Professional Documents
Culture Documents
Catena
journal homepage: www.elsevier.com/locate/catena
A R T I C L E I N F O A B S T R A C T
Keywords: Soil sodicity is best evaluated by the exchangeable sodium percentage (ESP); however, the determination of this
Exchangeable sodium percentage index is laborious and time consuming. Alternatively, the sodium adsorption ratio (SAR) is a simpler index that is
Sodium adsorption ratio commonly used to estimate soil sodicity. The objective of this research is to estimate ESP using four approaches:
Soil sodicity
(1) SAR of saturated paste (SARe), and SAR of 1:5 extracts (SAR1:5), (2) a conversion factor (CF) as a function of
Generalized regression neural networks
saturation percentage (θSP), (3) electrical conductivity of 1:5 extracts (EC1:5), and (4) Generalized Regression
Neural Networks (GRNN). Approximately 120 surface soil samples were collected from the Jordan Valley region
and ESP, SARe, SAR1:5, (θSP), soil texture, and soil hydraulic conductivity (HC) were determined. The GRNN
model (i.e., Approach 4) gave the most accurate estimates for the ESP and was able to handle the hetero
scedasticity of the data. Meanwhile the traditional dilution extracts (1) showed that soil ESP was highly related to
SARe and to SAR1:5; the CF- θSP approach (2) gave better estimates for prediction of ESP. Moreover, EC1:5 (3) gave
reasonably accurate estimation of ESP and could be used as a screening test for assessment of sodicity problems.
For the case study site investigations, a reduction of 20% in soil HC was observed when SARe increased from 0 to
3.5 or ESP increased from 0 to 6, indicating that this reduction occurred at ECe < 3 dS m− 1 for all soils. While the
θSP approach reduced the effect of heteroscedasticity of the data on the predictive model ability, the GRNN
models can accurately predict the ESP based on easy-to-obtain soil features. Our models represent a rapid and
accurate estimator of soil sodicity, and therefore offer a potentially valuable tool in managing soil landscapes
that are vulnerable to degradation.
* Corresponding authors.
E-mail addresses: mamoun@just.edu (M.A. Gharaibeh), a.elhanandeh@griffith.edu.au (A. El Hanandeh).
https://doi.org/10.1016/j.catena.2021.105466
Received 23 February 2021; Received in revised form 6 April 2021; Accepted 15 May 2021
Available online 26 May 2021
0341-8162/© 2021 Elsevier B.V. All rights reserved.
M.A. Gharaibeh et al. Catena 205 (2021) 105466
other words, the flocculating effect of salinity can offset the damaging In 1954, the United States Salinity Laboratory tested 59 surface soils in
effects of high sodium levels in soil (Levy and Shainberg, 2005). The nine western states in the United States and developed a widely used
earlier pioneering work by Quirk and Schofield (1955) introduced the relation between SAR and exchangeable sodium ratio: ESR = b + K6 ×
concept of threshold electrolyte concentration (TEC). The TEC is defined SAR, where b is the intercept and KG is Gapon’s coefficient. For most
as the EC of soil solution above which soil physical properties at a given tested soils KG is assumed to equal 0.01475. ESR is related to ESP by the
SAR-ESP are not affected by sodicity. following equation: ESP = [ESR/(1 + ESR)] × 100 (Richards, 1954).
Based on the latter, several studies proposed relating soil ESP-SAR Moreover, Suarez and Jurinak (2011) reported the following relation
and EC to either clay dispersivity or changes in hydraulic conductiv Ấ
ity, and identifying the TEC to maintain soil structural stability. More based on the modified Gapon equation: ESP/(100 − ESP) = kG * SARe,
Ấ
over, soil HC at high SAR values could be maintained provided that the where kG is the modified Gapon selectivity coefficient. The empirical
EC was sufficiently high. The TEC is defined as the concentration at equation relating SAR of soil saturation extracts to ESP of the soil
which either a 10% (Quirk and Schofield, 1955) or 20% (Arienzo et al., developed by the U.S. Salinity Laboratory (Richards, 1954) is shown in
2012; Dang et al., 2018; Ezlit et al., 2013) reduction in soil HC is Table 1. Many relationships between SAR and ESP for different soils
maintained, without any further soil structural degradation. around the world have been reported (Table 1). Other relations were
In addition, elevated Na+ levels in soil solution complicates the also reported for different clay contents, mineralogy, saturation per
leaching of salts (salinity) from the soil profile, reducing plant available centages, and different equilibrating solutions of varying SAR and or
water, root penetration and seedling emergence and increasing osmotic Na/Ca ratios (Endo et al., 2002; Nadler and Magaritz, 1981; Sumner,
stress and causing specific ion effects (nutritional imbalances) on plants, 1993; Toze, 2006).
leading to reduction in crop quality and yield (Marschner, 2011; Naidu Paliwal and Gandhi (1976) found this relation (SAR-ESP) was highly
and Rengasamy, 1993; Qadir et al., 2014, 2006; Qadir and Schubert, dependent on soil texture, and inversely related to Ca:Mg ratio. Un
2002). derestimation of ESP was reported in saline soils (Bower and Hatcher,
Over the past few decades, the availability of fresh water for irriga 1962; Richards, 1954), soils containing sodium carbonates and zeolites
tion has declined to alarming levels especially in arid regions (Elliott (Gupta et al., 1985; Nadler and Magaritz, 1981), and soils containing
et al., 2014). These shortages of fresh water may force many countries to smectites and vermiculites (Juang et al., 2001; Shen et al., 1997).
use poor quality irrigation water such as brackish and treated waste Dilution ratio (soil:water) was also reported to affect the SAR-ESP
water, which could lead to sodification/salinization and loss of pro relation in soils of Almeria, Spain (Faulkner et al., 2001). Further
ductive lands. more, the methods that rely on SAR analysis require the use of chemical
Globally, nearly 10% of the total land is affected by soil salinity and/ analysis which are expensive and involve specific technical expertise
or sodicity (Rengasamy, 2016; Szabolcs, 1989). Hence, it is essential to which makes them difficult to implement in developing countries where
collect and evaluate related information on the levels of salts in soils for farmers usually lack access to reliable chemical laboratories.
successful soil management. The best indicator of soil sodicity is the Therefore, simpler methods to estimate soil ESP are needed. These
exchangeable sodium percentage (ESP). The ESP is usually defined as methods may be used as screening techniques to identify potential
the ratio of exchangeable sodium (Exch Na) to the cation exchange ca ‘hotspots’ for further investigation using more advanced approaches.
pacity of the soil (CEC), both expressed in (cmolc kg− 1) (Richards, 1954). Measurements such electrical conductivity of saturated paste and soil:
ExchNa water extracts may serve as intermediate indicators for the estimation of
CEC *
ESP = 100 ESP, as these values are good indicators of the exchangeable ions in the
soil.
The CEC is defined as the sum of exchangeable cations (Ca2+, Mg2+,
This research aims to investigate the relationship between SAR-ESP
Na , and K+), and determination of exchangeable cations is useful since
+
in soils using four different relationships: (1) (conversion factors) and
they markedly influence the physicochemical properties of soils. To
the ESP and the SARe of saturated paste (SARe) as well as soil–water
calculate ESP; both CEC and exchangeable Na must be determined.
extraction ratios (1:5), (2) ESP and saturation percentage (θSP), (3) ESP
Traditional methods used to measure both parameters are laborious,
and EC of diluted extracts, and (4) factors – namely texture and EC –
time-consuming, and subjected to difficulty (error) especially when soil
developed into a Generalized Regression (GRNN) model. The ap
is saline and calcareous. The difficulty comes from the fact that these
proaches were then evaluated for their ability to predict hydraulic
soils have low permeability to aqueous extracting solutions
conductivity (HC). These models will help land managers to undertake
For the above reasons, a simpler and less labor-time-intensive index
the required amendments needed for reclamation of these soils, improve
for sodicity evaluation – the sodium adsorption ratio (SAR) – is more
sodicity management practices, and develop soil quality monitoring
appropriate and economical than the ESP index (Bower and Hatcher,
programs.
1962). The SAR is calculated using the following equation (concentra
To demonstrate the above mentioned concepts, the case of the Jor
tions are in meq L− 1):
dan Valley (JV) is used. Irrigation with saline-sodic water (i.e. Treated
Na wastewater-TWW) has been practiced in the JV since early 1980s. It is
SAR = √̅̅̅̅̅̅̅̅̅̅̅
Ca+Mg reported that salinity and sodicity related problems in the JV are likely
to increase in the future as a result of intensive farming practices,
2
The SAR is considered an approximation of ESP within the common excessive use of fertilizers and the use of poor quality irrigation water
ranges found in agricultural soils (ESP = 0–40) (Qadir et al., 2014, 2006; (Gharaibeh et al., 2021, 2016).
Qadir and Schubert, 2002). However, ESP is a measure for soil sodicity, Moreover, questions have been raised over the current approach by
while SAR a measure of sodicity of the soil solution. Irrigating soils with which ESP is determined in the case study site – where SAR values are
specific SAR water, for relatively long periods, yields proportionate used to estimate sodicity via the USDA equation. We interrogate the
relationship between the ESP and SAR. Therefore, ESP could simply be validity of this approach, taking the opportunity to use our developed
estimated from SAR using a linear regression model. models in an attempt to shed light on some anomalous results generated
The SAR of soil solutions can be measured using either saturation by the currently adopted USDA approach.
paste extracts (SARe) or soil:water extracts (e.g. 1:5). For example, in
Australia the SAR of 1:5 soil:water extract (SAR1:5) is the most
commonly used measure of soil sodicity (Rengasamy et al., 1984), while
in the US the SAR is commonly measured using saturated paste (SARe).
2
M.A. Gharaibeh et al. Catena 205 (2021) 105466
Table 1
Regression equations to convert SAR & SAR1:5 to ESP – Previous relationships reported between SAR of soil–water extracts and ESP.
Reference Regression equation ESP range Country/region
ESP: Exchangeable sodium percentage, SAR: sodium adsorption ratio (SAR is the saturated paste extract - SARe).
2. Materials and methods in a mechanical shaker for 1 h, then filtered. Collected solutions from
saturated pastes and 1:5 extracts were analyzed for EC and cations (Na,
2.1. Study area and soil sampling Ca, Mg, and K). Na, Ca, K were analyzed using flame photometer, and
Mg by titration method. SAR and ESP were then calculated using the
The study area, the Jordan Valley, has an arid climate [(32◦ 19′ equations described in the Introduction.
32.45′ ’ N, 35◦ 33′ 21.47′ ’ E), (31◦ 46′ 49.45′ ’ N, 35◦ 32′ 47.76′ ’ E)],
characterized by hot dry months during summer and mild low rainfall 2.3. Hydraulic conductivity
during winter with average annual temperature of 19 ◦ C. The majority of
soils of the JV are Aridisols (Typic Camborthids, Typic Calciorthids) Hydraulic conductivity measurements were determined using a mini
(Lucke et al., 2016). The southern parts of the JV are characterized by disk infiltrometer (MDI) (Decagon Devices, Inc., Pullman, WA, USA)
higher temperatures and lower precipitation; moving northward the filled with deionized water. After MDI was in full contact with the soil
average temperature decreases and precipitation progressively in surface, water started to pass into the soil.
creases. Average annual rainfall increases from <50 mm in the southern 3 cm suction was used as the optimal suction setting for the soil used
parts to 350–400 mm in the northern parts. The warm winter enables in this experiment. The drop of water level was recorded manually every
farmers in the JV to produce certain crops year around. 30 sec. until steady-state conditions were reached in two sequential
The southern parts are mainly planted with banana, palm trees, and measurements.
vegetables. The middle parts mainly grow vegetables (greenhouses and Hydraulic conductivity (K) can be calculated as:
open fields), while citrus trees are mainly found in the northern parts
(vegetables are also grown). Crops in the southern and middle regions C1
K=
receive blended water (TWW and fresh water), while the northern parts A
are irrigated with fresh water. The wastewater is treated in Khirbit As- C1 is the slope of the curve of the cumulative infiltration (I) versus
√̅̅
Samra and collected in King Talal Dam (mainly runoff water and the square root of time ( t ) and A can be calculated using the following
TWW) and then mixed with King Abdullah Canal (KAC) water (fresh equation (Zhang, 1997):
water).
A total of 120 composite surface (0–25 cm) soil samples were 11.65(n0.1 − 1)exp[2.92(n − 1.9)αh]
A= for n ≥ 1.9
randomly collected from agricultural areas along the Jordan River (αr)0.91
(Jordan Valley) spanning about 70 km long north of the Dead Sea and
covering a total area of about 300 km2 (Fig. 1). These soils are under A=
11.65(n0.1 − 1)exp[7.5(n − 1.9)αh]
for n < 1.9
intensive agricultural use. More details regarding sampling locations, (αr)0.91
crops, and irrigation water types are described in Gharaibeh et al.
(2020). where n and α are the van Genuchten parameters for the soil used, r is
the disk radius, and h is the suction used in the experiment.
2.2. Chemical analysis (Determination of CEC, soluble and exchangeable 2.4. Statistical analysis
cations)
Descriptive statistics (mean, standard deviation (SD), coefficient of
Cation exchange capacity (CEC) was determined using the ammo variation (CV), minimum, maximum, skewness, lower quartile (Q1),
nium acetate method (Polemio and Rhoades, 1977). Soil exchangeable median (Q2), and upper quartile (Q3) were used to describe SARe,
cations were determined as the difference in concentration between SAR1:5, ESP, and ECe for the collected soil samples. The Kolmogorov-
total extractable and soil solution cations. Extractable cations were Smirnov (K-S) with Lilliefors correction and Shapiro-Wilk tests were
determined using the standard ammonium acetate method (Richards, applied to evaluate the normality of the data using SigmaPlot 12.3.
1954). Soluble cations were determined from 300 g of air–dried soil Pearson correlation was conducted to determine the correlation be
using saturated soil paste extracts. Pastes were made by adding distilled tween major soil measured parameters (see supplementary materials for
water to the soil sample until saturation was reached, allowed to stand data).
overnight (criteria for saturation was rechecked after 1 h). Saturated Statistical models development
pastes were transferred to plastic cups fitted with Whatman filter paper,
and placed in a Centurion mechanical vacuum extractor(Centurion In 2.4.1. Calculating ESP from diluted extract with conversion factor (CF)
ternational, Inc., Lincoln, NE, USA) to obtain solutions. For the 1:5 soil: Traditionally, ESP is estimated through a simple linear regression
water ratio, 250 mL distilled water was added to 50 g of soil and agitated which correlates the diluted extract SAR (SARe or SAR1:5) to ESP. The
3
M.A. Gharaibeh et al. Catena 205 (2021) 105466
general equation can be expressed as ESP = CF*SARe + C, or ESP = 2.4.2. Estimation of ESP from the EC
CF*SAR1:5 + C; The ESP values of the soil may be estimated from the ECe and EC1:5
CF is the slope of the regression line; SARe is the saturated paste SAR, extracts by applying linear regression to estimate a conversion factor
SAR1:5 is the diluted extract (1:5), and C is the intercept. From a strict (CFEC), through the equation ESP = CFEC1:5*EC1:5 + C; where C is a
mathematical stand, the intercept should be 0. Therefore, the equation is constant.
reduced to ESP = CF*SARe, or ESP = CF*SAR1:5.
4
M.A. Gharaibeh et al. Catena 205 (2021) 105466
k
CF = m + ; 3.1. Physico-chemical characterization of studied soils
θSP
where m and k are regression coefficients. Summary statistics of saturated paste SAR (SARe), SAR of 1:5 extracts
(SAR1:5), exchangeable sodium percentage (ESP), and salinity of satu
2.4.4. Generalized Regression Neural Networks (GRNN) rated paste (ECe) are shown in Table 2.
The majority of studied soils have loam (30%) or clay loam (24%)
Generalized Regression Neural Networks belong to the Radial Basis
group of predictive models. They were first described by Specht in 1991 textures. Other textures are also present; clay (8%), silt loam (9%), and
sandy loam (12%). These soils varied considerably in texture and in
(Specht, 1991). The GRNN can be used for any regression problem
saturation percentage (SP) where about 41% of soils have SP% values >
where linearity assumption may not be justified. The GRNN is a one pass
45% and 30% ranged between 25 and 36%. Therefore, the great ma
network that estimates the output by converging to the underlying
jority of the soil samples correspond to medium to fine soil textures.
(linear or nonlinear) regression surface and may be used in problems
About 82% of the collected soil samples are irrigated with TWW, out of
that involve sparse multidimensional data (Specht, 1991).
which 75% have > 20% clay and 43% had clay contents > 30%. On the
The GRNN links the target values (Y) to the input values (X) by
estimating a continuous probability distribution function f(x,y). There other hand, 77% of FW irrigated soils have clay contents < 25% out of
which 73% have clay content between 24 and 26%.
fore, the output values may be thought of as the most likelihood
outcome of the input values. The associations are created by assigning Average values of SARe, SAR1:5, ESP, and ECe of FW were signifi
cantly higher than TWW irrigated soils (P < 0.05). Average SARe of all,
weights, these weights are weakened as the output value of the model
deviates away from the actual target. To estimate the distribution TWW, and FW irrigated soils was 9.6, 8.0, and 17.0, respectively.
Average ESP and SARe of FW were two-fold higher than TWW soils.
function, the following equations are used:
About 37% TWW soils had an ESP > 10 and 21% > 15, while 77% of FW
( )
∑n i D2i soils had an ESP > 10 and 60% had ESP value > 15. Moreover, soils with
Y exp −
SARe > 5 and ECe > 8 dS m− 1 were 95% for FW and 61% for TWW
i=1 2σ 2
̂ (X) =
Y ( )
∑n D2i irrigated soils. In addition, 34% of TWW soils had salinity values of < 4
i=1 exp − 2σ 2 dS m− 1, and only 7% were < 2 dS m− 1 (salinity threshold for sensitive
crops).
( )T ( )
D2i = X − X i X − X i Our findings are in general agreement with previous studies which
reported that ESP was correlated with SARe using different correlation
Table 2
Descriptive statistics of sodium adsorption ratio of saturated paste (SARe) and 1:5 extracts (SAR1:5), exchangeable sodium percentage (ESP), and salinity of saturated
paste (ECe) of all, treated wastewater (TWW), and fresh water (FW) irrigated soils.
Avg sd cv max min skew Q1 Q2 Q3
All
SARe 9.62 8.86 92.17 45.13 0.64 2.10 4.18 7.05 10.37
SAR1:5 3.73 2.19 58.75 12.84 0.30 1.66 2.29 3.44 4.39
ESP 14.50 16.40 113.10 89.80 0.10 2.61 5.28 8.58 15.57
ECe 17.11 22.35 130.61 134.40 1.45 2.56 3.93 7.03 20.50
TWW
SARe 8.02 7.63 95.13 40.06 0.64 2.73 3.84 6.29 8.79
SAR1:5 3.28 1.73 52.93 11.18 0.30 1.57 2.09 3.22 4.06
ESP 12.10 14.43 119.19 87.57 0.10 3.09 4.55 8.08 13.70
ECe 11.96 16.33 136.52 101.90 1.45 3.12 3.55 5.81 11.55
FW
SARe 17.05 10.51 61.66 45.13 4.34 1.14 9.54 14.16 20.65
SAR1:5 4.99 2.47 49.47 12.84 0.84 1.43 3.48 4.74 6.12
ESP 25.46 20.44 80.28 89.80 5.46 1.87 12.68 25.00 30.16
ECe 40.65 30.35 74.65 134.40 7.03 1.62 20.10 36.20 50.50
Avg: average, sd: standard deviation, cv: coefficient of variation, max: maximum value, min: minimum value, skew: skewness, Q1: lower quartile, middle quartile
(median), Q3: upper quartile, ECe (dS m− 1).
5
M.A. Gharaibeh et al. Catena 205 (2021) 105466
factors (Chi et al., 2011; Mohamed et al., 2008; Seilsepour et al., 2009) The ESP of sandy loam and silt loam (38.9 ± 31.3) soils were higher
other than the ones reported by Frenkel and Alperovitch (1984), Paliwal (but not statistically different) than loam (20.3 ± 10.8) soils in FW
and Gandhi (1976), and Richards (1954). irrigated soils. For TWW soils, clay (14.3 ± 7.6) soils had significantly
higher ESP than clay loam and loam soils (7.9 ± 6.9). Sandy loam and
silt loam soils had higher ESP but not significantly different form other
3.2. Sodicity levels in soils under different agricultural land uses, textures in both FW and TWW soils. For SARe values, FW sandy loam and
irrigation source, and main textural classes silt loam soils (22.0 ± 14.5) were significantly higher than TWW clay
(5.8 ± 2.3), clay loam, and loam soils (6.0 ± 3.4).
Soil sodicity (ESP, SARe) ranges and salinity levels (ECe) grouped
according to water source (FW and TWW), land use (tree vegetable) and
major soil textures are presented in box-whisker plots (Fig. 2). Kruskal- 3.3. Calculating ESP from SAR of saturated extracts (SARe), SAR of 1:5
Wallis One Way Analysis of Variance was used to determine significant extracts (SAR1:5), CF-θSP, and EC1:5
differences in sodicity values between different groups (P < 0.05). Soils
irrigated with FW had significantly higher ESP (25.5 ± 20.4), SARe A simple linear regression model was used to correlate measured ESP
(17.1 ± 10.5), and ECe (40.7 ± 30.4) than TWW irrigated soils -(12.1 ± to SARe and SAR1:5. The relationship between ESP-SARe and ESP-SAR1:5
14.4), (8.0 ± 7.6), and (12.0 ± 16.3), respectively. Moreover, FW- are presented in Fig. 3a (top and middle). In addition, the relationship
irrigated vegetable planted soils (30.1 ± 23.2) had significantly higher between SARe vs SAR1:5 is also shown in Fig. 3a (bottom). Highly sig
ESP than FW irrigated tree-planted soils (16.1 ± 8.3), while no signifi nificant linear relationships were obtained, the coefficients of determi
cant differences were observed between land uses in TWW soils. nation (R2) and regression equations are shown in Fig. 3a.
Fig. 2. Left: Exchangeable sodium percentages (ESP) of soils in FW vs TWW: all soils (top), tree vs vegetable cultivated soils (middle), major soil textures (bottom),
Middle: Sodium adsorption ratio (SARe) of soils in FW vs TWW: all soils (top), tree vs vegetable cultivated soils (middle), major soil textures (bottom), Right: salinity
of saturated paste (ECe) of soils in FW vs TWW: all soils (top), tree vs vegetable cultivated soils (middle), major soil textures (bottom).
6
M.A. Gharaibeh et al. Catena 205 (2021) 105466
( )
125.97
ESP = 0.888 +
θSP *SAR1:5
Table 3
Conversion factors of: (1) Sodium adsorption ratio (SAR) of saturated paste
extract (SARe) to exchangeable sodium percentage (ESP), (2) SAR of 1:5 extracts
SAR1:5 to ESP, and (3) SAR1:5 to SARe for different soil textures using measured
values.
Texture CF (SARe R2 CF (SAR1:5 R2 CF (SAR1:5 R2
to ESP) to ESP) to SARe)
Fig. 3a. Relationship between: (1) Saturated paste SAR (SARe) and measured All 1.69 0.84 5.09 0.71 2.71 0.81
ESP using traditional approach (open circles), SARe and predicted ESP using Clay 2.55 0.76 4.25 0.68 1.66 0.62
(θSP) approach (filled circles) (top), (2) SAR of 1:5 extracts (SAR1:5) and Clay Loam 1.41 0.72 2.42 0.70 2.00 0.71
measured ESP using traditional approach (open circles), SAR1:5 and predicted Loam 1.29 0.82 3.67 0.64 3.01 0.74
ESP using (θSP) approach (filled circles) (middle), and (3) SAR1:5 and measured Sandy Clay 1.35 0.78 3.72 0.76 2.70 0.81
SARe using traditional approach (open circles) and predicted SARe from SAR1:5 Loam
using (θSP) approach (filled circles) (bottom). Sandy 1.67 0.92 5.64 0.71 3.26 0.76
Loam
Silt Loam 1.98 0.94 6.59 0.80 2.82 0.82
Silty Clay 2.08 0.90 3.85 0.70 1.80 0.73
Loam
7
M.A. Gharaibeh et al. Catena 205 (2021) 105466
Fig. 3b. Relationship between EC of 1:5 extracts (EC1:5) and exchangeable Fig. 4b. Hydraulic conductivity (HC mm hr− 1) of all soils as a function of
sodium percentage (ESP). exchangeable sodium percentage (ESP). Dashed lines represent 20% reduction
in HC occurring from ESP = 2 to7.
8
M.A. Gharaibeh et al. Catena 205 (2021) 105466
Fig. 5a. Schematic view of the GRNN model constructed for the prediction of ESP and HC. Input = [Sand Clay EC1:5], Output = [ ESP HC]. W: weight and b: bias for
each neuron.
4. Conclusion
Soil ESP was estimated using four modeling approaches (1. tradi
tional extracts, 2. saturation percentage, 3. EC1:5, and 4. GRNN).
Incorporation of saturation percentage (θSP) in the estimation of the
conversion factors (CF) significantly improved the prediction of ESP of
different soil textures using saturated paste and diluted extracts over the
traditional CF approach. The Generalized Regression Neural Networks
(GRNN) approach offered accurate estimation for ESP as it considers soil
texture which improves the prediction. GRNN was also capable of pre
Fig. 5b. Actual vs predicted ESP values using GRNN network. dicting sodicity effects on soil hydraulic conductivity (HC). In the case
study area investigated here (the Jordan Valley), soil HC was reduced by
20% when ESP increased from 0 to 7?, highlighting critical impacts that
sodicity can exert on soil structure and function. Although the models
presented in this work relate to a regional case study area, they are based
on an extensive and diverse soil dataset. Hence, they represent a
promising tool to assist soil management particularly in the context of
arid landscapes.
Acknowledgement
57% of the variability in HC. Additionally, the inclusion of clay content Arienzo, M., Christen, E.W., Jayawardane, N.S., Quayle, W.C., 2012. The relative effects
of the soil improves the fit further (R2 = 0.996 and 0.961 for ESP and HC, of sodium and potassium on soil hydraulic conductivity and implications for winery
wastewater management. Geoderma 173-174, 303–310. https://doi.org/10.1016/j.
respectively). The network topography is shown in Fig. 5a and the
geoderma.2011.12.012.
predictions of the network are shown in Figs. 5b and 5c. The network Bardhan, G., Russo, D., Goldstein, D., Levy, G.J., 2016. Changes in the hydraulic
shown in Fig. 5a is composed of three layers: an input layer with a properties of a clay soil under long-term irrigation with treated wastewater.
neuron corresponding to each input parameters (in this case: sand%; Geoderma 264, 1–9. https://doi.org/10.1016/j.geoderma.2015.10.004.
Bower, C.A., Hatcher, J.T., 1962. Characterization of saltaffected soils with respect to
clay% and EC1:5); a hidden layer with 120 neurons, one corresponding to sodium. Soil Sci. 93 (4), 275–280.
each observed record and using a Gaussian transfer function and finally Chasset, P.O., 2013. GRNN: General regression neural network for the statistical software
R.
9
M.A. Gharaibeh et al. Catena 205 (2021) 105466
CHI, C.-M., ZHAO, C.-W., SUN, X.-J., WANG, Z.-C., 2011. Estimating exchangeable McNeal, B.L., Coleman, N.T., 1966. Effect of Solution Composition on Soil Hydraulic
sodium percentage from sodium adsorption ratio of salt-affected soil in the songnen Conductivity. Soil Sci. Soc. Am. J. 30 (3), 308–312. https://doi.org/10.2136/
plain of Northeast China. Pedosphere 21 (2), 271–276. https://doi.org/10.1016/ sssaj1966.03615995003000030007x.
S1002-0160(11)60127-6. McNeal, B.L., Norvell, W.A., Coleman, N.T., 1966. Effect of Solution Composition on the
CRAN, 2017. Collinearity Diagnostics, Model Fit & Variable Contribution. Tools for Swelling of Extracted Soil Clays. Soil Sci. Soc. Am. J. 30 (3), 313–317. https://doi.
Building OLS Regression Models. org/10.2136/sssaj1966.03615995003000030008x.
Dang, A., Bennett, J.M., Marchuk, A., Biggs, A., Raine, S.R., 2018. Quantifying the Mohamed, D.M., Ibrahim, S.I., Elamin, E.A., 2008. variability and correlation between
aggregation-dispersion boundary condition in terms of saturated hydraulic exchangeable sodium percentage and sodium adsorption ratio in vertisols of Sudan.
conductivity reduction and the threshold electrolyte concentration. Agric. Water Commun. Soil Sci. Plant Anal. 39 (19-20), 2827–2838. https://doi.org/10.1080/
Manag. 203, 172–178. https://doi.org/10.1016/j.agwat.2018.03.005. 00103620802432758.
Elliott, J., Deryng, D., Müller, C., Frieler, K., Konzmann, M., Gerten, D., Glotter, M., Nadler, A., Magaritz, M., 1981. Expected deviations from the ESP-SAR empirical-
Flörke, M., Wada, Y., Best, N., Eisner, S., Fekete, B.M., Folberth, C., Foster, I., relationships in calcium- and sodium-carbonate-containing arid soils: Field evidence.
Gosling, S.N., Haddeland, I., Khabarov, N., Ludwig, F., Masaki, Y., Olin, S., Soil Sci. 131 (4), 220–225. https://doi.org/10.1097/00010694-198104000-00005.
Rosenzweig, C., Ruane, A.C., Satoh, Y., Schmid, E., Stacke, T., Tang, Q., Wisser, D., Naidu, R., Rengasamy, P., 1993. Ion interactions and constraints to plant nutrition in
2014. Constraints and potentials of future irrigation water availability on australian sodic soils. Aust. J. Soil Res. 31, 801–819. https://doi.org/10.1071/
agricultural production under climate change. PNAS 111 (9), 3239–3244. https:// SR9930801.
doi.org/10.1073/pnas.1222474110. NeuPy, 2019. NeuPy Neural Networks in Python [WWW Document]. URL http://neupy.
Endo, T., Yamamoto, S., Honna, T., Eneji, A.E., 2002. Sodium-calcium exchange com/apidocs/neupy.algorithms.rbfn.grnn.html.
selectivity as influenced by clay minerals and composition. Soil Sci. 167 (2), Northcote, K.H., Skene, J.K.M., 1972. Australian Soils with Saline and Sodic Properties.
117–125. https://doi.org/10.1097/00010694-200202000-00004. Australia CSIRO soil Publication. Commonwealth Scientific and Industrial Research
Ezlit, Y.D., Bennett, J.M., Raine, S.R., Smith, R.J., 2013. Modification of the McNeal Clay Organization.
Swelling Model Improves Prediction of Saturated Hydraulic Conductivity as a Paliwal, K.V., Gandhi, A.P., 1976. Effect of salinity, SAR, Ca: Mg ratio in irrigation water,
Function of Applied Water Quality. Soil Sci. Soc. Am. J. 77 (6), 2149–2156. https:// and soil texture on the predictability of exchangeable sodium percentage. Soil Sci.
doi.org/10.2136/sssaj2013.03.0097. 122, 85–90.
Faulkner, H., Wilson, B.R., Solman, K., Alexander, R., 2001. Comparison of three cation Polemio, M., Rhoades, J.D., 1977. Determining Cation Exchange Capacity: A New
extraction methods and their use in determination of sodium adsorption ratios of Procedure for Calcareous and Gypsiferous Soils. Soil Sci. Soc. Am. J. 41 (3),
some sodic soils. Commun. Soil Sci. Plant Anal. 32 (11-12), 1765–1777. https://doi. 524–528. https://doi.org/10.2136/sssaj1977.03615995004100030018x.
org/10.1081/CSS-120000248. Qadir, M., Noble, A.D., Schubert, S., Thomas, R.J., Arslan, A., 2006. Sodicity-induced
Frenkel, H., Alperovitch, N., 1984. The effect of mineral weathering and soil solution land degradation and its sustainable management: problems and prospects. Land
concentration on ESR—SAR relationships of arid and semi-arid zone soils from Degrad. Dev. 17 (6), 661–676. https://doi.org/10.1002/(ISSN)1099-145X10.1002/
Israel. J. Soil Sci. 35, 367–372. https://doi.org/10.1111/j.1365-2389.1984.tb00292. ldr.v17:610.1002/ldr.751.
x. Qadir, M., Quillérou, E., Nangia, V., Murtaza, G., Singh, M., Thomas, R.J., Drechsel, P.,
Frenkel, H., Goertzen, J.O., Rhoades, J.D., 1978. Effects of Clay Type and Content, Noble, A.D., 2014. Economics of salt-induced land degradation and restoration. Nat.
Exchangeable Sodium Percentage, and Electrolyte Concentration on Clay Dispersion Resour. Forum 38 (4), 282–295. https://doi.org/10.1111/1477-8947.12054.
and Soil Hydraulic Conductivity 1. Soil Sci. Soc. Am. J. 42 (1), 32–39. Qadir, M., Schubert, S., 2002. Degradation processes and nutrient constraints in sodic
Ghafoor, A., Muhammed, S., Ahmed, N., Main, M., 1988. Indices for the estimation of soils. Land Degrad. Dev. 13 (4), 275–294. https://doi.org/10.1002/(ISSN)1099-
ESP from SAR of soil solution. Pak. J. Sci. 39–40, 89–98. 145X10.1002/ldr.v13:410.1002/ldr.504.
Gharaibeh, M.A., Albalasmeh, A.A., El Hanandeh, A., 2021. Estimation of saturated paste Quirk, J.P., 2001. The significance of the threshold and turbidity concentrations in
electrical conductivity using three modelling approaches: Traditional dilution relation to sodicity and microstructure. Aust. J. Soil Res. 39 (6), 1185–1217. https://
extracts; saturation percentage and artificial neural networks. CATENA 200, 105141. doi.org/10.1071/SR00050.
https://doi.org/10.1016/j.catena.2020.105141. Quirk, J.P., Schofield, R.K., 1955. The effect of electrolyte concentration on soil
Gharaibeh, M.A., Ghezzehei, T.A., Albalasmeh, A.A., Alghzawi, M.Z., 2016. Alteration of permeability. J. Soil Sci. 6 (2), 163–178. https://doi.org/10.1111/j.1365-
physical and chemical characteristics of clayey soils by irrigation with treated waste 2389.1955.tb00841.x.
water. Geoderma 276, 33–40. https://doi.org/10.1016/j.geoderma.2016.04.011. Rengasamy, P., 2016. Salt-affected soils in Australia. GRDC Grains Research and
Gharaibeh, M.A., Marschner, B., Heinze, S., Moos, N., 2020. Spatial distribution of metals Development Corporation.
in soils under agriculture in the Jordan Valley. Geoderma Regional 20, e00245. Rengasamy, P., Greene, RSB, Ford, G.W., Mehanni, A.H., 1984. Identification of
https://doi.org/10.1016/j.geodrs.2019.e00245. dispersive behaviour and the management of red-brown earths. Aust. J. Soil Res. 22
Gupta, R.K Gupta, R.K., Singh, C.P., Abrol, I.P., 1985. Determining cation exchange (4), 413. https://doi.org/10.1071/SR9840413.
capaclty and exchangeable sodium in alkali soils. Soil Sci. 139. Richards, L.A., 1954. Diagnosis and Improvement of, US Department of Agriculture
Juang, T.C., Wang, M.K., Chen, H.J., Tan, C.C., 2001. Ammonium fixation by surface Handbook, No. 60, Washington, DC.
soils and clays. Soil Sci. 166 (5), 345–352. Seilsepour, M., Rashidi, M., Khabbaz, B.G., 2009. Prediction of soil exchangeable sodium
Lado, M., Bar-Tal, A., Azenkot, A., Assouline, S., Ravina, I., Erner, Y., Fine, P., percentage based on soil sodium adsorption ratio. American-Eurasian J. Agric. &
Dasberg, S., Ben-Hur, M., 2012. Changes in Chemical Properties of Semiarid Soils Environ. Sci. 5 (1), 1–4.
under Long-Term Secondary Treated Wastewater Irrigation. Soil Sci. Soc. Am. J. 76 Shen, S., Tu, S.I., Kemper, W.D., 1997. Equilibrium and kinetic study of ammonium
(4), 1358–1369. https://doi.org/10.2136/sssaj2011.0230. adsorption and fixation in sodium-treated vermiculite. Soil Sci. Soc. Am. J. 61 (6),
Lado, M., Ben-Hur, M., 2010. Effects of Irrigation with Different Effluents on Saturated 1611–1618. https://doi.org/10.2136/sssaj1997.03615995006100060011x.
Hydraulic Conductivity of Arid and semiarid Soils. Soil Sci. Soc. Am. J. 74 (1), Specht, D.F., 1991. A General Regression Neural Network. IEEE Trans. Neural Netw. 2
23–32. https://doi.org/10.2136/sssaj2009.0114. (6), 568–576. https://doi.org/10.1109/72.97934.
Leuther, F., Schlüter, S., Wallach, R., Vogel, H.-J., 2019. Structure and hydraulic Suarez, D.L., Jurinak, J.J., 2011. The Chemistry of Salt-Affected Soils and Waters. In:
properties in soils under long-term irrigation with treated wastewater. Geoderma Wallender, W.W., Tanji, K.K. (Eds.), Agricultural Salinity Assessment and
333, 90–98. https://doi.org/10.1016/j.geoderma.2018.07.015. Management. American Society of Civil Engineers, Reston, VA, pp. 57–88. https://
Levy, G.J., Shainberg, I., 2005. Sodic Soils. In: Encyclopedia of Soils in the Environment. doi.org/10.1061/9780784411698.ch03.
Elsevier, pp. 504–513. https://doi.org/10.1016/B0-12-348530-4/00218-6. Sumner, M.E., 1993. Sodic soils: New perspectives. Aust. J. Soil Res. 31, 683–750.
Lucke, B., Ziadat, F., Taimeh, A., 2016. The Soils of Jordan. Atlas Jordan. https://doi. https://doi.org/10.1071/SR9930683.
org/10.4000/books.ifpo.4867. Szabolcs, I., 1989. Salt-affected soils. CRC Press, Boca Raton, Fla.
Marschner, P., 2011. Marschner’s Mineral Nutrition of Higher Plants, third ed. https://do Toze, S., 2006. Reuse of effluent water - Benefits and risks. Agric. Water Manag. 80 (1-3),
i.org/10.1016/C2009-0-63043-9. 147–159. https://doi.org/10.1016/j.agwat.2005.07.010.
Mathworks, C., 2020. Statistics and Machine Learning ToolboxTM User’s Guide R2020a. Zhang, R., 1997. Determination of Soil Sorptivity and Hydraulic Conductivity from the
MATLAB Mathworks Inc 2020a. Disk Infiltrometer. Soil Sci. Soc. Am. J. 61, 1024–1030. https://doi.org/10.2136/sss
aj1997.03615995006100040005x.
10