A GIS-Based Method of Deriving Spatially Distributed Unit Hydrographs

Abstract: Prior to using hydraulic and spatially distributed modelling softwares, the theory of the unit hydrograph was a commonly used tool for modelling of surface and runoff water. While distributed models often provide detailed results from extensive calculation durations, the unit hydrograph have been questioned for simplifying the physical characteristics of the watershed modelled. Typically, the unit hydrograph theory does not explicitly take the flow paths of the watershed in consideration during calculation. With the rise of geographical information systems, methods of deriving spatially distributed unit hydrographs have been developed. The aim of these have commonly been to find a spatially varied form of hydrological modelling, while still keeping the computation times low. The method is commonly built by calculating the travel time to the watershed outlet along the flow path. In this study, spatially distributed unit hydrographs are derived separately for the watershed’s pervious and impervious surfaces in a Python script using map algebra and the Esri’s Python wrapper module Arcpy. The travel times are generated from a velocity field calculated using Maidment and Olivera’s velocity equation. The velocity equation contains three unknown parameters; one for an average velocity and two calibration parameters. The excess precipitation is calculated of a 100 year return period Chicago Design Storm hyetograph using the SCS-CN method. The direct runoff hydrographs are calculated over three semi-urban watersheds in Smedby in southern Sweden, and the results are compared to MIKE 21 hydrograph data of each corresponding watershed and rain input. The result obtained showed to replicate the hydrograph response quite well, but only if the unknown parameters in the velocity equation were calibrated to match the MIKE 21 data. The unknown parameters of the velocity equations produces uncertainties of using the method without calibration data, which implies that the script is not well adapted to use for modelling predictions. It may be of interest to calculate the travel times of the locations within the watershed using a different formula. The script tool could be tested using different design storms as input, and areas of different characteristics compared to Smedby could be tested.