GRISTORM: GRID?BASED VARIABLE SOURCE AREA STORM RUNOFF MODEL

Abstract

A gridbased storm runoff model in a variable source area is described. This model predicts temporal variationsand spatial distributions of subsurface flowsaturation overland flow with various shallow soil depths. The model usesASCIIformatted map data supported by the GRASS Geographic Information Systems (GIS) and generates distributed resultssuch as discharge, flow depth, and soil moisture in overland flow areas. The model uses a single overland flowpath algorithmand simulates surface and/or subsurface water depth at each grid element for a given time increment. A combinedsurfacesubsurface kinematic modeling approach was used to simulate surface and subsurface flow. The model was appliedto a 170 ha watershed located in New York State. Predicted flows from six summer storm events in 1994 were compared withobserved flows at the watershed outlet. The initial soil moisture conditions for the storm events were based on calibratedvalues from Frankenbergers Soil Moisture Routing (SMR) model and were adjusted to include soil moisture variations withina day. The average NashSutcliffe efficiency for predicting runoff at the watershed outlet using calibrated parameters forthree storm events was 0.72. The initial soil moisture distribution proved to be the most sensitive parameter affecting streamflow at the watershed outlet and required calibration to obtain reasonable results. The second most sensitive parameter wasMannings roughness coefficient for stream and overland areas. It affected the time and magnitude of peak stream flow. Resultsshowing temporal variations and spatial distributions of overland flow areas were presented using GRASS. According to theresults showing the spatial effect with differing grid element sizes, the model behavior was also sensitive to DEM gridresolution.