An improved optimization scheme for representing hillslopes and depressions in karst hydrology

Understanding hydrological processes is essential for the management of water resources and for promoting catchment sustainability. In karst regions, high spatial heterogeneous landscapes, such as discontinuous soil distribution and complex network of matrices and conduits in hillslopes and depressions, result in different hydrological processes. However, most studies have mainly focused on the effects of the distribution of soil depth and the fast-slow flow in the matrices and conduits on hydrological processes, but they have ignored the different hydrological processes on hillslopes and depressions (HD). This study improved the VarKarst model by adding randomly distributed soil and epikarst depths (RSE), fast-slow flow (FS) and HD in six large catchments (1,213~5,454 km2) and one small catchment (1.25 km2). The combination of FS and HD (Scenario FS+HD) and the combination of RSE, FS, and HD (Scenario RSE+FS+HD) for the improved VarKarst model had the best performance (calibrated and validated KGE ranged from 0.54 to 0.89 and AIC ranged from -336.49 to 669.77) compared with other scenarios (original VarKarst, Scenario RSE, Scenario FS, Scenario HD, and Scenario RSE +FS). Particularly, these two scenarios performed better than the original VarKarst in reproducing the discharge of peaks and recessions. This study confirmed that the combination of HD, RSE, and FS improved VarKarst model for karst topography and the hillslopes. It also suggested that there is a need to separate the hillslopes and depressions for modeling karstic hydrological processes.

理解水文过程对于水资源管理与促进流域可持续性至关重要。在喀斯特地区，空间异质景观（如不连续的土壤分布、坡面与洼地内基质与管道构成的复杂网络）造就了各异的水文过程。然而，现有多数研究主要关注土壤深度分布以及基质与管道内的快慢流对水文过程的影响，却忽略了坡面与洼地（HD）间迥异的水文过程。本研究通过在6个大型流域（面积1213~5454 km²）与1个小型流域（面积1.25 km²）中加入随机分布土壤与表层喀斯特深度（RSE）、快慢流（FS）以及坡面与洼地（HD），对VarKarst模型进行了改进。相较于其他情景（原始VarKarst模型、情景RSE、情景FS、情景HD以及情景RSE+FS），改进后的VarKarst模型中，同时引入FS与HD的情景（FS+HD）以及同时引入RSE、FS与HD的情景（RSE+FS+HD）表现最优：其率定与验证期的Kling-Gupta效率系数（KGE）介于0.54~0.89之间，赤池信息准则（AIC）介于-336.49~669.77之间。尤为关键的是，这两种情景在重现径流峰值与退水过程方面的表现均优于原始VarKarst模型。本研究证实，将HD、RSE与FS相结合的改进VarKarst模型可适用于喀斯特地貌与坡面的水文模拟，同时表明在喀斯特水文过程模拟中，有必要将坡面与洼地分开进行建模。