SSF Model Benchmarking: Towards a robust parameterization of subsurface stormflow in hydrological models at the catchment scale

流域尺度的水文模型用不同的数学方法来描述非饱和带(地下风暴流，SSF)中的快速地下水流过程，从概念性的(线性存储)到简化的基于物理的方法。由于缺乏实测资料，控制SSF动态的模型参数通常是通过实测流量标定得到的。因此，由于错误的原因(模拟的SSF与模拟的地表径流、河岸带、地下水等不现实的相互作用)，很有可能是正确的(很好地模拟了流量)。现有方法对SSF的模拟效果如何仍是一个悬而未决的问题。以前的所有模型间比较研究都是在一个地点进行的，只使用一个水文模型，由于普遍缺乏对SSF过程的了解，到目前为止，在具有不同气候和水文特征的集水区的水文模型中，SSF例行程序难以系统地确定基准和加以改进。为了解决关于SSF程序可靠性的这一知识差距，我们将对两个集总(TopModel，HBV型)和三个更复杂的分布式水文模型(NASIM，WASIM-ETH)模拟SSF动态的能力进行基准测试，在研究单位的四个研究区域中的每一个区域中模拟流域尺度的SSF动态。我们将首先利用以前的流量观测来校准模型，并通过系统的灵敏度分析来确定SSF的相关参数。在研究单位的第一次实地活动之后，新收集的多变量SSF代理被用于通过多目标方法改进模型校准，以量化不同模型对SSF的模拟情况，并揭示所实施的数学过程表示中的缺陷。改进后的模型将被用来探索SSF阈值和集水区储存动态之间的关系，以帮助更好地理解集水区尺度上的SSF过程。

Catchment-scale hydrological models represent fast subsurface flow processes in the unsaturated zone (subsurface stormflow, SSF) by different mathematical methods ranging from conceptual (linear storages) to simplified physically based approaches. Due to a lack of measured data, model parameters that control SSF dynamics are commonly found by calibration with observed river discharge. Consequently, there is a high chance of being right (well-simulated discharge) for the wrong reasons (unrealistic interplay of simulated SSF with simulated surface runoff, riparian zone, groundwater, etc.). How well SSF is simulated by the existing approaches remains an open question. All of the previous model intercomparison studies were performed at one site using only one hydrological model and a general lack of understanding of SSF processes has so far prohibited the systematic benchmarking and improvement of SSF routines in hydrological models at catchments with different climatic and hydrological characteristics. To address this knowledge gap on the reliability of SSF routines we will benchmark the ability of two lumped (TOPMODEL, HBV) and three more complex distributed hydrological models (NASIM, WaSiM-ETH) to simulate SSF dynamics at the catchment scale in each of the four study areas of the research unit. We will first use previous discharge observations to calibrate the models and identify the SSF relevant parameters by a systematic sensitivity analysis. After the first field campaigns of the research unit, the newly collected multivariate SSF proxies are used to improve the model calibration by a multi-objective approach, to quantify how well SSF is simulated by the different models and to reveal deficits in the implemented mathematical process representations. The improved models will then be used to explore relationships between SSF thresholds and catchment storage dynamics to contribute to an improved understanding of SSF processes at the catchment scale.