GSSHA is a process-based model. Hydrologic processes that can be simulated and the methods used to approximate the processes with the GSSHA model are listed in Table 1. For several processes, there are multiple solution methods. A brief description of the processes and solution methods is presented. To obtain detailed information about the processes and methods, please refer to the GSSHA User’s Manual (Downer and Ogden in preparation).
|Snowfall accumulation and melting||Energy balance|
|Precipitation interception||Two-parameter empirical|
|Overland water retention||Specified depth|
|Overland flow routing||
2-D lateral diffusive wave
|Channel routing||1-D longitudinal, explicit, up-gradient, diffusive wave|
|Lake storage and routing||Level pool routing|
|Wetland peat layer hydraulics||Mixed Darcian and Manning's flow.|
Deardorff (1977) Penman-Monteith with seasonal canopy resistance
|Soil moisture in the Vadose zone||
Bucket, 1-D vertical Richards’ equation (RE)
|Lateral saturated groundwater flow||2-D vertically averaged|
|Stream/groundwater interaction||Darcy’s law|
Table 1. Process and Approximations Techniques in the GSSHA Model. (G&A – Green and Ampt (1911), GAR – Green and Ampt with Redistribution (Ogden and Saghafian 1997), RE – Richards’ equation (Richards 1931), ADE – alternating direction explicit, ADE-PC – alternating direction explicit with prediction-correction (Downer et al. 2002b)
- 1 Precipitation distribution
- 2 Snowfall accumulation and melting
- 3 Precipitation interception
- 4 Infiltration
- 5 Green and Ampt
- 6 Multi-layer Green and Ampt
- 7 Green and Ampt with redistribution
- 8 Richards’ equation
- 9 Overland flow routing
- 10 Channel routing
- 11 Lake and Detention Basin Routing
- 12 Wetland Hydraulics
- 13 Evapotranspiration
- 14 Soil moisture in the vadose zone
- 15 Lateral saturated groundwater flow
- 16 Stream/groundwater interaction
- 17 Exfiltration
- 18 Subsurface Drainage
- 19 Related Topics
In GSSHA, precipitation may be spatially distributed over the watershed by specifying a number of rain gages in the rainfall input file. Precipitation is distributed between the gages using either Thiessen polygons or an inverse distance square weighted method. Precipitation at each gage may vary in time, and non-uniform time increments may be used.
Snowfall accumulation and melting
Precipitation will automatically be treated as snowfall any time long-term simulations are conducted and the dry bulb temperature is below 0° C. Any accumulated snowfall is treated as a one-layer snowpack that melts as a result of heat sources including: non-frozen precipitation, net radiation, heat transferred by sublimation and evaporation, and sensible heat transfer as the result of turbulence.
Interception is the process of vegetation capturing precipitation and preventing it from reaching the land surface. Interception is modeled in GSSHA using an empirical two-parameter model that accounts for an initial volume of water that vegetation can hold plus the fraction of precipitation captured after the initial volume of water has been satisfied. The fate of intercepted water is not accounted for in GSSHA. The rainfall intercepted by vegetation is assumed to evaporate.
Infiltration is the process whereby rainfall and ponded surface water seep into the soil because of gravity and capillary suction. In GSSHA there are two general methods used to simulate infiltration. These are the Green and Ampt (1911) model and the Richards’ equation (1931) models. There are also two extended Green and Ampt models, making a total of four infiltration options to chose from.
Green and Ampt
The use of all the Green and Ampt based methods is limited to conditions where infiltration excess, or Hortonian runoff (Horton 1933), is the dominant stream flow producing mechanism. In the Green and Ampt model of infiltration, water is assumed to enter the soil as a sharp wetting front. Precipitation on initially dry soil is quickly infiltrated because of capillary pressure. As rainfall continues to fall and the ground becomes saturated, the infiltration rate will decrease until it approaches the saturated hydraulic conductivity of the soil.
Multi-layer Green and Ampt
The Green and Ampt model described assumes an infinitely deep, homogeneous, soil column. The GSSHA model also allows the user to specify Green and Ampt infiltration into soils with three defined layers. Changes in the hydraulic properties resulting from layering in the soil column always results in reduced infiltration capacity.
Green and Ampt with redistribution
When conducting long-term simulations, the Green and Ampt infiltration with redistribution (GAR) can be used (Ogden and Saghafian 1997). With GAR, multiple sharp wetting fronts can be simulated, and the water is redistributed in the soil column during non-precipitation periods.
Richards’ equation is currently the most complete method to compute soil water movement including hydrologic fluxes such as infiltration, actual evapotranspiration (AET), and groundwater recharge. The use of Richards’ equation is not limited to Hortonian runoff calculations. Richards’ equation is a partial differential equation (PDE) that is solved using finite difference techniques. In GSSHA three soil layers, each with independent parameters for each soil type and layer, are specified. Because the Richards’ equation is highly nonlinear, finding a solution can be difficult and time-consuming when Richards’ equation is used to simulate the highly transient conditions often found in hydrology, such as sharp wetting fronts and fluctuating water table. The GSSHA model employs powerful, mass conserving methods of solving the Richards’ equation and has been capable of simulating both soil moistures and associated hydrologic fluxes when the proper spatial discretization is employed (Downer 2002).
Overland flow routing
Water on the soil surface that neither infiltrates nor evaporates will pond on the surface. It can also move from one grid cell to the next as overland flow. The overland flow routing formulation is based on a 2-D explicit finite volume solution to the diffusive wave equation. Three different solution methods are available: point explicit, alternating direction explicit (ADE), and ADE with prediction-correction (ADE-PC). Through a step function, a depression depth may be specified. No water is routed as overland flow until the depth of water in the cell exceeds the depression depth. This depression depth represents retention storage resulting from microtopography.
When channel routing is specified, overland flow that reaches a user-defined stream section enters the stream and is routed through a 1-D channel network until it reaches the watershed outlet. Channel routing in GSSHA is simulated using an explicit solution of the diffusive wave equation. This simple method has several internal stability checks that result in a robust solution that can be used for subcritical, supercritical, and transcritical flows.
Lake and Detention Basin Routing
Lakes and detention basins are simulated as a lumped volume that can grow over the spatial domain of the model. As the lake grows or shrinks the connected streams shorten or lengthen as appropriate.
Wetlands are simulated through a conceptual model that includes lateral darcian flow through a peat layer, vertical infiltration through a peat layer, and a mixed darcian and manning's flow through the overlying vegetation.
Evapotranspiration (ET) is the combined effect of evaporation of water ponded on the soil surface and contained in the soil pores, as well as the transpiration of water from plants. GSSHA uses evapotranspiration to track soil moisture conditions for long-term simulations. Evapotranspiration can be modeled using two different techniques, the Deardorff (1977) and Penman-Monteith (Monteith 1965 and 1981). The Deardorff method is a simplified method used for formulations involving only bare soil. The Penman-Monteith method is a more sophisticated method used for vegetated areas.
Soil moisture in the vadose zone
During long-term simulations, the soil moisture in the unsaturated, or vadose, zone can be simulated with one of two methods: a simple fixed soil volume accounting method (Senarath et al. 2000) (bucket method), or simulation of soil moisture movement and hydrologic fluxes using Richards’ equation (Downer 2002). Evaporative demand is supplied to either method by the ET calculations.
Lateral saturated groundwater flow
Where groundwater significantly affects the surface water hydrology, saturated groundwater flow may be simulated with a finite difference representation of the 2-D, lateral, saturated groundwater flow equations. The saturated groundwater finite difference grid maps directly to the overland flow grid. The saturated groundwater zone resides below the unsaturated zone, which may be represented with either the GAR model or the Richards’ equation model. When simulating saturated groundwater flow, the additional processes of stream/channel interaction and exfiltration may occur.
When both saturated groundwater flow and channel routing are being simulated, water flux between the stream and the saturated groundwater can be simulated. By specifying that both overland flow and saturated groundwater flow grid cells containing stream network nodes be considered as river flux cells, water will move between the channel and the groundwater domain based upon Darcy’s law.
Exfiltration is the flux of water from the saturated zone onto the overland flow plane. You may have seen a seep at a change in slope on a hillside. This seepage is exfiltration. Exfiltration occurs when the water table elevation exceeds that of the land surface. Fluxes to the land surface are computed using Darcy’s law.
Subsurface drainage networks can be simulated in GSSHA using the SUPERLINK model. Surface inlets and subsurface tile drains can be simulated. Drains and tiles can discharge to the overland flow plane or to channeles. Multiple connected or unconnected networks can be simulated.