Catchment morphology and drainage network influences on runoff hydrographs

Hydrologic models are widely used to address many practical applications including flood forecasting, water supply management, floodplain delineation, etc. When modeling a watershed, two approaches are available. Lumped or semi-distributed models strategically divide a watershed into sub-watersheds and then describe each watershed using spatially constant characteristics. Distributed models divide the watershed into a large number of grid cells and allow different characteristics for every grid cell. While distributed models are considered to give a more accurate representation of a watershed, practitioners use lumped descriptions almost exclusively because they rarely have enough data to use the distributed approach. A critical component of lumped rainfall-runoff models is the so-called transform method, which is used to determine stream flow rates at the sub-watershed outlet from the excess precipitation rates that occur across a sub-watershed. This transformation depends on the paths that the water must travel to reach the outlet and the speed at which the water travels. Numerous so-called synthetic unit hydrographs are available to make this transformation, but such methods do not fully reflect the geometrical configuration of the given sub-watershed, which largely determines both the flow paths and the flow speeds. Topographic data is now available globally at resolutions that are suitable for this application, so currently available methods are now obsolete. However, rather than requiring detailed processing of the local topography, this research seeks to identify a smaller number of salient topographic characteristics that largely dictate the runoff hydrograph. The identification of such features is expected to allow more simplified incorporation of sub-watershed geometry into hydrologic models. The approach calculates and analyzes runoff hydrographs from synthetic topographies derived from a simple landscape evolution model and from real topographies with varying channel network types in order to understand the sensitivity of the hydrologic response to geomorphic measures. Hydrographs are estimated with a deterministic model based on the kinematic wave theory that incorporates local topographic characteristics and has been shown to perform better than older estimation methods. One measure used to quantify the characteristics' effects on the hydrograph is the time to return to baseflow divided by the time to peak flow. Using this methodology the synthetic topographies indicate that changes in the fluvial process parameters have greater effects on hydrograph timing than do changes to the hillslope diffusivity parameter. In addition, analyses of real catchments indicate that the drainage network classification can influence the type and strength of the observed relationships between surface characteristics and the hydrograph. For example, channel and hillslope roughness values both had stronger and more clearly defined effects on hydrograph timing for parallel networks than they did for dendritic networks. This work contributes to a better understanding of the connections between catchment physical characteristics and runoff hydrographs and, therefore, has application in estimating discharge from ungauged basins.