Agricultural field- and watershed-scale water quality models are used to assess the potential impact of management practices to reduce nutrient and sediment exports. However, observed data are often not available to calibrate and verify these models. Three years of data from the U.S. Department of Agriculture-Agricultural Research Service's 12 paired edge-of-field sites in northwest Ohio were used to calibrate and validate the Nutrient Tracking Tool. The goal of this study was to identify a single optimal parameter set for the Nutrient Tracking Tool in simulating annual crop yields, water balance, and nutrient loads across the Western Lake Erie Basin. A multi-site and multi-objective auto-calibration subroutine was developed in R to perform model calibration across the edge-of-field sites. The statistical metrics and evaluation criteria used in comparing the simulated results with the observed data were: Cohen's D Effect Size (Cohen's D < 0.20) and Percent bias (PBIAS ± 10% for crop yields, subsurface (tile) discharge, and surface runoff and ± 25% for dissolved reactive phosphorus (DRP) and nitrate‑nitrogen (nitrate-N) in tile discharge, and DRP, particulate phosphorus, and nitrate-N in surface runoff). In both calibration and validation, the Cohen's D and PBIAS for annual crop yields, tile discharge, surface runoff, DRP, particulate P, and nitrate-N showed that the average simulated results were similar to the average observed values for each variable. The calibrated model simulated well the annual averages of crop yields, flows, and nutrient losses across fields. The tile drainage and phosphorus transport subroutines in the Nutrient Tracking Tool should be further improved to better simulate the dynamics of discharge and phosphorus transport through subsurface drainage. Stakeholders can use the verified model to evaluate the effectiveness of conservation practices in improving the water quality across the Western Lake Erie Basin.

中文翻译：

---

伊利湖西部流域农业领域的作物生长，水文和水质动态：营养追踪工具（NTT）的多站点验证。

农业田间和流域尺度的水质模型用于评估管理措施对减少养分和沉积物出口的潜在影响。但是，观察到的数据通常无法用于校准和验证这些模型。来自美国农业-农业研究服务局位于俄亥俄州西北部的12个成对的边缘站点的三年数据用于校准和验证营养追踪工具。这项研究的目的是为营养跟踪工具确定一个最佳参数集，以模拟西伊利湖盆地的年度作物产量，水平衡和营养负荷。在R中开发了一个多站点和多目标自动校准子例程，以在整个场边缘站点执行模型校准。用于将模拟结果与观察到的数据进行比较的统计量度和评估标准是：Cohen's D效应量（Cohen's D <0.20）和百分偏差（PBIAS±10％，用于作物产量，地下（瓷砖）排放量以及地表径流和瓷砖排放物中的溶解性反应性磷（DRP）和硝酸盐-氮（硝酸盐-N）以及表面径流中的DRP，颗粒状磷和硝酸盐-N的浓度为±25％。在校准和验证中，Cohen的D和PBIAS在年度作物产量，瓷砖排放量，地表径流，DRP，颗粒物P和硝酸盐-N方面均显示，平均模拟结果与每个变量的平均观测值相似。校准后的模型很好地模拟了田间作物产量，流量和养分流失的年度平均值。营养跟踪工具中的瓷砖排水和磷转运子程序应进一步改进，以更好地模拟通过地下排水的排放和磷转运的动力学。利益相关者可以使用经过验证的模型来评估保护措施在改善伊利湖西部水质方面的有效性。