Catena 164 (2018) 107–115

Contents lists available at ScienceDirect

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

Deriving and evaluating hydraulics and detachment models of rill erosion T

for some calcareous soils
⁎
S. Mirzaee, S. Ghorbani-Dashtaki
Department of Soil Science, College of Agriculture, Shahrekord University, Shahrekord, Iran

A R T I C L E I N F O A B S T R A C T

Keywords: Rill erosion plays an important role in the amount of soil detachment and the transport sediment load on
Detachment rate models hillslope. Quantification of the soil erosion requires a more precise understanding of the processes and devel-
Flow and rill geometry opment of rill erosion models. The objective of this study was to derive and evaluate hydraulics and detachment
Hydrology models of rill erosion in calcareous soils of northwestern Iran. Rill erosion experiments were carried out at 55
Modeling
locations with three replications under field conditions. At each point, the rill plots were created with a 0.2 m
Soil erosion
width and 4 m length on agricultural soils. The inflow rates were 4, 12, 20 and 30 l min−1 with varying slope
from 4 to 25.5%. The results indicated that all conditions of flow regimes including sub- and super-critical
(laminar and turbulent) were observed in created rills by overland flow. The mean flow velocity and rill depth
have been described well by flow rate and slope gradient, while rill width and flow depth have been explained
well by flow rate. The prediction detachment rate by rill flow based on stream power model by non-linear
regression yielded the best results (R2 = 0.545 and RMSE = 0.00213 kg m−2 s−1) for all combinations of slope
classes. However, there are no significant differences between prediction accuracy of linear and non-linear
models, when individual slope classes were considered.

1. Introduction where Dr and Dc are detachment rate and detachment capacity

Soil erosion is a process of detachment of soil particles from the soil
mass and their transport by erosive factors (Elliot and Laflen, 1993;
Zhang et al., 2003; Wang et al., 2016). Physically soil erosion models
often divide soil erosion according to the sediment source into two
components: inter-rill and rill erosion (Wagenbrenner et al., 2010).
However, eroded rills are usual major concentrators of surface flow in
croplands where soil particles are non-layered and eroding loose at
farming depth. The rills on agricultural soils rapidly adapted their bed
topographic in response to flow and slope changes (Govers, 1992). On
the other hand, concentrated flow is deeper, faster and more energetic
than the shallow flow that happens in inter-rill areas. Therefore, it is
necessary that attention is focused on modeling rill geometry and de-
tachment rates especially on arid and semiarid calcareous soils condi-
tions.
Two important sub-processes in rill erosion are the flow detachment
erosion and sediment transport (Meyer et al., 1975). Foster and Meyer
(1972) generated a function to describe rill erosion:
q and S is slope gradient (m m−1) (Nearing et al., 1997).
Dr = Dc ⎡1 − s ⎤
⎢
⎣ ⎥
Tc ⎦
(kg m−2 s−1), respectively, qs represents the rate of sediment load
(g m−1 s−1) and Tc represents sediment transport capacity (g m−1 s−1).
The maximum Dr happens when the qs = 0 (Govers et al., 2007; Wang
et al., 2016). There are numerous relationships for predicting soil de-
tachment by rill flow within an eroding rill that used in soil erosion
models (Wagenbrenner et al., 2010). The Dc based on the flow shear
stress flow is usual relationship to predict soil detachment by con-
centrated (Govers et al., 2007). Eq. (2) is the following general form:
Dc = KHP (HP − HPc) (2)
where KHP is soil rill erodibility parameter, HP is hydraulic parameters
and HPc is the critical value below which no detachment occurs. HP can
be shear stress (τ) (Nearing et al., 1997), stream power (Ω) (Hairsine
and Rose, 1992; Elliot and Laflen, 1993), unit stream power (Ωu)
(Morgan et al., 1998) and unit length shear force (Γ) (Giménez and
Govers, 2002). The τ, Ω, Ωu and Γ parameters were calculated as:
τ = γRh sin[tan−1 (S)] (3)
−3
where γ is water specific weight (N m ), Rh is rill hydraulic radius (m),
(1) Ω = γRh Vsin[tan−1 (S)] (4)

⁎
Corresponding author.
E-mail address: ghorbani-sh@agr.sku.ac.ir (S. Ghorbani-Dashtaki).

https://doi.org/10.1016/j.catena.2018.01.016
Received 25 September 2017; Received in revised form 7 January 2018; Accepted 18 January 2018
Available online 20 February 2018
0341-8162/ © 2018 Elsevier B.V. All rights reserved.
S. Mirzaee, S. Ghorbani-Dashtaki
Fig. 1. Location of study area northwest Iran and distribution of studied points.
where V is the measured runoff velocity (m s−1) (Bagnold, 1966).
Ωu = Vsin[tan−1 (S)]
(Yang, 1972).
Γ = γAsin[tan−1 (S)]
2
where A is the rill cross-sectional area (m ) (Giménez and Govers,
2002).
Nearing et al. (1991) indicated that shear stress and stream power
were unsuitable predictors to estimate the soil detachment rate. They
found out that there was a logarithmic function between the dependent
variable (soil detachment) and independent variable (flow depth, bed
slope and mean weight diameter of the aggregates) on silt loam soil.
McIsaac et al. (1992) showed unit stream power was the best predictor
to predict soil detachment by concentrated flow. Elliot and Laflen
(1993) showed that the stream power model was highly efficient to
describe the detachment capacity rate on the 36 sites with a wide range
of geological, geomorphic, and geochemical properties from the WEPP
model database. Nearing et al. (1997, 1999) reported that both shear
stress or stream power were better suited to predict detachment rate
and concluded the stream power-based detachment model was favored
on silt loam and sandy loam soils. Zhang et al. (2003) showed that the
stream power by a power function was the best predictor to estimate
detachment rate at shallow flow conditions on the two silt loam soil
from rangeland and cropland. Giménez and Govers (2002) indicated
the superiority of the unit length shear force-based detachment model
Catena 164 (2018) 107–115
on silt loam and loamy sand soils with rough and smooth beds.
Wagenbrenner et al. (2010) compared rill erosion models in two dis-
(5)
turbed forest soils in the USA. It was observed that the stream power
was a feasible and reliable predictor. Similar results were also observed
for rangeland sites (Al-Hamdan et al., 2012). Li et al. (2015) quantified
(6)
land use effects on soil detachment in the Loess Plateau. In this study,
the detachment capacity was predicted by stream power, slope gra-
dient, soil bulk density, median diameter, silt content, cohesion, and
root density parameters. Wang et al. (2016) showed, in a loessial soil,
detachment predictions by flow velocity, unit energy and stream power
were better, but shear stress and unit stream power based detachment
models presented poor results. These contrasting results to determine
the best hydraulic parameter for estimating soil detachment rate by
concentrated flow could be a result of different conditions at the studied
sites.
It is clear that flow hydraulic parameters (i.e. τ, Ω, Ωu and Γ) are
functions of the flow velocity, flow depth, rill depth, rill width and
slope gradient. The results of more studies show that flow and rill
geometry (flow depth, flow velocity, rill depth and rill width) are re-
lated to the corresponding flow discharge for concentrated flow, but the
relationships are varied in the studies. Early research by Meyer et al.
(1975) and Lane and Foster (1980) found that flow velocity and rill
width predicted well from flow discharge. Gilley et al. (1990) using a
nonlinear regression method, predicted rill width from flow discharge.
Govers (1992) found that mean flow velocity depends on flow
108
S. Mirzaee, S. Ghorbani-Dashtaki Catena 164 (2018) 107–115

discharge, while no significant correlation was found between slope Table 1

and soil influence and flow velocities. Takken et al. (1998) and Nearing Descriptive statistics of studied soil.
et al. (1997) reported that accurate prediction for flow velocities was
Properties Min Max Mean Median Std. dev CV
obtained by flow discharge in rills eroding loose materials. Giménez
and Govers (2001) showed the flow velocity in rills on mobile beds is Clay (%) 10.0 54.0 31.3 32.0 9.10 29.1
slope independent and on fixed beds is clearly slope-dependent. Torri Silt (%) 22.0 70.0 36.9 36.0 7.70 20.9
Sand (%) 8.00 50.0 31.7 34.0 10.1 31.9
et al. (2006) found the flow discharge exponent varying from 0.43 to
CCE (%) 0.50 42.0 16.4 16.1 7.93 48.2
0.5 for small rills (with a width of 3 cm) and gullies (with a width of 50 OM (%) 0.17 3.02 1.72 1.81 0.59 34.3
to 100 cm), respectively. In contrast, some studies have showed that
slope gradient is another important factor affecting the flow and rill CV: coefficient of variation; OM: organic matter; CCE: calcium carbonate equivalent.
geometry (Foster et al., 1984; Abrahams et al., 1996; Di Stefano et al.,
2013; Peng et al., 2015). 0.1 ha and were mixed thoroughly. The soil was air dried and then
Connecting the hydrology routine data to the erosion models is sieved through a 2 mm sieve. Sand (0.05–2 mm) content was measured
necessary to predict erosion. At any point of watershed, hydrology by sieving. Silt and clay content was determined by hydrometer method
models can give flow discharges using rainfall intensity, infiltration rate (Gee and Bauder, 1986). The calcium carbonate equivalent (CCE)
and soil properties (Govers, 1992). Moreover, a quantitative equation contents were measured using back-titration methods (Nelson and
to predict detachment rate under calcareous conditions are still lacking Sommers, 1986). The organic carbon (OC) was measured by wet-oxi-
and need to be quantified. Hence, the objectives of this study were i) to dation (Nelson, 1982). Based on Jackson (1965), the organic matter
determine the relationship between flow and rill geometry and flow (OM) was accounted by OC as:
discharge-slope; and ii) to drive and evaluate flow shear stress, stream OM = OC × 1.724 (7)
power, unit stream power and unit length shear force-based detachment
models in calcareous soils of northwestern Iran. The descriptive statistics of soils in the region are shown in Table 1.
Generally, it was observed that soils in the southern part of the study
area have fine texture, while soils portrayed by coarse grained deposits
2. Material and methods are mostly found in the northern part. Based on investigations of soil
type, the main soil types are Entisols, Inceptisols and Mollisols (USDA,
2.1. Study area 2010).

The study area is located in the northeast part of Azerbaijan pro-
vince, Iran (38° 55.9′–39° 14.6′ N and 47° 9.4′–47° 22.4′ E) (Fig. 1). The
general landscape of the study region is shown in Fig. 1. The study area
is approximately 41,000 ha. Elevation ranges from 335 to 1690 m
above the sea level (Fig. 1). Dominated aspects in this area are north
and east. This region has parent materials of limestones and Quaternary
deposits which are enriched by carbonates.
Predominant land uses are agricultural and mostly used for wheat
and barley production. The study region was ploughed to 20 cm depth
using a moldboard and residues were returned in the late May (Mirzaee
et al., 2016).
The region is characterized by a semiarid climate and has an annual
average temperature and precipitation of 14.2 °C and 383.5 mm, re-
spectively. In the period 2003–2016 the driest month was August
(Fig. 2). Most of the heavy rainfall occurs in short episodes during
March and April (Fig. 2) and they lead to serious soil erosion.
Soil samples were taken using a random sampling method at 55 sites
(Fig. 1). Three sub-soil samples were collected at the depth of 0–15 cm.
At each site in three corners of a triangle with an approximate area of
2.2. Rill erosion simulation
The rill experiments were carried out at 55 sites the use of rills with
a width of 0.2 m and a length of 4 m on cultivated soil (Fig. 3). In
created rills on cultivated soil, stones and vegetation residue were
manually removed. Inflow rates including 4, 12, 20 and 30 l min−1
were added at the top of each rill as sequential by an energy dissipater
and controlled using a flow regulator. Each inflow rate was continued
to observe the steady state condition. A steady state condition was often
observed after 10-min. However, to ensure steady state condition, the
15-min duration was selected (Fig. 3). The procedure at each point was
replicated three times. The rill length was divided into 3 transects to
facilitate data measuring (Fig. 3). In each of these sections, the geo-
metric and hydraulic rill changes were recorded to obtain a mean value
(Cocharane and Flanagan, 1996).
During each inflow rate, the flow properties including flow velocity,
width and depth were determined four times. The flow width and depth
were measured by a scale ruler. Surface flow velocity was measured
using a dye method (Govers, 1992). A KMnO4 solution is used as a dye.
Surface flow velocity (V) was calculated as 2 m divided by the mean
travel time of the dye between the 1 and 3 m transects. The mean flow
velocity (ū) was accounted for laminar flow (Reynolds number <
2400) and turbulent flow (Reynolds number > 2400) as following
(Abrahams et al., 1985):
u = V × 0.7 for Laminar flow (8)

u = V × 0.8 for Turbulent flow (9)

After each run, the rill width and depth were measured by a scale
ruler. At 3 min interval, sediment and runoff samples were taken during
simulations. Runoff samples with sediments were settled to separate
sediments from the water. The wet sediments oven-dried at 105 °C for
24 h and then dry sediment weighed to determine the soil loss. The
detachment capacity rate (Dc) (kg m−2 s−1) was accounted as:
Wd
Dc =
twl (10)
where Wd is the dry eroded sediment weight (kg), t is the time (s), w is
Fig. 2. Average monthly precipitation and temperature (from 2003 to 2016).
the depth and l is the length of each rill (m).

109
S. Mirzaee, S. Ghorbani-Dashtaki Catena 164 (2018) 107–115

Fig. 3. Created rills on cultivated soil (a). Rill experiment in

unsteady state conditions (b). Rill experiment in steady
state conditions (c).

2.3. Estimation flow and rill geometry and detachment models of rill erosion determination) (Eq. (12)).
0.5
To estimate flow and rill geometry, the stepwise method was used to ⎡ ∑ (Y i − Y)
N 2
i ⎤
RMSE = ⎢ i = 1 ⎥
develop the regression models using the MatLab v.8.5. Hydraulic N (11)
⎣ ⎦
parameters including shear stress (τ), stream power (Ω), unit stream
power (Ωu) and unit length shear force (Γ) were calculated with Eq. (3), ∑i = 1 (Yi − 
N
Y) 2
i
(4), (5) and (6), respectively. The linear rill detachment model was R2 = 1 − N 2
N (∑i = 1 Yi)
applied to determine linear τ, Ω, Ωu and Γ-based detachment models as ∑i = 1 (Y2i ) − (12)
N
following:
where Yi and  Yi are measured and estimated values of parameters, re-
Dc = KHP (HP − HPc) (2)
spectively. N is the number of data (i.e. 55).
where KHP is soil rill erodibility parameter, HP is hydraulic parameters
and HPc is the critical value below which no detachment occurs. HP can 3. Results and discussion
be τ, Ω, Ωu and Γ. Also, a nonlinear method was used to develop the
regression models from hydraulic parameters (i.e. τ, Ω, Ωu and Γ). 3.1. The characteristics of hydraulic parameters

2.4. Performance criteria
In present study, the cross-validation method was used for vali-
dating regression models. The accuracy of estimations was investigated
by the RMSE (root mean square error) (Eq. (11)) and R2 (coefficient of
The descriptive statistics for hydraulic parameters are presented in
Table 2. Some hydraulic properties indicated a slight positive and ne-
gative skewness in the data. However, on the basis of Kolmogorov-
Smirnov test, all hydraulic parameters were normally distributed at the
5% level of significance. Fig. 4 indicates the Froude numbers versus

110
S. Mirzaee, S. Ghorbani-Dashtaki Catena 164 (2018) 107–115

Table 2 W = 3.104Q0.27 (RMSE=1.36 − R2 = 0.46) (15)

Hydraulic properties and sediment yield.
Hr = 0.149Q0.97 + 0.148S (RMSE=1.28 − R2 = 0.61) (16)
Parameters Min Max Mean Median Std. dev CV
−1
where ū is mean flow velocity (cm s ), Hw is flow depth (cm), W is rill
Flow velocity (cm s−1) 20.2 63.4 36.4 36.2 7.9 21.7
width (cm), Hr is rill depth (cm), Q is flow rate (l min−1) and S is slope
Flow depth (cm) 0.63 2.92 1.75 1.80 0.60 34.3
Rill width (cm) 2.73 12.41 6.17 6.00 1.78 28.8 (%). Fig. 6 revealed that the models derived from flow rate and slope
Rill depth (cm) 0.80 14.26 3.97 3.61 1.95 49.1 had consistent overestimated hydraulic geometry rills. In Eqs. (13) to
Slope (%) 4.0 25.5 12.8 12.0 4.60 35.9 (16), concentrated flow rate related to hydraulic geometry rill by a
power function. As can be seen from Eqs. (13) and (16) mean flow
velocity and rill depth were significantly correlated with slope. This
finding was difference in comparison with the results of Foster et al.
(1984), Abrahams et al. (1996), Di Stefano et al. (2013) and Peng et al.
(2015), that indicated the slope factor with a power model related to
hydraulic geometry rill.
The exponent of flow rate parameter in Eqs. (13), (14) and (16) is
agreement with those described in the other studies. According to the
results of studies world-wide the flow rate exponent for flow velocity,
rill width and flow depth ranging from 0.20 to 0.39, 0.14 to 0.51 and
0.21 to 0.48, respectively (Meyer et al., 1975; Lane and Foster, 1980;
Gilley et al., 1990; Govers, 1992; Abrahams et al., 1996; Takken et al.,
1998; Nearing et al., 1997; Giménez and Govers, 2001; Torri et al.,
2006). These studies were carried out in a wide range of geological,
geomorphic, and geochemical properties including field data on rill in
different land use type and soil texture classes from Belgium, Italy and
Spain (Torri et al., 2006), rills with freely erode a uniform soil layer and
with a fixed bed geometry (Giménez and Govers, 2001), a wide range of
Fig. 4. Hydraulic flow regime: sub-critical laminar, super-critical laminar, sub-critical hydraulic conditions, including slopes, flow rates, Froude numbers,
turbulent and super-critical turbulent for the experimental data. Reynolds numbers and bed roughness (Nearing et al., 1997). The flow
rate exponent in this study for mean flow velocity (Eq. 13) and rill
width (Eq. 15) is a relatively small and for flow and rill depth is a
relatively large. However, these exponents are very close to the ex-
ponents obtained by Foster et al. (1984), Di Stefano et al. (2013) and
Peng et al. (2015). On steep slope, deepening rills are more important
than the widening rills because of the higher shear stress. Moreover, as
the slope become steeper, the gravity influenced flow geometry in rill
and leading to a decrease in flow rate exponent to estimate flow velo-
city.

3.3. Predicting detachment rate (Dc)

3.3.1. Using hydraulic parameters by linear equations

Fig. 5. ū (mean flow velocity) and Hw (flow depth) versus Q (flow rate). The flow shear stress values varied from 3.05 to 32.56 Pa, the
stream power values ranged from 0.89 to 14.84 kg s−3, the unit stream
power values ranged from 0.01 to 0.11 m s−1, unit Length Shear force
Reynolds numbers under different flow discharges. The Reynolds
ranged from 0.23 to 4.56 kg s−2. Detachment rate ranged from 0 to
number is almost agreed on discharge, but no significant relationship
0.014 kg s−1 m−2.
was found between the Froude number and discharge (Fig. 4). This
In the linear relations (Table 3), for all combination of slope classes,
could be related to the influences of flow discharge and slope on the
the stream power (Ω) based detachment model was the best. Also, when
concentrated flow depth and velocity (Fig. 5). Flow depth increased and
individual slope classes were considered, the stream power was highly
decreased with increasing of flow rate and slope, respectively. How-
efficient to describe the detachment in the all slope classes (Table 3).
ever, flow velocity increases as either flow rate or slope. According to
However, the results were less conclusive for shear stress and unit
Fig. 4, all four flow regimes such as sub-critical laminar, super-critical
stream power. Based on the result in Table 3, in the slope classes of 4–8,
laminar, sub-critical turbulent and super-critical turbulent were ob-
12–16 and > 16%, the performance of unit stream power was better
served. Therefore, there was a broad range of data which were more
than shear stress. In the slope class of 8–12%, unit stream power based
important for deriving reliable soil hydraulic models.
detachment model followed closely by shear stress based detachment
model (Table 3). The unit length shear force had the worst performance
3.2. Models to estimate flow and rill geometry for the estimation of detachment rates in the all slope classes (Table 3).

A multiple regression models were derived to estimate mean flow
velocity, flow depth, rill width and rill depth from flow rate and slope
as following:
u = 18.21Q0.24 + 0.199S (RMSE=4.92 − R2 = 0.65)
Hw = 0.528Q0.459 (RMSE=0.25 − R2 = 0.84)
3.3.2. Using hydraulic parameters by non-linear equations
3.3.2.1. Flow velocity. In erosion modeling, mean flow velocity plays a
major role because effect of upon flow rate, soil surface roughness and
slope. The results showed that detachment rate increased with the
(13)
increasing flow velocity for each inflow rate and slope present (Fig. 7).
These can be showed using a power function in all slope classes (Fig. 7).
(14)
A power equation of flow velocity can be used to estimate the

111
S. Mirzaee, S. Ghorbani-Dashtaki Catena 164 (2018) 107–115

Fig. 6. Measured versus predicted hydraulic geometry rill

including ū (mean flow velocity), Hw (flow depth), Hr (rill
depth) and W (rill width).

Table 3
Rill erodibility based on shear stress (Kτ), stream power (KΩ), unit stream power (KΩu) and unit length shear force (KΓ) for linear model of Dc = KHP(HP − HPc) equation.

Slope class Kτ shear stress KΩ stream power KΩu unit stream power (kg m−1) KΓ unit length shear force (s)
(s m−1) (s2)

4–8% 4.4 × 10−4 (0.264) 1.2 × 10−3 (0.472) 2.2 × 10−1 (0.381) 2.6 × 10−3 (0.243)
8–12% 4.9 × 10−4 (0.444) 1.1 × 10−3 (0.530) 2.3 × 10−1 (0.449) 2.6 × 10−3 (0.343)
12–16% 4.8 × 10−4 (0.376) 9.7 × 10−4 (0.552) 1.8 × 10−1 (0.542) 2.7 × 10−3 (0.338)
> 16% 2.9 × 10−4 (0.332) 7.3 × 10−4 (0.556) 1.1 × 10−1 (0.401) 1.8 × 10−3 (0.339)
All 2.8 × 10−4 (0.303) 7.2 × 10−4 (0.486) 7.9 × 10−2 (0.285) 2.1 × 10−3 (0.324)

Fig. 7. Relationship between the detachment (Dc) and flow velocity (ū).

112
S. Mirzaee, S. Ghorbani-Dashtaki Catena 164 (2018) 107–115

Table 4 Table 7
Non-linear model to predict detachment rate (Dc) from flow rate (Q). Non-linear model to predict detachment rate (Dc) from unit length shear force (Γ).

Slope classes (%) Non-linear models R2 Slope classes (%) Non-linear models R2

4–8 Dc = 0.0011Q0.54 0.59 4–8 Dc = 0.00491Γ0.48 0.34

8–12 Dc = 0.0014Q0.49 0.49 8–12 Dc = 0.0049Γ0.54 0.39
12–16 Dc = 0.0015Q0.55 0.51 12–16 Dc = 0.0052Γ0.59 0.37
> 16 Dc = 0.0011Q0.61 0.50 > 16 Dc = 0.0043Γ0.54 0.43

Table 5 each slope class is shown in Table 5. Detachment rate by concentrated

Non-linear model to predict detachment rate (Dc) from shear stress (τ). flow increased with increasing shear stress (Table 5). A comparison of
the R2 value for linear (Table 3) and non-linear equations (Table 5)
Slope classes (%) Non-linear models R2
revealed that the R2 value of linear equation was low about 2% for
4–8 Dc = 0.001τ0.73 0.28 slope classes of 4–8 and > 16%. On the other hand, the R2 value
8–12 Dc = 0.0004τ1.04 0.43 slightly was high for linear equation in slope classes of 8–12 and
12–16 Dc = 0.0004τ1.02 0.32 12–16%.
> 16 Dc = 0.0005τ0.88 0.35
The results of investigation relationship between Dc and shear stress
for all combinations of slope classes indicated a power function, as
Table 6 following:
Non-linear model to predict detachment rate (Dc) from stream power (Ω).
Dc = 0.00097τ 0.67 (RMSE=0.00251 − R2 = 0.344) (19)
Slope classes (%) Non-linear models R2 2
The RMSE and R values indicated that non-linear equation (Eq.
0.73 (19)) for all combinations of slope classes showed the better perfor-
4–8 Dc = 0.0021Ω 0.48
8–12 Dc = 0.0017Ω0.79 0.53 mance than linear equation (Table 3) in estimating the detachment rate.
12–16 Dc = 0.0014Ω0.86 0.58
> 16 Dc = 0.0011Ω0.87 0.56 3.3.2.4. Stream power. Table 6 showed how the detachment rate is
explained by the stream power. In 4–8% slope class, the stream power
has explained about 48.3% of variations in detachment rate (Table 6).
detachment rate by concentrated flow as following:
Also, the non-linear regression model accounted for 53.0, 58.3 and
Dc = 0.000011u1.72 (RMSE=0.00217 − R2 = 0.504) (17) 55.7% of variability in detachment rate at slope class of 8–12, 12–16
and > 16% (Table 6). However, there is no significant difference
between the performance of linear (Table 3) and non-linear
3.3.2.2. Flow rate and slope. Table 4 shows the pairs flow rate (Q) and regression models (Table 6) to estimate detachment rate in slope
detachment rate (Dc) in different slope classes. As seen in Table 4, there classes.
was a power relation between Q and Dc in each slope classes. The nonlinear regression model derived to estimate detachment rate
Analysis demonstrated that a power equation for all combinations of that applied stream power as input variable for all combination of slope
slope classes showed a well performance for estimation detachment classes is shown in Eq. (20).
rates. Detachment rates by rill flow increased with both flow rate and
slope gradient. The following corresponding multiple regression equa- Dc = 0.0019Ω0.75 (RMSE=0.00213 − R2 = 0.545) (20)
tion were derived: The power equation improved the accuracy of detachment rate
Dc = 0.00053Q0.539S0.365 (RMSE=0.00217 − R2 = 0.519) (18) prediction about 6% (Eq. (19) and Table 3).

3.3.2.5. Unit stream power. The increasing unit stream power increased
3.3.2.3. Shear stress. The drag exerted by the flow on the soil is shear Dc by concentrated flow (Fig. 8). According to the R2 value (Table 3 and
stress. The relationship between detachment rate and shear stress for Fig. 8) linear and non-linear models were similar performance for slope

Fig. 8. Relationship between the detachment (Dc) and unit stream power (Ωu).

113
S. Mirzaee, S. Ghorbani-Dashtaki
Fig. 9. Comparison of detachment rate predicted by present study (a), Nearing et al. (1999) (b) and Zhang et al. (2003) (c).
classes, except for slope class of 4–8%.
At all combinations of slope classes, the relationship between de-
tachment rate by concentrated flow and unit stream power can be
shown by a non-linear regression model, as:
Dc = 0.044Ω u 0.676 (RMSE=0.00252 − R2 = 0.339)
The results showed that the non-linear model for all combinations of
slope classes performed better than the linear model (Table 3 and Eq.
(21)).
3.3.2.6. Unit length shear force. The relationship between Dc and unit
length shear at each slope classes are given in Table 7. These
relationships at all combinations of slope classes can be described by
a non-linear regression model, following as:
Dc = 0.0049Γ 0.51s (RMSE=0.0024 − R2 = 0.383)
Like the other hydraulic parameters, using the non-linear equation
had the better performance than the linear model for prediction de-
tachment rate.
3.3.3. Comparisons of hydraulic parameters based detachment rate (Dc)
The highest R2 and the smallest RMSE value in all combinations of
slope classes data was produced by the non-linear stream power based
model, while the highest RMSE and smallest R2 was obtained by the
linear unit stream power based model. Nonetheless, most physically
erosion models use the shear stress (Nearing et al., 1989) or the unit
stream power (Morgan et al., 1998) to estimate soil detachment rate by
overland flow. In present study, the superiority of stream power based
detachment model could be related to the flow velocity, which was
responsible for better performance stream power based detachment
model than the shear stress based detachment model. Torri et al. (1998)
reported that detachment rate by overland flow was closely related to
the stream power. Nearing et al. (1997, 1999) and Zhang et al. (2003)
studies for prediction detachment rate showed that non-linear stream
power based detachment model was more accurate. Li et al. (2015) and
Wang et al. (2016) found out that the predictions based on stream
power are powerful. However, the low R2 of relationship between de-
tachment capacity and shear stress, unit stream power and unit length
shear force indicate that detachment capacity in present study is not
well estimated by shear stress, unit stream power and unit length shear
force based detachment models. These results are similar to those re-
ported in previous studies (Nearing et al., 1999; Zhang et al., 2003;
Wang et al., 2016). These studies were carried out on a stony soil in a
semiarid environment (Nearing et al., 1999), on the natural, un-
disturbed, mixed mesic typical Udorthent soil (Zhang et al., 2003) and
on loessial soil (Wang et al., 2016).
As can be seen from the Fig. 9, Nearing et al. (1999) and Zhang et al.
(2003) models performed poorly in comparison to the function derived
here using non-linear regression model. The lack of precision and low
accuracy of the Nearing et al. (1999) and Zhang et al. (2003) can be
Catena 164 (2018) 107–115
somewhat attributed to the coefficients of stream power parameter in
stream power based detachment model, probably not being well suited
to conditions in the current study area. The difference between the
results of this study and other studies might be due to the unique in-
fluence of calcareous materials in soils (Ostovari et al., 2016). The high
(21)
calcium saturation tends to keep the calcareous soils in well aggregated
form. Aggregate stability is more important for infiltration, root growth
and resistance to water and wind erosion. Stable aggregates can also
provide a large range in pore space, including small pores within and
large pores between aggregates. Large pores associated with large,
stable aggregates favor high infiltration rates. However, surface crusts
and filled pores occur in weakly aggregated soils. Surface crusts prevent
infiltration and promote erosion (Algayer et al., 2014). This suggests
that erosion sub-models need to be calibrated for the variable field
conditions that exist in different areas.
(22)
4. Conclusion
In this study, the relationship between hydraulic-detachment
models and hydrodynamic parameters is investigated. Detachment rate
and hydraulic parameters were measured in calcareous soils in north-
western Iran. The models developed using flow rates and slope gra-
dients were only able to account for up to 65, 84, 46 and 61% of the
variation in mean flow velocity, flow depth, rill width and rill depth,
respectively. Moreover, the best predictor to estimate detachment rates
was stream power for each slope classes and all combinations of slope
classes. Non-linear equation between detachment rate and stream
power was the best to describe detachment rate for all combinations of
slope classes. The variation of the detachment rate explained up to
54.5% by non-linear stream power based detachment model. However,
there were no significant differences between linear and non-linear
performance when individual slope classes were considered. The results
could be useful to understanding the mechanism of rill erosion process
and to allocating soil conservation measures in northwestern Iran.
Future work could concentrate on quantifying at other land uses in-
cluding grasslands and forests on calcareous soils.
References
Abrahams, A.D., Parsons, A.J., Luk, S.H., 1985. Field measurement of the velocity of
overland flow using dye tracing. Earth Surf. Process. Landf. 11, 653–657.
Abrahams, A.D., Li, G., Parsons, A.J., 1996. Rill hydraulics on a semiarid hillslope,
southern Arizona. Earth Surf. Process. Landf. 21, 35–47.
Algayer, B., Wang, B., Bourennane, H., Zheng, F., Duval, O., Li, G., Le Bissonnais, Y.,
Darboux, F., 2014. Aggregate stability of a crusted soil: differences between crust and
sub-crust material, and consequences for interrill erodibility assessment. An example
from the Loess Plateau of China. Eur. J. Soil Sci. 65 (3), 325–335.
Al-Hamdan, O.Z., Pierson, F.B., Nearing, M.A., Williams, C.J., Stone, J.J., Kormos, P.R.,
Boll, J., Weltz, M.A., 2012. Concentrated flow erodibility for physically based erosion
models: temporal variability in disturbed and undisturbed rangelands. Water Resour.
Res. 48, 1–15.
Bagnold, R.A., 1966. An approach to the sediment transport problem from general phy-
sics. US Geol. Surv. Prof. Pap. 422 (1), 22–37.
114
S. Mirzaee, S. Ghorbani-Dashtaki
Cocharane, T.A., Flanagan, D.C., 1996. Detachment in a simulated rill. Trans. ASAE 40
(1), 111–119.
Di Stefano, C., Ferro, V., Pampalone, V., Sanzone, F., 2013. Field investigation of rill and
ephemeral gully erosion in the Sparacia experimental area, South Italy. Catena 101,
226–234.
Elliot, W.J., Laflen, J.M., 1993. A process-based rill erosion model. Am. Soc. Agric. Eng.
36 (1), 65–72.
Foster, G.R., Meyer, L.D., 1972. A closed-form equation for upland areas. In: Shen, H.
(Ed.), Sedimentation, Symp. to Honour Prof. H.A. Einstein, Fort Collins, pp.
12.1–12.17.
Foster, G.R., Huggins, L.F., Meyer, L.D., 1984. A laboratory study of rill hydraulics: I.
Velocity relationship. Trans. ASAE 27, 790–796.
Gee, G.W., Bauder, J.W., 1986. Particle size analysis. In: Klute, A. (Ed.), Methods of Soil
Analysis: Part 1 Agronomy Handbook No 9. American Society of Agronomy and Soil
Science Society of America, Madison, WI, pp. 383–411.
Gilley, J.E., Kottwitz, E.R., Simanton, J.R., 1990. Hydraulic characteristics of rills. Trans.
ASAE 33, 1900–1906.
Giménez, R., Govers, G., 2001. Interaction between bed roughness and flow hydraulics in
eroding rills. Water Resour. Res. 37 (3), 791–799.
Giménez, R., Govers, G., 2002. Flow detachment by concentrated flow on smooth and
irregular beds. Soil Sci. Soc. Am. J. 66, 1475–1483.
Govers, G., 1992. Relationship between discharge, velocity, and flow area for rills eroding
in loose, non-layered materials. Earth Surf. Process. Landf. 17, 515–528.
Govers, G., Gimenez, R., Oost, K.V., 2007. Rill erosion: exploring the relationship between
experiments, modeling and field observation. Earth Sci. Rev. 84, 87–102.
Hairsine, P.B., Rose, C.W., 1992. Modeling water erosion due to overland-flow using
physical principles. 2. Rill flow. Water Resour. Res. 28, 245–250.
Jackson, M.L., 1965. Soil Chemical Analysis. Englewood Cliffs, Prentice-Hall, NJ, USA.
Lane, L.J., Foster, G.R., 1980. Concentrated flow relationships. In: Knisel, W.G. (Ed.),
CREAMS: A Field Scale Model for Chemicals, Runoff, and Erosion from Agricultural
Management Systems, Conservation Research Report. 26. U.S. Dept. of Agriculture,
Washington D.C., pp. 174–485.
Li, Z.W., Zhang, G.H., Geng, R., Wang, H., Zhang, X.C., 2015. Land use impacts on soil
detachment capacity by overland flow in the Loess Plateau, China. Catena 124, 9–17.
McIsaac, G.F., Mitchell, J.K., Hummel, J.W., Elliot, W.J., 1992. An evaluation of unit
stream power theory for estimating soil detachment and sediment discharge from
tilled soils. Trans. ASAE 35 (2), 535–544.
Meyer, I.D., Foster, G.R., Romkens, M.J.M., 1975. Source of soil eroded by water from
upland slopes. In: Present and Prospective Technology for Predicting Sediment Yields
and Sources: Agricultural Research Service Report, ARS-S-40, pp. 177–189.
Mirzaee, S., Ghorbani-Dashtaki, S., Mohammadi, J., Asadi, H., Asadzadeh, F., 2016.
Spatial variability of soil organic matter using remote sensing data. Catena 145,
118–127.
Morgan, R.P., Quinton, J.N., Smith, R.E., Govers, G., Poesen, J.W.A., Auerswald, K.,
115
Catena 164 (2018) 107–115
Chisci, G., Torri, D., Styczen, M.E., 1998. The European soil erosion model
(EUROSEM): a dynamic approach for predicting sediment transport from fields and
small catchments. Earth Surf. Process. Landf. 23, 527–544.
Nearing, M.A., Foster, G.R., Lane, L.J., Finkner, S.C., 1989. A process-based soil erosion
model for USDA-Water Erosion Prediction Project technology. Trans. ASAE 32,
1587–1593.
Nearing, M.A., Bradford, J.M., Parker, S.C., 1991. Soil detachment by shallow flow at low
slopes. Soil Sci. Soc. Am. J. 55, 339–344.
Nearing, M.A., Norton, L.D., Bulgakov, D.A., Larionov, G.A., 1997. Hydraulics and ero-
sion in eroding rills. Water Resour. Res. 33 (4), 865–876.
Nearing, M.A., Simanton, J.R., Norton, L.D., Bulygin, S.J., Stone, J., 1999. Soil erosion by
surface water flow on a stony semiarid hillslope. Earth Surf. Process. Landf. 24,
677–686.
Nelson, R.E., 1982. Carbonate and gypsum. In: Page, A.L., Miller, R.H., Keeney, D.R.
(Eds.), Methods of Soil Analysis. Part 2, second ed. Agronomy Monograph 9. ASA,
Madison, WI, pp. 181–197.
Nelson, D.W., Sommers, L.P., 1986. Total carbon, organic carbon and organic matter. In:
Page, A.L. (Ed.), Methods of Soil Analysis: Part 2: Agronomy Handbook No 9.
American Society of Agronomy and Soil Science Society of America, Madison, WI, pp.
539–579.
Ostovari, Y., Ghorbani-Dashtaki, S., Bahrami, H.A., Naderi, M., Dematte, J.A.M., Kerry,
R., 2016. Modification of the USLE K factor for soil erodibility assessment on cal-
careous soils in Iran. Geomorphology 273, 385–395.
Peng, W., Zhang, Z., Zhang, K., 2015. Hydrodynamic characteristics of rill flow on steep
slopes. Hydrol. Process. 29, 3677–3686.
Takken, I., Govers, G., Ciesiolka, C.A.A., Silburn, D.M., Loch, R.J., 1998. Factors influ-
encing the velocity–discharge relationship in rills. In: Proceedings of the
International Symposium on Modelling Soil Erosion, Sediment Transport and Closely
Related Hydrological Processes, Vienna. 249. pp. 63–70.
Torri, D., Ciampalini, R., Accolti, G.P., 1998. The role of aggregates on soil erosion
processes. Presented at the “Global Changes: Modeling Soil Erosion by Water”
conference. NATO ARW, Oxford.
Torri, D., Poesen, J., Borselli, L., Knapen, A., 2006. Channel width flow discharge re-
lationships for rills and gullies. Geomorphology 76, 273–279.
USDA (Ed.), 2010. Keys to Soil Taxonomy, 11th ed. USDA National Resources
Conservation Service, Washington, DC.
Wagenbrenner, J.W., Robichaud, P.R., Elliot, W.J., 2010. Rill erosion in natural and
disturbed forests: 2. Modeling approaches. Water Resour. Res. 46, 1–12.
Wang, D., Wang, Z., Shen, N., Chen, H., 2016. Modeling soil detachment capacity by rill
flow using hydraulic parameters. J. Hydrol. 535, 473–479.
Yang, C.T., 1972. Unit stream power and sediment transport. J. Hydraul. Div. ASCE 98,
1805–1826.
Zhang, G.H., Liu, B.Y., Liu, G.B., He, X.W., Nearing, M.A., 2003. Detachment of un-
disturbed soil by shallow flow. Soil Sci. Soc. Am. J. 67 (3), 713–719.