Two-dimensional subsurface flow modeling for watersheds under spatially and temporally variable rainfall

Abstract

Subsurface flow constitutes an important part of the hydrologic cycle: it becomes especially important for land surfaces where most of the rain water infiltrates into the soil. Subsurface flow supplies water to the channels, or to the aquifer. In pervious soils, where the horizontal conductivity far exceeds the vertical conductivity, subsurface flow becomes the dominant process in the rainfall-runoff relationship of natural watersheds. A two-dimensional finite difference scheme has been formulated to simulate subsurface flow within the surface runoff model CASC2D. The subsurface flow formulation is based on Darcy's law, assuming that the vertical permeability is negligible. The subsurface flow is calculated for each square grid cell that represents spatially variable watershed characteristics. The physical properties, including surface roughness, soil infiltration parameters, hydraulic conductivity, and soil moisture, are assumed uniform within each grid cell. The modified model CASC2DSUB is subsequently integrated with the Geographical Information system(GIS) application Geographical Resource Analysis Support System(GRASS4.1) incorporating, two-dimensional graphic capabilities, allowing continuous observation of the ongoing hydrological processes resulting from a rainfall-runoff event as simulation progresses. The program is first tested on a virtual test plot under varying conditions of soil types, hydraulic conductivities, and uniform rainfall intensity (1.0E-5m/s). The results obtained for clayey soils, with low hydraulic conductivity (8.33E-7m/s), show that overland flow is the dominant process. For gravelly soils, with high hydraulic conductivity (3.0E-2m/s), all the water infiltrates into the ground, and subsurface flow is the dominant process. For sandy soils (K= 3.27E-5m/s), all the water infiltrates into the ground, but the subsurface flow drains the watershed slowly and most of the water is retained in the soil matrix. The model applicability has been tested on the Goodwin Creek watershed in Mississippi. The model is calibrated using the rainstorm event of October 17, 1981, and verified with the rainstorm event on December 2, 1983. In each case, the simulated hydrograph replicated the observed hydrograph. Finally, the range of rainfall intensities and soil types for which surface or subsurface flow is predominant is defined. It is observed that for the permeable sandy soils, with hydraulic conductivities ranging from 1.0E-5m/s to 1.0E-3m/s under rainfall intensities less than 1 in/hr(7.055E-06m/s), the response of a watershed in term s of rainfall runoff would mostly be influenced by subsurface flow. For clayey soils with hydraulic conductivities ranging from 1.0E-8m/s to 1.0E-7m/s, and rainfall intensities in excess of 0.1 in/hr(7.055E-07m/s), the response of the watershed would be governed by surface flow.