3.1. The soil erosion and sediment transport model
A modified version of the Morgan–Morgan–Finney (MMF) model (Morgan et al., 1984) was developed, as part of the wider project, which incorporated explicit simulation of the effects of vegetation and particle-size selectivity in the detachment, transport, and deposition of sediment. The conceptual representation of the modified MMF model is illustrated in Fig. 2. A detailed description of the modified MMF model, including sensitivity analysis and evaluation is given in Morgan and Duzant (2008).
Only points relevant to the development of the BST within the Buffers DSS will be discussed here. Unlike most existing soil erosion models, the modified MMF model allows measurable plant architecture to be used as model parameters, permitting soil erosion prediction based on site specific measurements.
3.2. Model parameters: Slope gradient, buffer width, soil texture, and vegetation characteristics
Within the modified MMF model, the slope gradient of the contributing field was set to 1, 5, and 10, which represented the midpoint of three defined classes 0–3, 3–7, and 7–12 used in the development of the Buffers Selection Table. The boundary of 3 represents the critical angle at which, for many soils, rill erosion begins (De Ploey, 1984), 7represents a moderately sloping landscape (e.g., Hudson, 1981), and 12 represents a moderately steep slope approaching the upper limits considered suitable for arable farming in many land capability classifications in the UK (Bibby and Mackney, 1969).
Five buffer widths (defined here as the length of buffer parallel to the direction of flow) were used in the model simulations: 2, 4, 6, 10, and 24 m.
These values were based on recommendations for vegetated buffers under Environmental Stewardship schemes (Defra, 2005b).
Soil type was classified according to the texture of the surface horizon.
This method was considered appropriate for this study because it has physical meaning in relation to the carrying capacity of surface runoff, and the relationship between the content of sediment-associated pollutants (e.g., P) and finer soil textures (Owens and Walling, 2002). It has also been documented that P content (total-P and Olsen-P) is greatest at the soil surface for agricultural land (Owens et al., 2008). Six broad soil texture classes were defined (Table 3) as it is unlikely that rapid field assessment can determine soil textures to any greater precision than this. These classes were
DIRECT THROUGHFALL
INTERCEPTION
INFILTRATION RUNOFF
FROM UPSLOPE
SLOPE ANGLE SEDIMENT
FROM UPSLOPE
SOIL
SOIL LOSS FROM SLOPE Soil particle
transport capacity
Deposition of soil particles when sediment load exceeds
transport capacity Soil particle
detachment by raindrop impact and
runoff
Immediate deposition of detached particles
Total detached material available
for transport
(moisture storage capacity, roughness, bulk density EVAPOTRANSPIRATION
PLANT COVER
(canopy cover, ground cover, effective hydrological depth, height, density, stem diameter) LEAF
DRAINAGE
OVERLAND FLOW
THROUGH FL OW
Figure 2 Schematic representation of the revised Morgan-Morgan-Finney model (from Morgan and Duzant, 2008, reproduced with permission of John Wiley and Sons Ltd).
based on those used by Defra (2005a) for assessing erosion risk and were expected to provide, by a rapid field assessment of soil texture, the precision needed for the BST.
From the evaluation of the calibrated MMF model (see Morgan and Duzant, 2008), it was concluded that the composition of the stem diameter and density of plants that make up the buffer vegetation are key design parameters affecting the trapping efficiency of a vegetated buffer. For soil loss the model was found to be highly sensitive to rainfall (average linear sensitivity (ALS): 1.6), the number of rain days (ALS: 1.3), and the diameter (ALS: 1.1) and density of the plant stems (ALS:1.6). Others have also shown the importance of these physical parameters to sediment trapping (e.g., Mudd et al., 2010). The BZIEF field survey records were used to provide an estimate of a “typical” buffer (in the Parrett catchment) in terms of average canopy architecture, for example, stem diameter and density, and a potential range in values to represent the uncertainty in buffer conditions (seeTable 4). Trapping efficiency of the “typical” buffer was also related to soil texture class, slope gradient, and buffer width through field observations (cf.Owenset al., 2007).
One source of variability lies in the fact that a similar percentage ground cover can be achieved by a combination of either very few, large stems or by a higher number of small stems. In order to provide a quantitative descrip- tion of this variability, the field sites were ranked according to stem diameter (Table 4). The two sites that represented the approximated quartile ranges, sites 7 and 5 (Table 4), were used to represent variability in ground cover composed of low density, large stems (14 mm diameter) and high density, small stems (1 mm diameter), respectively. These values were used in the model to provide an upper and lower limit in expected trapping efficiency (respectively) and, as such, to indicate uncertainty in buffer efficiency.
Table 3 Soil characteristics of the heavy, medium, and light soil texture classes used in the DSS
Soil texturea % clay % silt % sand MSb BDc LPd Class
Sandy Loam (SL) 10 25 65 0.28 1.2 50 Light
Silt loam (ZL) 15 66 19 0.35 1.3 14
Sandy clay loam (SCL) 28 14 58 0.38 1.4 88 Medium Silty clay loam (ZCL) 36 55 7 0.42 1.3 33
Silty clay (ZC) 48 45 7 0.30 1.3 22 Heavy
Clay (C) 64 18 18 0.45 1.1 11
a Guide values for % clay, silt and sand are based on mid-point values of the triangular soil texture graph used by USDA.
b MS, Soil moisture at field capacity (% w/w).
c BD, Bulk density of top soil layer (Mg/m3).
d LP, Lateral permeability (m/day) guide values fromMorgan and Duzant (2008).
The field data represented observations in established buffer features and as a consequence of this there was only a small range in ground cover. To increase the range of ground cover beyond that observed in the field, and in particular to consider lower ground cover fractions representative of estab- lishing or damaged buffers, values of stem diameter were varied proportion- ally from the quartile values observed in the field data. Number of stems per square meter ranged between 18.4 and 183.8 (nẳ10; 18.4 interval) for stem diameters of 14 mm and between 800 and 8000 (nẳ10; 800 interval) for stem diameters of 1 mm. This enabled the upper and lower limit of trapping efficiency to be predicted for ground covers ranging from 10% to 100%.
3.3. Model outputs
Paired output tables, constructed from output of the modified MMF model, for the two stem diameters (1 mm high density and 14 mm low density) and for each of the five buffer widths (2, 4, 6, 10, and 24 m) were generated for five ground cover fractions (< 20%, 20–40%, 40–60%, 60–80%, and 80–100%), three slope angles (1, 5, and 10) and for the six soil textural classes defined inTable 3. Rainfall was defined by the mean annual rainfall for the Parrett catchment; 1202 mm yr1. The output of these tables gave total soil (t ha1 yr1) transmitted through the buffer from a contributing field of 1 ha area (100 m100 m). The paired tables provided the upper and lower threshold values of soil loss within which vegetated buffers of a certain percentage cover are expected to function. Effective control was defined as an ability to reduce sediment loss (i.e., the amount of sediment
Table 4 Averaged, recorded field measurements of canopy architecture for 11 project sites in the Parrett catchment, ranked by stem diameter
Site
Average stem diameter (mm)
Number of stems in 1m2
Ground cover (%)
Canopy cover (%)
3 0.9 15,312 90 82
6 1.0 10,128 80 77
7 1.0 6400 82 82
9 1.0 7776 80 86
2 1.2 5728 90 90
13 1.2 1040 85 80
11 2.1 73 74 70
14 2.5 1136 85 82
5 14.0 148 82 78
10 24.0 1120 74 78
4 26.0 712 63 74
Highlighted figures are the quartile ranges.
leaving the field below the buffer) to less than 2 t ha1yr1, which repre- sents a tolerable rate of loss balanced against regeneration of top soil (Morgan, 1986). A full account of these output tables can be found in Woodet al. (2007). In summary, these tables can be developed to consider any threshold soil loss value. They can be used to interpret interrelationships between ground cover fraction, soil texture class, vegetated buffer width, and sediment loss. Further, this can be related to the loss from the field of sediment-associated pollutants (such as particulate P) using known relations between sediment particle size and pollutant content (e.g., Owens and Walling, 2002).
For any given width, the greatest factor affecting transmission is soil texture class followed by ground cover fraction. Figure 3 illustrates the relative effects of soil type, ground cover, and slope gradient on buffer efficacy for two buffer widths (4 and 24 m). Transmission rates are sensitive to ground cover by over a factor of two, making accurate measurements essential in order to instigate effective measures. For all soil textures, slope gradient typically affects buffer transmission by less than5%. This seems to be in line with previous studies that show soil loss to be less sensitive to slope in the presence of a plant or mulch cover (Lal, 1976; Quinn et al., 1980). Therefore, slope related values of soil transmitted (t ha1yr1) were averaged for each ground cover fraction.
3.4. The Buffer Selection Table
Logged values of buffer width (m) compared with sediment transfer (t ha1 yr1), through each ground cover class, were then interpolated through regression to predict sediment transfer at the intercept (INT; 0 m on the x-axis) and gradient of the slope (b). The minimum buffer width (BW) required to prevent >2 t ha1 yr1 sediment transfer (required erosion prevention, REP) was then calculated for each specific soil texture class (soil) and ground cover fraction (gc):
BWsoil;gcẳ10ðREPINTị=b
The lower (1 mm stem, high density) and upper (14 mm stem, low density) limits in buffer width preventing 2 t ha1yr1eroded soil leaving a field are presented in Table 5. To provide better guidance on appropriate design and placement of vegetated buffers a simplified version of Table 5 was produced, the BST (Table 6). The BST (Table 6) represents the greatest chance of controlling sediment and associated pollutant losses based on the upper limit buffer width (Table 5). Further simplification of Table 5 included rounding buffer widths 24 m to the nearest 2 m (to account
for practical field management factors, such as tillage equipment width) and buffer widths>24 m to the nearest 1 m (Table 6).
FromTable 6, it can be seen that heavier soils, that is soils with higher clay content, require a greater width of buffer than lighter soils. This is because finer sediments take longer to settle out of solution (Hjulstro¨m, 1935), and for clay-sized particles would require the water to infiltrate or evaporate before being deposited. While the quantity of sediment eroded from fields with heavier soils is usually less than that observed from lighter
0 2 4 6 8 10
<20% 20–40% 40–60% 60–80% 80–100%
Sediment transported through buffer (t/ha/yr)
% ground cover (buffer) 4 m width
0 2 4 6 8 10
<20% 20–40% 40–60% 60–80% 80–100%
Sediment transported through buffer (t/ha/yr)
% ground cover (buffer) 24 m width
C ZC ZCL
SCL ZL SL
Figure 3 The relationship between soil type, ground cover fraction, and soil loss through a 4 and 24m buffer width. The bars indicate upper and lower soil-loss values according to slope gradient (the upper bar represents a 10slope, and the lower bar is a 1slope).
Soil class Soil texture
Buffer condition (% ground cover)
<20% 20–40% 40–60% 60–80% 80–100%
Lower Upper Lower Upper Lower Upper Lower Upper Lower Upper
Heavy C 54.8 136.1 28.4 62.8 19.5 46.1 14.4 30.0 10.9 22.1
ZC 41.6 98.1 20.6 53.0 13.6 27.6 9.5 18.6 6.7 12.5
Medium ZCL 29.0 64.9 14.2 28.6 8.9 17.1 6.0 10.8 3.9 6.6
SCL 23.6 51.0 11.7 23.0 7.5 13.9 5.1 9.0 3.5 5.7
Light ZL 3.9 6.7 1.7 2.7 0.9 1.2 0.4 0.5 0.19 0.18
SL 1.4 2.2 0.6 0.8 0.3 0.3 0.14 0.14 0.06 0.05
soils (Abu-Zreig, 2001), the relative importance of the associated clay particles with sediment-associated pollutant (e.g., P) transfer is higher than for lighter soils (Barling and Moore, 1994). Longer buffer widths are therefore required to reduce sediment-associated pollution transfer from heavier soils. In lighter soils damage may be caused to the vegetated buffer through inundation by sediment. The width of the buffer in lighter soils must be sufficiently long to allow deposition without breach. The width of the buffer must therefore reflect the worst-case scenario of either sediment loss or associated pollutant transfer.
Having developed an approach to determine optimal buffer width for given landscape attributes (e.g., soil texture, vegetation cover), it is important to now consider protocols for buffer placement, design, and maintenance.