You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/253340741

Estimating the USLE Soil Erodibility Factor in Sicily, South Italy

Article  in  Applied engineering in agriculture · March 2012


DOI: 10.13031/2013.41347

CITATIONS READS

29 810

6 authors, including:

Vincenzo Bagarello Costanza di Di Stefano


Università degli Studi di Palermo Università degli Studi di Palermo
97 PUBLICATIONS   1,545 CITATIONS    82 PUBLICATIONS   895 CITATIONS   

SEE PROFILE SEE PROFILE

Vito Ferro Massimo Iovino


Università degli Studi di Palermo Università degli Studi di Palermo
227 PUBLICATIONS   3,332 CITATIONS    122 PUBLICATIONS   1,224 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Investigations about soil hydraulic characteristics of the experimental basin View project

Rainfall erosivity in soil erosion processes View project

All content following this page was uploaded by Vincenzo Pampalone on 02 June 2014.

The user has requested enhancement of the downloaded file.


ESTIMATING THE USLE SOIL ERODIBILITY
FACTOR IN SICILY, SOUTH ITALY
V. Bagarello, C. Di Stefano, V. Ferro, G. Giordano, M. Iovino, V. Pampalone

ABSTRACT. The Universal Soil Loss Equation (USLE) is used by professionals and technicians to predict soil loss by water
erosion and to establish soil conservation measures. One of the key elements of the USLE is the K factor, which is a measure
of the soil erodibility. Given the difficulty in collecting sufficient data to adequately measure K, early in the USLE's history
the soil erodibility nomograph method was developed to allow estimation of K based on standard soil properties. Since the
nomograph approach was developed based on a small number of soils in the United States, it is necessary for other contexts
to check the nomograph's ability to predict the soil's true erodibility. Considering that soil organic matter data are difficult
to obtain, an estimation procedure of the soil erodibility factor, K, based only on soil textural data is desirable. In this
investigation, the soil erodibility factor was first experimentally determined for the clay soil at the Sparacia (Sicily)
experimental station. A relatively low value (0.039 t ha h ha‐1MJ ‐1mm‐1) was determined, and summer erodibility was found
to be more than twice the value of winter erodibility. This measured K value was 1.85 times the nomograph K, which for many
practical applications is not a large difference. Finally, using 1813 data points, a procedure for estimating K using only soil
textural data was developed for Sicily. The errors of the predictions did not exceed a factor of two and three for 94.4% and
99.2% of the data points, respectively, suggesting a satisfactory ability of the developed procedure to yield an estimate of K
with a reduced input dataset.
Keywords. Soil erosion, USLE, Erodibility factor.

S
oil erosion is a major environmental threat to the on the basis of collecting storm and erosion data for many
sustainability and productive capacity of years. In the simpler alternative nomograph procedure from
agriculture, and available worldwide estimates Wischmeier et al. (1971), several measured soil properties
suggest that arable land is lost at a rate of more than are combined according to a pre‐established scheme to
10 million ha per year (Pimentel et al., 1995). Predicting soil determine K. In particular, the procedure needs knowledge of
erosion is necessary for implementing soil conservation soil particle size distribution (PSD), organic matter (OM)
measures and in obtaining reliable predictions of their content, and soil structure and permeability. The
effectiveness for an area of interest. experimental effort to determine these soil characteristics
The Universal Soil Loss Equation (USLE; Wischmeier may increase substantially with the spatial scale of interest
and Smith, 1978) and its revised version (RUSLE, Renard (plot, hillslope, watershed, region) in order to take into
et al., 1997) are still commonly used to predict plot scale soil account spatial variability of soil properties in the studied
erosion, despite the availability of many alternatives (Jetten area. In this case, reducing the number of input variables in
and Favis‐Mortlock, 2006). According to several authors, the the evaluation procedure of K can be practically attractive for
USLE/RUSLE are still popular because they i) combine limiting laboratory analyses and resulting costs. In particular,
acceptable accuracy with relative simplicity, ii) have the excluding OM content from prediction of K may be desirable
ability to use quite basic data, and iii) are the only models that for work on large scales, given that OM data are often missing
can rely on a global (i.e. worldwide distributed) dataset in regional soil maps (Zacharias and Wessolek, 2007).
(Risse et al., 1993; Rosewell, 1993; Liu et al., 2001; Hann and Attempts to simplify the K evaluation procedure have been
Morgan, 2006; Salvador Sanchis et al., 2008). carried out in the past and simplified relationships have been
One of the difficulties for those using the USLE is proposed for predicting K values of soils for which data are
estimation of the soil erodibility factor, K, which is defined limited (Römkens et al., 1986, 1997). One of these
relationships was included in the RUSLE manual (Renard
et al., 1997).
Estimating the K factor on the basis of reduced
Submitted for review in May 2011 as manuscript number SW 9197; experimental information makes sense if the procedure
approved for publication by the Soil & Water Division of ASABE in developed by Wischmeier et al. (1971) allows for satisfactory
December 2011. description of true soil erodibility. The nomograph was
The authors are Vincenzo Bagarello, Associate Professor, Costanza
Di Stefano, Researcher, Vito Ferro, Full Professor, Giuseppe Giordano, developed with specific reference to U.S. soils and therefore
Full Professor, Massimo Iovino, Associate Professor, and Vincenzo it needs testing in other areas of the world which requires the
Pampalone, Postdoctoral Fellow; Dipartimento dei Sistemi use of experimental data for those areas. A comparison
Agro‐Ambientali, Università di Palermo, Viale delle Scienze, 13, 90128, between the experimental erodibility factor and the one
Palermo, Italy. Corresponding author: Vincenzo Bagarello, Dipartimento
dei Sistemi Agro‐Ambientali, Università di Palermo, Viale delle Scienze,
estimated by the nomograph should be carried out for
13, 90128, Palermo, Italy; phone: 0039‐091‐7028108; e‐mail: different soils, but even a check limited to a single soil may
vincenzo.bagarello@unipa.it. contribute to the ability to better assess the applicability of

Applied Engineering in Agriculture


Vol. 28(2): 199‐206 E 2012 American Society of Agricultural and Biological Engineers ISSN 0883-8542 199
the nomograph, taking into account that a single procedure developed by Bagarello and Ferro (1998) and
measurement of K implies an onerous and prolonged (many Bagarello et al. (2004). A representative soil loss value at the
years) experiment at the plot scale. This check is particularly event temporal scale, Ae (t ha‐1), was obtained for a given plot
important for clay soils, which were not included in the group size (i.e., given width and length) by averaging the available
of soils for which the nomograph is expected to be well suited replicated data. Taking into account that, in the USLE, the
(Römkens et al., 1997). geometric characteristics of the plot are only represented in
An additional justification for development of an terms of plot length, l, and also considering that 2‐m wide
estimation procedure for K in Sicily is that, for 199 Sicilian plots are expected to give long‐term results representative of
samples, the soil physical quality experimentally determined wider plots (Bagarello et al., 2011), the measured soil loss
in terms of S‐index (Dexter, 2004) showed a tendency to from the narrow (2 m) and the wide (8 m) plots was averaged
decrease as the nomograph K increased (Bagarello et al., to obtain a single value of Ae for a given erosive event.
2009). In other terms, the nomograph K seems to have some The topographic factor, S (Wischmeier and Smith, 1978),
potential to describe soil physical quality with a reduced was calculated as (Nearing, 1997):
sampling and laboratory effort, independently of the soil
erosion predictive ability of the USLE. 17
S = −1.5 + (1)
The general objective of this investigation was to derive 1 + exp(2.3 − 6.1 sin β)
a method to predict the soil erodibility factor of the USLE in
where b (°) is the slope angle (S = 1.86 for a 14.9% slope
Sicily with limited soil data. The specific objectives were to:
steepness). A normalized soil loss, i.e., the soil loss from the
i) experimentally determine the K factor for the clay soil at
the “Sparacia” experimental station and compare it to the unit plot (Wischmeier and Smith, 1978), AeUP (t ha‐1), was
nomograph value; and ii) develop a new procedure to calculated from each individual Ae value according to the
following relationship:
estimate K based only on soil textural data for Sicily, using
a data base of 1813 individual measurements of PSD and OM Ae
AeUP = (2)
content distributed throughout the island. S
since the length of the considered plots was practically
coincident with the one of the unit plot.
MATERIALS AND METHODS The experimental value of the mean annual soil erodibility
MEASUREMENT OF THE SOIL ERODIBILITY FACTOR factor, K, was determined by the following relationship
The investigation was carried out at the experiment station (Foster et al., 1981):
for soil erosion measurement “Sparacia” of the Agricultural
Faculty of the Palermo University, located in western Sicily, N
Italy, approximately 100 km south of Palermo (Bagarello ∑ (AeUP )
et al., 2008, 2010; Bagarello and Ferro, 2010). The soil is a j =1 (3)
K=
Vertic Haploxerept (Soil Survey Staff, 2006) with a clay N
texture (62% clay, 33% silt, 5% sand). The soil is massive in
winter, being wet and fully swelled, but it develops a
∑ (Re )
j =1
polygonal pattern of surface shrinkage cracks in late spring
or early summer due to drying. The gravel content is where N is the number of erosive events occurred during a
negligible. Rainfall data were measured at 1‐min time multiyear period. This equation was actually the one used to
intervals by a recording rain gauge located in the determine K by the designers of the USLE, and it was also
experimental area. The area has a typical Mediterranean applied in other investigations to determine the temporal
semi‐arid climate with a mean annual rainfall of variability of the soil erodibility factor (Mutchler and Carter,
approximately 700 mm and a dry period extending for up to 1983; Bagarello et al., 2008). Equation 3 was also used to
seven months (April‐October). determine the seasonal values of the soil erodibility factor
Data for this investigation were obtained from plots (Kw, wet season; Kd, dry season). In this case, N in equation 3
installed on a 14.9% slope, maintained periodically in a was the number of erosive events occurring in the considered
cultivated fallow using hand implements or a powered season (November to April for Kw and May to October for Kd)
cultivator operated in the up and down direction. In during the multi‐year period. Only two periods of the year
particular, soil loss data were collected from six plots of 8 m were considered since sample sizes were not large enough to
width × 22 m length. These plots produced 42 events of allow analysis at a more detailed temporal scale, such as the
measurable runoff and soil loss during the period from monthly one. Annual values of K (Ka), Kw (Kwa), and Kd (Kda)
January 2000 to January 2009. An additional two 2‐m wide were also calculated by equation 3. These last two variables
× 22‐m long plots produced seven measurable runoff/soil were determined for six years, characterized by one or more
loss events from September 2007 to January 2009. Rills were erosive events in each season.
obliterated at the end of each erosive event, mainly using The USLE scheme was considered in this investigation to
hand implements, in an attempt to establish similar determine the soil erodibility factor because the K calculation
conditions (i.e., reasonably smoothed soil surface) at the depends on the estimating procedure of Re and almost all
beginning of each event. available data on rainfall erosivity in Sicily were expressed
For a given erosive event, the rainfall erosivity index, Re, in terms of USLE's R factor (Ferro et al., 1991, 1999;
was calculated (Wischmeier and Smith, 1978) and the soil Bagarello and D'Asaro, 1994; D'Asaro et al., 2007). A more
loss of each individual plot was determined by measuring important reason from a scientific point of view was that
sediment concentration in the storage tanks according to the recent investigations carried out in Sicily confirmed the

200 APPLIED ENGINEERING IN AGRICULTURE


USLE's ability to predict rainfall energy on the basis of the sand and sandy loam soils; SS = 2 (fine granular) was used for
measured intensities (Ferro et al., 2008). sandy clay, sandy clay loam, loam, silt loam and silt soils;
SS = 3 (medium or coarse granular) was applied to clay loam
ESTIMATION OF THE SOIL ERODIBILITY FACTOR and silty clay loam soils; and SS = 4 (blocky, platy or massive)
A data set including surface soil samples collected at was used for clay and silty clay soils. For establishing the
1813 sampling points distributed throughout Sicily (fig. 1) permeability index, PP, of each sampled soil, the average
was used for this investigation. For each sampling point, the class values of saturated soil hydraulic conductivity, Ks,
PSD was measured following H2O2 pretreatment to eliminate proposed by Carsel and Parrish (1988) were used. The USDA
organic matter and clay deflocculation by sodium soil texture classes were grouped according to decreasing
metaphospate and mechanical agitation. Fine size fractions values of Ks. For each considered textural group, table 1 lists
were determined by the hydrometer method, whereas the the expected range of Ks values and the corresponding PP
coarse fractions were obtained by mechanical sieving using index.
mesh sizes of 2000, 860, 425, 250, 106, and 75 mm. Eight fine The soil erodibility factor, K (t ha h ha‐1MJ‐1mm‐1), and its
fraction data points were obtained by the hydrometer method first approximation, K' (Wischmeier et al., 1971), were
measuring the suspension density after 2, 5, 15, 30, 60, 180, calculated by the nomograph of Wischmeier et al. (1971).
1440, and 2880 min. For a few samples, a laser diffraction Relating both SS and PP to soil texture might induce some
technique was applied and the conversion relationships criticism since the actual soil structural and permeability
developed by Di Stefano et al. (2010) were used to assure characteristics are not considered in the determination of K.
homogeneity within the data set in terms of considered However, a soil survey is generally considered to be enough
particle size fractions for the analysis. Particle size fraction to estimate structure and permeability categories, since SS
data were classified according to the USDA standards (Gee and PP can be estimated using qualitative information
and Bauder, 1986). For each soil sample, the percentage of (Wischmeier et al., 1971). Soil texture is known to be an
silt + very fine sand particles (equivalent particle diameter important factor controlling soil structure (e.g., Hillel, 1998)
ranging from 0.002 mm to 0.1 mm), f, and the percentage of and the SS and PP indices have a reduced impact on the
coarse sand (0.1 to 2 mm), g, were also determined using the estimated erodibility factor, mainly depending on soil
measured PSD. The geometric mean particle diameter, Dg textural and OM characteristics. Therefore, the applied
(mm), was calculated according to the following relationship procedure was considered to be physically reasonable and
(Shirazi and Boersma, 1984): objectively repeatable. A similar approach was used in the
⎛ Ns ⎞ RUSLE (Römkens et al., 1997).
D g = exp⎢ 0.01 ∑ Fi ln mi ⎟ (4) The nomograph K values were compared with the
⎢ ⎟ additional erodibility values, KR86 and KR97 (both in SI
⎝ i =1 ⎠ units), estimated by the following relationships (Römkens
where Fi (%) is the primary particle size fraction, mi (mm) is et al., 1986; 1997):
the arithmetic mean of the particle size limits of that size, i ⎡ ⎛ log D + 1.519 ⎞ 2 ⎤
is the size fraction number and Ns is the total number of size ⎪ 1 g
KR86 0.0035 + 0.0388 exp − ⎢⎢
= ⎟ ⎥ (5)
fractions. Following Verstraeten et al. (2002), the particle ⎪ 2⎝ 0.7584 ⎟ ⎥
⎣ ⎠ ⎦
size classes clay, silt and sand (<0.002, 0.002 to 0.05, and
0.05 to 2.0 mm) were considered to calculate Dg. The total ⎡ ⎛ log D + 1.659 ⎞ 2 ⎤
organic carbon content, TOC (%), was determined by the 1 g
K R97 = 0.0034 + 0.0405 exp ⎪− ⎢ ⎟ ⎥ (6)
Walkley‐Black method and the organic matter, OM, content ⎪ 2 ⎢⎝ 0.7101 ⎟ ⎥
⎠ ⎦
was estimated to be equal to 1.724 times the measured TOC ⎣
value (Nelson and Sommers, 1996).
For each sample, the structure index, SS, of the nomograph
by Wischmeier et al. (1971) was estimated using the
available soil texture information and the classification
reported in figure 2, based on the USDA texture triangle
(Giordano et al., 2004). In particular, the structure index SS % silt
= 1 (very fine granular) was associated with sandy, loamy

% clay

% sand
1 very fine granular 3 medium or coarse granular
2 fine granular 4 blocky, platy or massive
Figure 1. Sampling points in Sicily for determination of the USLE's soil
erodibility factor.
Figure 2. Soil structure index, SS, classification.

Vol. 28(2): 199‐206 201


Table 1. Proposed ranges of saturated soil hydraulic conductivity, Ks, considered reasonably representative for the sampled site.
values according to Carsel and Parrish (1988) and permeability The annual values of the soil erodibility factor, Ka, varied
index, PP, for the considered groups of soils.
between 0.00058 and 0.075 t ha h ha‐1MJ‐1mm‐1 with a ratio
Soil Type Ks (cm s‐1) PP between the two extremes (130.7) much higher than the one
Sandy loam, Loamy sand, Sand 10‐3 - 10‐2 2 (6.7) obtained by Bagarello et al. (2008). Therefore, a
Silt loam, Loam, Sandy clay loam 10‐4 - 10‐3 3 reasonably stable mean annual soil erodibility factor was
Clay, Silty clay loam, Clay loam, Sandy clay, Silt 10‐5 - 10‐4 4 obtained notwithstanding that variability of the annual values
Silty clay 10‐6 - 10‐5 5 was not fully represented in the available dataset.
The experimental erodibility factor obtained in this
Equation 5 was obtained by using data from 249 soils investigation was considered to be reliable since it was
worldwide (Verstraeten et al., 2002). Equation 6, which was slightly lower than the maximum possible value (0.042 in SI
included in the RUSLE manual (Renard et al., 1997), was units), estimated for this soil with the relationship by Torri
deduced on the basis of measured K values for 225 soils. et al. (1997) as a function of the clay content of the soil,
According to Römkens et al. (1997), equations 5 and 6 yield developed with reference to a global dataset of soil
values of K for classes of soils. However, individual K values erodibility factor values. If the range of possible K values is
may have more practical interest than a mean representative partitioned into four classes (i.e., low erodibility, K < 0.0225;
value for a given textural class, since USLE and RUSLE were relatively low, K = 0.0225‐0.045; relatively high, K =
developed to be applied at a local (plot) scale, and a correct 0.045‐0.0675; and high, K > 0.0675 t ha h ha‐1MJ‐1mm‐1), it
estimate of individual values implies a correct estimate of the can be suggested that this clay soil was characterized by a
mean. Therefore equations 5 and 6 were applied at the relatively low erodibility, probably because the high
sampling point scale and their performances were assessed cohesive forces between particles oppose particle
using the Nash‐Suttcliffe efficiency index, NSEI (Nash and detachment.
Suttcliffe, 1970): The discrepancy between the experimental and the
predicted erodibility by a factor of 1.85 was considered not
N
to be substantial. Given the multiplicative nature of the
∑ (Ym − Yes ) 2 USLE, a true erodibility 1.85 times higher than the one given
i =1 by the nomograph implies an equal error in the estimated soil
NSEI = 1 − (7)
N loss by a professional or a technician using the nomograph.
∑(Ym − Ym )
2
According to Foster et al. (1981), “low” erosivity (R = 300 MJ
i =1 mm ha‐1h‐1year‐1) and erodibility (K = 0.01 t ha h
ha‐1MJ‐1mm‐1) values yield a unit plot soil loss prediction of
in which N is the number of considered data points, Ym and 3 t ha‐1year‐1, whereas “high” R and K values (8000 and 0.06,
Yes are the measured (calculated, in this investigation; i.e. the respectively, in SI units) yield 480 t ha‐1year‐1. The ratio
nomograph K) and the corresponding estimated (KR86 or between high and low unit plot soil loss is 160 or probably
KR97) soil erodibility factor values, respectively, and Y is much more, given for example that possible K values for
m
the mean of the measured values. A value of NSEI = 1 is agricultural soils can theoretically vary from near zero to
indicative of a perfect correspondence between measured 0.09 t ha h ha‐1MJ‐1mm‐1 (Foster et al., 1981). It therefore
and predicted values while a value of NSEI = 0 suggests that seems reasonable to suggest that an uncertainty in the K
the model prediction is no more accurate than the mean of the factor prediction by a factor of less than two does not
observed values. A NSEI value < 0 suggests that the observed compromise estimation of, at least, the order of magnitude of
mean is better than the predictions. soil loss. The discrepancy observed in this investigation
In another analysis, however, the correspondence between might depend on the fact that, in the Sicilian context, the
the two approaches (nomograph, eqs. 5 or 6) was determined USLE did not work perfectly because of differences in
with reference to soils grouped by textural characteristics. In rainfall, dominant hydrological processes and landscape
particular, the experimental range of log Dg was subdivided diversity as compared to the eastern United States
into 10 intervals of equal magnitude and the median of the (Boardman, 2006). Other possible factors are the
nomograph K, calculated for each established interval, was unsuitability of the nomograph to represent erodibility of
compared with the corresponding medians of KR86 and K R97. clay soils, or the circumstance that the sampled plots had a
greater slope (14.9%) than the unit plot (9%), on which K
should be measured (Wischmeier and Smith, 1978). This last
RESULTS AND DISCUSSION difference is irrelevant only in the case that cannot still be
verified, of a perfect description of slope steepness effects on
MEASUREMENT OF THE SOIL ERODIBILITY FACTOR
soil loss by equation 1. Plots of 22 m have been recently
For the considered investigation period, R was equal to
realized on other slopes differing in steepness at Sparacia,
779 MJ mm ha‐1h‐1year‐1. For the Sparacia's soil, the
with the objective to test slope steepness effects in the near
nomograph by Wischmeier et al. (1971) yielded a value of K
future, once a reasonably representative number of erosive
equal to 0.021 t ha h ha‐1MJ‐1mm‐1 and the experimentally
events has been sampled.
determined soil erodibility factor was equal to 0.0389 t ha h
The fact that a single station for soil erosion measurement
ha‐1MJ‐1mm‐1. In a previous investigation carried out on the
was operating in Sicily precluded the possibility to generalize
22‐m long plots with five years of data, Bagarello et al.
the testing results. However, locally collected data were
(2008) obtained K = 0.0405 t ha h ha‐1MJ‐1mm‐1. Therefore,
encouraging, since they suggested that the nomograph was
including additional events in the database did not
practically usable for a clay soil, frequent in hilly areas of
appreciably modify the measurement of K, which can be
Sicily, and not included in the original database. In other

202 APPLIED ENGINEERING IN AGRICULTURE


words, the only possible experimental check of the procedure ESTIMATION OF THE SOIL ERODIBILITY FACTOR
by Wischmeier et al. (1971) was compatible with the decision Table 2 provides the summary statistics of the data used
to simplify the estimation procedure of K of the nomograph. in this investigation and also summarizes both the first
The Kd/Kw ratio was equal to 2.26 (Kd = 0.0504, Kw = approximation, K', and the final value, K, of the nomograph
0.0222, SI units), suggesting that the soil was more erodible soil erodibility. According to the K results, the Sicilian soils
in the dry period than the wet one. Erodibility was less are characterized, on average, by a relatively low erodibility.
variable in the wet period (ratio between the maximum and The ratio between the maximum and the minimum K factor
the minimum annual value = 70.7) than the dry one (ratio = was equal to 13.4, suggesting a moderate variability of the
229.0). The Kda/Kwa ratio varied widely from year to year soil erodibility factor in Sicily.
(from 0.024 to 129.9), but a dry soil erodibility higher than Equations 5 and 6 have a similar form but they do not
the wet soil one was the most common result, occurring in coincide. In particular, in the experimental range 0.002 < Dg
four of the six considered years. The ratio between Kda and < 0.99 mm of this investigation, equation 5 yielded different
Kwa was correlated with the rainfall erosivity of the wet K predictions as compared with equation 6 by a maximum
period (Rwa; coefficient of determination, R2 = 0.54, percentage of 45%. In addition, the maximum predicted soil
coefficient of correlation > 0, P = 0.05) by a decreasing erodibility factor (KR86 = 0.0423 for Dg = 0.030 mm and KR97
relationship (fig. 3) but not with the erosivity of the dry period = 0.0439 for Dg = 0.022 mm) was appreciably lower than the
(Rda; R2 = 0.03) or the Rda/Rwa ratio (R2 = 0.06). Note that, maximum possible K value, equal to 0.09 t ha h ha‐1MJ‐1mm‐1
for this period, the mean erosivity was higher in the dry (Wischmeier et al., 1971; Foster et al., 1981).
season (Rd = 784, SI units) than the wet one (Rw = 493, SI Figure 4 compares equations 5 and 6 with the (Dg , K of the
units). Variation of soil erodibility during the year has been nomograph) data pairs of this investigation. A noticeable
detected in several investigations (Mutchler and Carter, scattering of the data points was detected on the Dg , K plane
1983; Zanchi, 1988; Torri et al., 2006), but the factors causing and very low values of NSEI (0.028 for KR86 and ‐0.004 for
this form of variability have not been completely established. KR97) were obtained, showing that the proposed relationships
According to Salvador Sanchis et al. (2008), soil aggregate cannot be used to obtain reliable estimates of the soil
stability is probably the most important parameter erodibility factor of the nomograph at a selected sampling
determining seasonal changes in soil erodibility. Soil point. This result may be explained by the origin of the tested
aggregates are more stable in a wetter situation, and high air equations, based on values of the erodibility factor for classes
temperatures and lack of rain influence aggregate stability of soil. The generally high variation of K for a given Dg value
negatively (Boix‐Fayos et al., 1998, 2001). In the eastern (fig. 4) suggested that a point K estimation cannot be obtained
Iberian Peninsula, aggregates tested from an initially moist by empirically re‐fitting a relationship of the form of
condition were more stable than air dried aggregates (Cerdà, equations 5 and 6 to the Sicilian data.
1998). However, data by Mutchler and Carter (1983) in the In the analysis of the medians for each log D g interval,
United States, and by Zanchi (1988) and Torri et al. (2006) both KR86 and KR97 were found to be significantly correlated
in Italy, suggested a higher erodibility in winter than in with nomograph K (fig. 5). For KR86, the 95% confidence
summer. A possible explanation is that soil shear strength interval for the slope included unity (0.67‐1.31) but the
generally decreased as soil moisture increased, which intercept was higher than zero (0.0005‐0.016). For KR97, the
resulted in an increase in soil erodibility (Al‐Durrah and 95% confidence interval for the intercept included zero
Bradford, 1981; Hussein et al., 2007). Therefore, the (‐0.002‐0.004) but the slope was >1 (1.16‐1.40). Therefore,
conclusion for the investigated clay soil was that soil the two regression lines did not coincide with the 1:1 line. In
aggregate stability had a larger impact on erodibility as other words, grouping soils by textural characteristics was
compared with soil shear strength. A condition for the not sufficient to obtain an estimate of the soil erodibility
occurrence of Kwa > Kda was a high rainfall erosivity in the factor coinciding with the one obtained by the nomograph in
wet period, probably because soil shear strength was greatly the considered range of Dg values.
reduced in this case.
Table 2. Summary statistics of the measured soil characteristics and
the first approximation (K') and final value of the soil
erodibility factor of the nomograph, K (sample size, N = 1813).
1000 Standard
y = 73.982e-0.007x Variable Minimum Maximum Mean Median Deviation
100 R@ = 0.5385
Clay (< 0.002 mm) 0.0 76.3 30.0 28.3 15.7
10 Silt 0.6 92.7 39.4 39.6 14.3
K da / K wa

(0.002‐0.05 mm)
1
Sand 1.1 99.2 30.6 24.3 21.4
0.1 (0.05‐2.0 mm)

0.01 f (0.002‐0.1 mm) 0.8 93.4 46.6 47.1 14.2


0 200 400 600 800 1000 g (0.1‐2.0 mm) 0.3 99.0 23.7 16.9 20.3
Rwa (MJ mm ha-1h-1) Organic matter, 0.0 18.0 2.9 2.3 2.2
OM (%)
Figure 3. Annual values of the ratio between erodibility of dry and wet
periods, K da/K wa, plotted against rainfall erosivity of the wet period, R wa. K' (SI units) 0.0004 0.0887 0.0250 0.0228 0.0132
K (SI units) 0.0065 0.0868 0.0291 0.0285 0.0126

Vol. 28(2): 199‐206 203


an estimate of K' obtained without any information on soil
OM content and the one based on both textural and OM data.
Although data appeared rather scattered in figure 6, a
noticeable discrepancy between the two estimates of K' was
a very rare occurrence given that the two K' values differed
by a factor of less than two for 1747 data points,
corresponding to 96.4% of the complete data set. The
discrepancy factor was less than three for 98.9% of the cases.
The K/K'es ratios, being the soil nomograph erodibility K
and K'es the estimate of K' obtained by equation 8, were
grouped according to the value of the SS × PP index, and the
Figure 4. Relationship between the soil erodibility factor, K, determined median of K/K'es, denoted by the symbol MR, was calculated
by the nomograph and the estimates obtained by equations 5 (KR86) and for each SS × PP value. Plotting MR against SS × PP showed
6 (KR97), as a function of the geometric mean particle diameter, D g . that the two variables were related by the following linear
relationship (fig. 7):
Figure 6 shows that the relationship between K' and M =
MR = 0.731 + 0.068 × SS × PP (9)
f × (f + g) (Wischmeier and Smith, 1978) can be expressed
by the following equation: with R2 = 0.94. Finally, the following relationship was
(8) obtained by combining equations 8 and 9:
K '= 7.0 × 10 −6 × M 0.998
which is characterized by R2 = 0.67 and an exponent K es = 7.0 × 10 −6 × M × (0.731 + 0.068 × SS × PP) (10)
practically equal to one. A discrepancy factor, equal to the where Kes is the estimate of the soil erodibility factor, in SI
ratio between K' obtained by equation 8 and the nomograph units, based exclusively on soil textural data. The comparison
K', was calculated as an indicator of the departure between between Kes and K of the nomograph (fig. 8) suggested a
minimal bias of the predictions, perhaps related to the small
0.05
number of high erodibility values in the considered dataset,
KR86 (SI units)

y = 0.9915x + 0.0081
0.04 R@ = 0.8636 and it yielded a NSEI value of 0.589. Moreover, the
discrepancies between Kes and K of the nomograph did not
0.03
exceed a factor of two for 1711 data points, corresponding to
0.02 94.4% of the complete dataset, with a discrepancy of less than
three in 99.2% of the cases.
0.01 In terms of medians, Kes was found to be significantly
a)
0 correlated with nomograph K (fig. 5c), and the regression line
0 0.01 0.02 0.03 0.04 0.05
K (SI units) 0.1
y = 7E -06x0.9977
K' (SI units)

0.08
0.05 R@ = 0.6691
0.06
KR97 (SI units)

y = 1.2792x + 0.0012
0.04 R@ = 0.987
0.04
0.03 0.02
0.02 0
0.01 0 2000 4000 6000 8000 10000
b)
M
0
0 0.01 0.02 0.03 0.04 0.05 Figure 6. Relationship between the first approximation of the soil
K (SI units) erodibility factor of the nomograph, K', and the M parameter.

0.05 2.5
Median of K/K'es

y = 1.1382x-0.0014 2
K es (SI units)

0.04 R@ = 0.9199
1.5
0.03
1 y = 0.0679x + 0.7308
0.02
R@ = 0.9385
0.5
0.01
c) 0
0 0 5 10 15 20 25
0 0.01 0.02 0.03 0.04 0.05 SS x PP
K (SI units) Figure 7. Relationship between the median of the K/K'es (K = erodibility
factor determined by the nomograph; K'es = first approximation of K
Figure 5. Comparison between the medians of K of the nomograph and the estimated by eq. 8) values grouped according to the SS (soil structure
corresponding medians obtained by a) eq. 5 (KR86); b) eq. 6 (KR97), and code) by PP (soil permeability code) product and SS × PP.
c) eq. 10 (Kes).

204 APPLIED ENGINEERING IN AGRICULTURE


Estimated K (SI units)
0.1 This investigation gave support to early suggestions that
0.08 the USLE can be viewed as a satisfactory compromise
between reliability and simplicity of the estimate, at least in
0.06
particular cases such as the ones corresponding to the unit
0.04 plot condition, and it added a new experimental measurement
0.02 of the soil erodibility factor for a still unexplored
Mediterranean area. Improving interpretation of the soil
0
erosion processes at different spatial and temporal scales and
0 0.02 0.04 0.06 0.08 0.1 testing or developing physically oriented modeling
K (SI units) approaches have an obvious scientific importance. However,
simple and physically plausible empirical methods for
Figure 8. Comparison between the soil erodibility factor, K, values
determined by the nomograph and the ones estimated by eq. 10.
predicting soil erosion should not be abandoned because they
allow soil erosion prediction by professionals and
was not significantly different from the 1:1 line because the technicians. More soil erosion data are necessary in all cases.
95% confidence intervals for the slope (0.86‐1.41) and the In particular, other experimental installations for soil erosion
intercept (‐0.008‐0.005) included one and zero, respectively. measurement should be realized in Sicily to test applicability
Taking into account the multiplicative nature of the of the USLE and other models in other locations of the island.
USLE, an error in the predicted erodibility by a factor of two
or three implies an equal approximation in the predicted soil ACKNOWLEDGEMENTS
loss per unit area as compared with a prediction also using This study was supported by grants of the University of
OM data. It seems reasonable to suggest that this level of Palermo (Dottorato di Ricerca in Idronomia Ambientale) and
approximation for a model yielding satisfactory results in the the Sicilian Region (Progetti MOFEROS and CISS), Regione
original formulation is acceptable, at least in practical Sicilia. All authors set up the research, analyzed the results
applications where the order of magnitude of the soil erosion and wrote the paper. We wish to thank the Associate Editor
phenomenon has to be estimated. Therefore, equation 10 is and all Reviewers since their constructive comments helped
expected to have interest for professionals and technicians. us to improve the quality of the manuscript.

CONCLUSIONS REFERENCES
Al‐Durrah, M., and J. M. Bradford. 1981. New methods of studying
The general objective of this investigation was to derive
soil detachment due to waterdrop impact. Soil Sci. Soc. America
a method to predict the soil erodibility factor of the USLE, K, J. 45(5): 949‐953.
in Sicily with limited soil data. For the Sparacia clay soil, K Bagarello, V., and F. D'Asaro. 1994. Estimating single storm
was experimentally determined because the nomograph is erosion index. Trans. ASAE 37(3): 785‐791.
not expected to be well suited for this type of soil. The Bagarello, V., and V. Ferro. 1998. Calibrating storage tanks for soil
experimental K factor (0.0389 t ha h ha‐1MJ‐1mm‐1) was erosion measurement from plots. Earth Surf. Proc. Land.
higher than the nomograph estimation (0.021 t ha h 23(13): 1151‐1170.
ha‐1MJ‐1mm‐1) but the detected level of discrepancy was not Bagarello, V., and V. Ferro. 2010. Analysis of soil loss data from
considered substantial from a practical point of view. plots of differing length for the Sparacia experimental area,
Possible reasons for this discrepancy were the unsuitability Sicily, Italy. Biosyst. Eng. 105(3): 411‐422, doi: 10.1016/j.
biosystemseng.2009.12.015.
of the nomograph to represent erodibility of clay soils or the
Bagarello, V., G.V. Di Piazza, and V. Ferro. 2004. Manual sampling
circumstance that the sampled plots had a greater slope and tank size effects on the calibration curve of plot sediment
(14.9%) than the unit plot (9%). A higher erodibility in storage tanks. Trans. ASAE 47 (4): 1105‐1112.
summer than in winter was measured at Sparacia, suggesting Bagarello, V., G.V. Di Piazza, V. Ferro, and G. Giordano. 2008.
that soil aggregates were less stable in a drier condition. Predicting unit plot soil loss in Sicily, south Italy. Hydrol.
Taking into account that OM data are often missing in Process. 22 (5): 586‐595, doi: 10.1002/hyp.6621.
regional soil databases, a procedure to obtain an estimate of Bagarello, V., M. Castellini, M. Iovino, and A. Sgroi. 2009.
K using exclusively soil textural information was considered Indagine sulla qualità fisica di alcuni suoli siciliani. In Proc. IX
to have practical importance for professional and Convegno Nazionale dell'Associazione Italiana di Ingegneria
technicians. Existing simplified methodologies were found Agraria , Ischia Porto, 12‐16 Settembre, memoria n.3‐46 (in
Italian).
not to work well in Sicily. Therefore, an alternative
Bagarello, V., V. Ferro, and G. Giordano. 2010. Testing alternative
methodology was developed. This methodology, also using erosivity indices to predict event soil loss from bare plots in
a criterion to estimate structure and permeability classes on Southern Italy. Hydrol. Process. 24 (6): 789‐797, doi:
the basis of soil texture data, was found to be reasonably 10.1002/hyp.7538.
accurate for most practical purposes, given that the errors of Bagarello, V., V. Ferro, G. Giordano, F. Mannocchi, V. Pampalone,
the predictions did not exceed a factor of two and three for F. Todisco, and L. Vergni. 2011. Effect of plot size on measured
94.4% and 99.2% of the considered 1813 data points, soil loss for two Italian experimental sites. Biosyst. Eng. 108(1):
respectively. The large sample size and the wide range of 18‐27, doi: 10.1016/j.biosystemseng.2010.10.006.
values for both the soil nomograph erodibility and the Boardman, J. 2006. Soil erosion science: Reflections on the
variables used for its calculation gave plausible cause to limitations of current approaches. Catena 68(2‐3): 73‐86, doi:
10.1016/j.catena.2006.03.007.
believe that the developed procedure should be usable at a
larger scale than the Sicilian one.

Vol. 28(2): 199‐206 205


Boix‐Fayos, C., A. Calvo‐Cases, A. C. Imeson, M. D. Soriano‐Soto, Nash, J. E., and J. E. Suttcliffe. 1970. River flow forecasting
and I. R. Tiemessen. 1998. Spatial and short‐term temporal through conceptual models: Part 1. A discussion of principles.
variations in runoff, soil aggregation and other soil properties J. Hydrol. 10(3): 282‐290.
along a Mediterranean climatological gradient. Catena 33(2): Nearing, M. A. 1997. A single continuous function for slope
123‐138. steepness influence on soil loss. Soil Sci. Soc. America J. 61(3):
Boix‐Fayos, C., A. Calvo‐Cases, A. C. Imeson, and M. D. 917‐919.
Soriano‐Soto. 2001. Influence of soil properties on the Nelson, D. W., and L. E. Sommers. 1996. Total carbon, organic
aggregation of some Mediterranean soils and use of aggregate carbon, and organic matter: Part 3: Chemical methods. In
size and stability as land degradation indicators. Catena 44(1): Methods of Soil Analysis, 961‐1010. D. L. Sparks et al., eds.
47‐67. Madison, Wis.: ASA and SSSA.
Carsel, R. F., and R. S. Parrish. 1988. Developing joint probability Pimentel, D., C. Harvey, P. Resosudarmo, K. Sinclair, D. Kurz, M.
distributions of soil water retention characteristics. Water Resour. McNair, S. Crist, L. Shpritz, L. Fitton, R. Saffouri, and R. Blair.
Res. 24(5): 755‐769. 1995. Environmental and economic costs of soil erosion and
Cerdà, A. 1998. Soil aggregate stability under different conservation benefits. Science 267(No.5201): 1117‐1123.
Mediterranean vegetation types. Catena 32(2): 73‐86. Renard, K. G., G. R. Foster, G. A. Weesies, D. K. McCool, and D.
D'Asaro, F., L. D'Agostino, and V. Bagarello. 2007. Assessing C. Yoder. 1997. Predicting Soil Erosion by Water: A Guide to
changes in rainfall erosivity in Sicily during the twentieth Conservation Planning with the Revised Universal Soil Loss
century. Hydrol. Process. 21(21): 2862‐2871, doi: Equation (RUSLE). USDA Agriculture Handbook No.703.
10.1002/hyp.6502. Washington, D.C.: USDA.
Dexter, A. R. 2004. Soil physical quality: Part I. Theory, effects of Risse, L. M., M. A. Nearing, A. D. Nicks, and J. M. Laflen. 1993.
soil texture, density, and organic matter, and effects on root Error assessment in the Universal Soil Loss Equation. Soil Sci.
growth. Geoderma 120(3‐4): 201‐214. Soc. America J. 57(3): 825‐833.
Di Stefano, C., V. Ferro, and S. Mirabile. 2010. Comparison Römkens, M. J. M., S. N. Prasad, and J. W. A. Poesen. 1986. Soil
between grain‐size analyses using laser diffraction and erodibility and properties. In Proc. of the 13th Congress of the
sedimentation methods. Biosyst. Eng. 106(2): 205‐215, doi: Intl. Soil Science Society 5: 492‐504.
10.1016/j.biosystemseng.2010.03.013. Römkens, M. J. M., R. A. Young, J. W. A. Poesen, D. K. McCool,
Ferro, V., G. Giordano, and M. Iovino. 1991. Isoerosivity and S. A. El‐Swaify, and J. M. Bradford. 1997. Soil erodibility factor
erosion risk map for Sicily. J. Hydrol. Sci. 36(6): 549‐564. (K). In Predicting Soil Erosion by Water: A Guide to
Ferro, V., P. Porto, and B. Yu. 1999. A comparative study of rainfall Conservation Planning with the Revised Universal Soil Loss
erosivity estimation for southern Italy and southeastern Equation (RUSLE) 65‐99. K. G. Renard et al., eds. Washington,
Australia. J. Hydrol. Sci. 44(1): 3‐24. D.C.: USDA.
Ferro, V., M. Birtone, F. G. Carollo, V. Pampalone, and S. Pomilla. Rosewell, C. J. 1993. SOILOSS: A Program to Assist in the
2008. Una nuova installazione per la misura contemporanea Selection of Management Practices to Reduce Erosion. Tech.
delle caratteristiche energetiche della precipitazione e della Handbook 11. Sidney, New South Wales: Soil Conservation
perdita di suolo parcellare. Quaderni di Idronomia Montana Service of New South Wales.
28(1): 185‐212 (in Italian). Salvador Sanchis, M. P., D. Torri, L. Borselli, and J. Poesen. 2008.
Foster, G. R., D. K. McCool, K. G. Renard, and W. C. Climate effects on soil erodibility. Earth Surf. Proc. Land. 33(7):
Moldenhauer. 1981. Conversion of the universal soil loss 1082‐1097, doi: 10.1002/esp.1604.
equation to SI metric units. J. Soil Water Conserv. Nov.‐Dec.: Shirazi, M. A., and L. Boersma. 1984. A unifying quantitative
355‐359. analysis of soil texture. Soil Sci. Soc. America J. 48(1): 142‐147.
Gee, G. W., and J. W. Bauder. 1986. Part 1: Particle‐size analysis. In Soil Survey Staff. 2006. Keys to Soil Taxonomy. 10th ed.
Methods of Soil Analysis, 383‐411. 2nd ed. A. Klute ed. Washington, D.C.: USDA‐Natural Resources Conservation
Madison, Wis.: ASA and SSSA. Service.
Giordano, G., V. Ferro, V. Bagarello, C. Di Stefano, M. Iovino, and Torri, D., J. Poesen, and L. Borselli. 1997. Predictability and
M. Minacapilli. 2004. Studi applicativi per la realizzazione della uncertainty of the soil erodibility factor using global dataset.
carta dell'erosione potenziale del territorio siciliano e del Catena 31 (1‐2): 1‐22.
relativo sistema informativo territoriale. Edizioni Anteprima Torri, D., L. Borselli, F. Guzzetti, C. Calzolari, P. Bazzoffi, F.
s.r.l., Palermo (in Italian). Ungaro, D. Bartolini, and M. P. Salvador Sanchis. 2006. Soil
Hann, M. J., and R. P. C. Morgan. 2006. Evaluating erosion control erosion in Italy: An overview. In Soil Erosion in Europe,
measures for biorestoration between the time of soil 245‐261. J. Boardman and J. Poesen, eds. New York, N.Y.:
reinstatement and vegetation establishment. Earth Surf. Proc. Wiley.
Land. 31(5): 589‐597. Verstraeten, G., K. Van Ost, A. Van Rompaey, J. Poesen, and G.
Hillel, D. 1998. Environmental Soil Physics . San Diego, Calif.: Govers. 2002. Evaluating an integrated approach to catchment
Academic Press. management to reduce soil loss and sediment pollution trough
Hussein, M. H., T. H. Kariem, and A. K. Othman. 2007. Predicting modelling. Soil Use Mgmt. 18(4): 386‐394.
soil erodibility in northern Iraq using natural runoff plot data. Wischmeier, W. H., and D. D. Smith. 1978. Predicting Rainfall
Soil Till. Res. 94(1): 220‐228. Erosion Losses: A Guide to Conservation Planning. USDA
Jetten, V., and D. Favis‐Mortlock. 2006. 2.16. Modelling soil Agriculture Handbook No. 537. Washington, D.C.: USDA.
erosion in Europe. In Soil Erosion in Europe. J. Boardman and Wischmeier, W. H., C. B. Johnson, and B. V. Cross. 1971. A soil
J. Poesen, eds. Chichester, UK: Wiley. erodibility nomograph for farmland and construction sites. J.
Liu, B. Y., M. A. Nearing, P. J. Shi, and Z. W. Jia. 2001. Slope Soil Water Cons. 26: 189‐193.
length effects on soil loss for steep slopes. In Sustaining the Zacharias, S., and G. Wessolek. 2007. Excluding organic matter
Global Farm, 784‐788. D. E. Stott, R. H. Mohtar, and G. C. content from pedotransfer predictors of soil water retention. Soil
Steinhardt, eds. Selected papers from the 10th Intl. Soil Sci. Soc. America J. 71(1): 43‐50.
Conservation Organization Meeting held May 24‐29, 1999 at Zanchi, C. 1988. Soil loss and seasonal variation of erodibility in
Purdue University and the USDA‐ARS National Soil Erosion two soils with different texture in the Mugello valley in Central
Research Laboratory. Washington, D.C.: USDA‐ARS. Italy. Catena Supplement 12(1): 167‐174.
Mutchler, C. K., and C. E. Carter. 1983. Soil erodibility variation
during the year. Trans. ASAE 26(4): 1102‐1104.

206 APPLIED ENGINEERING IN AGRICULTURE

View publication stats

You might also like