Hydrological drought simulations: How climate and model structure control parameter sensitivity

This folder contains the output files from:

L.A. Melsen and B. Güse (2019), Hydrological drought simulations: How climate and model structure control parameter sensitivity, Water Resources Research, doi: 10.1029/2019WR025230

The data contain hydrological drought indicators (e.g. median drought duration) for three different models (SAC, VIC, HBV), where these models were run with a sample of parameters for 605 basins in the US. Sensitivity analysis was applied to the indicators.

Study abstract:
Hydrological drought, defined as below average streamflow conditions, can be triggered by different mechanisms which are to a large extent dictated by the climate. Moreover, the simulation of hydrological droughts highly depends on the model structure and how drought triggering mechanisms are parameterized. In this large-sample hydrological study, we investigate how climate and model structure control hydrological drought simulations. We conducted sensitivity analysis on parameters of three frequently used hydrological models (HBV, SAC, and VIC) for the simulation of drought duration and drought deficit over 605 basins covering more than ten different K\"oppen-Geiger climates. The sensitivity analysis revealed that, as anticipated, different parameter are sensitive in different climates. However, not all expected drought mechanisms were reflected in the parameter sensitivity: especially the sensitivity of ET parameters does not align with the theory, and the role of snow parameters in snow-related droughts shows a distinction between degree-day based models and energy-balance models. Besides parameter sensitivity being different over climates, we also found that parameter sensitivity differed over the different models. Where HBV and SAC did display fairly similar behaviour, in VIC other model mechanisms were triggered. This implies that conclusions on driving mechanisms in hydrological drought cannot be based on hydrological models only, as different models would lead to different conclusions. Hydrological models can have heuristic value in drought research, to formulate new theories and identify research directions, but formulated theories on driving processes should always be backed up by observations.

本文件夹包含来自L.A. Melsen和B. Güse（2019年）的研究成果，该研究题为《水文干旱模拟：气候与模型结构如何控制参数敏感性》，发表在《水资源研究》期刊上，DOI：10.1029/2019WR025230。数据集包含了水文干旱指标（例如，干旱持续时间的中位数）的样本，涉及三种不同的模型（SAC、VIC、HBV），这些模型针对美国605个流域的参数样本进行了运行。对指标进行了敏感性分析。

研究摘要：水文干旱，定义为低于平均的河流流量条件，其触发机制多种多样，在很大程度上受气候所决定。此外，水文干旱的模拟高度依赖于模型结构以及干旱触发机制的参数化。在本项大规模水文研究中，我们探讨了气候和模型结构如何控制水文干旱的模拟。我们对三种常用水文模型（HBV、SAC、VIC）的参数进行了敏感性分析，以模拟605个流域的干旱持续时间和干旱亏缺，这些流域覆盖了超过十个不同的K"oppen-Geiger气候类型。敏感性分析揭示，如预期的那样，不同的参数在不同气候下表现出敏感性。然而，并非所有预期的干旱机制都在参数敏感性中得到了反映：特别是ET参数的敏感性并未与理论相符，而在与雪相关的干旱中，雪参数在基于度日法和能量平衡模型中的角色表现出显著差异。除了参数敏感性在气候间的差异外，我们还发现参数敏感性在不同模型之间也存在差异。HBV和SAC在行为上显示出相当相似的模式，而在VIC中则触发了其他模型机制。这表明，关于水文干旱驱动机制的研究结论不能仅基于水文模型，因为不同的模型会导致不同的结论。水文模型在干旱研究中具有一定的启发价值，可用于制定新理论和确定研究方向，但关于驱动过程的理论应始终以观测结果为支撑。