Time-varying parameters of hydrological models play a crucial role in capturing the dynamic nature of hydrological processes and nonpoint source pollution under changing environments. In this study, the SWAT-DynamicParam model was developed by integrating the Soil and Water Assessment Tool (SWAT) model with an ensemble Kalman filter (EnKF) to identify and analyze time-varying parameters. The results showed that the SWAT-DynamicParam framework is capable of effectively identifying multiple time-varying characteristics of parameters. The combination of flow and evapotranspiration (ET) data can accurately identify changes in CANMX (Maximum canopy storage), CN2 (Initial SCS runoff curve number for moisture condition II) and ALPHA_BF (Baseflow alpha factor), with the Relative Absolute Error 5%. Compared with using static parameters, the model simulation effect was improved by more than 30% when using time-varying parameters. In addition, the variation in parameters showed significant spatio-temporal heterogeneity. The change trend of parameters at Weijiabao station showed the largest fluctuation, with a range greater than 100. Temporally, CN2 and ALPHA_BF both reached peak values in 2008 while the trend of CANMX was the opposite. At the monthly scale, the trends of ALPHA_BF and CANMX were similar: Values were at a minimum in April and May, with the range is 2.57 times the minimum value. The CN2′s lowest value was recorded in August, whereas December saw its highest, reaching 82. In summary, the SWAT-DynamicParam model enhanced simulation accuracy by over 30%, demonstrating the pivotal role of accurately identified time-varying parameters in capturing the dynamic nature of hydrological processes and their response to environmental changes.

中文翻译：

---

变化环境下水文模拟模型的时变参数


水文模型的时变参数在捕捉变化环境下水文过程和非点源污染的动态性质方面起着至关重要的作用。在本研究中，SWAT-DynamicParam 模型是通过将土壤和水评估工具 （SWAT） 模型与集成卡尔曼滤波 （EnKF） 集成来识别和分析时变参数而开发的。结果表明，SWAT-DynamicParam 框架能够有效地识别参数的多个时变特性。流量和蒸散 （ET） 数据的组合可以准确识别 CANMX（最大冠层存储）、CN2（水分条件 II 的初始 SCS 径流曲线编号）和 ALPHA_BF（基流 α 因子）的变化，相对绝对误差为 5%。与使用静态参数相比，使用时变参数时，模型仿真效果提高了 30% 以上。此外，参数的变化表现出显著的时空异质性。魏家堡站参数变化趋势波动最大，范围大于 100。从时间上看，CN2 和 ALPHA_BF 均在 2008 年达到峰值，而 CANMX 的趋势则相反。在月度尺度上，ALPHA_BF 和 CANMX 的趋势相似：4 月和 5 月的值处于最小值的最小值，范围是最小值的 2.57 倍。CN2 的最低值出现在 8 月，而 12 月最高，达到 82。总之，SWAT-DynamicParam 模型将仿真精度提高了 30% 以上，证明了准确识别的时变参数在捕获水文过程的动态性质及其对环境变化的响应方面的关键作用。