Incorporating Spatial and Temporal Variation of Watershed Response in a GIS-based Hydrologic Model
Files
TR Number
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The hydrograph at the watershed outlet was simulated using the time-area curve concept implemented in a geographic information system (GIS). The goal of this study was to determine if hydrograph prediction accuracy would be improved by accounting for spatial and temporal variation of excess rainfall. Three models with different methods of estimating excess rainfall were developed: the Distributed Curve Number (DCN) model uses a CN for each cell, generating spatially distributed excess rainfall using the Soil Conservation Services curve number method (SCS, 1972); the Uniform Curve Number (UCN) model uses a single "average" CN for the whole watershed, thus generating a uniform excess rainfall; the Phi index model which uses the Phi-index method to generate uniform excess rainfall.
With the aid of a GIS, the cumulative flow time to the watershed outlet is estimated for each cell in the watershed and the isochrones of equal travel time are developed. The time-area curve is developed in the form of an S curve. The spatially distributed 1-hr unit hydrograph is derived from the S curve as the difference between the S curve and its value lagged by 1-hr. The models used in this study describe the physical processes and flow mechanisms. They also reflect effects of watershed characteristics (slope, landuse, soil drainage potential) and excess rainfall intensity on the resulting hydrograph at the watershed outlet. Surface flow is divided into channel flow and overland flow based on the upstream drainage area. Flow is routed to the watershed outlet through a channel network derived from the watershed Digital Elevation Model (DEM).
The models developed were tested against observed rainfall-runoff data from the 1153-ha Virginia Piedmont watershed (Owl Run). A total of 30 storms were simulated, with statistical comparison of peak flow rate, time to peak flow rate, and the hydrograph shape. The hydrograph shape was compared both visually and statistically. Results indicated that the two models which account for temporal variation in excess rainfall (DCN and UCN) predicted the output hydrograph much more accurately than the Phi model which lacks the ability to capture the temporal variation of excess rainfall. For this watershed, results showed that the spatial variability in excess rainfall which was accounted for by the DCN model did not improve the prediction accuracy over the UCN model which lacks that ability. However, a sensitivity analysis for the effect of the spatial distribution of the excess rainfall indicated that can be a significant effect of spatial distribution on the predicted hydrograph.