一个新的分布式水文模型在鄱阳湖赣江流域的验证

VERIFICATION OF A NEW DISTRIBUTED HYDROLOGIC MODEL IN GANJIANG RIVER CATCHMENT,POYANG LAKE WATERSHED

考虑同类模型在模拟大尺度流域时存在的一些弱点，开发了基于单元网格，并且结构较为简单、参数数量较少的分布式水文模型WATLAC。为了验证该模型在大尺度流域长时间序列的模拟效果，选取鄱阳湖流域的最大子流域——赣江流域（8．17万km^2）为验证区域。流域的空间非均匀性由单元网格离散并模拟。根据下垫面数据精度，选取4km×4km的离散网格尺寸，并应用ArcGIS等软件建立流域网格信息，生成模型输入数据。采用1960～1989年外洲站和峡江站的实测资料对模型进行了验证。率定期（1960～1969年）两水文站日、月和年径流确定性系数在0．81～0．93，Ens系数在0．79～0．99，年径流统计量相对误差均低于8％；校核期（1970～1989年）两水文站日、月和年径流确定性系数在0．76～0．92，Ens系数在0．75～0．91，年径流统计量相对误差均低于2％。结果显示，模拟的径流量与实测值吻合良好，模型在赣江流域得到了很好的验证。模型对赣江流域具有较强的适用性，在流域水资源管理中具有较好的应用前景。

After reviewing the weaknesses of some of the existing hydrological models in simulating the hydrological processes for large-scale catchments,a new distributed hydrologic model--WATLAC was developed. The model is grid-based adopting relatively simple mathematical formulations with a reduced number of parameters. In order to verify the capabilities of the model in terms of simulating large-scale catchment with a long simulation time,Ganjiang River catchment （8.17 × 10^4 km^2） ,the largest sub-catchment of Poyang Lake watershed,was selected as the study area in this paper. The spatial heterogeneity of the catchment was simulated by discretized grids. Based on the resolution of the surface data, a grid size of 4 km ×4 km was selected in the model. Input data including surface and subsurface grid information for the model were prepared by using ArcGIS. The model was calibrated against the observed daily stream flows at Waizhou and Xiajiang stations from 1960 to 1989. In the calibration period for 1960-1969,the correlation coefficients for daily, monthly and annual stream flows at the two gauging stations varied from 0.81 to 0.93 ,and the Nash-Suttclife efficiency coefficients （Ens） were in the range of 0.79-0.99. The relative errors between the simulated and the observed runoff volume were less than 8%. In the validation period of 1970-1989,the correlation coefficients for daily, monthly and annual stream flows varied from 0.76 to 0.92,and the Nash Suttclife efficiency coefficients （Ens） were in the range of 0.75-0.91. The relative errors between the simulated and the observed runoff volume were tess than 2%. Results indicate that the modeled stream flows match well with the observed ones,and the model is satisfactorily tested in Ganjiang River catchment. The model may be suitable for large-scale hydrological modeling,and has a good application prospect in assisting water resources management.