Proposed Standard for Automatic Calculation of Rainfall Erosivity
WATER EROSION: EROSIVITY AND ERODIBILITY (In...
Transcript of WATER EROSION: EROSIVITY AND ERODIBILITY (In...
1
WATER EROSION: EROSIVITY AND ERODIBILITY
(In Encyclopedia of Environmental Management. Taylor & Francis, New York.,
Published online: 29 May 2013: 980-990)
P.I.A. Kinnell, Institute of Applied Ecology, University of Canberra, Australia.
ABSTRACT
Conceptually, rainfall erosivity is the capacity of rain to produce erosion where as soil
erodibility is the susceptibility of the soil to be eroded. However, no absolute numerical
values can be determined for them because rainfall erosion results from various forms of
erosion (splash erosion, sheet erosion, rill erosion, interrill erosion) that are driven by
different forces. Originally, the terms erosivity and erodibility were associated with the
rainfall factor (R) and the soil factor (K) in the Universal Soil Loss Equation where the event
erosivity factor was given by the product of the kinetic energy of the storm (E) and the
maximum 30-minute rainfall intensity (I30). Soil loss at the scale at which the USLE applies
results from the discharge of sediment with runoff but, because the USLE does not include
any direct consideration of runoff in the event erosivity factor, it is not good at accounting for
event soil loss at some geographic locations. Erodibility values have units of soil loss per unit
of the erosivity index and cannot be used with erosivity indices other than the one that was
associated in their determination. Some models ignore this fact and so ignore certain
fundamental rules that should apply to the modelling of water erosion. So called processed
based models attempt to account for the effects of the various forms of erosion and do
consider the effect of runoff on erosivity associated with each form. Particles travel across the
soil surface at virtual velocities that vary from the velocity of the flow to near zero. Little
regard is given to this fact in the determination of soil erodibility values.
Keywords: Soil loss, splash erosion, sheet erosion, rill erosion, interrill erosion, rainfall kinetic energy,
runoff, sediment composition.
Revised 8 Jan 2011
2
WATER EROSION: EROSIVITY AND ERODIBILITY
(In Encyclopedia of Environmental Management. Taylor & Francis, New York.,
Published online: 29 May 2013: 980-990)
INTRODUCTION
Conceptually, rainfall erosivity is the capacity of rain to produce erosion where as soil
erodibility is the susceptibility of the soil to be eroded. Historically, the terms erosivity and
erodibility were originally associated with the R and K factors in the Universal Soil Loss
Equation (USLE),
A = R K L S C P (1)
where A is the long term (eg 20 years) annual average soil loss per unit area from sheet and
rill erosion, R is the rainfall (erosivity) factor defined as the average annual value of the
product of the total storm kinetic energy (E) and the maximum 30-minute intensity (I30, twice
the maximum amount of rain that falls in any 30 minute period during a storm), K is the soil
(erodibility) factor, L is the slope length factor, S is the slope gradient factor, C is the crop
(vegetation) and crop management factor, and P is the conservation support practice factor
(Wischmeier and Smith, 1978). Numerically, soil erodibilty is the mass of soil eroded per
unit of the erosive index. This means that numerical values of K can only be used when R is
as it was originally defined, the average annual value of the product of E and I30.
The USLE was designed to predict sheet and rill erosion from field sized areas and only R
and K have units. The L, S, C and P factors each have values of 1.0 for the so called “unit”
plot, a bare fallow area 22.1 m long on a 9 % slope with cultivation up and down the slope.
Consequently, the soil loss for the “unit” plot (A1) is given by
A1 = R K (2)
and, for any other situation,
A = A1 L S C P (3)
Although traditionally K is calculated by dividing A1 by R, basically K can be perceived as
the regression coefficient in the direct relationship between event soil loss on the unit plot
(A1e) and EI30.
To reduce the need to run long term experiments to determine K values for soils where K is
unknown, Wischmeier (1971) developed a nomograph for determining K from soil properties
for soils with less that 70 % silt in the USA. Alternatively, K values in customary US units
for soils where the nomograph can be used may be obtained using
K = (2.1 X11.14
10-4
(12 – X2 ) + 3.25 (X3 – 2) + 2.5 (X4 – 3)) / 100 (4)
where X1 is % silt multiplied by 100 – % clay, X2 is % organic matter, X3 is the soil structure
code used in the US soil classification, and X4 is the profile permeability code. A number of
other equations have been developed for soils at various geographic locations (eg. El-Swaify
and Dangler , 1976; Young and Mutchler, 1976; Loch et al, 1998; Zhang et al., 2008) but Eq.
4 is frequently used outside the USA without being validated for the soils involved.
3
The two staged mathematical approach shown by Eqs 2 and 3 results from the fact that the
USLE is an empirically based model. It was developed in the 1960s and 1970s from more
than 10,000 plot years of data. Mathematically, the Revised Universal Soil Loss Equation
(RUSLE, Renard et al (1997)) uses Eqs. 2 and 3 in the same way as the USLE but changes
were made to how some of the factors in the model are calculated. Originally, in the USLE,
the events used to calculate R were restricted to those that produced more than 12.5 mm of
rain or at least 6.25 mm of rain in 15 min. That rule was abandoned in the RUSLE when R
values were calculated for the western part of the USA because it was argued that the rule
had no appreciable effect on R values. Yu (1999) noted that changing the threshold to zero
increased the R factor by 5 % in the tropical region of Australia.
VARIANTS OF THE UNIVERSAL SOIL LOSS EQUATION
Eq.2 operates on the assumption that a direct linear relationship exists between event erosion
(Ae) and EI30. Although this assumption is appropriate at some geographic locations, it is not
appropriate at others (Fig. 1). Soil measured as a loss from the plots used to develop the
USLE was discharged in runoff. When runoff occurs, the amount of soil discharged (Qs) in
the runoff can be considered to be the product of the amount of water discharged (Qw) and the
sediment concentration (cs), the amount of soil per unit quantity of water,
Qs = Qw c (5)
It follows from Eq. 5 that, in the USLE, the sediment concentration is directly proportional to
EI30 divided by runoff amount. However, it has been observed (Kinnell, 2010) that the
sediment concentration for an event at some geographic locations is correlated to the product
of the kinetic energy per unit quantity of rain and I30 (Fig. 2). This results in an event
erosivity index (Re) that is given by
Re = QRe EI30 (6)
where QRe is the runoff ratio (runoff amount divided by rainfall amount) for the event. This
index is more effective in accounting for event soil losses at some locations where that EI30
index does not work well (Fig. 3).
A number of other event erosivity indices have been proposed. Williams (1975)
proposed a modification of the USLE to predict event sediment yield (SYe),
SYe = 11.8 (qe qp)0.56
K L S Ce Pe (7)
where qe is the volume of runoff (m3) for the event, qp is the peak flow rate (m
3/s), K, L, and
S are standard USLE factors, and Ce and Pe are event C and P factors. This model is
commonly known as the Modified Universal Soil Loss Equation (MUSLE). In APEX
(Williams et al, 2008),
SYe = Xe K L S Ce Pe [RKOF] (8)
Xe is selected from
4
Xe = EI30 (9a)
Xe = 1.586 (qeqpe)0.56
DA0.12
(9b)
Xe = 0.65 EI30 + 0.45 (qeqpe)0.33
(9c)
Xe = 2.5 (qeqpe)0.5
(9d)
Xe = 0.79 (qeqpe)0.65
DA0.009
(9e)
Xe = b5 qeb6
qpeb7
DAb8
(9f)
where DA is drainage area expressed in ha, b6 – b8 are user selected coefficients (Williams et
al, 2008), and RKOF is the coarse fragment factor as defined by Simanton et al. (1984).
However, Eqs. 7 and 8 with 9b – 9f all use USLE K factor values and not ones that are
associated with the different erosivity indices used and so do not conform with the
mathematical modeling rules upon which the USLE model is based. Consequently, the
MUSLE and APEX models are not valid variations of the USLE. In addition, any model that
uses event erosivity index values that are calculated using runoff from a vegetated area or any
area that is not cultivated up and down the slope will violate the mathematical rules if it uses
USLE C and P factor values.
Mathematical models like the USLE and its derivatives are largely designed to aid
management decisions and operate at a level where spatial and temporal variations in the
various forms of erosion (splash erosion, sheet erosion, rill erosion, interrill erosion) are not
considered in any appreciable detail. However, rill erosion is driven by flow energy while
sheet and interrill erosion are associated more closely with rainfall kinetic energy. In order to
better account for this, Onstad and Foster (1975) used the equation
Re = 0.5EI30 + 0.5α(qe qp)0.333
(10)
Eq. 10 indicates that the sediment concentration for an event (ce) in the data set considered by
Onstad and Foster (1975) is given by
ce = 0.5[EI30/qe] + 0.5[α( qp)0.333
/qe0.666
] (11)
The value of α in Eq. 10 was set so that the long-term average value of Re produced by Eq.
10 was the same as the long-term average value produced when Re = EI30. Given that K is, by
definition, the amount soil loss per unit of R when Re = EI30, this enabled Eq. 10 to be used
with USLE K values to predict soil loss from bare fallow areas. Although Williams et al.
(1984a, b) indicated that Onstad and Foster (1975) was the source of Eq. 9c, there is no
provision in APEX to do the same.
Bagarello et al. (2010) observed that
Re = (QRe EI30)β (12)
with β > 1.0 applied to soil losses from a number of simultaneously operating plots of
different length (λ) established at the experimental station of Sparacia, Sicily. Subsequent
analysis (Bagarello et al. (in review)) established that β > 1.0 appeared to be the result of an
interaction between event runoff (Qe) and slope length (λ) on sediment concentration so that
5
Re = QRe EI30 Qe0.0207λ
(13)
when λ is in metres.
Arguably, from a physical viewpoint, since slope length and gradient influence the erosive
stress, they can be considered to be factors that influence erosivity but the USLE model is not
designed to model the physical processes themselves. It is designed to predict rainfall erosion
based on climate, soil, topography, crops and management factors. Normally, in the USLE
model, the effect of slope length on soil loss is expressed through the L factor but in the case
of Eq. 13, L would remain at 1.0 irrespective of variations in slope length. The positive effect
of runoff on sediment concentration may have been associated with the fact that the data were
collected on a 15 % slope. Rainfall erosion on much lower slope gradients is not associated
with as high an erosion stress from runoff as likely to have been the case at the Sparacia site.
In fact, the RUSLE makes provision for the cushioning effect of water depth on the energy of
raindrop impact on low slopes.
RAINFALL KINETIC ENERGY
Data on storm rainfall kinetic energy is not measured at many geographic locations and often,
kinetic energy values are obtained indirectly. If data on storm rainfall intensities are available
then it is common for storm kinetic energies to be estimated from rainfall kinetic energy –
intensity relationships. These relationships are based on raindrop size data collected during
rainfall events at a geographic location assumed to have rainfall characteristics that are
consistent with those at the geographic location of interest. Various techniques have been
employed to do this and these have generated rainfall intensity – kinetic relationships which
are often used at locations which lie outside the climatic zones where the measurements were
originally made. In the USLE, the relationship recommended between the kinetic energy per
unit quantity of rain (e) and rainfall intensity (I) for the USA is expressed by
e = 916 + 331log10 I , for I < 3in/h (14a)
e = 1074 , for I ≥ 3in/h (14b)
where e is in units of ft-ton/acre/in, and I is in in/h, following analysis of drop size data
collected by Laws and Parsons (1943). In the RUSLE, Eq. 14 is replaced by
e = 1099 (1 - 0.72 exp[- 1.27I]) (15)
whose metric equivalent is
em = 0.29 (1 - 0.72 exp[0.05Im]) (16)
where em has units of MJ/ha/mm and Im has units of mm/h. Eqs. 15 and 16 use the
mathematical form proposed by Kinnell (1981, 1987). Yu (1999) observed that replacing Eq.
13 by Eq. 14 reduced the R factor by about 10 % in the tropical region of Australia. A
number of other mathematical equations have been reviewed by van Dijk et al. (2002). Some
of these produce negative values of rainfall kinetic energy at low rainfall intensities. Eq. 16
produces a response curve that increases non-linearly from low values at low intensities with
em values remaining close to 0.29 MJ/ha/mm for intensities beyond 75 – 100 mm/h.
6
However, short term values measured during rainstorms vary greatly about the values
produced by Eq. 16 or any other rainfall intensity – e relationship. Also, although Eqs. 15 and
16 were developed for use in the USA, they have been used in other parts of the world
without validation. In many cases, that results in erroneous values. In effect, storm kinetic
energies calculated from rainfall intensity – kinetic energy relationships produce numerical
values that are biased towards high intensity rainfall at the expense of low intensity rainfall.
In many geographic locations, there is a lack of rainfall intensity data so that it is not possible
to determine storm kinetic energies using rainfall kinetic energy – intensity relationships.
Event or daily rainfall amounts are more commonly recorded. One approach that has been
used in a number of geographic areas such as Canada (Bullock et al., 1989), Finland
(Rekolainen and Posch, 1993; Posch and Rekolainen,1993), Italy (Bagarello and D’Asaro,
1994), and Australia (Yu and Rosewell, 1996) considers that EI30 is related to a power of
event rainfall amount, (Xp)
EI30 = a1Xpb1
(17)
where a1 and b1 are empirical constants. The value of a1 may show seasonal variation
(Richardson et al., 1983; Posch and Rekolainen,1993). Given that, in the context of the
criteria associated with the USLE/RUSLE model, daily rainfall provides a reasonable proxy
for storm rainfall amount (Hoyes et al., 2005), Eq. 17 using daily rainfall amount provides a
practical approach to extending observed EI30 values to areas where appropriate rainfall
intensity data are lacking.
MORE PROCESS BASED MODELS
Although the USLE/RUSLE is the most widely used method of predicting soil losses from
the land worldwide, rainfall erosion results from various forms of erosion (splash erosion,
sheet erosion, rill erosion, interrill erosion) that are driven by different forces so that there is
no absolute measure of either rainfall erosivity or soil erodibility. Consequently, more
process based models have been developed in order to predict the contributions of the various
forms of erosion more directly. Often, the forms vary in a topographic sequence with splash
erosion dominating erosion at the upper end of a slope, sheet erosion further down before
areas of rill and interrill erosion. Particles detached at the top of the slope may be transported
by a number of different transport mechanisms before being finally discharged from the
eroding area. The Water Erosion Prediction Program in the USA generated the WEPP model,
a more process based model than the USLE/RUSLE designed to model the spatial
contributions of rill and interrill erosion in agricultural landscapes (Flanagan and Nearing,
1995). Flow Shear stress (τ) is used as the erosivity factor in rill erosion model,
Dr = kr (τ - τc) ( 1 – qsr / Tr) , τ > τc (18)
where Dr is rill detachment, kr is the rill erodibility factor, τc is the critical shear stress that
has to be exceeded before detachment occurs, qsr is the sediment load in the flow, and Tcr is
the sediment load at the transport limit. For erosion to occur, particles must be plucked from
within the soil surface where they are held by cohesion and inter-particle friction.
Detachment is the term used to refer to this process. For detachment in rills to occur, the flow
must have shear stress that exceeds τc. Also, for erosion to occur, detached particles must be
7
transported away from the site of detachment. Flows are known to have a limited capacity to
transport soil material and that limit is represented by Tr in Eq. 18. Consequently, the term 1
– qsr / Tr causes Dr = 0 when the sediment load in the rill equals the transport limit. As a
result, Dr may vary along the length of a rill. Interrill erosion contributes to qsr so that rill
erosion may be completely suppressed if the discharge of sediment from the interrill areas is
high enough.
Originally, the erosivity factor in the WEPP interrill model was assumed to be the square of
rainfall intensity (I) so that basically
Di = ki I2 (19)
where Di is interrill detachment and ki is the interrill soil erodibility factor. A series of rainfall
simulation experiments (Elliot et al, 1989) was undertaken to determine ki values for soils in
the USA. Subsequent analysis (Kinnell, 1993) of the data generated by these experiments
established that it was more appropriate to use
Di = kiq qi I (20)
where qi is the runoff rate from the interrill area, and kiq is the interrill erodibility factor. The
ki values determined using Eq. 19 cannot be used in Eq. 20 (Kinnell, 1993).
Di is determined using data on the amounts of soil material discharged from interrill areas.
The use of the term “detachment” in relation to Di is not absolutely correct because the
transport processes involved in moving soil material over interrill areas are not 100 %
efficient. Also, detachment varies with raindrop size and velocity but, in using Eq.20, that
effect is ignored. However, provision is made in WEPP to take account of differences in drop
size distribution associated with different rainfall simulators in experiments designed to
determine values of kiq.
Eqs. 18 and 20 provide the basic equations that enable WEPP to model the contributions of
rill and interrill erosion to soil loss more directly than when USLE/RUSLE model is used but,
in WEPP, the spatial distribution of rill and interrill areas needs to be well defined. WEPP
was originally designed to operate in situations where ridge tillage is used. With ridge tillage,
interrill erosion is well defined as being on the ridge slopes and rill erosion in the furrows. In
other situations, the distributions of rill and interrill areas is more difficult to specify. In
rangelands, broad areas of surface water flow occur more commonly than rills so that splash
and sheet erosion dominate. An alternative to the WEPP cropland interrill erosion model,
Dss = kss I1.05
qss 0.59
(21)
where DSS “detachment” associated with splash and sheet erosion, kss is the erodibility factor
and qss is the runoff rate from the associated eroding surface, has been developed for use in
rangelands (Wei et al.,2009).
EUROSEM is another erosion model that is more process based than the USLE/RUSLE. In
EUROSEM, detachment by raindrop impact (Dr) is modeled using
8
kD
Dr = _____
KE exp (-zh) (22)
ρs
were kD is an index of detachability that is determined experimentally, ρs is particle density,
KE is the rainfall kinetic energy of the raindrops impacting the ground, z is a factor that
varies with soil texture, and h is the mean depth of the water layer on the soil surface.
Detachment by flow is modeled using
Df = β w vs (CT – C) (23)
where β is an index of detachability, w is flow width, vs is particle settling velocity, C is the
sediment concentration, and CT is the sediment concentration at the transport capacity of the
flow.
The coefficients used in Eq. 22 result from measurement of soil material transported by
splash under ponded conditions in the experiments of Torri et al. (1987) where increases in
water depth were observed to reduce splash erosion. The effect of water depth on splash
erosion results from (a) dissipation of raindrop energy in the water layer and (b) the effect of
water depth on splash trajectories. Considering that splash does not transport 100 % of the
material detached by raindrops impacting water, Eq. 22 does not actually model detachment.
Also, erosion by rain-impacted flows where sediment is transported by rolling, saltation and
suspension in the flow is the real focus of Eq. 22 and the effect of flow depth on erosion by
rain-impacted flows is quite different to its effect on splash erosion (Kinnell, 2005).
As a general rule, the values of soil erodibility factors have to be determined experimentally
although there are cases where they are predicted from soil properties. In areas where
detachment results from raindrop impact, erodibility is affected by modification of the soil
surface generated by the impacts. Raindrop impacts on surfaces not covered by water may
break soil aggregates and generate surface crusts which affect soil erodibility. In areas where
soil crusts occur, particles are held more tightly within the soil surface than in areas that are
not crusted and this reduces detachment. With splash erosion, loose particles sit and wait on
the soil surface between drop impacts. The transport efficiency of splash increases with slope
gradient but, over time, a layer of loose particles builds up on the soil surface and energy has
to be expended in moving them before detachment can occur. Consequently, this layer
provides a degree of protection against detachment. When raindrops impact a soil surface
covered by water, the protection provided by loose particles sitting on the surface is in
addition to that provided as the result of the dissipation of raindrop energy in the water layer.
Although splash erosion may dominate erosion for considerable periods during a rainfall
event, rain-impacted flows are usually more important in moving soil material across the soil
surface because the transport mechanisms in rain-impacted flows are much more efficient
than splash transport. In rain-impacted flows, particles move across the soil surface by
rolling, saltation and in complete suspension. Shallow low velocity flows often do not have
the capacity to cause particles to move by rolling and saltation by themselves but rolling and
saltation can be stimulated to occur when raindrops impact the soil surface through the flow.
Under these circumstances, each rolling or saltation event is of limited duration and is
9
associated with individual raindrop impacts. Fig.4 illustrates how raindrop and flow factors
influence the detachment and transport processes associated with the erosion of fine particles,
silt and sand by rainfall. Particles moving by raindrop induced rolling and saltation move
across the surface at rates that depend on raindrop size, impact frequency, particle size and
density and the velocity of the flow. Consequently, depending on the rain and flow
conditions, particles in rain impacted flows travel across the surface at virtual velocities that
vary from near zero to the velocity of the flow. In effect, particles are winnowed from the soil
surface at various rates. Large gravel particles may not move at all and so, over time, soils
that have high gravel contents may become highly resistant to erosion through the formation
of erosion “pavements”. Particles moving by raindrop induced rolling and saltation sit on the
soil surface between drop impacts and so provide a degree of protection against detachment
as described in the case of splash erosion. As a result, soil resistance to erosion may vary
considerably during a rainfall event. If H is the degree of protection provided by the loose
material, then all the material discharged from the soil surface comes from the layer of loose
material when H = 1.0. Consequently, it follows from Eq. 20 that, for any given rainfall
event,
Di = (kiq.U (1 – H) + H kiq.P) qi I (24)
where kiq.U is the interrill erodibility factor when no loose material exists on the surface and
kiq.P is the soil erodibility factor when loose material completely protects the soil surface from
detachment. As noted earlier, detachment of soil particles from a cohesive surface varies with
cohesion and interparticle friction so that factors such as the development of surface crusts
can cause kiq.U to vary with time. The soil erodibility term in Eq. 24 is kiq.U (1 – H) + H kiq.P
and values of kiq obtained in rainfall simulator experiments like those undertaken by Elliott et
al (1989) lie between kiq.U and kiq.P. Where exactly they do lie is unknown because H is
unknown.
THE EFFECT OF PARTICLE TRAVEL RATES ON
SEDIMENT COMPOSITION AND ERODIBILITY
As noted above, erodibilities have units of soil loss per unit of the erosivity index used in the
model that is being considered in the analysis of the data. Factors such as cohesion, particle
size and aggregates stability have been observed to influence these erodibilities. Data on the
physical and chemical nature of the soil involved may, in some cases, be used to predict
erodibility but often little attention is given to the composition of the sediment discharged in
experiments undertaken to determine erodibility.
The composition of the soil transported from an eroding area by rain-impacted flow during an
experiment tends to be finer than the original soil (eg. Meyer et al., 1980; Miller and
Baharuddin, 1987; Palis et al, 1990; Parsons et al., 2006) and particle travel rate is one of the
factors influencing sediment composition. Fast moving particles detached at the top of an
eroding area arrive at the downstream boundary well before slow moving particles detached
at the same place at the same time. Consequently, once sediment transport in rain-impacted
flow starts, the fine material dominates the early discharge of sediment but the composition
of the sediment becomes coarser with time as the more of the slower moving particles reach
to downstream boundary. If the particles are stable, then at the steady state, the composition
of the sediment discharge must be the same both physically and chemically as the original
soil. This is demonstrated by the results of laboratory experiments on erosion of sand by rain-
10
impacted flows by undertaken by Walker et al (1978). In these experiments, 3 m long sloping
beds of sand were eroded for 1 hour by rain-impacted flows generated by artificial rain at
three rainfall intensities (45, 100 and 150 mm/h). The data shown in Fig. 5 were generated by
rain made up of a single drop size (2.7 mm) falling on rain generated flows over beds of sand
with 2 different slope gradients (5 %, 0.5 %). The least erosive situation is the case where 45
mm/h rain fell when the slope gradient was 0.5 %. The most erosive situation is the case
where 150 mm/h rain fell when the gradient was 5 %. The enrichment factor is the ratio of
the proportion of the material in the sediment discharge to the proportion in the original
material. In all cases, the discharge of sediment was dominated by fine material at 2 minutes
and, in some cases, coarse material at 1 hour (Fig. 5). In the most erosive situation (150 mm/h
rain falling on 5 %.slope ), the sediment composition at 1 hour was close to the composition
for the steady state, the composition that occurs when the enrichment factor for all particles
sizes equals 1.0. Sediment composition generated by the rain impacted flows varies with the
intensity of the rain, the slope gradient, slope length and time because particles of different
size and density travel at different rates. The differential rate of transport of particles in rain-
impacted flows has consequences with respect to erodibility because the actual area
contributing to the soil loss does not stabilize until the slowest moving particles detached at
the farthest point from downslope boundary at the start of the rainfall event are discharged.
This fact is seldom considered when experiments are undertaken to determine erodibility.
Eq. 24 is essentially targeted at situations where raindrop induced saltation controls sediment
discharge in situations like that illustrated in Fig. 6A. Particles of silt and sand may travel
over the soil surface in more than one mode before being discharged. For example, as may be
perceived from Fig. 6B, they may leave the point of detachment traveling in splash (ST), then
move further downslope by raindrop induced saltation (RIS) and finally by flow driven
saltation (FDS) as flow energy increases down the slope. Transitions between raindrop
induced saltation and flow driven saltation may vary in time and space during a rainfall event
and will depend on the intensity of the rain, the infiltration characteristics of the soil, the
length of the slope and the slope gradient. The effect of the transition can have an appreciable
effect on both soil loss and the composition of the sediment discharged. Fig. 7B shows
modeled flow velocities at the brink (discharge boundary) of planar bare soil surfaces of
various length on a 9 % slope resulting from the rainfall event shown in Fig. 7A. Fig. 8 shows
the loss of materials of various size and density from those bare soil surfaces when a
mechanistic model of erosion by rain-impacted flow was used by Kinnell (2009). For the
surfaces up 15 m in length, particles larger than 0.1 mm in size traveled over the whole length
by raindrop induced saltation. Under these conditions more of the 0.11 mm sand was lost
during the event that 0.46 mm coal. On the 20 m long area, 0.46 mm coal which had been
traveling slower than 0.11 sand on shorter areas, traveled for a short period of time by flow
driven saltation during the high intensity bursts of rain and, as a result, more of the 0.46 mm
coal was lost than the 0.11 mm sand. On areas 25 m long and more, flow driven saltation also
contributed to the discharge of 0.11 mm sand but the amount lost was always less than the
amount of 0.46 mm coal. Little regard is given to the effect of temporal and spatial changes
in transport mechanism when experiments are undertaken to determine values for soil
erodibility factors in sheet and interrill erosion areas even though these changes may have a
major impact on the amount of soil lost from an area.
11
CONCLUSION
Although conceptually, rainfall erosivity is the capacity of rain to produce erosion where as
soil erodibility is the susceptibility of the soil to be eroded, the factors controlling the erosive
stress applied to the soil surface and the factors influencing the resistance of the soil to them
vary in time and space in complex ways. In all existing predictive models of rainfall erosion,
numerous simplifications and assumptions have to be made for practical reasons and, as a
result, these models do not have the capacity to deal with such complexity. Consequently,
erosivity and erodibility values are specific to the model in which they are parameterised and
to the scale that the model operates within.
REFERENCES
Bagarello V , and D'Asaro F. V. Estimating single storm erosion index, Transactions of the
American Society of Agricultural Engineers 1994, 37 , 785–791.
Bargarello, V. Di Stefano; C., Ferro, V., Kinnell, P.I.A, Pampalone, V.; Porto, P.; Todisco, V.
Predicting soil loss on moderate slopes using an empirical model for sediment concentration.
In review. Hydrological Processes 2010.
Bullock, P.R.; de Jong, E.; and Kiss, J.J... An assessment of rainfall erosion potential in
southern Saskatcheuian , from daily rainfall records. Canadian Agricultural Engineering
1987, 29, 109-115
Elliot, W.J.; Liebenow, A.M.; Laflen, J.M.; Kohl, K.D. A compendium of soil
erodibility experiments. Publication No. 3. USDA-ARS National Soil Erosion
Research Laboratory, West Lafayette, IN. 1989
El-Swaify, S.A., and Dangler, E.W. Erodibilities of selected tropical soils. In Soil Erosion:
Prediction and Control. Foster, G.R., Ed..: Soil Conservation Society of America, Ankeny,
Iowa. 1976, 105-114.
Flanagan, D.C., and Nearing, M.A., Eds. USDA-Water Erosion Prediction Project: Technical
documentation. NSERL Rep. No. 10. National Soil Erosion Research Laboratory, West
Lafayette, IN. 1995.
Hoyes, N., Waylen, P.R., Jaramillo, A. Seasonal and spatial patterns of erosivity
in a tropical watershed of the Colombian Andes. Journal of Hydrology 2005, 314, 177–
191.
Kinnell, P.I.A. Rainfall intensity-kinetic energy relationships for soil loss prediction.
Soil Sci. Society of America Journal 1981, 45, 153-155
Kinnell, P.I.A.. Rainfall energy in Eastern Australia: Intensity-kinetic energy relationships for
Canberra,A.C.T. Australian Journal Soil Research 1987, 25, 547-553
Kinnell, P.I.A. Interrill erodibilities based on the rainfall intensity – flow discharge erosivity
factor. Australian Journal Soil Research, 1993, 31, 319- 332.
12
Kinnell , P.I.A. Raindrop impact induced erosion processes and prediction: A review
Hydrological Processes 2005,19, 2815-2844
Kinnell , P.I.A. The influence of raindrop induced saltation on particle size distributions in
sediment discharged by rain-impacted flow on planar surfaces. Catena 2009, 78; 2-11
Kinnell , P.I.A. Event soil loss, runoff and the Universal Soil Loss family of models: A
review. Journal of Hydrology 2010, 385, 384-397
Law,J.O.; Parsons, D.A. The relation of raindrop size to intensity. Transactions of the
American Geophysical Union 1943, 24, 452-460
Loch, R. J. , Slater, B. K., and Devoil, C. Soil erodibility (Km) values for some Australian
soils. Australian Journal of Soil Research 1998, 36, 1045 - 1056
Meyer, L.D, Harmon, W.C., and McDowell, LL. Sediment sizes eroded from crop row
sideslopes. Transactions of the American Society of Biological Engineers 1980, 23, 891-898.
Miller, W.P, and Baharuddin, M.K. Particle Size of Interrill-eroded Sediments from Highly
Weathered Soils. Soil Science Society of America Journal 1987, 51,1610-1615.
Moore,I.D., and Kinnell, P.I.A. Kinematic overland flow - generalization of Rose's
approximate solution, part II. Hydrology 1987, 82, 351-362
Nash, J.E., Sutcliffe, J.E. River flow forecasting through conceptual models.
Part 1 – a discussion of principles. Journal of Hydrology 1970, 10, 282–290.
Onstad, C. A., and Foster, G. R. Erosion modelling on a watershed. Transactions of the
American Society of Agricultural Engineers 1975, 18, 288-92.
Palis, R.G., Okwach, G., Rose, C.W., and Saffigna, P.G’. Soil Erosion Processes and Nutrient
Loss. I. The interpretation of enrichment ratio and nitrogen loss in runoff sediment.
Australian Journal of Soil Research 1990, 28, 623-639.
Parsons, A.J., Abrahams, A.D, and Luk, S-H. Size characteristics of sediment in interrill
overland flow on a semiarid hillslope , Southern Arizonia. Earth Surface Processes and
Landforms 2006, 16, 143-152.
Posch, M., and Rekolainen, S. Erosivity factor in the Universal Soil Loss Equation
Estimated from Finnish rainfall data, Agricultural Science in Finland 1993, 2, 271-279
Renard, K.G., Foster, G.R., Weesies, G.A., McCool, D.K., and Yoder, D.C. Predicting soil
erosion by water: A guide to conservation planning with the Revised Universal Soil Loss
Equation (RUSLE). U.S. Department of Agriculture Agricultural Handbook. No. 703. US
Department of Agriculture, Washington, DC.1997
Richardson, C.W., Foster, G.R., Wright, D.A. Estimation of Erosion Index from
Daily Rainfall Amount. Transactions of the American Society of Agricultural
13
Engineers 1983, 26, 153–160.
Rekolainen, S., and Posch , M. Adapting the CREAMS Model for Finnish Conditions. Nordic
Hydrology 1993, 24, 309-322
Simanton, J.R., Rawizt, E., and Shirley, E.D.. Effects of rock fragments on erosion of semi-
arid rangeland soils. In: Erosion and Productivity of Soils Containing Rock Fragments. Krai,
D. M.; Hawkins, S. L., Eds. Soil Science Society Amererica, Madison, WI. 1984, 65-67.
van Dijk, A.I.J.M, Bruijnzeel, L.A., and Rosewell, C.J. Rainfall intensity–kinetic energy
relationships: a critical literature appraisal. Journal of Hydrology 2002 , 261, 1-23
Walker,P.H., Kinnell, P.I.A., and Patricia Green. Transport of a non-cohesive sand mixture in
rainfall and runoff experiments. Soil Science Society of America Journal. 1978, 42, 793-801
Wei, H., M.A. Nearing, J.J. Stone, D.P. Guertin, K.E. Spaeth, F.B. Pierson, M.H. Nichols,
and C.A. Moffett.. A new splash and sheet erosion equation for rangelands. Soil Science
Society of America Journal 2009, 73:1386–1392.
Williams, J.W., Izaurralde, R.C., and Steglich, EM. Agricultural Policy/ Environmental
Extender Model Theoretical Documantation. BRC Report # 2008-17. Blackland Research
and Extension Center, Temple, Texas. 2008.
Wischmeier,W.C. A soil erodibility nomograph for farmland and construction sites. Journal
of Soil and Water Conservation 1971, 26, 189-193
Wischmeier,W.C., and Smith, D.D. Predicting rainfall erosion losses – a guide to
conservation planning. Agricultural Handbook No. 537. US Department of Agriculture ,
Washington, DC. 1978.
Young, R.A, and Mutchler, C.K. Erodibility of some Minnesota soils. Journal of Soil and
Water Conservation 1997, 32,180-182
Yu, B. A comparison of the R-factor in the Universal Soil Loss Equation and the Revised
Universal Soil Loss Equation. Transactions of the American Society of Agricultural and
Biological Engineers 1999, 1615-1620.
Yu, B.; Rosewell, C.J. An Assessment of a Daily Rainfall. Erosivity Model for New South
Wales. Aust. J. Soil Res. 1996, 34, 139–152.
Zhang, K.L., Shu, A.P., Xu, X.L., Yang , Q.K. and Yu, B. Soil erodibility and its estimation
for agricultural soils in China. Journal of Arid Environments 2008, 72,
1002-1011.
14
Figure Captions
Figure 1. The relationships between event erosion (Ae) and EI30 obtained on bare fallow plots
in the USA: (A) Plot 5 Experiment 3 at Holly Springs, MS and (B) Plot 5 Experiment 1 at
Morris, MN. Eff(ln) is the Nash–Sutcliffe (1970) efficiency factor for logarithmic transforms
of the data.
Figure 2. The relationship between event sedment concentration (ce) and the product of I30
and the kinetic energy per unit quantity of rain (E / re) for Plot 5 Experiment 1 at Morris, MN.
Figure 3. The relationship between event erosion and the product of event runoff ratio (QRe)
and EI30 for Plot 5 Experiment 1 at Morris, MN.
Figure 4. Schematic of how variations raindrop kinetic energy and flow shear stress affect the
detachment and transport of silt, sand and fine particles during rainfall erosion. Ec is the
critical raindrop kinetic energy required to cause raindrop detachment (RD). Its variation
during rain with no runoff signifies changes in resistance to detachment caused by, for
example, the development of a soil crust. Its variation with flow shear stress signifies the
cushioning effect of increasing water depths. τc(bound) is the critical shear stress required to
cause flow detachment (FD). τc(loose) is the critical shear stress required to cause flow driven
saltation (FDS). FS is continuous suspension in the flow. RIS is raindrop induced saltation.
ST is splash transport.
Figure 5. Enrichment factor curves at for sediment discharged at 2 minutes and 60 minutes
when beds of sand were eroded by rain impacted flows in experiments undertaken by Walker
et al (1978) using 2.7 mm raindrops. (Extracted from Figure 3 of Walker et al (1978)).
Figure 6. Schematic representation of how detachment and transport processes associated
with rainfall erosion may vary spatially on planar inclined bare soil surfaces.
Figure 7. Brink (downstream boundary) flow velocities (B) for bare soil of various lengths
inclined at 9 % produced by the rainfall-runoff model described by Moore and Kinnell (1987)
and the rainfall intensities recorded during a rainfall event at the Ginninderra Experiment
Station, Canberra, Australia (A).
Figure 8. Amounts of coal, sand and fine particles discharged for the rainfall and runoff
conditions shown in Figure 7 when a mechanistic model of erosion by rain-impacted flow
was used by Kinnell (2009).
15
0.01
0.1
1
10
100
0.01 0.1 1 10 100observed event soil loss (t/ha)
soil loss estimated by R
e = EI 30 (t/ha)
1 : 1
Eff(ln) = 0.592
1
2
3
4
Ae = 0.0302 EI30 5 6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
Figure 1 47
48
49
50
0.01
0.1
1
10
100
0.01 0.1 1 10 100
observed event soil loss (t/ha)
soil loss estimited by R
e = EI 30 (t/ha)
1 : 1
Eff(ln) = 0.007
Ae = 0.0203 EI30
A
B
16
1
2
0.01
0.1
1
10
0.01 0.1 1 10
observed event sediment concentration (t/ha/mm)
estimated sedim
ent concentration (t/ha/m
m)
Eff(ln) = 0.303
1 : 1
3 4
5
Figure 2 6
7
ce = 0.0839 I30 (E / re)
17
1
0.01
0.1
1
10
100
0.01 0.1 1 10 100
observed event soil loss (t/ha)
estimated soil loss (t/ha)
Eff(ln) = 0.835
1 : 1
2 3
4
5
Figure 3. 6
7
Ae = 0.0839 QRe EI30
18
RAIN WITH
NO RUNOFF
RAIN WITH
RUNOFF
Fine Particles
RD-FS
Silt & Sand
RD-RIS
Silt & Sand
RD-FDS
Clay, Silt & Sand
RD-ST
Clay, Silt & Sand
FD-FDS,FS
NO EROSION E < Ec, τ < τc (bound)
A B
0
Ec Ec
0 τc (bound) τc (loose)
Raindrop
Energy (E)
Flow Shear Stress (τ) RAIN WITH
NO RUNOFF
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
46
47
48
49
Figure 4 50
51
19
1
2
3
4
5
6
7
8
9
10
11 12
13
14
Figure 5 15
16
17
18
19
20
20
1
2
3
4
5
6 7
8
Figure 6 9
10
11
12
13
14
RD - ST
RD – ST
RD – RIS
RD – RIS
RD - ST
RD – ST
RD – RIS
RD – RIS
RD – FDS
A B
RD - ST
RD – ST
RD – RIS
RD – RIS
RD – FDS
RD – FDS
FD – FDS
C
Flow
21
0
25
50
75
100
125
150
0 20 40 60 80 100 120 140
time (mins)
rainfall intensity (mm/h)
0
50
100
150
200
250
0 20 40 60 80 100 120 140
time (mins)
brink velocity (mm/s)
2 m5 m10 m15 m20 m25 m30 m
9% slope
critical for FDS of coal
1
2
3
4
A 5
6
7
8
9
10
11
12
13
Figure 7A 14
15
16
17
18
19
20
21
22
23
24
25
B 26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Figure 7. 45
46
22
1
2
0
5000
10000
15000
20000
25000
0 5 10 15 20 25 30 35
plot length (m)
coal and sand discharged
during event (g/m
)
0
500
1000
1500
2000
fine sedim
ent discharged
during event (g/m
)
coal p=0.11 p=0.2 p=0.46 p=0.9 fine
3 4
Figure 8 5
6
7
8
9
10