以地形为基础的HSC产流模型在辽西半干旱丘陵流域的适用性

HSC topography-driven runoff generation model′s applicability in a semi-arid watershed in western Liaoning Province

采用以地形为基础的HSC产流模型(HAND-based storage capacity curve),以典型半干旱丘陵区的辽宁省叶柏寿流域为研究区,检验模拟效果。在只改变产流模块的情况下,把HSC线型同HBV(hydrologiska byrans vattenbalansavedlning)模型的beta函数、新安江模型的蓄水容量曲线、基于雨强阈值的超渗模型进行比较后发现,在进行长时间序列模拟时,极端高洪峰流量过程对径流的模拟结果有很强的影响。传统模型依赖参数率定,普遍存在参数过拟合现象,即为了提高洪峰流量模拟精度,导致优化的产流参数严重背离物理意义和假设。研究提出的产流模型基于HSC线型并耦合根系区蓄水容量反演的MCT方法,参数物理意义明确且无需率定,可避免传统模型存在的为提高对少数洪峰流量的模拟精度而产生的参数过拟合现象,因此在率定期和验证期其洪水模拟精度表现更为可靠和稳定。

In arid and semi-arid regions,water scarcity poses prominent constraints to socio-economic development and threatens nature ecosystem conservation.Hydrological models represent integrated and systematic knowledge of catchment hydrology,which is essential to forecast floods and droughts,and decisions-makings on water resources management.However,the model performance in arid and semi-arid regions is far from satisfactory and the runoff generation mechanism is much more complex due to heterogeneous landscapes and complex hydrometeorological processes in humid regions.The topographic-driven HSC(HAND-based storage capacity curve)runoff generation model using the topographic information to derive the non-linear relationship between the relative soil moisture and the extent of saturated areas in catchment scale.Compared with HBV and TOPMODEL,HSC model performed better in hundreds of catchments in the United States,especially in the semi-arid catchment with gentle slopes,less forest coverage.However,the model performance in China′s arid and semi-arid hilly catchments is still unknown.The HSC topography-driven runoff generation model uses the topographic index HAND(height above the nearest drainage)to establish the HAND-based storage capacity curve,which allows to estimate the spatial distribution of storage capacity without parameter calibration.Besides,through the mass curve technique(MCT)method,the parameter of root zone storage capacity(Sumax)can be derived from climate data.Coupling HSC with MCT,the HSC-MCT runoff generation model is established which does not require parameter calibration.To test HSC′s performance,three benchmark models:the HBV′s beta function,the Xin′anjiang model′s water storage capacity curve,and the infiltration excess model which is based on rainfall threshold were used for comparison.The Yebaishou watershed,a typical semi-arid hilly region in Liaoning Province of China,was used as a study site.The Kling-Gupta efficiency(IKGE),the root mean square error(ERMS),and the coefficient of determination(R2)was used to evaluate model performance.The hydrological and meteorological data of the Yebaishou watershed from 1961 to 1970 was used to calibrate the extra free parameters,including routing module and three benchmark runoff generation modules,by the MOSCEM-UA(multi-objective parameter optimization algorithm),and period from 1971 to 1980 was used for validation.Five models could reproduce the timing of the main peak flows very well.The HSC-MCT performed well in both calibration and validation for simulating the volume of peak flow,but the other four models overestimated peak flows.However,for the maximum peak flood simulation in 1962,the HSC-MCT and the infiltration excess model underestimated the peak flood,but the HSC,HBV,and Xin′anjiang could reproduce this event well.In terms of evaluation metrics,such as IKGE,ERMS,and R2,it was found that despite the other four models were of good results during calibration,the simulation results in validation significantly deteriorated.The calibrated parameters of these models were far beyond their range of physical reality,for example,the parameter of maximum interception capacity(Simax),and the parameter of root zone storage capacity(Sumax).The HSC-MCT model had the best performance in validation,and maintain robustness thoroughly.The extreme peak flow had a strong influence on model performance and parameter calibration.But the HSC-MCT model allows us to derive the root zone storage capacity(Sumax)by climate data and its spatial pattern by topography information,thus free of parameter calibration,and avoid parameter overfitting.In summary,the robustness and stability of HSC-MCT model to simulate hydrological processes is satisfactory in this semi-arid Yebaishou watershed.