流氮同位素丰度值较大，汛期降水对实验流域硝酸氮的分布影响较大，降水过程对河流氮素迁移有较强的驱动作用，且不同监测子流域区域内均表明土壤有机氮和化肥氮为主要氮素来源。通过搭建流域分布式水文与非点源氮素输移耦合模拟模型HYPE，同时开展了参数敏感性分析，并通过DREAMZS单站点（SSC）、多目标参数、不确定性分析等方法对模型进行率定验证，研究结果表明指出汛期为花山实验流域氮素流失的关键时段，比重约占84.92%，农田特别是旱田为关键源区。通过构建PLOAD模型研究提出了该流域最佳管理措施。