洞庭湖流域分布式水文模型

流域出口径流观测序列是水文模型参数率定重要依据,不受水文站控制区域的模型应用是水文研究关注点之一。首先根据水文站观测资料建立洞庭湖流域四水控制站之上基于水文响应单元的分布式水文模型,在此基础上结合实验、同质移植和虚拟水库等方法,将分布式水文模型拓展到包含无径流站控制区域的丘陵区间和平原圩垸区,最终实现了洞庭湖全流域水文过程模拟。结果表明:在较完备土壤、地形、土地利用等空间数据支持下,通过合理的流域划分和水文响应单元定义,建立的流域分布式水文模型可以较好地在水文响应单元尺度反应降水发生后蒸散、地表径流、土壤和地下水的响应特征。而基于观测实验及基流分割等方法获取的关键水文过程特征对模型参数优化的认识,可以提高模型参数率定效率,在较少优化迭代运算后既可使月径流模拟的效率系数NSE和确定性系数R²值高于0.81(日过程高于0.62)。借助参数同质移植和虚拟水库解决了区间和圩垸区无控制站区域水文过程模拟。在全流域水文过程的模拟中,基流指数和蒸散比例与实际过程具有较好的一致性。说明相关参数较好地反映了其物理机制,具备在相似气候及下垫面条件区域进行同质移植的基础,圩垸区径流交换采用虚拟水库的处理方式也合理可行。

PARAMETER OPTIMIZATION FOR SWAT MODEL AND ITS EXTENDED APPLICATION IN THE DONGTING LAKE BASIN,YANGTZE RIVER

Distributed hydrological model becomes an important tool for revealing the temporal and spatial characteristics of water cycles at ranging scale. But the application of the model at regions without runoff observations is still a major obstacle to overcome. In this study, a HRU-based distributed hydrological model was developed with SWAT for a large lake basin based on DEM, land use, reservoir, climate and hydrological data. To obtain a deterministic and physical mechanism clear parameter system some field work on understanding of parameters was performed, then the calibrated and validated with SUFI-2 algorithm were conducted within the hydrological gauging station control regions by using SWAT-CUP. The results show that the application of the model could be more efficient with field knowledge support about rational value ranges of the parameters. The efficiency coefficient values NSE and the correlation of determination coefficient R² values were easily reaching 0.81 at monthly step (0.62 for daily simulation), and the number of iteration progress was also largely reduced. In this distributed model, the hydrological response unit (HRUs) reflects the heterogeneity of hydrological behaviors of the catchment in space. Thus, some of the hydro-parameters are transportable between HRUs with similar conditions of terrain, land use and soil attributes. Base on the experimental knowledge and output of parameters from validation, the hydrological model was extended covering the whole basin by homogeneous transplantation of hydro-parameters in HRUs. For the ponder regions, the flow exchange process was dealt by introducing a virtual reservoir for each independent ponder. The output of the extended model successfully reveals the characters of ratios among ET, surface runoff, baseflow at a HRU, sub-basin and basin scales. The simulated baseflow, and ET processes are well consist with BFI and remote sending retrieved ET data. From the application of this study, a detailed database covering land use, soil properties and meteorology is an essential pre-requisite condition to start a distributed hydrological model. A logical parameter assembling is not only important for the set-up of distributed hydrological model, it is also crucial for its extending application. And the extendable of the model is also largely decided by the reasonable delineation for watershed and HRUs. The application has especially contributed for hydrological modeling in regions of gauging flow is partly absent in space. The parameters of HRUs which were generated from the modeling processes could be valuable support information for the study of water retention, vulnerability analysis, and spatial planning in view of the hydrological and ecological effects, which in turn may largely reduce the uncertainty analysis in land hydrology.