DISTRIBUTED RUNOFF MODELING BASED ON WATERSHED GEOMORPHOLOGICAL CHARACTERISTICS
Xiao-meng SONG,Jian-yun ZHANG,Fan-zhe KONG,Cui-shan LIU
DOI: https://doi.org/10.11870/cjlyzyyhj201504008
2015-01-01
Abstract:Hydrologists have attempted to relate the hydrologic response of watersheds as runoff production from rainfall to watershed topographic structures for many years.Unit hydrograph explains many characteristics of the basin response to rainfall input.A number of approaches have been developed to derive a unit hydrograph from watershed geomorphology,but essentially they are all involved in using measurements of distance,velocity and time to derive physical characteristics of the watershed to parameterize a unit hydrograph in the absence of observed runoff and rainfall data.In order to solve the problem of being dependent on the rainfall and runoff data,taking the Yanduhe catchment of Three Gorges in Yangtze River as the study area,this paper proposed a new method to derive distributed unit hydrograph and a revised Muskingum method based on watershed geographic characteristics and hydrodynamic conditions.First,the basin is divided into a set of sub-basins according to natural watershed line and topographic information is extracted based on DEM and GIS tools.Kriging interpolation method was used to estimate the precipitation depth at the centroid of each sub-basin.Secondly,the area-time histograms derived by means of digital elevation model were translated to discharge-time histograms.And then the discharge-time histograms were routed through a linear reservoir and unit hydrographs were obtained.In addition,the Muskingum parameters K and x were estimated based on the physical characteristics of river reach,including slope,width and length.Finally,the HEC-HMS model is used to simulate and analyze the rainfall-runoff response processes coupling with proposed distributed unit hydrograph and Muskingum model.The seven evaluation indices are used to estimate and validate the simulation results involving water balance,goodness-of-fit,peak-flows and low-flows.The results showed that,80%of the 25 floods has a relative error of peak flow of less than 20%,96%has a relative error of runoff amount of less than 20%,and 84% has a Nash-Sutcliffe coefficient of greater than 0.8is.The simulated flow agrees well with the observed flow at the outlet of the Yanduhe catchment.The proposed distributed runoff modeling method can be used to simulate the rainfall-runoff processes in the mountain basins or these ungauged basins based on DEM data and GIS information.