Comparison of methods for estimating sediment and nitrogen loads from a small agricultural watershed
Comparison of methods for estimating sediment and nitrogen loads from a small agricultural watershed
复制标题
小型农业流域沉积物和氮负荷估算方法的比较
DOI：
--
复制
发表时间：
2007
期刊：
影响因子：
--
通讯作者：
J. Duchemin
中科院分区：
文献类型：
--
作者：
A. Zamyadi;J. Duchemin
Zamyadi， A, Gallichand， J.和Duchemin， M. 2007。估算农业小流域泥沙和氮负荷方法的比较。加拿大生物系统工程/加拿大生物系统学报49:1.27-1.36。流域出水口养分质量负荷的认识是水质管理工程的重要工具。然而，由于缺乏频繁的浓度测量，质量负荷的精确估计是不可能的。本研究采用6个采样频率（每日至每月）、2种采样方案（固定和随机）和7种计算方法（平均、比率估计和插值）对曲海省5.3 km农业流域的沉积物和3种氮组分进行了质量负荷估算。每小时的流量、总氮、硝态氮、铵态氮和沉积物的值是通过为流域校准的HSPF模型在两年的时间内生成的。假设基于小时值质量累积的负荷估计代表真实负荷，使用偏差、标准差和均方根误差（RMSE）将其他方法估计的负荷与真实负荷进行比较。各水质参数的均方根误差随采样频率的增加而减小。固定抽样方案的均方根误差小于随机抽样方案的均方根误差。固定采样方案的良好性能归因于所有水质参数在时间上的自相关。沉积物自相关图显示出24小时的周期性，这可以用融雪和更频繁的傍晚和夜间降雨来解释。所有的负荷评估通常会导致对真实负荷的低估，这很可能是由于所研究的小流域的水文响应很短暂。在研究的7种负荷计算方法中，线性插值法（用于获得每个可用小时流量值的浓度估计值）系统地产生了最低的RMSE值。然而，比率估计方法表现不佳，在测试的7种方法中仅排名第5和第6。
Zamyadi, A., Gallichand, J. and Duchemin, M. 2007. Comparison of methods for estimating sediment and nitrogen loads from a small agricultural watershed. Canadian Biosystems Engineering/Le génie des biosystèmes au Canada 49: 1.27-1.36. The knowledge of nutrient mass load at the outlet of watersheds is a key tool in water quality management projects. However, because of the lack of frequent concentration measurements, a precise estimation of mass load is not possible. This study was conducted to determine the quality of mass load estimation for a combination of six sampling frequencies (daily to monthly), two sampling schemes (fixed and random), and seven calculation methods (averaging, ratio estimator, and interpolation) for sediment and three nitrogen components at the outlet of a 5.3 km agricultural watershed in the province of Québec. Hourly values of flow, total nitrogen, nitrate, ammonium, and sediment were generated for a two-year period by the HSPF model calibrated for the watershed. Load estimates based on the mass accumulation of hourly values were assumed to represent the true loads to which loads estimated by other methods were compared using bias, standard deviation, and root mean square error (RMSE). For all water quality parameters, the RMSE decreased with the sampling frequency. Fixed sampling schemes always resulted in RMSE values less than those of random schemes. This better performance of the fixed sampling scheme is attributed to the autocorrelation in time for all water quality parameters. The sediment autocorrellogram showed a 24-hour periodicity that is explained by snowmelt and more frequent evening and night rainfalls. All load evaluations generally resulted in an underestimation of the true load, most likely due to the flashy hydrological response of the small watershed studied. Of the seven load calculation methods studied, linear interpolation, which is used to obtain an estimate of concentration for each available hourly flow value, systematically yielded the lowest RMSE value. However, ratio estimator methods did not fare well, ranking only 5 and 6 out of the seven methods tested.