Faculty of Forestry and Environmental Management, University of New Brunswick, P.O. Box 4400, 28 Dineen Drive, Fredericton, NB, E3B 5A3, Canada; Charlottetown Research and Development Center, Agriculture and Agri-Food Canada, 440 University Avenue, Charlottetown, PE C1A 4N6, Canada.
Charlottetown Research and Development Center, Agriculture and Agri-Food Canada, 440 University Avenue, Charlottetown, PE C1A 4N6, Canada.
Sci Total Environ. 2020 Aug 10;729:138793. doi: 10.1016/j.scitotenv.2020.138793. Epub 2020 Apr 19.
Excessive nitrate loading from agricultural non-point source is threatening the health of receiving water bodies at the global scale. Quantifying the drivers/sources of water and nitrate flux in watersheds and relating them to spatial and temporal land uses is essential for developing effective mitigation strategies. This study investigated the impact of land use on water yield and nitrate loading to surface water in a typical agricultural watershed in Prince Edward Island (PEI), Canada. We used historical streamflow and water quality records to calibrate the comprehensive hydrological model Soil and Water Assessment Tool (SWAT), which was setup with detailed annual land use records. The SWAT model performed well in predicting both daily streamflow and nitrate load. Land use demonstrated little impact on water yield but affected nitrate load significantly. Annual nitrate load ranged from 5.6 to 44.4 kg N ha yr for forest and soybean, respectively. Potato rotated land contributed 84.5% of annual nitrate load to the watershed. Source of water yield demonstrated high variability between the growing season and non-growing season. About 90% of water yield was contributed by groundwater during growing season, while runoff contributed over 60% of water yield during the non-growing season. Groundwater was the dominant source of nitrate loading for both seasons. The watershed estuary faced the highest threats from subbasins in the south western area due to the high nitrate load and proximity to the watershed outlet. Results by the machine learning algorithm random Forest analysis indicated that the climatic variables of temperature and precipitation were the top two factors affecting water yield, with a combined relative importance of 61%. Land use was the dominant factor affecting nitrate load, the relative importance of land use alone was ~50%. The results of this study provided critical insights for watershed management in Atlantic Canada.
过量的农业非点源硝酸盐负荷正在威胁全球范围内受纳水体的健康。量化流域中水流和硝酸盐通量的驱动因素/来源，并将其与时空土地利用联系起来，对于制定有效的缓解策略至关重要。本研究调查了土地利用对加拿大爱德华王子岛（PEI）典型农业流域地表水产水量和硝酸盐负荷的影响。我们使用历史径流量和水质记录来校准综合水文模型 SWAT，该模型是使用详细的年度土地利用记录建立的。SWAT 模型在预测日径流量和硝酸盐负荷方面表现良好。土地利用对产水量的影响不大，但对硝酸盐负荷的影响显著。森林和大豆的年硝酸盐负荷分别为 5.6 至 44.4 kg N ha yr。轮作马铃薯的土地对流域年硝酸盐负荷的贡献为 84.5%。水源产水量在生长季和非生长季之间表现出高度的可变性。生长季地下水贡献了约 90%的产水量，而非生长季径流量贡献了超过 60%的产水量。地下水是两个季节硝酸盐负荷的主要来源。由于硝酸盐负荷高且靠近流域出口，流域河口面临来自西南部子流域的最大威胁。随机森林分析的机器学习算法结果表明，温度和降水等气候变量是影响产水量的前两个因素，其相对重要性合计为 61%。土地利用是影响硝酸盐负荷的主要因素，其单独的相对重要性约为 50%。本研究的结果为加拿大大西洋地区的流域管理提供了重要的见解。
