Dep. of Biological and Environmental Engineering, Riley Robb Hall, Cornell Univ., Ithaca, NY 14853, USA.
J Environ Qual. 2010 Jul-Aug;39(4):1338-48. doi: 10.2134/jeq2010.0013.
The contribution of runoff from various land uses to stream channels in a watershed is often speculated and used to underpin many model predictions. However, these contributions, often based on little or no measurements in the watershed, fail to appropriately consider the influence of the hydrologic location of a particular landscape unit in relation to the stream network. A simple model was developed to predict storm runoff and the phosphorus (P) status of a perennial stream in an urban watershed in New York State using the covariance structure of runoff from different landscape units in the watershed to predict runoff in time. One hundred and twenty-seven storm events were divided into parameterization (n = 85) and forecasting (n = 42) data sets. Runoff, dissolved P (DP), and total P (TP) were measured at nine sites distributed among three land uses (high maintenance, unmaintained, wooded), three positions in the watershed (near the outlet, midwatershed, upper watershed), and in the stream at the watershed outlet. The autocorrelation among runoff and P concentrations from the watershed landscape units (n = 9) and the covariance between measurements from the landscape units and measurements from the stream were calculated and used to predict the stream response. Models, validated using leave-one-out cross-validation and a forecasting method, were able to correctly capture temporal trends in streamflow and stream P chemistry (Nash-Sutcliffe efficiencies, 0.49-0.88). The analysis suggests that the covariance structure was consistent for all models, indicating that the physical processes governing runoff and P loss from these landscape units were stationary in time and that landscapes located in hydraulically active areas have a direct hydraulic link to the stream. This methodology provides insight into the impact of various urban landscape units on stream water quantity and quality.
流域内各种土地利用类型对河道的径流量贡献常常被推测出来，并被用于支持许多模型的预测。然而，这些贡献往往基于流域内很少或根本没有测量，未能适当考虑到特定景观单元在与河道网络的水力位置关系方面的影响。本文开发了一个简单的模型，利用流域内不同景观单元的径流量协方差结构来预测时间上的径流量，从而预测纽约州一个城市流域内的永久性溪流的暴雨径流量和磷（P）状况。将 127 次暴雨事件分为参数化（n = 85）和预测（n = 42）数据集。在三个土地利用类型（高维护、无维护、林地）的九个地点、流域的三个位置（出水口附近、流域中部、上游流域）以及流域出口处的溪流中测量了径流量、溶解磷（DP）和总磷（TP）。计算了流域景观单元（n = 9）径流量和 P 浓度之间的自相关和景观单元与溪流测量值之间的协方差，并用于预测溪流响应。使用留一交叉验证和预测方法验证的模型能够正确捕捉溪流流量和溪流 P 化学的时间趋势（纳什-苏特克利夫效率，0.49-0.88）。分析表明，所有模型的协方差结构都是一致的，这表明控制这些景观单元径流量和 P 损失的物理过程在时间上是稳定的，位于水力活跃区域的景观与溪流之间具有直接的水力联系。这种方法为了解各种城市景观单元对溪流水量和水质的影响提供了思路。
