Hillslope runoff and erosion on duplex soils in grazing lands in semi-arid central queensland. III. USLE erodibility (K factors) and cover-soil loss relationships.
Soil erosion by water is of concern in the grazing lands of Queensland due to loss of productivity (Chilcott et al. 2004) and off-site impacts, for example to water quality in the Great Barrier Reef Lagoon (Roth et al. 2003; O'Reagain et al. 2005). The Universal Soil Loss Equation (USLE) of Wischmeier and Smith (1978) or the revised USLE (RUSLE; Renard et al. 1997) and the Australian version (SOILOSS; Rosewell 1993) are widely used in Australia for broad-scale assessments of hillslope erosion. Thc RUSLE was used in the National Land and Water Resources Audit (NLWRA) assessment of hillslope soil erosion for the Australian continent (NLWRA 2001; Lu et al. 2003). The SedNet model, developed to implement the erosion assessment for the NLWRA (Prosser et al. 2001), was then used in finer scale assessments of various catchments (e.g. Murray Darling Basin, Prosser et al. 2003; Mary River, Qld, DeRose et al. 2002; Tully-Murray, Qld, Armour et al. 2007; Queensland Great Barrier Reef catchments, Cogle et al. 2006). USLE-based spatial soil-loss estimates were also used in water-quality modelling at river-basin scale with the EMSS model (e.g. Searle 2005) and E2 model (CondamineBalonne-Maranoa catchments, Waters and Webb 2007). SedNet, EMSS, and E2 modelling is occurring in most coastal and some inland fiver basins in Queensland. A large proportion of these catchments are pastures or rangelands dyslexic.
Measured values of the soil erodibility K factor used in the USLE are rare in Australia (Loch and Rosewell 1992), particularly for non-cultivated situations. The USLE, the method for estimating erodibility from soil properties (Wischmeier et al. 1971 ; Loch et al. 1998), and the USLE covermanagement factor (C)-cover relationship were all originally developed for cultivated croplands and construction sites. Silburn et al. (2011) found soil losses decreased rapidly with increasing cover in Queensland grazing lands, which may not be captured with the default USLE C factor-cover relationship. Confidence in the use of the USLE in Queensland grazing lands could be much improved by testing it with local soil erosion data and comparing it to alternative models.
Studies of runoff and erosion in Queensland grazing lands are briefly reviewed in a companion paper (Silburn et al. 2011). These grazing erosion trials, other than Carroll (2004), did not include bare plots that can be used to define soil erodibility. Most have limited data for low covers, which means that extrapolation to low cover, and the soil erodibility, will be dependent on the choice of cover relationship used. This results in uncertainty in the K estimates. However, Silburn et al. (2011) had plots with a wide range of cover, and the data are suitable for deriving K factors and the C factor-cover relationships. Because soil loss was measured as bedload and suspended loads separately, the data can be used to build models for each component.
In this paper, USLE soil erodibility K factors and cover-soil loss relationships are derived using data from the hillslope runoff plots at the 'Springvale' catchment study, in central Queensland. Runoff, soil erosion, and sediment composition from the hillslope plots, for the period 1987-94, for three main soils and a range of pasture cover were described by Silburn et al. (2011). A simple model was applied to these data to estimate suspended and bedload sediment concentrations separately by Silburn (2011 ).
Data were collected on a cattle-grazing property, Springvale (23.693[degrees]F and 147.455[degrees]F), 75 km west of Emerald, in the Nogoa subcatchment of the Fitzroy basin, central Queensland. The site, measurements, and methods are described in detail by Ciesiolka (1987) and Silburn et al. (2011). For seven water years (Oct. 1987-Aug. 1994), rainfall, runoff, suspended load, bedload, and cover were measured on 12 hillslope plots (Table 1). The plots had a range of pasture cover (10 80%), with and without grazing by cattle, with and without tree canopy cover (tree basal area up to 11 me/ha), on the two main soils derived from sandstone (SS) or mudstone (MS). Areas with low cover (scalds) were caused by overgrazing in the previous decade and did not recover during the study. Two plots were on severely eroded mudstone (MSe) areas (Table 1). Measurement methods were similar to those described by Ciesiolka et al. (1995), using bedload troughs, large tipping buckets to measure runoff, and splitters to sample suspended load. Plots slopes varied from 4 to 8%. Plot areas varied from 14.2 to 640.4m2. Hillslope length of plots ranged from 7.5 to 60 m. The USLE length-slope factor (LS) varied from 0.21 to 1.13. Plots are referred to as Tr (for trough), followed by a letter/number combination (Table 1).
Climate is classified as subtropical with a moderately dry winter (Stem et al. 2003), or the E4 class of Hutchinson et al. (2005). Mean annual rainfall over the 7-year period was 610mm, compared with 648mm for 100 years at Emerald and 678mm at Bogantungan (-25 km west of Springvale) (Ciesiolka 1987). About 75% of rain falls in the six warmer months. Rainfall is highly variable, with a coefficient of variation of annual rainfall of 35 and 47% at Emerald and Bogantungan, respectively, and is characterised by storms with high rainfall intensities and large rain depressions (Silburn et al. 2011).
The site had open or very open woodland (Carnahan 1990) dominated by silver-leaved ironbark (Eucalyptus melanophloia) with a grassy understorey dominated by Bothriochloa ewartiana (desert blue grass), Heteropogon contortus (black spear grass), and Themeda triandra (kangaroo grass), all tussock grasses. There were some areas of denser ironbark trees with up to 400trees/ha (Ciesiolka 1987) (tree basal area 10[m.sup.2]/ha). Tree canopy cover in these thick patches was <30% and is considered to affect hydrology and erosion mainly via soil water use by trees and tree litter input (Owens et al. 2003; S ilburn et al. 2011). The USLE C factor for a 6-m canopy height and 20% canopy cover is 0.97 (Renard et al. 1997); this small effect of tree canopy was ignored, especially as the K factor was mainly influenced by data from plots without trees.
Soils are derived from either mudstone or sandstone (Table 1). The soils on mudstone are moderately deep, texture-contrast (duplex) soils (Sodosols, Isbell 1996), or very shallow to shallow, poorly developed profiles in heavily eroded areas (Leptic Rudosols). Soils on SS plots are Chromosols, Kandosols, Orthic Tenosols, and Rudosols. Surface texture is sandy clay loam on MS soils and sandy loam on SS soils, overlying clay subsoil except on Rudosols. The surface of soils developed on SS and MS is hard-setting. Bulk density of the A horizon averaged 1550 kg/m3 and was not significantly different between soils or cover levels (Silburn et al. 2011). Soils will be discussed grouped by the geology from which they are derived (SS, MS, or MSe), as soil classifications (e.g. Australian Soil Classification) do not provide useful discrimination of the critical soil attributes affecting hydrologic behaviour. Owens et al. (2003) found that all soils fitted on the same relationship between runoffparameters and cover, once differences in profile water capacity were considered in the runoff model.
One important requirement of the study was to determine total and annual average runoffand soil losses, to allow calculation of the soil erodibility K factor. Bedload measurements were rarely missed; however, suspended load data were available only for 1991 to 1994. To infill suspended sediment concentration data, the Rose (1985) model was fitted to the measured data and was used to estimate missing data, as described by Silburn (2011).
The model for suspended sediment concentrations, which considers differences in soil type, slope, and cover between the plots, accounted for >95% of the variance in the measured average annual suspended sediment soil losses (Silburn 2011, figure 3 therein). It was important to use a record of soil losses that was as long as possible in deriving the K factor, so that the resulting estimated soil losses were as close as possible to the long-term average, even though this meant using an estimated 15-20% in-filled soil-loss data.
Annual average USLE
The USLE equation (Wischmeier and Smith 1978) is:
A = R K LS C P (1)
where A is the average annual soil loss (t/[ha.year]), R is the average annual rainfall erosivity factor EI30 ([MJ.mm]/[ha.b. year]), K is the soil erodibility ([t/hal/E130), LS is the length--slope factor, C is the cover-management factor, and P is the supporting practice factor. In deriving K values for Springvale, it is important to use the same methods for deriving the other factors (R, LS, C, P) that are used when the USLE is applied in practice (e.g. SedNet, RUSLE, SOILOSS). Some differences occur, in some factors, between the various applications of the USLE, in particular, in deriving LS and C factors. These differences are described below.
Rainfall erosivity factor El30 was derived from rainfall intensity data from the site and compared with a nearby site (Ciesiolka 1987), long-term estimates for Emerald (Rosenthal and White 1980), and for the site (from SedNet, Brough et al. 2004) based on Yu and Rosewell (1996). LS was derived for each plot, using the measured plot slope and length, by two methods: (i) those in the original USLE (Wischmeier and Smith 1978), and (ii) in RUSLE (Renard et al. 1997), which considers the likely proportion of rill/interrill erosion. For the RUSLE LS factor, interrill erosion was assumed to dominate, and the '[beta]' value was divided by 2 in the equation (m= [beta]/[1+ [beta]]; eqn 4.2 in Renard et al. 1997) was used to calculate the exponent 'm' in the slope length equation (L = [[length(ft)/72.6].sup.m]; eqn 4.1 in Renard et al. 1997), as recommended by Renard et al. (1997). The LS factors calculated using RUSLE were greater than USLE LS factors (Table 1) for plots with low LS (e.g. 0.44 v. 0.32), were similar when LS = 1.0, and were less than USLE LS factors for plots with high LS (e.g. 1.01 v. 1.13). The use of [beta]/2 instead of [beta] made only a small difference (<10%) in RUSLE LS values.
The C factor is expressed in RUSLE and SOILOSS by the multiplication of four subfactors: LU, land use; CC, canopy cover; SC, surface cover; and SR, surface roughness. The LU factor equals 0.45 in RUSLE for consolidated soil with no roots or incorporated plant residues (i.e. an uncultivated bare plot). The SR factor is equal to 1.0 for smooth surfaces (random roughness = 6 mm). The CC factor was ignored at Springvale, because for the typical canopy for plots under trees (canopy cover 15-30%, height 3-5 m), the CC value was [greater than or equal to] 0.95 (i.e. had little effect). Also, plots with low cover largely determine the K factor and these had no tree cover. Therefore, the C factor is equal to the surface cover factor SC multiplied by LU (0.45), for conditions at Springvale. The cover sub-factor SC is calculated (for a smooth surface) as:
SC = exp(bcov.cover%) (2)
where bcov is 0.035 in SOILOSS. In RUSLE, bcov is 0.035 for typical croplands in the USA, 0.025 when interrill erosion is predominant, and 0.05 when rill erosion is predominant (Renard et al. 1997). For US rangelands, Simanton et al. (1984) recommended a bcov value of 0.039.
For undisturbed lands including native pastures and woodlands, SedNet (Lu et al. 2003) and SOILOSS do not use the C subfactors and instead use a relationship between cover and C from Rosewell (1997):
C = exp(a + b.cover% + c.cover% (2) + d.cover% (3)) (3)
where a is -0.7986, b is -0.047384, c is 0.0004488, d is -5.2035E-06 (Rosewell 1997, see errata), and C-0.45 when cover%= 0, which adjusts for consolidation. This equation is shown later to be reasonably similar to Eqn 2 with bcov = 0.035. The supporting practice factor P is equal to 1.0 for the Springvale plots.
To derive the K factor:
K = A/(R LS C P) (4)
where A, R, and LS are measured, C = LU.SC = 0.45.exp(- bcov. cover%) where cover% is measured and P = 1.0. The 'measured' K factor (= A/R LS when CP = 1.0) is for undisturbed soil and is denoted [K.sub.[upsilon]. Because of the use of LU = 0.45 for undisturbed lands in application of RUSLE and SOILOSS, and Eqn 3 in SedNet and SOILOSS, the K factor that must be used when running the models is the measured K for undisturbed soil ([K.sub.upsilon]) divided by 0.45. In this paper, we derive values of [K.sub.upsilon] and K (using both USLE and RUSLE LS factors) and bcov, and evaluate the RUSLE and SOILOSS cover-soil loss relationships (Eqns 2 and 3), for undisturbed (consolidated), hardsetting soils.
The parameters for the USLE were fitted by two methods: (i) fitting an exponential equation to the soil loss--cover data (combining Eqns 2 and 4), and (ii) optimising the parameters K and bcov to minimise the sum of squares of errors (SSE) for measured and modelled average annual soil losses. The two methods of fitting apply different weightings to the data. Optimisation to derive parameter values was performed using the SOLVER function in Microsoft Excel 2003.
Results and discussion
Rainfall erosivity (EI30)
Choice of the EI30 value will influence the K factor derived from the soil-loss data. Data in Fig. 1 indicate that annual EI30 and rainfall data for the study period are consistent with other data from the Nogoa (Ciesiolka 1987: Medway 1974-84 and Springvale 1980-84) and with regional estimates of Rosenthal and White (1980) (based on Emerald pluviometer data) and of Brough et al. (2004). The study period had slightly lower average annual average rainfall and 27% lower EI30 than the long-term at Emerald and 14% lower EI30 than from Brough et al. (2004). The study period did not include any years with >800 mm rainfall and El30 >3200 as observed by Ciesiolka (1987). Long-term average soil loss is thus likely to be somewhat greater than that measured, according to the USLE. The EI30 measured during the study period (2189) was used to determine K factors, rather than a long-term estimate, e.g. 2982 from Rosenthal and White (1980) or 2538 from Brough et al. (2004).
[FIGURE 1 OMITTED]
Annual total and bedload soil losses were poorly correlated with annual EI30, for three plots with low cover (relationships not significant for linear regression). Thus, the USLE will not give good estimates of annual soil losses and should be restricted to use as an average annual equation (Wischmeier 1976).
Effects of slope on soil loss
Silburn et al. (2011) found that bedload soil loss for plots with low cover had a strong linear relationship with slope at Springvale. The relationship between soil loss and cover was stronger once slope was accounted for by adjusting soil losses to LS = 1.0. The slope and shape of the USLE and RUSLE LS factor equations fit the data for low cover well (Fig. 2) (assuming a slope length of 20 m, the average for the three plots with low cover). Soil losses for the low-cover plots increase with plot LS; the USLE LS explains 79% of the variation in soil loss, and the RUSLE LS explains 87% of the variance. Slope also had a slight effect on soil loss for plots with high cover (Fig. 2).
Effects of cover and defining the soil loss for bare soil
The estimate of the K factor is dependent on the soil loss for bare soil (by definition). To derive this, the RUSLE C factor equation was fitted to the soil-loss data (adjusted to LS = 1, e.g. using USLE LS) (Fig. 3a). Soil losses were not significantly different for SS and MS plots (Silburn et al. 2011) and these data were treated together. A different soil loss-cover relationship was hand-fitted for eroded mudstone plots. This resulted in bareplot soil losses of 38 and 45 t/ha, and bcov coefficients of 0.076 and 0.065, respectively, for SS and MS combined and eroded mudstone (Table 2).
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
The bcov values are greater than the RUSLE default for cropland of 0.035. The measured soil loss--cover relationship has greater curvature than the RUSLE default and Eqn 3 from SOILLOSS (Rosewell 1997) (Fig. 3b). Freebairn et al. (1989) and Loch (2000) also found that measured reductions in erosion due to cover were greater than those predicted by the USLE, in cropland and rehabilitated mined land, respectively. The soil loss--cover relationship for eroded mudstone was based on only two plots and is a first approximation. The authors' observation in the Nogoa is that exposed mudstone is more erodible than other soils and is prone to dispersion and rill and gully erosion; the soil loss estimated for bare soil may, if anything, be an underestimate. The RUSLE bcov value for bedload losses (~0.080, Table 2) is slightly higher than for total soil losses because cover is more effective in reducing bedload than total soil loss.
Soil losses for bare soil derived using USLE LS and RUSLE LS factors, and by fitting soil loss--cover relationships as an exponential or by adjusting Ku values to minimise the SSE of soil losses are shown in Table 2. In all cases, fitting soil loss-cover relationships by minimising SSE resulted in lower root mean square error (RMSE), regression slopes closer to 1.0, and total soil loss closer to total measured soil loss than fitting an exponential equation. The use of USLE or RUSLE LS factors gave models of about equal quality, when fitted by minimising SSE. For SS and MS plots, use of the USLE LS gave a higher Ku value than with RUSLE LS factors. However, for eroded mudstone, use of the USLE LS gave a lower Ku value. Fitting soil loss--cover relationships by minimising SSE also resulted in preferred models for bedload losses (Table 2).
Soil erodibility K values
The [K.sub.U] values of SS and MS plots (Table 2), 0.024 (USLE LS) or 0.019 (RUSLE LS), are in the low--medium part of the range of K factors for croplands (Loch et al. 1998, 0.01-0.06). The [K.sub.U] factors for eroded MS were somewhat higher, 0.025 (USLE LS) or 0.028 (RUSLE LS). However, K values for disturbed soil (K.sub.U]/0.45), which are needed in RUSLE, SOILOSS, and SedNet, were in the high part of the range of K factors for cropland. The K factors derived with RUSLE LS are probably more useful, as RUSLE LS is generally used in current soil-loss estimation (e.g. RUSLE, SOILOSS, and SedNet).
The K values for SS and MS plots were greater than those used for the Springvale location in SedNet (Cogle et al. 2006, K = 0.039), by 38%, if USLE LS factors (K = 0.054) were used to derive K. However, when RUSLE LS values (which were used in SedNet) were used to derive K, the measured K (0.042) was only 7% greater. Thus, Cogle et al. (2006) use of the modified monograph equation of Loch et al. (1998) and generalised soil properties to derive K values (Lu et al. 2003; Brough et al. 2004), and the correction for undisturbed soil, appear to give a reasonable estimate of the K value for the main soils at Springvale. However, soil loss from bare, eroded mudstone areas would be underestimated by a factor of 0.63.
Soil loss--cover relationships were similar for plots under trees and in the open, for plots with cover greater than ~40% (Silburn et al. 2011); therefore, C factors were the same for plots under trees and in the open. That is, grass cover (in the open) and tree litter and light grass cover (under trees) had similar effects on runoff and erosion. However, this may not necessarily be the case with low groundcover under trees (Moss and Green 1987). Low cover under trees occurs in the Nogoa catchment due to cattle grazing and tree litter being washed away by run-on water. A plot exposed to grazing cattle (plot G2) gave somewhat higher soil loss and K values of 0.057 and 0.050 (Table 2), for USLE and RUSLE LS factors, or 1.06 and 1.20 times greater than for SS and MS plots. This is due to the cattle disturbing the soil surface and making loose soil available for erosion.
Obtaining more soil erodibility values for grazing lands
There is a great need for reliable erodibility values across the whole of Australia, as they are now the basis for hillslope erosion assessment in many catchments. The need is particularly great in northern Australia, where the contribution of hillslope erosion, as opposed to gully and stream bank erosion, is larger. (This does not mean that gully and stream bank erosion are insignificant, e.g. Bartley et al. 2007.) In this study, the same USLE model parameters applied for SS and MS plots, irrespective of Australian Soil Classification (Chromosol, Kandosol, Rudosol, Sodosol, Tenosol). Only MSe plots (Leptic Rudosol) had different parameters. The MSe plots were also the only plots with a substantially different surface soil. Soil classification is too broad to be a useful indicator of erodibility for the Springvale soils and is probably not the preferred method for estimating and spatially extrapolating erodibility values. The use of surface soil properties, as in the USLE erodibility nomograph (Loch and Rosewell 1992; Loch et al. 1998; Lu et al. 2003), is a better approach to the estimation of erodibility. However, this still requires measured erodibility values that can be used to test and build new predictive equations based on soil properties.
Silburn et al. (2011) point out the lack of soil erosion data for bare plots, and limited data at low cover, from which to derive erodibility values in past grazing erosion studies in northern Australia. Those studies are also generally of short duration (i.e. may not represent the long-term average) and missing data for runoff, suspended loads, or bedload for some events, so that annual values of soil loss cannot always be determined. A suggested approach for obtaining more erodibility estimates for northern Australian grazing lands is to:
(a) Derive sediment concentration--cover models from collections of event data (e.g. Silburn 2011); this will to some extent require extrapolation to low covers and possibly adoption of a default shape of the cover-sediment concentration relationship.
(b) Use soil moisture balance models to estimate long-term sequences of runoff, cover, and soil losses (using the sediment concentration--cover models) in simulated 'grazing trials' with a range of pasture utilisation rates and, thus, cover, in northern Australia. The GRASP model (McKeon et al. 1990) has been widely parameterised for native pastures and has had some parameterisation for runoff (e.g. Scanlan et al. 1996; Owens et al. 2003).
(c) Fit USLE K and C--cover relationships to the long-term simulated soil loss--cover data.
(d) Measure standard soil properties (e.g. morphology, particle size/texture, organic matter, soil structure, and permeability; Wischmeier et al. 1971), percentage of particles <0.125 mm diameter in rainfall-wet soil surfaces (Loch and Rosewell 1992; Loch et al. 1998), and mid-infrared spectroscopy (Janik et al. 2009) for soils at the original erosion study sites.
(e) Test the revised USLE nomograph-type equations (Wischmeier et al. 1971; Loch et al. 1998; Lu et al. 2003) and develop improved estimation equations as required, using the data from the sites characterised above.
There are at least 11 suitable, pasture erosion study sites available (reviewed by Silburn et al. 2011, Silburn 2011), with some 14 soils, plus the data of Carroll et al. (2000) for disturbed soils on coal mines.
Soil erosion parameters were derived for the revised USLE, using data from pasture hillslope erosion plots in the Nogoa catchment, in central Queensland. Methods used to fit parameters affected the results; optimising parameters to obtain the lowest sum of squares of errors for soil losses gives better results than fitting an exponential equation.
The soil erodibility (K factor) was 0.042 for all soils, irrespective of Australian Soil Classification (Chromosol, Kandosol, Rudosol, Sodosol, Tenosol), except for plots on exposed, decomposing mudstone, where K = 0.062. The measured K factor (0.042) is reasonably close to that used in catchment-wide soil-loss estimation for the site (0.039), indicating that the method used for estimating K from soil properties (derived from cultivated cropping soils) gave a reasonable estimate of K for the main soils at the study site. The land-use factor for undisturbed/uncultivated soil (C factor = 0.45 for bare soil) must be used both in deriving K and in applying the USLE model. A 20% increase in K (0.050) for sandstone and mudstone soils may be warranted where heavy grazing by cattle occurs. The C factor-cover relationship was different from the standard RUSLE relationship, requiring a greater exponent ('bcov') of 0.075, rather than the default for cropland of 0.035. Thus, cover was more effective in reducing soil loss than predicted by the RUSLE. However, the default bcov value does give a conservative estimate of the amount of cover needed to minimise soil loss. An approach for deriving USLE parameters from other erosion studies in northern Australia, and for developing a method for estimating erodibility from soil properties, is outlined.
This paper is a contribution to the DERM QSCAPE, Paddock to Reef Monitoring, Modelling and Reporting Program and eWater CRC Catchments and Climate projects. The data used in this paper were the result of efforts by several dedicated DERM start, whom we gratefully acknowledge, in particular Chris Carroll, Ralph deVoil, Peter Burger, and Cyril Ciesiolka. Ross Searle and Dr David Freebairn (RPS) and three referees provided helpful comments on the manuscript. Funding support for the field work included the LAMSAT project funded by Land and Water Resources Research and Development Corporation (now Land and Water Australia) and its predecessor.
Armour J, Hateley L, Pitt G (2007) 'SedNet ANNEX modelling in the Tully Murray catchments.' (Department of Natural Resources and Water: Mareeba, Qld)
Bartley R, Hawdon A, Post DA, Roth CH (2007) A sediment budget in a grazed semi-arid catchment in the Burdekin basin, Australia. Geomorphology 87, 302-321. doi:10.1016/j.geomorph.2006.10.001
Brough D, Lawrence P, Fraser G, Rayner D, Le Grand J (2004) Improved inputs for prediction of regional-scale soil erosion potential for Queensland. In 'Conserving soil and water for society: sharing solutions. ISCO 2004--13th International Soil Conservation Organisation Conference'. Brisbane, July 2004. (Eds SR Raine, AJW Biggs, NW Menzies, DM Freebairn, PE Tolmie) (ASSSI/IECA: Brisbane, Qld)
Carnahan JA (1990) Vegetation. Vol. 6. In 'Atlas of Australian resources. Third series.' (Australian Surveying and Land Information Group, Department of Administrative Services: Canberra, ACT)
Carroll C (2004) Temporal changes to runoff and erosion processes on disturbed landscapes under natural rainfall. PhD Thesis, University of Queensland, Brisbane, Qld, Australia.
Carroll C, Merton L, Burger P (2000) Impact of vegetative cover and slope on runoff, erosion and water quality for field plots on a range of soil and spoil materials on central Queensland coal mines. Australian Journal of Soil Research 38, 313-327. doi:10.1071/SR99052
Chilcott CR, Owens JS, Silburn DM, McKeon GM (2004) How long will soil resources last in semi-arid grazing systems? Paper 249. In 'Conserving soil and water for society: sharing solutions. ISCO 2004. 13th International Soil Conservation Organisation Conference'. Brisbane, July 2004. (Eds SR Raine, AJW Biggs, NW Menzies, DM Freebairn, PE Tolmie) (ASSSI/IECA: Brisbane, Qld)
Ciesiolka CAA (1987) Catchment management in the Nogoa watershed. Research Project Completion Report, AWRC Project 80/128, Australian Water Resources Council, Department of Resources and Energy.
Ciesiolka CAA, Coughlan KJ, Rose CW, Escalante MC, Mohd. Hashim G, Paningbatan Jr EP, Sombatpanit S (1995) Methodology for a multi-country study of soil erosion management. Soil Technology 8, 179-192. doi:10.1016/0933-3630(95)00018-6
Cogle AL, Carroll C, Sherman BS (Eds) (2006) The use of SedNet and ANNEX models to guide GBR catchment sediment and nutrient target setting. Department of Natural Resources, Mines and Water, Brisbane, Qld, Report QNRM06138.
DeRose RC, Prosser IP, Wilkinson LJ, Hughes AO, Young WJ (2002) Regional patterns of erosion and sediment and nutrient transport in the Mary River Catchment, Queensland. CSIRO Land and Water, Canberra, Technical Report 37/02.
Freebairn DM, Silburn DM, Loch RJ (1989) Evaluation of three soil erosion models for clay soils. Australian Journal of Soil Research 27, 199- 211. doi:l0.1071/SR9890199
Hutchinson ME, McIntyre S, Hobbs RJ, Stein JL, Garnett S, Kinloch J (2005) Integrating a global agro-climate classification with bioregional boundaries in Australia. Global Ecology and Biogeography 14, 197-212. doi:10.111l/j.1466-822X.2005.00154.x
Isbell RF (1996) 'The Australian Soil Classification'. Australian Soil and Lands Survey Handbook, Vol. 4. (CSIRO Publishing: Melbourne)
Janik LJ, Forrester ST, Rawson A (2009) The prediction of soil chemical and physical properties from mid-infrared spectroscopy and combined partial least-squares regression and neural networks (PLS-NN) analysis. Chemometrics and Intelligent Laboratory Systems 97(2), 179-188. doi:10.1016/j.chemolab.2009.04.005
Loch RJ (2000) Effects of vegetation cover on runoff and erosion under simulated rain and overland flow on a rehabilitated site on the Meandu Mine, Tarong, Queensland. Australian Journal of Soil Research 38, 299 312. doi:10.1071/SR99030
Loch RJ, Rosewell CJ (1992) Laboratory methods for measurement of soil erodibilities (K Factors) for the Universal Soil Loss Equation. Australian Journal of Soil Research 30, 233-248. doi: 10.1071/SR9920233
Loch RJ, Slater BK, Devoil C (1998) Soil erodibility (Km) values for some Australian soils. Australian Journal of Soil Research 36, 1045-1056. doi:10.1071/S97081
Lu H, Prosser IP, Moran CJ, Gallant JC, Priestley G, Stevenson JG (2003) Predicting sheetwash and rill erosion over the Australian continent. Australian Journal of Soil Research 41, 1037-1062. doi:10.1071/SR02157
McKeon GM, Day KA, Howden SM, Mott JJ, Orr DM, Scattini WJ, Weston EJ (1990) Management of pastoral production in Northern Australian savannas. Journal of Biogeography 17, 355-372. doi:10.2307/2845365
Moss AJ, Green TW (1987) Erosive effects of the large water drops (gravity drops) that fall from plants. Australian Journal of Soil Research 25, 9-20. doi:10.1071/SR9870009
NLWRA (2001) Australian agriculture assessment. National Land and Water Resources Audit, Canberra. Available at: www.nlwra.gov.au/
O'Reagain PJ, Brodie J, Fraser G, Bushell JJ, Holloway CH, Faithful JW, Haynes D (2005) Nutrient loss and water quality under extensive grazing in the upper Burdekin fiver catchment, North Queensland. Marine Pollution Bulletin 51, 37-50. doi:10.1016/j.marpolbul.2004.10.023
Owens JS, Silburn DM, McKeon GM, Carroll C, Willcocks J, deVoil R (2003) Cover-runoff equations to improve simulation of runoff in pasture growth models. Australian Journal of Soil Research 41, 1467-1488. doi:10.1071/SR03047
Prosser IP, Hughes AO, Rustomji P, Young W, Moran CJ (2001) Assessment of fiver sediment budgets for the National Land and Water Resources Audit. CSIRO Land and Water, Canberra, Technical Report 15/01.
Prosser IP, Moran CJ, Lu H, Olley J, DeRose R, Cannon G, Croke B, Hughes A, Jakeman A, Newham L, Scott A, Weisse M (2003) Basin-wide mapping of sediment and nutrient exports in dryland regions of the Murray-Darling Basin. CSIRO Land and Water, Canberra, Technical Report 33/03.
Renard KG, Forster GA, Weesies DK, McCool DK, Yoder DC (1997) 'Predicting soil erosion by water: A guide to conservation planning with the Revised Universal Soil Loss Equation.' Agricultural Handbook 703. (United States Department of Agriculture: Washington, DC)
Rose CW (1985) Developments in soil erosion and deposition models. Advances in Soil Science 2, 1 63.
Rosenthal KM, White BJ (1980) Distribution of a rainfall erosion index in Queensland. Department of Primary Industries, Queensland, Division of Land Utilisation Report 80/8.
Rosewell CJ (1993) 'SOILOSS--a program to assist in the selection of management practices to reduce erosion.' Technical Handbook No. 11, 2nd edn (Soil Conservation Service of New South Wales: Sydney)
Rosewell CJ (1997) 'Potential sources of sediments and nutrients: sheet and rill erosion and phosphorus sources, Australia.' State of the Environment Technical Paper Series (Inland Waters) (Department of the Environment, Sport and Territories: Canberra, ACT)
Roth CH, Prosser IP, Post DA, Gross JE, Webb MJ (2003) Reducing sediment export from the Burdekin catchment. CSIRO Land and Water, Canberra, NAP3.224.
Scanlan JC, Pressland AJ, Myles DJ (1996) Run-off and soil movement on mid-slopes in north-east Queensland grazed woodlands. The Rangeland Journal 18, 33-46. doi:10.1071/RJ9960033
Searle R (2005) Modelling of runoff, sediment and nutrient loads for the Maroochy river catchment using EMSS. Cooperative Research Centre for Catchment Hydrology. (ISBN 1-920813-27-6).
Silburn DM (2011) Hillslope runoff and erosion on duplex soils in grazing lands in semi-arid central Queensland. II. Simple models for suspended and bedload sediment. Soil Research 49, 118 126.
Silburn DM, Carroll C, Ciesiolka CAA, deVoil RC, Burger P (2011) Hillslope runoff and erosion on duplex soils in grazing lands in semiarid central Queensland. I. Influences of cover, slope, and soil. Soil Research 49, 105-117.
Simanton JR, Rawitz E, Shirley ED (1984) The effect of rock fragments on erosion of semiarid rangelands. In 'Erosion and productivity of soils containing rock fragments'. Soil Science Society of America, Special Publication 13. pp. 65-72. (Soil Science Society of America: Madison, WI)
Stern H, de Hoedt G, Ernst J (2003) Objective classification of Australian climates. Commonwealth Bureau of Meteorology, Melbourne, Vic. Available at: www.bom.gov.au/climate/environ/other/koppen_explain.shtml (accessed Oct. 2009).
Waters DK, Webb P (2007) The application of the E2 water quality model for regional NRM planning. In 'MODSIM 2007: International Congress on Modelling and Simulation'. December 2007. (Eds L Oxley, D Kulasiri) pp. 888-894. (Modelling and Simulation Society of Australia and New Zealand)
Wischmeier WH (1976) Use and misuse of the Universal Soil Loss Equation. Journal of Soil and Water Conservation 31(1), 5-9.
Wischmeier WH, Smith DD (1978) 'Predicting rainfall erosion losses a guide to conservation planning.' United States Department of Agriculture, Agriculture Handbook No. 537. (US Government Printing Office: Washington, DC)
Wischmeier WH, Johnson CB, Cross BV (1971) A soil erodibility nomograph for farmland and construction sites. Journal of Soil and Water Conservation 26, 189-193.
Yu B, Rosewell CJ (1996) A robust estimator of the R-factor for the Universal Soil Loss Equation. Transactions of the American Society of Agricultural Engineers 39, 559-561.
Manuscript received 21 April 2009, accepted 13 September 2010
D. M. Silburn (A,B)
(A) Agricultural Production Systems Research Unit, Department of Environment and Resource Management, PO Box 318, Toowoomba, QId 4350, Australia. Email: email@example.com
(B) eWater Cooperative Research Centre, Innovation Centre, Building 22, University of Canberra, Canberra, ACT 2601, Australia.
Table 1. Description of experimental plots (modified from table 1, from Silburn et al. 2011) USLE, Universal Soil Loss Equation (Wischmeier and Smith 1978); LS, length-slope factor; RUSLE, revised USLE (Renard et al. 1997); MS, mudstone; SS, sandstone; MSe, severely eroded mudstone; n.d., not determined Basal area ([m.sup.2]/ha) Plot Cover Plot area Slope (Tr #) (%) Tree Grass ([m.sup.2]) length (m) Open, ungrazed 8 76 3.3 4.0 43.4 18 C 13 5.0 0.1 73.9 35.8 2 9.7 5.0 0.1 14.2 7.4 D 44 5.3 2.0 340.0 36 B 22 5.0 1.0 640.4 58.5 E 40 4.5 n.d. 89.1 41 A 60 2.4 n.d. 30 15.5 Tree plots, ungrazed 6 72 9.9 1.0 37.8 18.8 9 56 9.6 n.d. 58.6 18.9 5 52 10.8 1.5 50.2 24.7 12 52 7.4 2.0 62.1 26.2 Grazed G2 11 1.3 0.5 61.0 16.3 g4 15 >1 0.5 33.4 15 g4* 27 Plot Slope USLE RUSLE Geology Australian (Tr #) (%) LS LS soil classification (B) Open, ungrazed 8 4.0 0.32 0.44 MS Black Sodosol C 5.2 0.60 0.67 MS/SS (C) Brown Sodosol 2 4.0 0.21 0.37 MS Leptic Rudosol D 3.7 0.40 0.48 MSe Leptic Rudosol (D) B 6.7 1.13 1.01 MSe Leptic Rudosol (D) E 6.6 0.91 0.9 SS Rudosol/Tenosol A 17.3 2.26 2.03 SS Rudosol Tree plots, ungrazed 6 6.2 0.55 0.68 SS Kandosol 9 7.0 0.65 0.76 SS Kandosol 5 8.2 0.94 0.96 SS Chromosol 12 5.5 0.56 0.66 SS Chromosol Grazed G2 8.2 0.74 0.84 SS Orthic Tenosol g4 7.6 0.64 0.77 SS Orthic Tenosol g4* Plot Pasture and soil (Tr #) surface conditions (all hard surfaced, all bare areas hardset) Open, ungrazed 8 High grass cover C Mostly bare, scalded 2 Mostly bare, scalded D Medium cover, eroded B Bare areas, severely eroded E Medium cover, stony A Steep, shallow, stony (only bedload data and 4 years of runoff) Tree plots, ungrazed 6 High tree litter cover 9 High tree litter cover (adjacent to Tr6) 5 Medium cover, small gully 12 Medium cover Grazed G2 Grazed heavily, bare areas g4 Grazed heavily, bare areas (adjacent to G2) g4* Cover in last 1 m of g4 (A) Average total cover (standing dry matter and litter) 1987-94. (B) Isbell (1996). (C) Mudstone/sandstone banded. (D) Severely eroded Brown Dermosol. Table 2. USLE soil erodibility K and C factors for total soil loss and for bedload alone, using two methods of fit (all E130=2189, SI units) Plots SS and MS are without Tr g4; exp fit, parameters fitted by fitting an exponential equation to soil loss cover data; min SSE, parameters fitted by adjusting [K.sub.U] to minimise the SSE of soil losses. Value in bold are recommended as best values. n.d., Not determined Plots Method Soil loss RUSLE C of fit for bare factor 'bcov' soil (t/ha) coefficient Using USLE LS factors SS & MS exp fit 38.0 0.076 min SSE 52.9 0.075 MSe exp fit 45.0 0.065 min SSE 54.2 0.075 Plot G2 (SS grazed) n.d. 55.7 0.075 Bedload only SS & MS exp fit 20.2 0.080 min SSE 37.1 0.080 Using RUSLE LS factors SS & MS exp fit 35.2 0.0764 min SSE 41.0 0.075 MSe exp fit 47.8 0.0642 min SSE 60.4 0.075 Plot G2 grazed 48.9 0.075 Bedload only SS & MS exp fit 16.3 0.079 min SSE 36.5 0.079 USLE soil credibility Plots K factor (SI units) Measured Adjusted undisturbed [K.sub.U] ([K.sub.U]/0.45) Using USLE LS factors SS & MS 0.0174 0.0387 0.0242 0.0538 MSe 0.0210 0.0466 0.0250 0.056 Plot G2 (SS grazed) 0.0254 0.057 Bedload only SS & MS 0.0092 0.021 0.017 0.038 Using RUSLE LS factors SS & MS 0.016 0.036 0.0187 0.0416 MSe 0.0218 0.0466 0.028 0.062 Plot G2 0.022 0.050 Bedload only SS & MS 0.0074 0.017 0.0167 0.037 Model v. measured Plots RMSE P/O (A) total Slope of (t/ha) regression (intercept 0) (B) Using USLE LS factors SS & MS 2.17 0.77 0.754 (0.917) 0.94 1.10 1.00 (0.971) MSe 0.50 1.00 0.997 (N = 2) 0.40 0.97 0.997 (N = 2) Plot G2 (SS grazed) n.d. n.d. n.d. (N = 1) Bedload only SS & MS 2.34 0.59 0.527 (0.952) 0.86 1.09 0.968 (0.952) Using RUSLE LS factors SS & MS 1.83. 0.85 0.769 (0.968) 1.09 1.02 0.910 (0.968) MSe 0.52 1.00 1.00 (N = 2) 0.23 0.99 1.013 (N = 2) Plot G2 n.d. n.d. n.d. (N = 1) Bedload only SS & MS 2.12 0.49 0.431 (0.950) 0.88 1.095 0.967 (0.950) (A) Total modelled soil loss divided by total measured soil loss. (B) All intercepts were not significantly different from zero.
|Printer friendly Cite/link Email Feedback|
|Date:||Mar 1, 2011|
|Previous Article:||Hillslope runoff and erosion on duplex soils in grazing lands in semi-arid central Queensland. II. simple models for suspended and bedload sediment.|
|Next Article:||Effect of combined soil water and external load on soil compaction.|