The use of GIS to derive distributed unit hydrographs for streamflow prediction

Abstract

In many regions where rainfall-runoff models are required, there is a lack of streamflow data available to calibrate the model parameters. This can restrict the use of commonly used lumped parameter models, such as AWBM, SimHyd or the Sacramento model. Some studies suggest that the parameters of these models have limited physical basis, and do not regionalise reliably. An alternative approach to determining the timing of runoff involves the use of unit hydrographs, provided an estimate of excess rainfall can be obtained using a suitable approach (of which many are available). A great deal of data relevant to hydrologic modelling, such as catchment topology, soil type and land use, are now stored in Geographical Information Systems (GIS). Much of these data are readily available, or can be obtained in a reasonable time frame through remote sensing methods, for example. Analysis of these data allows estimates of travel times over the catchment to be obtained, and hence the unit hydrograph to be derived, without any historical record of streamflow. In order to derive the unit hydrograph, estimates of the flow velocities throughout the catchment are required. A number of methods have been developed to allow the velocity to be calculated on a grid cell basis over the catchment, allowing the velocity to change based on the flow type (hillslope flow or channel flow) and local catchment characteristics, such as slope and roughness. A number of uncertainties are involved in the determination of these velocities, such as the flow accumulation required before channel flow commences, Manning's n values, and the approach used to account for the hydraulic radius of flow. The purpose of this paper is to investigate the influence of the parameters involved in the velocity estimates on the unit hydrograph produced, and to determine if these parameters can be calibrated to provide an accurate representation of the observed unit hydrograph. Estimates of travel times and velocities have been obtained by using Visual Basic ArcObjects in ArcGIS® 9.2. The SimLab package (Saltelli et al., 2000) with extended Fourier Amplitude Sensitivity Testing (FAST) method was used to investigate the importance and relationships between parameters involved on the unit hydrographs produced. The results indicate that the threshold for channel flow and the Manning's n value had little impact on the hydrographs produced, while the maximum channel velocity and velocity when flow transition from hillslope to channel flow occurs had the most significant effect on the shape of the hydrograph. These two velocities can be used to produce the desired shape of the hydrograph, altering the peak flow and timing of the peak flow, while the minimum hillslope velocity can be used to fine tune the length of the recession limb. The results suggest that using only remotely sensed elevation data in a GIS framework is a feasible approach to approximate the expected unit hydrograph for a given catchment. Further work will investigate if values for the velocity parameters can be derived from catchment characteristics, such as the channel cross-sectional area or catchment slope, to enable the approach to be applied to ungauged catchments. This would allow streamflow to be estimated for ungauged catchments, where a lack of streamflow data prohibits the calibration of more traditional models, but where suitable GIS data are available.