Huang Zhi-lin, Tian Yao-wu, Xiao Wen-fa, Zeng Li-xiong, Ma De-ju
Key Laboratory of Forest Ecology and Environment, State Forestry Administration China, Research Institute of Forest Ecology, Environment and Protection, Chinese Academy of Forestry, Beijing 100091, China.
Huan Jing Ke Xue. 2009 Oct 15;30(10):2872-8.
Watershed models provide a cost-effective and efficient means of estimating the pollutant loadings entering surface waters, especially when combined with traditional water quality sampling and analyses. But there have often been questions about the accuracy or certainty of models and their predictions. The main goal of this study was to evaluate the performance of AnnAGNPS (Annualized AGricultural NonPoint Source)Pollution Model, in simulating runoff, sediment loading and nutrient loadings under Three Gorges Reservoir area. Most of model input parameters were sourced from Zigui Forest Ecology Station in Three Gorges Reservoir area, State Forestry Administration. Data year 2003 was used for calibration while data year 2004 was used for validation of the model. The whole evaluation consisted of determining the coefficient of determination (R2), Nash-Sutcliffe coefficient of efficiency (E), and the percentage volume error (VE). Results showed that the model predicted the daily runoff volume within the range of acceptable accuracy. The runoff on a daily basis was underpredicted by 5.0% with R2 of 0.93 (p < 0.05) during calibration and underpredicted by 6.7% with R2 of 0.90 (p < 0.05) during validation. But sediment loading was able to produce a moderate result. The model underpredicted the event-based sediment loading by 15.1% with R2 of 0.63 (p < 0.05) during calibration and 26.7% with R2 of 0.59 (p < 0.05) during validation. For the events of small magnitude, the model generally overpredicted sediment loading, while the opposite was true for larger events. Nitrogen loading prediction was slightly better with R2 = 0.68 (p < 0.05), and phosphorus loading performance was slightly poor with R2 = 0.65 (p < 0.05). In general, the model performs well in simulating runoff compare to sediment loading and nutrient loadings, and as a watershed management tools it can be used for Three Gorges Reservoir area conditions that with mixed types of land uses and steep slopes.
流域模型提供了一种经济高效的方法来估算进入地表水的污染物负荷，特别是与传统水质采样和分析相结合时。但人们常常对模型及其预测的准确性或确定性存在疑问。本研究的主要目标是评估AnnAGNPS（年度农业非点源）污染模型在模拟三峡库区径流、泥沙负荷和养分负荷方面的性能。模型的大部分输入参数来自国家林业局三峡库区秭归森林生态站。2003年的数据用于模型校准，2004年的数据用于模型验证。整个评估包括确定决定系数（R2）、纳什-萨特克利夫效率系数（E）和体积误差百分比（VE）。结果表明，该模型在可接受的精度范围内预测了日径流量。在校准期间，日径流预测值低了5.0%，R2为0.93（p < 0.05）；在验证期间，预测值低了6.7%，R2为0.90（p < 0.05）。但泥沙负荷模拟结果一般。在校准期间，该模型对基于事件的泥沙负荷预测值低了15.1%，R2为0.63（p < 0.05）；在验证期间，预测值低了26.7%，R2为0.59（p < 0.05）。对于小量级事件，该模型通常高估了泥沙负荷，而对于大量级事件则相反。氮负荷预测稍好，R2 = 0.68（p < 0.
