Abstract It is critical for monthly water balance models (MWBMs) to achieve realistic hydrological modelling of total flow and its components (i.e. quick flow and baseflow) in practical application. Various methods have been developed to improve the performances of the three flow components by focusing on calibration procedures. However, the understanding of runoff partitioning structure in MWBMs for better performances is still very limited, especially whether the storage-discharge relationship is linear or nonlinear at monthly time scale. In this study, model structures for baseflow simulation in 5 widely used MWBMs are reviewed and modified from a linear storage-discharge relationship to a nonlinear exponential storage-discharge relationship to achieve realistic baseflow simulation in 443 catchments from Australia with diverse hydro-climatic conditions. The performances of original and modified models are evaluated and compared through four assessment criteria including Nash-Sutcliffe efficiency (NSE), logarithmic form of NSE (NSE(log)), Pearson correlation coefficient (r) and Bias (B). Basically, the original models with linear storage-discharge relationship perform satisfactorily in simulating total streamflow and quick flow, but degrade remarkably for simulating baseflow with an underestimation of −60 ± 36% in all study catchments. The modified MWBMs with nonlinear storage-discharge relationship significantly outperform the original ones for simulating both total streamflow and baseflow. The assessment criteria NSE, NSE(log), r and B of total streamflow improve in 82 ± 4.0% (mean ± 1 standard deviation of 5 MWBMs), 72 ± 4.7%, 76 ± 4.5% and 51 ± 2.4% study catchments, respectively. The NSE(log) and r of baseflow simulated using the modified MWBMs have improved in 68 ± 4.6% and 83 ± 4.1% catchments with median improvement of 0.17 ± 0.03 and 0.14 ± 0.03, respectively. It suggests that the exponential nonlinear storage-discharge relationship is more capable for MWBMs to capture storage-discharge dynamics than the linear one at monthly time scale. This study highlights that, at monthly time scale, the nonlinearity in catchment storage-discharge relationship is a very important factor for MWBMs performance and more studies are required to reveal catchment monthly runoff generation mechanisms.

中文翻译：

---

使用澳大利亚 443 个流域评估月水量平衡模型中的基流模型结构

摘要 月水量平衡模型（MWBMs）在实际应用中实现总流量及其组成部分（即快流和基流）的真实水文建模至关重要。已经开发了各种方法来通过专注于校准程序来提高三个流量组件的性能。然而，对于 MWBMs 中径流分区结构以获得更好性能的理解仍然非常有限，尤其是在月时间尺度上的储排关系是线性还是非线性。在这项研究中，审查了 5 个广泛使用的 MWBM 中用于基流模拟的模型结构，并将其从线性存储-排放关系修改为非线性指数存储-排放关系，以在澳大利亚 443 个具有不同水文气候条件的流域中实现真实的基流模拟。通过纳什-萨特克利夫效率 (NSE)、NSE 的对数形式 (NSE(log))、皮尔逊相关系数 (r) 和偏差 (B) 四个评估标准，对原始模型和修改模型的性能进行评估和比较。基本上，具有线性储排关系的原始模型在模拟总流量和快速流量方面表现令人满意，但在模拟基流时性能显着下降，在所有研究流域中低估了 -60 ± 36%。在模拟总流量和基流方面，具有非线性存储-放电关系的改进型 MWBM 显着优于原始 MWBM。总流量的评估标准 NSE、NSE(log)、r 和 B 在 82 ± 4.0%（5 MWBM 的平均值 ± 1 标准偏差）、72 ± 4.7%、76 ± 4.5% 和 51 ± 2.4% 研究集水区改善，分别。使用改进的 MWBM 模拟的基流的 NSE(log) 和 r 在 68 ± 4.6% 和 83 ± 4.1% 的流域中分别提高了 0.17 ± 0.03 和 0.14 ± 0.03 的中值改进。这表明指数非线性存储-放电关系比线性关系在月时间尺度上更能够使 MWBM 捕获存储-放电动态。这项研究强调，在每月的时间尺度上，