Erosion and Sediment Transport: Finding Simplicity in a Complicated Erosion Model - Environmental Modelling: Finding Simplicity in Complexity - page 248

average hillslope erosion ( x w ) for each 1 km 2 grid cell is
then calculated on the following basis to allow mapping
of the average hillslope erosion in each grid cell:

that hillslope-scale erosion will occur under a variety of
scenarios at different magnitudes, the simulated results
of which are mapped out to the sites in question (see
Zhang et al ., 2002 and Chapter 5 for an example of this
approach). MIRSED also differs from other established
methodologies such as the 'erotop' approach of Kertesz
and Markus (1995), which applies USLE to all hillslopes
that occur within an area, and is consequently less com-
putationally efficient than MIRSED, which may model
widespread hillslope types only once, but use the results
numerous times. Further discussion of this modelling
approach is made in Brazier et al . (2001a, 2001b).

w i x i

x w = w i (15.8)

where:

w i = weight based upon percentage of grid cell

occupied by hillslope;

x i =

soil erosion from each hillslope

by referring back to the distribution of each hillslope type
within each grid cell andmaking a weighted average based
on the area of each grid cell occupied by each hillslope,
in units of t ha − 1 yr − 1 (Figure 15.2h). The cumulative
distribution of hillslope erosion in each grid cell therefore
underlies the simplified, average results that are visualized
in the MIRSED maps of erosion. Model output in the
form of averaged results for each grid cell or distributions
of results from each grid cell can then be evaluated
against observed data. In this example, data from the
SSEW/MAFF overflight study (Evans, 1988) are used for
evaluation.
As the MIRSED matrix represents all possible (mod-
elled) interactions of soil, slope and land use that can
occur within the (predicted) landscape, it is a parsimo-
nious method of applying complex models to extensive
areas. Widespread crop/soil associations such as winter
wheat on a silt-loam soil such as the Andover series, a
common occurrence in the South Downs area of Sussex,
for example (Boardman and Favis-Mortlock, 1993), once
simulated in the model can be recalled numerous times
to produce output. The matrix is then queried for soil
erosion and runoff values according to the characteristics
of the hillslopes that exist in each grid cell in terms of
soil type, crop type and slope gradient, as all the other
data related to soil erodibility, crop cover, planting and
harvesting dates for example are contained within the
matrix of original WEPP runs. Therefore, once model
runs are completed for all combinations of soil, slope
and land use in a region, the complex input requirements
of WEPP can be put aside and the straightforward and
readily available spatial data can be employed to map the
potential hillslope erosion for each grid cell using a GIS.
In this way, the model differs from the scale-sensitive
modelling of projects such as MEDALUS (Kirkby et al .,
1996, 1998; also see discussion in Chapter 5), which take
into account different dominant processes at different
scales. No attempt is made to 'upscale' from the hillslope
scale, to the catchment or region; rather, it is recognized

15.5 Data requirements

15.5.1 Soilsdata

The soils data are held in the National Soils Research
Institute (NSRI) database (NSRI, 2011). These data and
their underlying attributes are stored as dominant soil
per 1 km 2 and all parameters required for WEPP runs are
available either directly from the database or indirectly via
transfer functions derived from regression analysis (see
Flanagan and Nearing, 1995).

15.5.2 Topography

The 10 10m DTM that was queried for hillslope data

at the study sites is generated from the OS Panorama
database, which is freely available to academic users via
EDINA Digimap (EDINA, 2011).

×

15.5.3 Land-use/managementdata

In the UK, land-use data are collected annually by MAFF
through an agricultural census which is then held at the
parish scale of c. 10 km 2 (Miles et al ., 1996; although
parishes vary greatly in size). These data have been inte-
grated with the ITE (Institute of Terrestrial Ecology)
land-cover map of Great Britain (LCMGB) to improve
the spatial resolution of the census land-use data (Miles
et al ., 1995, 1996). Six key types of land cover are iden-
tified within the database: arable, grassland, rough land,
woodland, water and urban. These classes are subdi-
vided, so that arable land cover (for example) is split
into 14 land uses including: winter wheat, winter bar-
ley, spring barley, potatoes and sugar beet. Management
data that are required for parameterization of WEPP
are taken either directly from the GIS database in the
case of planting/harvesting dates for instance, or for