Geoscience Reference
In-Depth Information
units (HRUs) can be obtained (e.g. Figure 2.7). This is a relatively easy task with a modern GIS - at
least, relatively easy once all the different sources of information have been stored and properly spatially
registered in the GIS database (which can be very time consuming). The HRUs defined in this way
may be irregular in shape where overlays of vector data are used, or based on regular elements where a
raster (grid or pixel) database is used. A variation on a grid-based calculation scheme is the hexagonal
discretisation used in the DHMS system of Vinogradov
et al.
(2011). Calculations might be made on
every grid element in space, e.g. LISFLOOD (De Roo
et al.
, 2000; van der Knijff
et al.
, 2008), or similar
HRUs within the catchment discretisation will often be grouped together into a single unit for calculation
purposes, as in the Grouped Response Units of the SLURP model (Kite, 1995) and the Runoff Formation
Complexes of the DHMS model (Vinogradov
et al.
, 2010). It is these groupings that are then used to
predict the distribution of responses within the catchment.
The difficult part of this type of modelling is working out how to represent the hydrological response
of each HRU and its contribution of discharge to the stream network. This varies significantly between
different models. Some models use a conceptual storage model to represent each HRU element (e.g
the SLURP, LISFLOOD and DHMS models; the USGS PRMS system (Leavesley and Stannard, 1995;
Fl ugel, 1995); the HYPE model (Lindstrom
et al.
, 2010); the ARC/EGMO model (Becker and Braun,
1999); the ECOMAG model (Motovilov
et al.
, 1999); the HYDROTEL model (Fortin
et al.
, 2001); the
ARC Hydro model (Maidment, 2002); the PREVAH model (Viviroli
et al.
, 2009); and the WATFLOOD
model (Cranmer
et al.
, 2001; Bingeman
et al.
, 2006), which is included in the Green Kenue modelling
system of Environment Canada).
Other models use a loss function to calculate a rainfall excess which is then routed to the catchment
outlet, in some cases assuming a distribution of storage capacities within each HRU (e.g. Schumann
and Funke, 1996). As the HRU element scale becomes finer, and the hydrological description becomes
more continuum physics-based, this type of model approaches the fully distributed models of Chapter
5; the distinction we draw here, by including HRU models in this chapter, is that they do not explicitly
aim to solve the continuum flow equations for surface and subsurface flow but allow the grouping of
elements to reduce the number of calculations required. Finally, there have been attempts to define HRUs
directly in terms of the dominant hydrological processes in each unit of the landscape (e.g. Uhlenbrook
and Leibundgut, 2002; Scherrer and Naef, 2003).
One widely used example is the USDA Soil and Water Assessment Tool (SWAT) model of Arnold
et al.
(1998) which is based on the USDA Soil Conservation Service (SCS) curve number method for
fast runoff generation (although other options are also provided). A particular advantage of SWAT is
that, because it was developed by the United States Department of Agriculture, it can be freely down-
loaded (this is also the case with the Green Kenue system from Environment Canada, see Appendix
A). It also has many additional components than just predicting runoff, including the transport of sedi-
ments, nutrients, pathogens and pesticides, and the prediction of crop growth based on the soil moisture,
temperature and nutrient predictions. The model is also provided with databases of default parame-
ter values for different types of soils, crops and natural vegetation. Many other organisations have
taken advantage of this to organise workshops and training courses as a way into using the model.
The SWAT model has been so widely used in the last decade that a more complete description is provided
in Box 6.2.
In fact, there are a number of examples of this type of model that use the SCS curve number method
for predicting runoff generation, for example AFFDEF (Brath
et al.
, 2004, 2006). The SCS method has
an interesting history and it will continue to be used because of the way in which databases of the SCS
curve number can be related to distributed soil and vegetation information stored within a GIS. The SCS
method has its origins in empirical analyses of rainfall-runoff data on small catchments and hillslope
plots. It is commonly regarded as a purely empirical method for predicting runoff generation with no
basis in hydrological theory. It is also commonly presented in hydrology texts as an infiltration equation
