[MODFLOW-WHaT] what mtbe merl leap isco iron home
Documentation Vadose Zone Conjunctive Simulation Sleepers River Watershed Flowpath

MODFLOW-Watershed Hydrology and Transport:

2-Dimensional Hypothetical Watershed Flowpath Simulations


In order to verify its functionality in simulating watershed response, a numerical modeling system must be tested under complex boundary conditions at realistic spatial and temporal scales. [Surface Runoff] The watershed flowpath simulations of Johnson et al. (2003) provided an opportunity to compare the performance of this modeling system (MFWHT) with the well-established variably-saturated groundwater flow and transport model, MODFLOW-SURFACT 99 (MFS99) (Hydrogeologic, 1999). Johnson et al. (2003) examined the flow and transport of two different contaminants within a 2-dimensional hypothetical watershed flowpath using 10 years of precipitation and evapotranspiration data from six different sites. Two of the simulations (New Brunswick, New Jersey and Fort Collins, Colorado) exhibiting fundamentally different behavior are chosen for this comparison study. The two objectives of the comparison are to (1) evaluate and verify the model's performance against an equivalent commercial model for two climatically different scenarios and (2) examine the simulation of overland flow using two different approaches for the top grid layer.

The model domain consisted of a vertical aquifer slice representing a flowpath from the watershed divide at the up-slope end to a constant stage stream downgradient. [Water Table Positions] The slice is discretized into 20 horizontal layers and 36 vertical columns with 15 meter wide grid cells varying from 0.8 meters thick near the surface to 45 meters thick at the bottom of the domain. The top layer is modified to represent overland flow using the method described below. The ground surface is set at the top of the second layer. Layers 2 through 19 of the domain represent a typical alluvial, valley-fill acquifer with a hydraulic conductivity of 0.3 meters/day and the bottom layer represents the acquifer's bedrock (Ksat=0.01 m/day). The sides and bottom of the domain are set as no-flow boundaries and the head in the surface soil cell (layer 2) on the down-gradient end is held constant at stream level. The model parameters used in this study are listed in table 1. Johnson et al. (2003) adjusted the soil hydraulic parameters in the variably saturated groundwater flow component of MFS99 to approximate overland flow in the top grid layer. The top layer was assigned an effective hydraulic conductivity that is four orders of magnitude greater than that of the underlying soil. The soil characteristic parameters were fitted so that the soil drainage and permeability functions approximate laminar surface storage-runoff response. The top layer thickness was set to 1 meter based on the assumption that the surface flow would not exceed this height.

[Surface Runoff] For the simulations using the model presented here (MFWHT), overland flow is assumed to be laminar and the surface roughness parameter is set to a value (kd = 12000) that corresponds with a steep, rough surface (Chen, 1976). It was necessary to increase the model grid resolution (106 columns and 24 layers) for the New Jersey simulation using the MFWHT model because of numerical difficulties caused by the increased magnitude of precipitation boundary fluxes.


The model results for the New Jersey case are shown in figures 1a and 1b and the Colorado case is shown in figures 2a and 2b. The surface runoff from the down-slope model cell (adjacent to the stream) in the top layer is normalized by the total surface area of the flowpath (figures 1a and 2a). The water table position is determined relative to the ground surface for the grid column 90 meters up slope from the stream (figures 1b and 2b). The MFWHT simulation of the New Jersey case exhibits runoff characterized by large peaks that dissipate quickly (figure 1a) and are of the same order of magnitude as the precipitation events that triggered them. [Water Table Positions] This contrasts with the MFS99 runoff results which remain non-zero throughout the entire ten years, rise and fall gradually with the seasonal precipitation and do not exhibit the large peak amplitudes of the MFWHT results. The MFS99 runoff behavior reveals the shortcoming of the overland flow approximation method that was used. Specifically, the soil permeability function that was used to control the top layer's hydraulic conductivity was non-zero for the majority of the simulation due to the shallow position of the watertable (figure 1b) and resultant low capillary pressure. This resulted in more total runoff over the course of the simulation, causing less flow to the subsurface and thus a slightly lower water table than the MFWHT results. The MFS99 model results from the Colorado case (figure 2a) show minimal surface flow (an order of magnitude less than the precipitation events) that fluctuates seasonally between down slope and up slope directions while the MFWHT case does not exhibit any runoff during the simulation. Again this slight difference in surface flow results in a slightly lower water table position for the MFS99 results (figure 2b). The minimal surface flow influence on the system and strong similarity in water table behavior for this case confirms the MFWHT model's subsurface performance against the more established MFS99 subsurface model.

Overall, the MFWHT results confirm the model's two dimensional subsurface performance by the close similarity in water table behavior with the MFS99 results for both cases (figures 1b and 2b). The assumption that the surface runoff method used by Johnson et al. (2003) would have minimal impact on subsurface behavior appears valid, as the discrepancy in water table positions for both cases is minimal and is far less than the smallest grid cell thickness of 0.8 m. However, the surface runoff behavior of this method does not reflect the daily pulses (or peaks) of runoff inherent to natural systems, especially for systems with shallow water tables and high annual precipitation. Therefore, although the MFS99 results were appropriate in the context of that particular study, the MFWHT results are more representative of a watershed system.


Johnson, R.L., R.B. Thoms, J.S. Zogorski. "Effects of Daily Precipitation and Evapotranspiration Patterns on Flow and VOC Transport to Groundwater Along a Watershed Flowpath", Environmental Science and Technology, (In Press).

HydroGeoLogic, 1999, MODFLOW-SURFACT Software (Version 2.1), HydroGeoLogic, Inc. Herndon, VA, http://www.hgl.com.

Chen, C, 1976. Flow resistance in broad shallow grassed channels. Journal of the Hydraulics Division. Proceedings of the American Society of Civil Engineers. 102 (HY3): 307-322.


Thoms, R.B. Simulating fully-coupled overland flow and variably saturated subsurface flow using MODFLOW, Masters Thesis, OGI School of Science and Engineering at OHSU, 136 p., 2003.

Return to: Top | MODFLOW-WHaT | MF-WHaT Documentation |