Environmental Engineering Reference
In-Depth Information
plant specific parameters that are not available from the
database, the Crop Parameter Intelligent Database System
(CPIDS, 2011).
the closest match to WEPP hillslope predictions in terms
of scale and are therefore used elsewhere (Brazier
et al
.,
2000) to evaluate directly predictions in an uncertainty
framework.
Ideally, plot data from a wide range of situations would
be available for evaluation, but as this is not the case
in the UK, use of large-scale survey data (Evans, 1988)
has been made in order to provide a context for the
MIRSED results. Evans (1988) reported the results of
a collaboration between MAFF and the Soil Survey of
England and Wales (SSEW), whereby 17 areas of 'known
erosion risk' were aerially photographed at a scale of
1:10 000 (1:15 000 for the Bedfordshire/Cambridgeshire
site) over three years beginning in 1982. Survey transects
of 2.3 km width with areas varying from 31-105 km
2
were sampled permitting an assessment of soil-erosion
volumes. Images were used to identify erosion events and
field visits made to measure volumetric soil losses by
measurement of rill and ephemeral gully features. Results
were worked up to provide a variety of statistics including:
median and maximum soil losses for each of the years
1982-1984, including an assessment of areas of erosion
and specific erosion rates under different crop types.
Eleven transects were chosen to evaluate MIRSED results,
those sites which provided observed data from at least two
contrasting survey techniques - the SSEW/MAFF surveys
and
137
Cs surveys - were favoured (but see discussion in
Parsons and Foster, 2011). Approximate locations are
shown in Figure 15.3 and site characteristics are detailed
in Table 15.1. As a working rule, any grid cell that was
encroached upon by the flight path was included in the
analysis, providing in general, slightly larger transects
than the MIRSED predictions.
15.5.4 Climatedata
The WEPP hillslope model requires detailed breakpoint
data for parameters such as rainfall inorder to characterize
the shape of the daily hyetograph and daily input data for
all other climate variables. In this example, time-series
data that were available from the Environmental Change
Network (ECN) database, were used to parameterize
rainfall characteristics for each site. Later steps in the
MIR procedure (Figure 15.2e), incorporate the 'local'
rainfall data from the study sites, generated in this case as
average annual rainfall (AAR) over the period 1961-1990
(Spackman, 1993), illustrated in Figure 15.2d. These data
were originally sourced froma network of over 10 000 rain
gauges collecting on a daily time step, and interpolated
between gauges to provide an AAR figure for each 1 km
2
grid cell (Spackman, 1993).
15.6 Observed data describing
erosion rates
Evaluation data are limited because of the absence of
appropriate observed data, collected in a consistent man-
ner, across the UK. The following section summarizes
the data available to evaluate the MIRSED predictions.
It is recognized that these data are not ideal in either a
spatial or temporal sense to compare model predictions.
However, these are the available data and the different
drawbacks associated with these data are discussed below.
Ideally, further observations will be provided in the future
to test model output more rigorously.
Plot data are valuable for calibration; however, they
are often confined to one crop on one specific soil type
over one slope gradient for each year, as for the Woburn
Experimental Farm plots (Catt
et al
., 1994) or the USDA
experimental plots for WEPP or USLE (Wischmeier and
Smith, 1978; Zhang
et al
., 1996). In reality, for any 1-km
2
grid cell there may be numerous crops on a variety of
different slope gradients providing a range of soil erosion
rates that site specific plot measurements do not capture.
Evans (1995) points to the unrepresentative nature of
plots in relation to the wider landscape as plots are often
sited on sites of known erosion, where they may logically
be assumed to record 'extreme' rates relative to the sur-
rounding hillslopes. However, data at this scale represent
15.7 Mapping predicted erosion rates
The highest average predicted erosion rates were found in
the Dorset, Gwent, Herefordshire and Sussex West sites.
The loamy and silty loam soils of Gwent, Herefordshire
and Sussex were described as being 'at moderate/high risk
of erosion' (Evans, 1990) and would therefore be expected
to rank high relative to other sites. The Dorset transect
included both erodible soils such as sandy loams and less
erodible clays and clay loams. The maximum predicted
erosion rates were highest on these sandy loam soils; thus,
the high average values for Dorset reflect the erodible
nature of these soils. The influence of the more erodible
soils is clearly shown by the area of each transect under
erosion. Though between 5 and 10%of theDorset transect
Search WWH ::
Custom Search