Sie sind auf Seite 1von 9

Catena 145 (2016) 257–265

Contents lists available at ScienceDirect

Catena

journal homepage: www.elsevier.com/locate/catena

A review of the design and operation of runoff and soil loss plots
P.I.A. Kinnell
Institute for Applied Ecology, University of Canberra, Canberra, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Rainfall erosion is a complex process and good understanding of the mechanisms involved is necessary if inap-
Received 12 October 2015 propriate designs and procedures are not unwittingly used in rainfall erosion experiments. Numerous runoff
Received in revised form 6 June 2016 and soil loss plots of various sizes have been installed in many parts of the world. It is essential that on any erod-
Accepted 8 June 2016
ing surface water flows across the surface without interference from any equipment designed to collect runoff
Available online 17 June 2016
and sediment. Examples of designs that do not conform to this requirement are presented. Because sedimenta-
Keywords:
tion occurs in tanks, coarse material needs to be collected and measured separately from fine material which can
Rainfall erosion be subsampled when suspended in the runoff water.
Inappropriate plot designs Despite attempts to operate in areas where soil properties are uniform, replicates are shown to produce consid-
Runoff collectors erable variation in the soil losses produced by the same event. Slope length and gradient influence the type of ero-
Sheet erosion sion that occurs on a plot. Slope lengths less than one metre encourage erosion where detachment and transport
Rill erosion is controlled by the expenditure of the energy generated by raindrop impact but as slope lengths and gradients
increase, detachment by flow may result in the development of rills. Experiments on slopes of one or two metres
in length do not provide data that can be used to parameterise models like the USLE that operate on slopes up to
300 m long.
Rainfall simulators have been widely used in rainfall erosion experiments on plots shorter than 10 m. Given the
fact that raindrop induced bed load transport is stimulated by individual raindrop impacts, spatial variations in
raindrop size and impact frequency in artificial rainfall produced by sprays can lead to erroneous results. Similar-
ly, temporal and spatial variations in flow depths in rain-impacted flows on inclined surfaces in experiments
leads of variations in erosive stress that are seldom taken into account when analysing the results produced
using artificial rainfall. This presents difficulties in respect to apply the results to other situations. The WEPP
interrill erosion model was designed to predict soil movement from interrill areas to rills but it is shown that
the ranking of soils according to their interrill erodibility varies depending on whether eroding surface is flat
plot or on a ridge tillage sideslope.
© 2016 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
2. The USDA-ARS runoff and soil loss plots. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
3. The influence of detachment and transport mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
4. Sheet erosion, rill and interrill erosion on runoff and soil loss plots. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
5. Collection, storage and measurement of runoff and soil loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
6. Experiments using artificial rainfall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265

1. Introduction

Rainfall erosion is a major problem if it causes soil degradation lead-


ing to a decline in agricultural production and if it produces sediment
E-mail address: peter.kinnell@canberra.edu.au. that has serious negative offsite effects on land use, dwellings,

http://dx.doi.org/10.1016/j.catena.2016.06.013
0341-8162/© 2016 Elsevier B.V. All rights reserved.
258 P.I.A. Kinnell / Catena 145 (2016) 257–265

infrastructure and water quality. Research to determine soil erosion where R is the rainfall-runoff “erosivity” factor (climate), K is the soil
losses caused by rainfall has been undertaken in the field under both ar- “erodibility” factor. R is defined as the long term average annual value
tificial and natural rainfall for a considerable period of time. Numerous of the product of storm kinetic energy (E) and the maximum 30-min
laboratory experiments have also been undertaken to gain understand- rainfall intensity (I30). This definition of R results from the observation
ing of the mechanisms involved in rainfall erosion. While laboratory ex- of Wischmeier and Smith (1958) that event soil losses from a plot
periments have been undertaken at the small scale (slope lengths b 1 m with Shelby soil on an 8% slope measured over a 10 year period were di-
to about 5 m), field experiments have been undertaken at both the rectly related to EI30. R does not include direct consideration of runoff
small and larger scales (slope lengths N 20 m). These experiments but provisions were made in the USLE to determine an R equivalent
have provided data that facilitated the development of soil erosion pre- for runoff that occurs from melting snow (Wischmeier and Smith,
diction technology that aid management decisions to conserve soil. 1978). In the second step, the soil loss from the unit plot is multiplied
Over time, a wide range of plot sizes and equipment have been devel- by appropriate values of factors that account for variations in slope
oped in order to collect data on rainfall erosion in the field and the lab- length (L), slope gradient (S), crops and crop management (C) and
oratory. However, many field experiments undertaken by various soil conservation practice (P) to predict soil loss from areas which differ
people have produced results that perhaps unbeknown to them have from the unit plot;
been compromised by the designs that they have used. Consequently,
this review examines some of the designs and procedures that have A ¼ A1 L S C P ð2Þ
been used with a view to ensuring that past mistakes are not propagat-
ed in the future. As can be seen from Eqs. (1) and (2), the primary physical model in
the USLE/RUSLE approach is the unit plot and the factors R and K that ac-
2. The USDA-ARS runoff and soil loss plots count for geographic variation in climate and soil on predicted soil loss
depend on this being so. Consequently, it is essential that the bare fallow
Field erosion research in the USA began in 1917 at the Missouri Ag- condition be included when installing runoff and soil loss plots to deter-
ricultural Experiment Station in Columbia (http://www.ars.usda.gov/ mine local values for some of the factors used in the USLE approach.
Research/docs.htm?docid=18093). Later, 10 experiment stations Given that R is the average annual value of EI30 for events that cause
were set up (Guthrie, OK, Temple, TX, Hays, KS, Tyler, TX, Bethany, soil loss, it follows from Eqs. (1) and (2) that
MO, Statesville, NC, Pullman, WA, Clarinda, IA, LaCrosse, WI, and Zanes-
ville, OH). The experimental studies used a plot design based on the A1e ¼ EI30 K e ð3Þ
work of Miller at the University of Missouri. The most common plots
where A1e is the event soil loss from the unit plot and Ke is the soil erod-
were 6 ft (1.83 m) wide by 72.6 ft (22.1 m) long, which comprised 1%
ibility for that event, and as a result, event soil loss from areas that differ
of an acre (0.00405 ha). The data from these plots and others provided
from the unit plot is given by
the data source for the development of so called Universal Soil Loss
Equation (USLE; Wischmeier and Smith, 1965, 1978). The USDA-Purdue Ae ¼ A1e L S C e P e ð4Þ
rainfall simulator, or “rainulator”, was developed by Meyer and McCune
(1958) as a tool to conduct experiments to supplement the USLE natural where Ce and Pe event values of C and P respectively. In the USLE, short
rainfall database. An extensive 5-year erodibility experiment using the term variations in soil erodibility are not taken into account directly so
rainulator was conducted on 55 Corn Belt soils in the 1960s. This that Ke can be considered to be constant and equal to K. Consequently,
study (Wischmeier and Mannering, 1969) was critically important for for any given plot, the combination of Eqs. (3) and (4) can be written as
development of the soil erodibility nomograph (Wischmeier et al.,
1971) that made USLE relatively easy to apply to any soil in the USA Ae ¼ k EI30 ð5Þ
(http://www.ars.usda.gov/Research/docs.htm?docid=18093). The
Universal Soil Loss Equation approach has been widely used not only where k = K L S Ce Pe. k can be considered constant for an individual bare
in the USA but also throughout the world, often without validation. fallow plot (C = 1) with cultivation up and down the slope (P = 1).
However, runoff and soil loss plots under natural rainfall have been Arguably, the best physical model is given by a duplicate plot
installed in many places outside the USA in order to determine local (Nearing, 1998) but, as shown by Fig. 1, and the results for 40 replicates
values for some of the factors used in the USLE approach (e.g. Australia; obtained by Wendt et al. (1986), the relationship between duplicate
Edwards and Rosewell, 1990). plots is not perfect. Soil properties across field sites are seldom homoge-
The USLE operates mathematically in two steps. In the first step, soil neous and some degree of natural variation can be expected. For exam-
loss is predicted for the unit plot, a bare fallow area, 22.1 m long on a 9% ple, in the case of the bare fallow plots at Presque Isle, plot 1–18 predicts
slope; soil losses that are about 60% of those observed from plot 1–3 and, al-
though plot 1–8 at Presque Isle predicts losses that are the same order
A1 ¼ R K ð1Þ of magnitude as those observed from plot 1–3, there is considerable

Fig. 1. Relationships between event losses from plot 1–8 at Presque Isle and the losses predicted by (A) replicate plot 1–18 and (B) replicate plot 1–3. Note that soil losses predicted by plot
1–3 are about 60% of those predicted by plot 1–18.
P.I.A. Kinnell / Catena 145 (2016) 257–265 259

necessarily valid at other geographic locations. This issue has been


recognised in RUSLE2 (USDA, 2008) where the effective value of K at a
location depends on the long term rainfall and temperature variations
that exist at that location. Although that approach is used in the USA,
it is not necessarily a good substitute for K values obtained from bare fal-
low plots at locations outside the USA.

3. The influence of detachment and transport mechanisms

Soil erosion results from the transport of soil material away from the
site where an erosive force detaches soil particles from the soil surface.
It is well known that the energy required to produce rainfall erosion
comes from two sources, raindrop energy and flow energy, and that
soil erosion can occur as a result of both raindrop impact and surface
water flows acting together or by themselves. Initially, at the beginning
Fig. 2. Relationship between event soil losses predicted by multiplying event soil losses of a rainstorm, surface water flows are usually not present on the soil
from a nearby bare fallow plot by RUSLE period Soil Loss Ratios (fortnightly C factor
surface, and detachment of particles from where they are held by cohe-
values) and event soil losses observed for conventional corn at Clarinda, Iowa plus
0.0001 t ha−1 to enable predicted losses to be displayed when observed losses are zero.
sion and inter-particle friction on bare fallow runoff and soil loss plots of
the scale used to develop the USLE is associated with raindrop impact
and the detached particles are subsequently transport by splash. As
stochastic variation between the observed and predicted losses. In addi- time goes by and runoff develops on the soil surface, less soil material
tion, the relationship Eq. (4) is not necessarily valid for all storms. As in- is transported by splash as more is transported within the rain-impact-
dicated by Fig. 2, differences in runoff producing capacity between bare ed flow (Moss and Green, 1983). In addition to material traveling in
fallow and cropped areas may result in no soil loss when indicated oth- continuous suspension, transport of particles detached by raindrop im-
erwise by Eq. (4). Similarly, as observed by Bagarello et al. (2011), het- pact in rain-impacted flow can, depending on flow conditions, occur by
erogeneity in soil properties over an experimental site may result in raindrop induced saltation, raindrop induced rolling, flow driven salta-
runoff being produced on one replicate and not on another. Despite con- tion, and flow driven rolling (Kinnell, 2005). Further increases in runoff
siderable effort to manage replicates exactly the same, considerable var- can lead to detachment by flow and sediment transport of coarse mate-
iation in event soil losses is not easily avoided. Also, the runoff rial dominated by flow driven saltation and flow driven rolling (Kinnell,
producing capacity of the soil at a site may vary with slope length 2009). Often, bare fallow plots are initially planar with little micro-to-
where as it is assumed not to do in the USLE/RUSLE model (Bagarello pography where detachment is dominated by raindrop impact but
et al., 2011). when flow detachment occurs, rills develop. Rainfall intensities vary in
As shown by Fig. 3A, the assumption that event erosion (Ae) on bare time so that flow conditions on the eroding surface also vary in time,
fallow areas is directly related to EI30 that underlies Eqs. (3) and (5) is and spatial variations in the detachment and transport mechanisms
not valid for all locations. Including direct consideration of runoff by that result (Fig. 4) lead to variations in soil loss that cannot be predicted
multiplying EI30 by the runoff ratio in the event erosivity index well using erosive indices like EI30 and QREI30. Part of the variation in soil
(Kinnell and Risse, 1998) reduces the model error when runoff is loss observed in the relationships between event soil loss and EI30 and
known (Fig. 3B) but a considerably amount of random error remains. QREI30 result from temporal and spatial changes in these detachment
Despite the failure to directly consider runoff causing Eqs (3), (4) and transport mechanisms on the eroding surface during rainstorms.
and (5) to not necessarily be valid for all rainfall events, knowledge of
the two stepped mathematical structure of the USLE/RUSLE means 4. Sheet erosion, rill and interrill erosion on runoff and soil loss plots
that, as noted above, it is important to ensure that the bare fallow con-
dition be part of a set of runoff and soil loss plots experiments because As noted above, initially, at the beginning of a rainstorm, surface
data from plots with the bare fallow condition provide a capacity to de- water flows are usually not present on the soil surface, and detachment
termine geographic variations in R and K that cannot be evaluated well of particles from the soil surface on bare fallow runoff and soil loss plots
when the bare fallow condition is missing from the data set at a given of the scale used to develop the USLE is associated with raindrop impact
location. Although methods to determine long term values K from soil and the detached particles are subsequently transport by splash. As time
properties exist (Wischmeier et al., 1971; Wischmeier and Smith, goes by, runoff develops on the soil surface and this results in fine par-
1978; Auerswald et al., 2014), the values of K so obtained are deter- ticles moving in suspension in the flow while coarse particles move by
mined for the climate in central USA (Römkens, 1985) and so are not raindrop impact induced bed load transport or flow driven bed load

Fig. 3. The relationship between observed event soil loss from plot 1–5 at Morris, MN and soil losses predicted using (A) Eq. (5) and (B) replacing EI30 by the product of the runoff ratio (QR)
and EI30.
260 P.I.A. Kinnell / Catena 145 (2016) 257–265

can be determined by experiments using field plots specifically de-


signed to obtain that data (Elliot et al., 1989).

5. Collection, storage and measurement of runoff and soil loss

As a general rule, bare fallow runoff and soil loss plots with planar
surfaces encourage sheet flow to occur. Because flow depth affects the
ability of raindrop impact to detach and transport soil material in rain-
impacted flows, soil losses are influenced by variations in flow depth
(Kinnell, 1993a). Sediment transport is also dependent on flow velocity
so that a runoff and soil loss plot of any size must have a design that en-
sures that water flows freely over the eroding surface without any ap-
preciable interference from the lateral sides of the plot, and the flow
of water over the downstream boundary and through the runoff collec-
tion system must not have any appreciable influence of the water
depths on the eroding surface. Many designs achieve this by having a
vertical drop at the boundary at the bottom end of the eroding surface
when that boundary is orthogonal to the slope gradient. Fig. 5A shows
an example but although the design worked appropriately during
many storms, it did not in some cases. Replacement by a design that
concentrated flows vertically and produced supercritical flow in the
channel leading to collection tanks (Fig. 5B and C) overcame the prob-
lem. However, there are a number of designs where flows are forced
Fig. 4. 3 examples (A, B, C) of the spatial variation in detachment and transport to concentrate on the eroding area so that they produce results that
mechanisms associated with the erosion of silt and fine to medium sand by rain on a
planar surface which may occur depending on the spatial variation in flow conditions on
are open to question. Fig. 6A shows an example where flow was forced
a surface. RD-ST = raindrop detachment followed by splash transport, RD-RIS/RIR = to concentrate on the eroding surface and sedimentation occurred on
raindrop detachment followed by raindrop induced saltation or rolling, RD-FDS/FDR = the plot just upslope of a flume. The kite shaped plot shown in Fig.
raindrop detachment followed by flow driven saltation or rolling, FD-FDS/FDR = 5.1.1 in Kuhn et al. (2015) provides another example where flows are
detachment by flow followed by flow driven saltation or rolling.
forced to concentrate on the eroding surface.
Fig. 6B shows schematics of other designs that have been used in ex-
transport on part of the plot as indicated in Fig. 4. However, subsequent periments reported in literature (e.g. Vaezi et al., 2008; Zhang et al.,
increases in flow shear velocities may lead to detachment by raindrops 2015; Strohmeier et al., 2015) but produce results that are open to ques-
giving way to detachment by flow leading to rills developing on an area tion. Ensuring that surface water flows freely over the whole of the
previously eroding entirely by sheet erosion on plots with planar sur- eroding surface is essential in all rainfall erosion experiments no matter
faces. Rilling increases the overall rate of detachment not just by the ad- what the scale or whether the experiments are done in the field or in the
dition of flow driven detachment but also by reducing the depths of laboratory. In addition to the inappropriate designs shown in Fig. 6,
flow on areas that are now between the rills – the so called interrill round plots have been used in experiments using rainfall simulators
areas. Surface water in interrill areas absorbs some of the raindrop ener- (Iserloh et al., 2013a, b). In experiments comparing round and rectangu-
gy so that a reduction in flow depth in the interrill areas enables more of lar plots with the same area (0.28 m2), Iserloh et al. (2016) observed
the raindrop energy to be applied in eroding the soil surface. Although that the rectangular plot produced more runoff (33 l cf. 23 l) and soil
the USLE is considered to predict soil loss from sheet and rill erosion, loss (92 g cf. 45 g) than the round plot when subject to artificial rain
the EI30 index is essentially directed at accounting for situations domi- at 40 mm h−1 for 30 min using a small portable spray type rainfall sim-
nated by raindrop detachment and part of the variation observed in ulator. Fig. 7 shows two examples of round plots, one of the type often
the relationships between event soil loss and EI30 results from this used in field conditions, the other, which looks from above like a
fact. The WEPP model (Flanagan and Nearing, 1995) was developed to donut, has been used in laboratory experiments. In the case of the
overcome this by linking specific equations that predict soil movement donut plot, runoff and soil loss are collected in the middle of the plot.
from interrill erosion areas with equations designed to predict detach- Other examples of poor plot design are shown in Fig. 1 of Boix-Fayos
ment and transport with rills. This has led to small plots with slopes of et al. (2006).
b1 m in length such as those described by Elliot et al. (1989) being Some papers simply provide plot dimensions in terms of the area or
used to parameterise the WEPP interrill erosion model. Apart from the width and length of plots without giving any good information on
interrill erodibility data, data on rill erodibility and the critical shear the equipment used to collect the runoff and soil loss data (e.g.
stress to cause detachment by rill flow necessary for running WEPP Lopez-Bermundez et al., 1998; Paningbatan et al., 1995) so that there

Fig. 5. Runoff collection equipment used on 40 m long 2.6 m wide bare fallow plots at Gunnedah, New South Wales, Australia: (6 A) shows the design commonly used by the then Soil
Conservation Service Service of New South Wales and (6B) and (6C) show the design that replaced it.
P.I.A. Kinnell / Catena 145 (2016) 257–265 261

Fig. 6. Examples of designs of runoff collection systems that should not be used on runoff and soil loss plots.

is no good guide to the quality of the data. Also, given the acknowledge first tank, that cannot be done because the fast settling nature of the
effect of slope length in the USLE/RUSLE, just providing data on plot area coarse sediment results in underestimation of the sediment concentra-
is insufficient. tion (Ciesiolka et al., 2006). Consequently, the most accurate method in-
Plot area influences the amount of runoff that needs to be collected, volves collecting the coarse sediment from the water in the first tank to
stored and measured after a rainfall event or series of rainfall events. enable it to dried and weighed separately. The fine material in the water
The runoff collection system should not overflow between the end of from the first tank can then be determined through subsampling the liq-
the plot and tanks designed to collect runoff. A relatively shallow trough uid once the coarse material has been removed. This approach was
put to collect runoff across the width of the plot with a small tube feed- adopted by Kinnell (1983) who ensured that subsamples were taken
ing to a container downslope may overflow during a number of rainfall while stirring the mixture after removal of coarse material. This resulted
events and lead to a loss of data. 1 mm of runoff on a square metre of in variations in pairs of subsamples about their mean being frequently
plot will result in 1 l of water for collection and storage so that the runoff within ± 1% and seldom exceeding ± 3%. Bagarello and Ferro (1998)
collection and storage system must be tailored to the runoff producing and Todisco et al. (2012) report on a technique where a rectangular
capacity of the plots in the climate at the location being studied. Gener- tank with a set of 10 vertically spaced access ports on the side is used
ally, the greatest storm soil loss is produced by high intensity – high run- to reportedly overcome the problem of incomplete mixing during sam-
off events so that it is just as important that the storage system be pling. Samples of the suspension are taken after mixing at different
sufficient to obtain good data for those events as well as others. levels in the downward direction after the contents of the tank have
As plot size increases, the ability to store all runoff and soil loss be- been stirred and the time taken to collect all the samples depends on
comes increasingly difficult. In some cases, a series of connected tanks the water level in the tank but was always b5 min. No separation of
is installed with the overflow from one tank feeding to another being the coarse material from the water in the tank was done before the sus-
subsampled by devices called multislot divisors. Outflow from the pension was sampled and sedimentation occurred during sampling. In
tank flows through a number of slots with only one being used to pass cases where a tap was clogged, the sediment concentration for the
a known portion of the outflow to the next tank (Brakensiek et al., clogged tap was obtained by interpolating/extrapolating the adjacent
1979; Hudson, 1993). Deposition of coarse sediment usually takes concentrations (Todisco et al., 2012). The technique requires calibration
place in the first tank so that the flow into the downstream tank usually of the relationship between the sediment concentration in the set of
contains only fine suspended load. Cascading to additional tanks with samples collected and the sediment concentration in the tank because
the outflows passing through multislot divisors provides the means of it is soil and water depth dependent. For example, Todisco et al. ob-
storing information on runoff and soil loss for large runoff events that served that, for the soil at Masse in Italy, the ratio between the actual
would be impossible to obtain otherwise. The separation of coarse and sediment concentration to the measured sediment concentration varied
fine loads also enables the runoff in the downstream tanks to be from 3.94 when the height of water in the tank was 0.13 m to 2.66 when
mixed and subsampled to obtain sediment concentrations that can be the height was 0.7 m. Bargarello and Ferro observed the ratio between
used to determine the sediment load in the tanks. However, in the the actual sediment concentration to the measured sediment

Fig. 7. Round plots used in erosion experiments cause flows to concentrate unnaturally and lead to questionable results.
262 P.I.A. Kinnell / Catena 145 (2016) 257–265

Fig. 8. (A) An early version of the Coshocton Wheel used by the USDA-ARS. (B) An example of a Coshocton Wheel installed on a runoff and soil loss plot.

concentration to vary from 2.81 for a sandy soil to 1.80 for a clay soil storm applied the next day was applied under what was considered to
when the height of water in the tank was 0.56 m. Given that sedimenta- be a wet condition, and that was then followed by another storm on
tion is taking place during sampling, time of sampling is critical and the the same day on a surface that was considered to be in very wet condi-
calibration may not apply correctly when sediment size composition tion. The different values of erodibility recorded for the different ante-
varies between rainfall events as is more often than not the case. A sam- cedent moisture conditions were then used in an equation that
pling cylinder with a closing valve at the bottom has been proposed by calculated the K for the climate in central USA (Römkens, 1985).
Carallo et al. (2015) as a replacement for the Bagarello and Ferro meth- The rainulator used downward facing nozzles which, while produc-
od. A device reported by Ye et al. (2005) is similar to the one proposed ing raindrops having sizes and velocities similar to natural rainfall, pro-
by Carallo et al. (2015) but has frame with a thin fixed base plate that duce sprays with very high rainfall intensities so that they have to be
slides under sediment deposited in the tank. The frame is put in place applied intermittently in order to achieve time averaged intensities
in the tank before the tube is inserted in the frame to sample both the that are close to those in natural rainfall. The rainulator moved nozzles
liquid and the deposited sediment (Wang et al., 2016). horizontally above the plot and was a large and cumbersome piece of
Apart from multislot devisors, some other devices have been devel- equipment as shown in Fig. 1 of Meyer (1988). Ridge tillage is common-
oped to reduce the quantities of runoff and soil loss that have to be ly used in agriculture in the USA and provides separation between rain-
stored during one or more erosion events. With the Coshocton Wheel drop driven erosion on the sideslopes from flow driven erosion in the
sampler (Carter and Parsons, 1967; Brakensiek et al., 1979) flow from preformed channels between them. This practice led to the develop-
the runoff collector rotates a slot through the outflow collect know ali- ment of a smaller rainfall simulator that could be used to study erosion
quots with the flow controlling the speed of rotation by falling on to on the sideslopes (Meyer and Harmon, 1979). In this simulator, the
vanes (Fig. 8A). Runoff measuring devices like H-flumes can cause depo- spray moves intermittently from side to side across the plot in a manner
sition to occur upslope of Coshocton Wheels and care needs to be taken that enables rainfall intensity to be varied. Using this simulator, Meyer
to ensure that such devices do not cause ponding and deposition on the and Harmon observed that sediment discharged from the sideslopes
eroding surface such as shown in Fig. 7A. A H-flume – Coshocton Wheel was directly related the square of rainfall intensity. This then led to
system (see Fig. 2 in Mutchler et al., 1988) was used in the evaluation of the concept later used WEPP (Nearing et al., 1989) that interrill erosion
the product of runoff rate and rain kinetic energy flux as an alternative was directly related to a soil factor multiplied by rainfall intensity
to the EI30 index at Holly Springs, MS in the USA (Kinnell et al., 1994; squared. However, based on the concept that sediment discharge is
Kinnell, 1995). Coarse material deposited upslope of the flume was the product of runoff and sediment concentration, Kinnell (1993b) pro-
measured separately from the material collected by the Coshocton posed that the intensity squared term should be replaced by the product
Wheel. of the runoff rate and rainfall intensity. This proposal which suggests
Another approach to measuring runoff and sampling sediment load that sediment concentration is directly related to rainfall intensity was
centers about the use of the tipping bucket method to measure runoff. later adopted in WEPP (Flanagan and Nearing, 1995). Consequently, it
Tipping buckets were initially developed as a meteorological device is essential that runoff rates are measured during experiments using
for measuring rainfall but have been expanded in size to measure runoff rainfall simulators to determine interrill erosion in respect to
(Pillsbury et al., 1962; Edwards et al., 1974; Khan and Ong, 1997). Sed- parameterising WEPP.
iment sampling is achieved in some cases by using slots to collect runoff The parametrization of WEPP for the soil effects on interrill and rill
and sediment as the bucket empties (Deasy et al., 2009; Silgram et al., erosion was undertaken at some 30 locations during 1987 and 1988
2010). Some designs are better suited to situations where only fine ma- (Elliot et al., 1989) using the “rotating-boom” simulator (Swanson,
terial is discharged with runoff. 1965). This simulator uses TeeJet 2HH-SS50WSQ nozzles on four
7.6 m long arms that rotate around the central position. Ten 750 mm
6. Experiments using artificial rainfall long by 500 mm wide plots for determining infiltration and interrill ero-
sion together with 6 rills erosion plots 9 m long were all located within
Experiments using runoff and soil loss plots under natural rainfall the wetted area. 6 of the 750 mm long, 500 mm wide interrill erosion
such as those used to develop the Universal Soil Loss Equation are still plots were ridged with sides sloping at approximated 50% towards a
undertaken from time to time in various parts of the world but, as central collecting trough. 4 were flat. The rotating-boom simulator has
noted earlier, the USDA-Purdue rainfall simulator, or “rainulator”, was been used above two 10 m long by 3 m wide plots in experiments to pa-
developed by Meyer and McCune (1958) as a tool to conduct experi- rameterize WEPP for in rangelands (Simanton et al., 1991). However,
ments to supplement the USLE natural rainfall database. Wischmeier some of the concepts behind the development of WEPP for predicting
and Mannering (1969) used the rainulator on 55 soils to examine the erosion in croplands do not apply to rangelands and Wei et al. (2009)
relationship of soil properties to erodibility. Each test consisted of 3 used the data from these and other rainfall simulator experiments un-
storms. The first was for 60 min on an initially dry soil. The second dertaken in the rangelands in 1990, 1991, and 1992 (Franks et al.,
P.I.A. Kinnell / Catena 145 (2016) 257–265 263

1998) to produce a splash and sheet erosion model to replace the at the top of a slope where flows are very shallow is different from that
interrill erosion model normally used in WEPP. Fundamentally, this at the bottom. That difference will be increased by increasing slope
equation reduced the effect of the runoff rate on raindrop driven erosion length and reducing slope gradient. A soil that has a higher infiltration
from being linear (ie to a power of 1.0) to a power of 0.59 and is consid- rate than another will be subjected to a different erosive stress on a pla-
ered more appropriate than the original WEPP because the slope nar slope when both eroding surfaces have the same slope length and
lengths and gradients in rangelands differ from the cropland situation. gradient. Increasing rainfall intensity increases the frequency of drop
The Rangelands Hydrology and Erosion Model (RHEM; Nearing et al., impact and so increases the rate of detachment and transport, but the
2011) uses this power function and reportedly incorporates the up-to- change in flow depth also influences detachment at various points
date scientific understanding of hydrology and erosion processes into along the surface. Most field experiments using artificial rainfall do
modelling erosion in rangelands. not take these issues into account when designing the experiment or
There are many different rainfall simulator designs using a wide va- analysing the results. Generally, each experiment is conducted in an en-
riety of nozzles reported in the literature (Meyer, 1988; Iserloh et al., vironment which makes it difficult to extend the results to conditions
2013a). The rainfall simulators that have been the development of that differ appreciably from those that operated during the
WEPP and RHEM have used TeeJet 2HH-SS50WSQ nozzles. In WEPP experiments.
interrill erosion model given in Flanagan and Nearing (1995), there is Another issue that needs to be considered in designing and
a term (Fnozzle) to take account of the use of other nozzles. That provi- analysing experiments on erosion by rain-impacted flow is that parti-
sion is not documented well in many other places and consequently, cles of different size travel at different rates. Very small particles that
the data on interrill erodibilities published without taking account of travel in continuous suspension travel at about the some velocity as
differences between the nozzles used and TeeJet 2HH-SS50WSQ noz- the flow. However, aggregates and particles of silt, sand and gravel
zles (Bajracharya et al., 1992; Romero et al., 2007; Mahmoodabadi and that travel by raindrop induced saltation and rolling travel at rates
Cerdà, 2013). Apart from Iserloh et al. (2013b) there appears to be little that depend on their size, the velocity of the flow and the frequency
or no work reported on the effect of different rainfall simulators on sed- that they are stimulated to roll or saltate. As a result, particles of differ-
iment discharged from interrill areas. ent size detached at the top of an eroding area at any particular time ar-
Apart from differences in drop size and drop velocity characteristics rive at the downstream boundary at different times and the steady state
produced by different nozzles, spatial variations in drop size, drop ve- will not be achieved until the slowest moving particles detached at the
locity and impact frequency can influence sediment discharges associat- top of the area cross the bottom boundary. An initial flush of fast moving
ed with rain-impacted flows. Once detached by raindrop impact, as material will be counter balanced later by slow moving material
noted above, some of the soil material may be transported by raindrop (Kinnell, 2012). Premature termination of an experiment will produce
induced saltation or rolling. These raindrop induced bed load transport a result that indicates that the sediment discharged is rich in faster mov-
processes rely on successive impacts to move detached soil material ing material compared with the original soil.
from the point of detachment down along the line of flow and over Field experiments are undertaken for various reasons but are usually
the boundary of the eroding area. Consequently, deviations from uni- underpinned by a conceptual model of some sort. The aim of the exper-
form erosive raindrop input, particularly in the area close to the down- iments may be to test one or more hypotheses, test the relevance of a
slope boundary, can have a considerable influence on sediment particular process, or generate input data from a quantitative model
discharge. In tests of the Wageningenan Rainfall Simulator designed to (Kuhn et al., 2015). Rainfall simulator experiments are often used to de-
rain on a 6 m long by 2.5 m wide plot under two Lechler nozzles (nr termine values for interrill erodibilities for use with WEPP. As noted
460–788 for low intensities, or nr 461–008 for high intensities), Lassu above, the WEPP interrill erodibility experiments undertaken by Elliot
et al. (2015) observed Christiansen coefficients of uniformity et al. (1989) used ridged plots 750 mm long, 500 mm wide with sides
(Christiansen, 1942) values of about 70% for the nr 460–788 nozzle sloping at approximated 50% towards a central collecting trough.
pair with flow rates of 0.075 and 0.092 l s−1, and the nr 461–008 pair The750 mm long, 500 mm wide infiltration plots were flat and although
with a flow rate of 0.167 l s−1. The uniformity increased to about 83% not intended for determining interrill erodibilities, runoff and sediment
for the nr 462–008 nozzles when the flow rate was 0.2 l s−1. However, were collected in the experiments at the same time as the ridged plots.
the range of intensities within the 6 m by 2.5 m wide area was 7 to 77, Consequently, erodibilities could be determined for both the ridged
12 to 105, 3 to 156, and 23 to 145 mm hr.−1 respectively for the 4 flow (Kridged) and flat (Kflat) plots using
rates used with the higher intensities being observed in zones close to
the nozzles. The spatial distribution of rainfall intensity as determined K ridged ¼ Dridge Q w −1 I−1 ð1:05–0:85 expð−4 sinðθÞÞ−1 ð6Þ
using the Thies Clima Laser Precipitation Monitor was appreciably dif-
ferent from the intensity distribution measured using rain gauges. Con-
and
sequently, the Laser Precipitation Monitor is not particularly suited to
determining the kinetic energy and intensity characteristics of artificial
rainfall produced by some types of rainfall simulator. However, many K flat ¼ Dflat Q w −1 I−1 ð1:05–0:85 expð−4 sinðθÞÞ−1 ð7Þ
rainfall simulators may have similar characteristic to those reported
by Lassu et al. and produce results from experiments on erosion by where Dridge and Dflat are the rates sediment is discharged from the
rain-impacted flow that may be open to question. ridged and flat plots respectively, Qw is the runoff rate, I is rainfall inten-
Although maintaining spatially uniform rainfall characteristics of ar- sity and θ is the slope angle. Kinnell (1993b) observed appreciable dif-
tificial rainfall is important, the erosive stress applied to the soil surface ferences between the values of Kridged and Kflat for a given soil.
in rain-impacted flows varies with flow depth. Laboratory studies show Consequently, as shown in Table 1, the ranking of the soils in terms of
that the effect of flow depth on the sediment discharge produced by a their erodibilities differed depending on which type of plot was used.
raindrop in rain-impacted flow when flow velocity is held constant ini- Many of the experiments focussing on determining WEPP interrill erod-
tially increases when flow is very shallow, peaks at some drop-size de- ibilities use flat plots of various sizes (Bajracharya et al., 1992;
pendent depth, before the declining as flow depth increases further Mahmoodabadi and Cerdà, 2013) but the results obtained using these
(Kinnell, 2005). While it is common for artificial rainfall of constant values in WEPP may be quite different from those obtained had the
rainfall intensity to be applied in field experiments, flow depths and ve- ridge-furrow plot design originally used by Elliot et al. been used. Con-
locities on the eroding surface vary in space and time and consequently, sequently, there is a need to develop standard protocols that should
so too does the erosive stress applied to the soil surface even if the rain- be applied when experiments are undertaken to parameterise a given
fall characteristics remain constant in time and space. The erosive stress model.
264 P.I.A. Kinnell / Catena 145 (2016) 257–265

Table 1
The effect of plot design on the ranking of soils according to interrill erodibility determined by Kinnell (1993b) from the WEPP interrill erodibility experiments undertaken by Elliot et al.
(1989).

Interrill erodibility (10−6 kg s m−4) Ki relative to Heiden

Rank Location Ki.ridge Location Ki.flat Ki.ridge Ki.flat

1 Heiden 1.62 Portneuf 1.79 1.00 0.53


2 Portneuf 2.31 Sverdrup 2.00 1.43 0.59
3 Sharpsberg 2.51 Barnes-ND 2.02 1.55 0.60
4 Barnes-MN 2.58 Pierre 2.72 1.59 0.81
5 Pierre 2.72 Barnes-MN 2.84 1.68 0.84
6 Barnes-ND 3.06 Woodward 3.04 1.89 0.90
7 Los Banos 3.27 Whitney 3.10 2.02 0.92
8 Williams 3.55 Academy 3.32 2.19 0.99
9 Zahl 3.84 Heiden 3.37 2.37 1.00
10 Academy 3.97 Keith 3.82 2.45 1.13
11 Sverdrup 3.97 Hersh 3.89 2.45 1.15
12 Keith 4.26 Los Banos 4.13 2.63 1.23
13 Whitney 4.38 Zahl 4.19 2.70 1.24
14 Nansene 4.93 Sharpsberg 4.53 3.04 1.34
15 Palouse 5.20 Nansene 4.64 3.21 1.38
16 Amarillo 6.19 Williams 5.10 3.82 1.51
17 Hersh 6.80 Amarillo 5.51 4.20 1.64
18 Woodward 7.56 Palouse 6.41 4.67 1.90

As noted above, K values from rainfall simulator experiments on and the revised versions of it (RUSLE, RUSLE2). Because the model is
11 m long plots were used in developing the soil erodibility nomograph. not physically based, similar experiments are often needed to obtain
The approach adopted involved adjusting soil loss values obtained to local data if reliable estimates of soil loss are to be obtained using
those for the unit plot using the USLE equations for L and S. USLE technology outside the USA. Temporal and spatial variations in
the detachment and transport mechanisms that operate on
L ¼ ðλ=22:1Þm ð8Þ plots N 10 m long, particularly those which encourage sheet erosion to
dominate during some rainstorms and rill erosion to occur in others,
where λ is slope length (m) measured along the projected horizontal lead to variations in sediment discharge which are not well predicted
plane and m a factor that varies with slope gradient, by rainfall factors like storm kinetic energy and maximum rainfall inten-
sity. Despite attempts to provide situations where soil characteristics
S ¼ 65:4 sin2 θ þ 4:65 sinθ þ 0:0654 ð9Þ are uniform, replicates may show considerable differences in event
soil loss between them despite being management being the same.
where θ is the slope angle. Other experiments have been conducted on
Rainfall simulators have been widely used in rainfall erosion exper-
shorter slopes. According to Renard et al. (1997), the L factor equation
iments on plots shorter than 10 m in both the field and the laboratory.
commonly applied in the RUSLE is fixed for slope lengths b 4.6 m,
Given the fact that raindrop induced bed load transport is stimulated
L ¼ ð4:6=22:1Þm λb4:6 m ð10Þ by individual raindrop impacts, spatial variations in raindrop size and
impact frequency in artificial rainfall produced by sprays can lead to er-
where m is varies with the ratio of rill to interill erosion, and roneous results. Similarly, temporal and spatial variations in flow
depths in rain-impacted flows on inclined surfaces in experiments
S ¼ 3:0ð sinθÞ0:8 þ 0:56 λb4:6 m ð11Þ leads of variations in erosive stress that are seldom taken into account
when analysing the results produced using artificial rainfall. This pre-
However, given that the relative contributions of raindrop driven de- sents difficulties in respect to apply the results to other situations.
tachment and transport process and flow driven detachment and trans- The manner in which particles are detached and transported in sur-
port processes varies spatially with slope length and gradient, K values face water flow dictates that surfaces water must be allowed to flow
determined on slopes longer that 10 m but b 300 m are almost certainly over the whole of the surface without being influenced by any equip-
more reliable than those obtained when slopes are shorter than 4.6 m. ment installed at the bottom of a plot. A number of designs have been
Experiments on slopes of one or two metres are unlikely to generate K reported in the literature that have not met this requirement and conse-
values that can be reliably applied over 10 m to 300 m long slopes. Con- quently have produced erroneous results. In addition, equipment used
sequently, experiments that focus on parameterising RUSLE should not needs to have the capacity to collect all the runoff and all the soil loss
be undertake on such short slopes. Vaezi et al. (2016) provide an exam- for all erosion events. Once collected, coarse sediment should be sepa-
ple where an 80 cm long plot was used to determine RUSLE K values in rated and measured separately from fine sediment so that subsampling
experiments that should have been directed at modelling interrill ero- of any soil material suspended in the runoff water is directed at measur-
sion using a model like WEPP, not the RUSLE. ing the fine material.
It is well known that rainfall erosion is caused by the expenditure of
7. Conclusion energy resulting from raindrop impact and surface water flow acting
singly or together. Temporal and spatial changes in flow depth and ve-
Numerous laboratory and field experiments on rainfall erosion have locity result in transitions from raindrop impact dominated erosion to
been reported in the literature. The aim of some of these experiments flow dominated erosion. It is important that experiments focusing on
has been to test one or more hypotheses, or test the relevance of a par- parameterising a particular model are performed at the right scale. For
ticular process, or generate input data from a quantitative model. Field example, experiments on slopes one or two metres long do not generate
experiments under natural rainfall such as those setup at the Missouri data than can be used to calculate USLE/RUSLE Ks. Rainfall erosion is a
Agricultural Experiment Station in Columbia in 1917 and then at a num- complex process and good understanding of the mechanisms involved
ber of other locations in the USA proved to be an invaluable source of is necessary if inappropriate designs and procedures are not unwittingly
data which led to the development of the Universal Soil Loss Equation, used in experiments in rainfall erosion experiments.
P.I.A. Kinnell / Catena 145 (2016) 257–265 265

References Lopez-Bermundez, F., Romero-Diaz, A., Martinez-Fernandez, J., Martinez-Fernandez, J.,


1998. Vegetation and soil erosion under semi-arid Mediterranean climate from Mur-
Auerswald, K., Fiener, P., Martin, W., Elhaus, D., 2014. Use and misuse of the K equation in cia (Spain). Geomorphology 24, 51–58.
soil erosion modelling: an alternative equation for determining USLE nomograph soil Mahmoodabadi, M., Cerdà, A., 2013. WEPP calibration for improved predictions of interrill
erodibility values. Catena 119, 220–225. erosion in semi-arid to arid environments. Geoderma 204, 75–83.
Bajracharya, R.M., Lal, R., Elliot, W.J., 1992. Interrill erodibility of some Ohio soils based on Meyer, L.D., 1988. Rainfall simulators for soil erosion research. In: Lal, R. (Ed.), Soil Erosion
field rainfall simulation. Soil Sci. Soc. Am. J. 56, 267–272. Research Methods. Soil Conservation and Water Conservation Society of America, An-
Boix-Fayos, C., Martinez-Mena, M., Arnau-Rosalen, E., CalvO-Cases, A., Castillo, V., keny, Iowa, pp. 75–95.
Albaladjeo, J., 2006. Measuring soil eroion by field plots: understanding the sources Meyer, L.D., Harmon, W.C., 1979. Multiple-intensity rainfall simulator for soil erosion re-
of variation. Earth-Sci. Rev. 78, 267–285. search on row sideslopes. Trans. Am. Soc. Agric. Eng. 22, 100–103.
Bagarello, V., Ferro, V., 1998. Calibrating storage tanks for soil erosion measurement from Meyer, L.D., McCune, D.L., 1958. Rainfall simulator for runoff plots. Agric. Eng. 39,
plots. Earth Surf. Process. Landf. 23, 1151–1170. 644–648.
Bagarello, V., Di Stefano, C., Ferro, V., Kinnell, P.I.A., Pampalone, V., Porto, P., Todisco, F., Moss, A.J., Green, P., 1983. Movement of solids in air and water by raindrop impact. Effects
2011. Predicting soil loss on moderate slopes using an empirical model for sediment of drop-size and water-depth variations. Aust. J. Soil Res. 21, 257–269.
concentration. J. Hydrol. 400, 267–273. Mutchler, C.K., Murphee, C.E., McGregor, K.C., 1988. Laboratory and field plots for soil ero-
Brakensiek, D.L., Osborn, H.B., Rawls, W.J., 1979. Field manual for research in agricultural sion studies. In: Lal, R. (Ed.), Soil Erosion Research Methods. Soil and Water Conser-
hydrology. USDA Agricultural Handbook 224. USDA, Washington DC. vation Society, Ankney, Iowa.
Carallo, F.G., Di Stefano, C., Ferro, V., Pampalone, V., Sanzone, F., 2015. Testing a new sam- Nearing, M.A., 1998. Why soil erosion models over-predict small soil losses and
pler for measuring soil loss. Earth Surf. Process. Landf. http://dx.doi.org/10.1002/esp. underpredict large soil losses. Catena 32, 15–22.
3866 (online). Nearing, M.A., Foster, G.R., Lane, L.J., Finkner, S.C., 1989. Erosion prediction project tech-
Carter, C.E., Parsons, D.A., 1967. Field tests of Coshocton-type wheel runoff sampler. Trans. nology. Trans. Am. Soc. Agric. Eng. 32, 1587–1593.
Am. Soc. Agric. Eng. 10, 133–135. Nearing, M.A., Wei, H., Stone, J.J., Pierson, F.B., Spaeth, K.E., Weltz, M.A., Flanagan, D.C.,
Ciesiolka, C.A.A., Yu, B., Rose, C.W., Ghadiri, H., Lang, D., Rosewell, C.J., 2006. Improvement Hernandez, H., 2011. A rangeland hydrology and erosion model. Trans ASABE 54, 1–8.
in soil loss estimation in USLE type experiments. J. Soil Water Conserv. 61, 223–229. Paningbatan, E.P., Ciesolka, C.A., Coghlan, K.J., Rose, C.W., 1995. Alley cropping for manag-
Christiansen, J.E., 1942. Irrigation by sprinkling. University of California Agricultural Ex- ing soil erosion on hilly lands in the Philippines. Soil Technol. 8, 193–204.
perimental Station vol. 670. Berkley, CA (Bulletin). Pillsbury, A.F., Pelishek, R.E., Osborn, J.F., Szuszkiewicz, T.E., 1962. Effects of vegetation ma-
Deasy, C., Quinton, J.N., Silgram, M., Bailey, A.P., Jackson, B., Stevens, C.J., 2009. Mitigation nipulation on the disposition of precipitation on chaparral-covered watersheds.
options for sediment and phosphorour loss from winter-sown arable crops. J. Geophys. Res. 67, 695–703.
J. Environ. Qual. 38, 2121–2130. Renard, K.G., Foster, G.R., Weesies, G.A., McCool, D.K., Yoder, D.C., 1997. Predicting soil
Edwards, K., Rosewell, C.J., 1990. Evaluation of alternative land management and erosion by water: a guide to conservation planning with the revised universal soil
cropping practices for soil conservation. Soil Use Manag. 6, 120–124. loss equation (RUSLE). U.S. Department of Agriculture Agricultural Handbook. No.
Edwards, I.J., Jackson, W.D., Fleming, P.M., 1974. Tipping bucket gagues for measuring 703. US Department of Agriculture, Washington, DC.
run-off from experimental plots. Agric. Meteorol. 13, 189–201. Romero, C.C., Stroosnijder, L., Baigorria, G.A., 2007. Interrill and rill erodibility in the
Elliot, W.J., Liebenow, A.M., Laflen, J.M., Kohl, K.D., 1989. A compendium of soil erodibility northern Andean Highlands. Catena 70, 105–113.
data from WEPP cropland soil field erodibility experiments 1987 & 88. NSERL Report Römkens, M.J.M., 1985. The soil erodibility factor: a perspective. In: El-Swaify, S.A.,
3. Ohio State Univ. and US Department of Agriculture Agricultural Research Service, Moldenhauer, W.C., Lo, A. (Eds.), Soil Erosion and Conservation. Soil and Water Con-
Washington, DC. servation Society of America, Ankeny, Iowa, pp. 445–461.
Flanagan, D.C., Nearing, M.A., 1995. USDA-water erosion prediction project: hillslope pro- Silgram, M., Jackson, D.R., Bailey, A., Quinton, J., Stevens, C., 2010. Hillslope scale runoff
file and watershed model documentation. NSERL Report 10. National Soil Erosion Re- sediment and nutrient losses associated with tramline wheelings. Earth Surf. Process.
search Laboratory, Purdue, IA. Landf. 35, 699–706.
Franks, C.D., Pierson, F.B., Mendenhall, A.G., Spaeth, K.J.E., Weltz, M.A., 1998. interagency Simanton, J.R., Weltz, M.A., Larsen, H.D., 1991. Rangeland experiments to parameterize
rangeland water erosion project report and state data summaries. NWRC 98-1. the water erosion prediction project model: vegetation canopy cover effects.
USDA-ARS Northwest Watershed Research. Center, Boise, ID. J. Range Manag. 44, 276–282.
Hudson, N.W., 1993. Field measurement of soil erosion and runoff. FAO Soils Bulletin No. Strohmeier, S., Laaha, G., Holzmann, H., Klik, A., 2015. Magnitude and occurance probabil-
68. FAO, Rome. ity of soil loss: a risk analytical approach for the plot scale for two sites in lower Aus-
Iserloh, T., Ries, J.B., Cerdà, A., Echeverría, M.T., Fister, W., Geißler, C., Kuhn, N.J., León, F.J., tria. Land Degrad. Dev. http://dx.doi.org/10.1002/ldr.2354 (early view).
Peters, P., Schindewolf, M., Schmidt, J., Scholten, Seeger, M., 2013a. Comparative mea- Swanson, N.P., 1965. Rotating-boom rainfall simulator. Trans. Am. Soc. Agric. Eng. 8,
surements with seven rainfall simulators on uniform bare fallow land. Zeitschrift für 71–72.
Geomorphologie 57, 11–26 (Supplementary Issues). Todisco, F., Vergni, L., Mannocchi, F., Bomba, C., 2012. Calibration of the soil loss measure-
Iserloh, T., Ries, J.B., Arnáez, J., Boix-Fayos, C., Butzen, V., Cerdà, A., Echeverría, M.T., ment method at the Masse experimental station. Catena 91, 4–9.
Fernández-Gálvez, J., Fister, W., Geißler, C., Gómez, J.A., Gómez-Macpherson, H., USDA, 2008. Draft Science Documentation: Revised Universal Soil Loss Equation Version 2
Kuhn, N.J., Lázaro, R., León, F.J., Martínez-Mena, M., Martínez-Murillo, J.F., Marzen, (RUSLE2) USDA-Agricultural Research Service. Washington, D.C. (http://
M., Mingorance, M.D., Ortigosa, L., Peters, P., Regüés, D., Ruiz-Sinoga, J.D., Scholten, www.ars.usda.gov/Research/docs.htm?docid=6028).
T., Seeger, M., Solé-Benet, A., Wengel, R., Wirtz, S., 2013b. European small portable Vaezi, A.R., Hasanzadah, H., Cerda, A., 2016. Developing an erodibility triangle for soil
rainfall simulators: a comparison of rainfall characteristics. Catena 110, 100–112. lextures in semi-arid regions, NW Iran. Catena 142, 221–232.
Iserloh, T., Bäthke, L., Ries, J.B., 2016. Round versus rectangular: does plot shape matter? Wang, W., Yin, S., Xie, Y., Liu, B., Liu, Y., 2016. Effects of four storm patterns on soil loss
Geophys. Res. Abstr. 18 (EGU2016-9197-1). from five soils under natural rainfall. Catena 141, 56–65.
Khan, A.H., Ong, C.H., 1997. Design and calibration of tipping bucket system for field run- Wendt, R.C., Alberts, E.E., Hjelmfelt Jr., A.T., 1986. Variability of runoff and soil loss from
off and sediment concentration. J. Soil Water Conserv. 52, 437–443. fallow experimental plots. Soil Sci. Soc. Am. J. 50, 730–736.
Kinnell, P.I.A., 1983. The effect of kinetic energy of excess rainfall on soil loss from non- Wei, H., Nearing, M.A., Stone, J.J., Guertin, D.P., Spaeth, K.E., Pierson, F.B., Nichols, M.H.,
vegetated plots. Aust. J. Soil Res. 21, 445–453. Moffet, C.A., 2009. A new splash and sheet erosion equation for rangelands. Soil Sci.
Kinnell, P.I.A., 1993a. Sediment concentrations resulting from flow depth - drop size inter- Soc. Am. J. 73, 1386–1392.
actions in shallow overland flow. Trans. Am. Soc. Agric. Eng. 36, 1099–1103. Wischmeier, W.H., Mannering, J.V., 1969. Relation of soil properties to its erodibility. Soil
Kinnell, P.I.A., 1993b. Interrill erodibilities based on the rainfall intensity - flow discharge Sci. Soc. Am. Proc. 33, 131–137.
erosivity factor. Aust. J. Soil Res. 31, 319–332. Wischmeier, W.H., Johnson, C.B., Cross, B.V., 1971. A soil erodibility nomograph for farm-
Kinnell, P.I.A., McGregor, K.C., Rosewell, C.J., 1994. The IxEA index as an alternative to the land and construction sites. J. Soil Water Conserv. 26, 189–193.
EI30 erosivity index. Trans. Am. Soc. Agric. Eng. 37, 1449–1456. Wischmeier, W.C., Smith, D.D., 1958. Rainfall energy and its relationship to soil loss. Trans.
Kinnell, P.I.A., 1995. The IxEA index: an index with the capacity to give more direct con- Am. Geophys. Union 39, 285–291.
sideration of hydrology in predicting short-term erosion in the USLE modeling envi- Wischmeier, W.C., Smith, D.D., 1965. Predicting rainfall erosion losses from cropland east
ronment. J. Soil Water Conserv. 50, 507–512. of the rocky mountains. Agricultural Handbook No. 282. US Dept Agric., Washington,
Kinnell, P.I.A., 2005. Raindrop impact induced erosion processes and prediction: a review. DC.
Hydrol. Process. 19, 2815–2844. Wischmeier, W.C., Smith, D.D., 1978. Predicting rainfall erosion losses – a guide to conser-
Kinnell, P.I.A., 2009. Impact of slope length on the discharge of sediment by rain impact vation planning. Agricultural Handbook No. 537. US Dept Agric., Washington, DC.
induced saltation and suspension. Earth Surf. Process. Landf. 34, 1393–1407. Vaezi, A.R., Sadeghi, S.H.R., Bahrami, H.A., Mahdian, M.H., 2008. Modeling the USLE-K fac-
Kinnell, P.I.A., 2012. Raindrop-induced saltation and enrichment of sediment discharged tor for calcareous soils in northwestern Iran. Geomorphology 97, 414–423.
form sheet and interill erosion areas. Hydrol. Process. 26, 1449–1456. Ye, Z.H., Lui, B.Y., Lu, B.J., Zeng, X.Q., Fu, S.H., 2005. Design and testing of depth profile sed-
Kinnell, P.I.A., Risse, L.M., 1998. USLE-M: Empirical modelling rainfall erosion through iment sample for runoff plots. J. Sediment. Res. 3, 24–29.
runoff and sediment concentration. Soil Sci. Soc. Am. J. 62, 1667–1672. Zhang, L., Wang, J., Bai, Z., Lv, C., 2015. Effects of vegetation on runoff and soil erosion on
Kuhn, N.J., Greenwood, P., Fister, W., 2015. Use of field experiments in soil erosion re- reclaimed land in an opencast coal-mine dump in a loess area. Catena 128, 44–53.
search. 2014. Geomorphological Fieldwork. 18, pp. 175–200.
Lassu, T., Seeger, M., Peters, P., Keesstra, S.D., 2015. The Wageningen rainfall simulator:
set-up and calibration of an indoor nozzle-type rainfall simulator for soil erosion
studies. Land Degrad. Dev. 26, 604–612.

Das könnte Ihnen auch gefallen