1 s2.0 S0341816222002740 Main

Catena 214 (2022) 106288
Contents lists available at ScienceDirect
Catena
journal homepage: www.elsevier.com/locate/catena
Pedological assessments along an arid and semi-arid transect using soil

spectral behavior analysis
Sahar Taghdis, Mohammad Hady Farpoor *, Majid Mahmoodabadi
Department of Soil Science, Faculty of Agriculture, Shahid Bahonar University of Kerman, P. O. Box 76169-14111, Kerman, Iran
A R T I C L E I N F O A B S T R A C T
Keywords: Reflectance spectroscopy as an efficient proximal sensing method, can be employed in soil genesis, monitoring,
Calcareous soils and management studies. However, limited data about the performance of this technique in different pedoen
Pedogenesis vironmental conditions is available. This study was conducted to investigate the capability of VIS-NIR-SWIR in
Proximal soil sensing
detection of soil variations along a transect, monitoring the vertical changes of soil properties in relation to soil-
Soil spectral characterization
VIS-NIR-SWIR spectroscopy
forming factors and processes, and determination of the relationship between soil variables and their effects on
the spectral features. Nine representative pedons were selected on different geomorphic positions including al
luvial fan, piedmont plain, flood plain, and lowland. Soil samples were collected from different diagnostic ho
rizons and analyzed for routine physico-chemical properties, iron forms, clay mineralogy, and VIS-NIR-SWIR
spectroscopy. The results of the spectral interpretations revealed that the differences between soil horizons, clay
mineralogy and the variations in key soil properties can be detected regarding the reflectance intensity, shape
and absorption features. Smectite, chlorite, illite, kaolinite, and palygorskite were identified in the studied soils
based on X-ray diffraction (XRD) analyses and their detection ranges were presented using spectral library and
the second-derivative spectrum analyses. OC, CEC, clay, Fed and ᵡlf were negatively correlated with the reflec
tance wavelengths, while CCE, EC, sand, silt, and the evolution index (Feo/Fed) showed positive correlations. The
most important factors affecting spectral behavior of the studied soils were found as particle size distribution,
clay minerals, iron oxides, organic carbon, gypsum, CaCO3, soluble salts, and soil drainage conditions. Moreover,
quantitative estimation of soil properties using partial least squares regression (PLSR) model showed VIS-NIR-
SWIR to be accurate for estimation of CEC (RPD > 2), and reliable for OC, clay, Fed and ᵡlf (1.4 < RPD < 2).
Overall, the results highlighted the significant role of VIS-NIR-SWIR on pedological assessments through soil
characterization and differentiation of the horizons.
1. Introduction Demattê, 2006; Viscarra Rossel et al., 2006b; Viscarra Rossel et al.,
2016). The diffuse reflectance spectroscopy technique is invaluable to
As an important natural resource, soil plays a vital role in regulation achieve this goal because of its distinct advantages over complicated
of various terrestrial ecosystems on the earth. It has consisted of a very laboratory procedures (Cohen et al., 2007; Silva et al., 2020). This
heterogeneous dynamic environment, involving different complex technique results in high quality and easily available data collections to
mechanisms such as chemical and biological reactions and physical be used in soil and environmental monitoring and assessments (Viscarra
processes which simultaneously occur in the soil. The type, rate, and Rossel et al., 2006b). As one of the proximal sensing methods, soil data
behavior of these mechanisms cause different soil properties to be will be evaluated without any direct contact using the diffuse reflectance
developed. In other words, soil characteristics are a function of different spectroscopy technique (Farifteh et al., 2006). Based on the reflectance
processes which vary at both spatial and temporal scales (Summers curves of the samples, this technique examines the compositional and
et al., 2011). In order to better understand soil properties and processes, structural behavior of the molecules in the visible (VIS- 350–780 nm),
and to overcome the limitations of conventional methods, it is necessary near-infrared (NIR-780–1100 nm) and shortwave infrared (SWIR-
to employ effective measuring and monitoring techniques, especially 1100–2500 nm) wavelengths, because the basic molecular frequencies
modern, simple, nondestructive, and cost-effective ones (Nanni and of the soil constituents mostly exist in these spectral regions (Stoner and
* Corresponding author.
E-mail addresses: staghdis@gmail.com (S. Taghdis), farpoor@uk.ac.ir (M.H. Farpoor), mahmoodabadi@uk.ac.ir (M. Mahmoodabadi).
https://doi.org/10.1016/j.catena.2022.106288
Received 13 November 2021; Received in revised form 27 March 2022; Accepted 4 April 2022
Available online 9 April 2022
0341-8162/© 2022 Elsevier B.V. All rights reserved.
S. Taghdis et al. Catena 214 (2022) 106288
Baumgardner, 1981; Clark, 1999; Demattê, 2002; Ben-Dor et al., 2008; induced alterations and weathering, and pedogenetic processes. In
Stenberg et al., 2010). another research, Demattê and da Silva Terra (2014) stated that soil can
Molecular vibrations and electronic transitions are two general be identified by reflectance spectroscopy through variations in reflec
processes responsible for the absorption features in the soil spectra. tance intensity and absorption features which are affected by different
Following the radiation of the electromagnetic energy with all related weathering rates and soil forming processes. They called this link
frequencies in a specific range to a soil sample, the molecular bonds of “spectral pedology”. Demattê et al. (2017) also studied the genesis and
the soil constituents will be individually vibrated in bending or properties of wetland soils using VIS-NIR-SWIR and reported that dif
stretching states, resulting in various degrees of energy absorption by ferentiation of soil horizons and identification of some pedogenetic
the soil constituents. The quantum of this energy is directly related to the processes could be easily performed by interpretation of soil spectral
frequency and inversely related to the wavelength, and the generated morphologies. Viscarra Rossel et al. (2009) investigated the in situ
absorption spectrum presents a diagnostic morphology that can be determination of soil color, mineralogy, and clay content using VIS-NIR-
specifically used in soil spectral analyses (Miller, 2001; Stenberg et al., SWIR spectroscopy and reported that spectroscopic assessments of soil
2010). In the electronic process, the absorption features occur in the color were in an acceptable consistency with the Munsell soil color
spectra as a result of electron transitions between the energy levels of the charts. In addition, the soil mineral composition was effectively deter
atomic orbitals. This process predominantly occurs in the VIS-NIR mined by VIS-NIR-SWIR, and a reliable correlation was found between
spectral region, whereas the vibrational one occurs mainly in SWIR the results of this method and XRD analyses especially for iron oxides
and mid-IR regions (Coblinski et al., 2020). The analytical performance detection. de Souza et al. (2020) identified the mid-IR spectral patterns
of the spectroscopy technique depends largely on the broad and frequent for characterization of soil taxonomic classes and reported that organic
energy absorption by N-H, C-H, and O-H bonds (Xuemei and Jianshe, matter, sand, clay and iron oxides were of the most effective soil prop
2013). All of the compounds containing covalent bonds, either organic erties impacting its spectral behavior. Their results approved the effi
or inorganic, can absorb different wavelengths in the VIS-NIR-SWIR ciency and capability of the spectro-radiometric method in the
range, but regarding their specific molecular composition, structure, characterization and classification of soils in the Northeast of Brazil.
and properties, they represent unique spectral behaviors (Clark, 1999). Quantitative spectroscopy has already been applied to model soil
Due to broad overlapping and weak vibrational modes in VIS-NIR- properties (Soriano-Disla et al., 2014). However, for rapid and efficient
SWIR region, the interpretation of the soil spectra is usually compli soil identification, it is necessary to use qualitative assessments through
cated (Stenberg et al., 2010). Useful data on soil components can be the investigation of spectral features (Viscarra Rossel et al., 2016; Zeng
found in the visible region, which is primarily associated with iron- et al., 2016). Due to overlapping absorption of soil constituents, the
containing minerals such as goethite, hematite, etc. (Mortimore et al., diffuse reflectance spectra of soils in the VIS-NIR-SWIR region are
2004; Heller Pearlshtien and Ben-Dor, 2020). The broad absorption largely nonspecific (Ogen et al., 2017). For instance, different minerals
bands in this region are also related to the chromatic and dark nature of with similar absorption bands are simultaneously present in the soil,
soil organic matter (SOM) (Heller Pearlshtien and Ben-Dor, 2020). Clark resulting in interferences in their absorption bands. Among them, it can
(1999) attributed the absorption bands in the NIR-SWIR (780–2500 nm) be cited to the similarities and proximities in absorption features of
region to the overtones of OH, SO4, and CO3 groups, as well as combi carbonates with Illite (Khayamim et al., 2015), and the interaction of
nations of fundamental features of H2O and CO2. Clay minerals usually iron oxides with organic carbon (Marques et al., 2019). Moreover, some
represent absorption peaks in the NIR-SWIR region due to metal-OH soil constituents such as gypsum, soluble salts, soil texture, soil color,
bonds plus OH stretching combinations (Viscarra Rossel et al., 2006a). etc, influence and change the position of soil absorption bands by
Carbonates also show weak absorption bands in this region (Hunt and increasing or decreasing its spectral reflectance (Khayamim et al., 2015,
Salisbury, 1971). The soil VIS-NIR-SWIR spectra can be strongly affected Pessoa et al., 2016, Demattê et al., 2017, Marques et al., 2019). There
by water, so the absorption bands around 1400–1900 nm are related to fore, the changes in spectrally active constituents cannot be expected to
soil moisture (hydration, hygroscopic and free water), while weaker be globally stable. On the other hand, vertical and horizontal variations
bands exist in other parts of the VIS-NIR-SWIR region (Weidong et al., in soil properties occur over short distances, making it necessary to
2002). regular monitoring of soil variations (Stenberg et al., 2010; Marques
In recent years, the reflectance spectroscopy technique has been et al., 2019). Although VIS-NIR-SWIR spectra have already been used in
effectively used in soil characterization and monitoring the variations of pedological studies including effective methods for optimizing soil sur
soil properties (Kim et al., 2014; Stenberg et al., 2010), estimation of soil vey, and soil mapping, classification and management, there is still a gap
physico-chemical properties (Araújo et al., 2015; Davari et al., 2021; in the pieces of research about spectral pedology especially in arid and
Minasny et al., 2008; Pinheiro et al., 2017; Viscarra Rossel and Webster, semi-arid regions. About 90% of soils in Iran are arid and/or semi-arid
2012; Seifi et al., 2020; Shahrayini et al., 2020), soil organic carbon soils. However, the spatio-temporal variability of soil properties espe
predictions (Ben-Dor et al., 1997; Daniel et al., 2004; Islam et al., 2003; cially in saline, calcareous, and gypsiferous soils differently influence on
Viscarra Rossel et al., 2006a; Zhang et al., 2020), particle size distri soil spectrum (Weindorf et al., 2016). In fact, these studies are vital for
bution (Coblinski et al., 2020), soil erodibility investigations (Mina environmental concerns including its risk factors, monitoring and
et al., 2021; Ostovari et al., 2018), gypsum and calcium carbonate management plans. That is why the present study was conducted based
equivalent assessments (Asgari et al., 2020; Khayamim et al., 2015; on the following three main objectives: 1) investigating the capability of
Naimi et al., 2021), soil color determination (Ramos et al., 2020; Vis VIS-NIR-SWIR method to detect soil variations along a transect, 2)
carra Rossel et al., 2009), and soil mineralogy (Viscarra Rossel et al., monitoring the vertical changes of soil properties using VIS-NIR-SWIR
2009; Zhao et al., 2020). related to environmental conditions and soil forming processes, and 3)
An important stage in soil behavior analysis is to link pedogenesis determining the relationship between soil properties and investigating
with reflectance spectroscopy (Demattê et al., 2017). A “spectral the extent to which spectral features are affected by soil variations.
signature” is defined as a graphical presentation of reflectance data
along wavelength and is dependent on the inherent characteristics of the 2. Material and methods
objects (Jensen, 2009). Soil physico-chemical and mineralogical char
acteristics, pedogenetic processes, and environmental conditions 2.1. Study area and soil sampling
strongly influence its spectral signature (Demattê et al., 2017). In their
study on soil evaluation by VIS-NIR-SWIR reflectance along nine top The study area is a transect from Ghatrooye to Estahban regions, at
osequences, Demattê et al. (2004) reported this technique to have a the east of Fars Province, Iran. The soil moisture regimes are Aridic, Dry-
great potential for soil classification, detection of parent material Xeric, and soil temperature regime is Thermic (Banaie, 2001). The mean
2
annual precipitation rate varies from 168 mm in Ghatrooye to 226 mm capacity (CEC) measurement (Chapman, 1965). The soil pH and elec
in the Estahban region. Calcareous parent materials with sandstone, trical conductivity (EC) were determined in soil saturation paste and
limestone, marl, shale, and conglomerate are dominated in the study extract, respectively using Jenway pH and EC-meter instruments. Acid
area from the geological point of view (Iran Geology Organization, ammonium oxalate with pH = 3 was used to extract poorly crystalline
1995). Flood plain (pedons 1 and 5), alluvial fan (pedon 2), lowland and organic iron compounds (Feo) in dark conditions (Schwertmann,
(pedon 4), and piedmont plain (pedons 3, 6, 7, 8 and 9) were among 1973). Pedogenic iron oxides (crystalline, amorphous and organic) were
dominant different geomorphic surfaces studied. Overall, a total of 9 determined using citrate-bicarbonate-dithionate (CBD) (Mehra and
representative pedons were excavated, described, sampled and analyzed Jackson, 1958). The Fe concentration in the extracts was determined
(Fig. 1). Soil classification was performed using Soil Taxonomy (Soil using a Vario Atomic Absorption Spectroscopy instrument (Table 2). A
Survey Staff, 2014) and WRB (IUSS Working Group WRB, 2015) sys Bartington MS2 Dual Frequency Sensor magnetometer equipped with
tems. Dominant soil orders in the study area were identified as Incep MS2B sensor was used to measure the magnetic susceptibility (MS) of
tisols, Aridisols, Vertisols, Mollisols, and Alfisols (Soil Survey Staff, the soil samples at 0.46 kHz frequency (ᵡlf). The clay mineralogy ana
2014) and Vertisols, Calcisols, Solonchaks, Kastanozems, Cambisols, lyses of selected soil samples were carried out according to Jackson
and Luvisols. (IUSS Working Group WRB, 2015). (1975) and (Kittrick and Hope, 1963) methods using a D8 ADVANCE X-
ray diffractometer (XRD) with Cu kα target under 40 kV and 30 mA.
2.2. Soil laboratory analyses

2.3. Soil spectral measurements
Laboratory analyses were carried out after air drying, crushing, and
sieving of soil samples through 2 mm mesh size. Standard soil analysis The spectral analyses of the soil samples were carried out in the
methods were used as follows: the pipette method for particle size dis range of 350–2500 nm (VIS-NIR-SWIR) using a field spectro-radiometer
tribution analyses (Gee and Bauder, 1986), HCl back titration method (FieldSpec®3, ASD, FR, USA) in controlled dark conditions. The sam
for calcium carbonate equivalent (CCE) determination (Page et al., pling intervals and spectral resolutions were 1.4 and 3 nm at 350–1000
1982), acetone method for gypsum percentage (Nelson, 1982), wet nm, and 2 and 10 nm at 1000–2500 nm, respectively, and data were
oxidation method for organic carbon (OC) content (Walkley and Black, resampled to 1 nm output values. All of the spectral analyses were
1934), and sodium acetate (NaOAC) with pH = 8.2 for cation exchange performed using a contact probe with 13 mm spot size and a 150 W
Fig. 1. Geographical location of the study area and studied pedons.
3
halogen light source. The soil samples were first air-dried, crushed and and RPIQ < 1.4 refers to very poor models for accurate prediction of
passed through a 2-mm sieve and were then placed in aluminum petri properties (Nawar and Mouazen, 2017).
dishes with 10 cm diameter and 1 cm height. In order to obtain the The Pearson’s correlation coefficient was used to evaluate the matrix
maximum variance on the surfaces of the soil samples, the reflectance of the correlations between measured soil properties and spectral
spectra were recorded at five positions on each sample through main reflectance values. The correlation coefficients between soil properties
taining the contact probe at the successive positions. The instrument was were determined using SPSS version 16.0 and graphs were plotted using
calibrated on every three samples using a spectralon white reference GraphPad Prism software version 6.
panel. The final reflectance spectrum for each soil sample was obtained
from the average of five recorded spectra using View Spec Pro software 3. Results and discussion
version 6.
The interpretation of spectral curves and differentiation of soil ho 3.1. Description of soil properties
rizons were carried out following the seven steps suggested by Demattê
et al. (2014), including the reflectance intensity, morphology of spectral The descriptive statistics of the physicochemical properties and
curves, evaluation of absorption features related to analytical data, magnetic susceptibility of the studied soils are summarized in Table 1.
comparison of spectral curves between horizons of a pedon, quantitative The average values of gypsum and EC in the study area were about 9.8 g
view of spectral data, comparison of spectra with an already known kg− 1 and 84.2 dS m− 1, respectively, and they showed the highest CV
database and finally, comparison between quantitative and qualitative values (200 and 177%) among the measured soil parameters. The pH
data. The second-derivative analyses + Savitsky-Golay filter were per values of the studied soils were in the neutral to the alkaline range and
formed using The Unscrambler X 10.4 software to improve the spectral had the lowest CV (3%). The OC content varied in the range of 0 to 14.5
features of minerals and better description of the spectral morphologies. g kg− 1 with an average value of 5 g kg− 1. Seven soil texture classes
including clay, silty clay, clay loam, silty clay loam, silt loam, loam and
2.4. Data analyses sandy loam were found in the studied pedons. The amounts of CCE were
relatively high in the studied soils (mean 360 g kg− 1 and CV 20%) which
The partial least squares regression (PLSR) model, was used to could be attributed to the calcareous parent rocks dominated in the
determine the best correlation between analytical and spectral data. study area. The amorphous iron (Feo) content varied from 0.18 to 2.54 g
This model is a multivariate calibration method based on principal kg− 1 and had an average value of 0.73 g kg− 1, while the content of the
components on both independent (X) and dependent (Y) variables pedogenic form (Fed) was in the range of 0.2 to 6.05 g kg− 1 with an
(Sacko et al., 2018). It links a response variable (Y) to a block of pre average value of 3.65 g kg− 1. The lower amounts of amorphous (Feo)
dictor variables (X) through successive latent variables which are linear and higher amounts of pedogenic (Fed) iron represent the high rate of
combinations of the initial predictor variables. This model provides soil forming processes in the study area.
good approaches in processing data with extreme co-linearity in the The average value of magnetic susceptibility (ᵡlf) of the studied soils
independent variables, especially in small data sets (Viscarra Rossel was about 50 × 10-8 m3 kg− 1. The low ᵡlf values of the studied soils could
et al., 2006b). From the total soil data set (n = 36), 26 soil samples were be related to the lithology of parent materials in the study area, which
randomly selected to calibrate the model, and the remaining samples were mainly derived from calcareous sediments. The continuous erosion
(10 samples) were used in model validation. Selected soil properties and sedimentation processes in the study area have led the studied soils
including OC, CEC, Fed, ᵡlf, and clay contents were quantified by the to be affected by the surrounding parent materials with low magnetic
spectral model. To reduce the non-linearity in the spectrum, the minerals contents including marl and limestone. The increased MS
reflectance spectra (R) were converted into absorbance (A) spectra values in different horizons of these soils refer to the pedogenic forma
through A = Log (1/R) relation. Validation of the model was carried out tion of ferromagnetic minerals followed by soil formation processes
using leave-one-out cross-validation through calculating the coefficient (Hanesh et al., 2007).
of determination (R2), Root Mean Square Error (RMSE), Residual Pre
dictive Deviation (RPD) and Ratio of Performance to Inter-Quartile 3.2. General spectral characteristics of the studied soils
range (RPIQ), which their mathematical equations are presented as
follows (Chang et al., 2001; Williams et al., 2017): Fig. 2 shows the different characteristics of absorption spectra of the
∑n studied pedons in the VIS-NIR-SWIR region. In order to better under
̂ − Y)2
R2 = ∑i=1
( Yi
(1) stand the relationship among the studied soil variables, their correlation
coefficients were calculated and presented in Table 3. In addition, the
n 2
i=1 (Yi − Y)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ correlations between soil properties and the corresponding absorption

√∑ n
√ ( Yi ̂ − Y)2 spectra are schematically shown in Fig. 3. Correlation coefficients be
√ tween soil variables and absorption spectra indicate the positive or
RMSE = i=1
(2)
n negative interactions at different spectral wavelengths. Due to different
spectral reflectance and various soil conditions, the absorption bands of
SD
RPD = (3) the soils were different (Pinheiro et al., 2017).
RMSE The soil spectral curves were analyzed based on three important
RPIQ = Q3 − Q1/RMSE (4) topics proposed by Demattê (2002) including reflectance intensity (al
bedo), absorption features (depth and amplitude), and spectral
where Y, Ŷ and Y are the measured, predicted and mean of measured morphology. Various soil properties such as particle size distribution,
values, respectively, “SD” is the standard deviation, “n” is the number of organic carbon content, amounts of iron oxides, soil mineralogy and
the soil samples, and Q3 and Q1 are third and first quartiles of the different amounts of gypsum, salts and carbonates influence the
validation set, respectively. Based on the RPD values, the prediction morphology and intensity of the absorption bands (Khayamim et al.,
accuracy of the model was categorized as precise (RPD > 2), reliable 2015; Zhang et al., 2013). As the studied soils spectra show, three
(1.4 ≤ RPD ≤ 2), and non-reliable (RPD < 1.4) (Chang et al., 2001). The distinct absorption features exist around 1410, 1910, and 2210 nm
RPIQ values were interpreted according to the categories proposed by wavelengths with different intensities and weaknesses (Fig. 2). Ab
Terra (2012) regarding the quality to reliability ratio of the prediction. sorption features at 1400 and 1900 nm wavelengths are related to the
Accordingly, RPIQ > 2.5 refers to excellent, 2 ≤ RPIQ ≤ 2.5 refers to OH groups and vibrations of water molecules including symmetric and
very good, 1.7 ≤ RPIQ < 2 refers to good, 1.4 ≤ RPIQ < 1.7 refers to fair asymmetric stretching and bending of the OH bonds (Hunt, 1977; Terra
4
Table 1
Descriptive statistics of selected soil properties in the study area.
Variable Unit Min Max Mean Skew Kurt SD CV%
− 1
EC dS m 4.2 530 84.2 2.1 3.17 149 177
pH – 6.5 7.85 7.3 − 1 1.6 0.26 3
OC g kg− 1 0 14.5 5 0.88 0.33 3.5 70
CCE g kg− 1 240 500 360 0.4 − 0.9 70 20
Gypsum g kg− 1 0 71 9.8 2.1 3.67 19.6 200
Sand g kg− 1 16 676 247 0.98 0.66 159.7 64
Silt g kg− 1 216 680 480 − 0.27 − 0.26 121.4 25
Clay g kg− 1 68 672 270 0.75 − 0.56 166.8 61
CEC Cmol+ kg− 1 6.2 32.8 20 − 0.04 − 0.1 6.7 33
Fed g kg− 1 0.2 6.05 3.65 − 0.35 − 0.21 1.46 40
Feo g kg− 1 0.18 2.54 0.73 2 5 0.47 64
ᵡlf 10-8 m3 kg− 1 3.5 94.5 50 0.21 0.94 20 40
Fed-Feo g kg− 1 − 1.3 5.3 2.9 − 1 1.4 1.5 51
Feo/Fed – 0.06 7.66 0.5 4.5 21 1.4 275
EC: electrical conductivity, OC: organic carbon, CCE: calcium carbonate equivalent, CEC: cation exchange capacity, Fed: dithionate – citrate – bicarbonate extractable
iron, Feo: oxalate extractable iron, ᵡlf: low frequency magnetic susceptibility, SD: standard deviation, CV: coefficient of variation.
et al., 2018). The absorption feature at 2200 nm refers to the vibration of originated from mica in insoluble limestone residues (Norish and Pick
water molecules, hydroxyl groups and silicates as well as octahedral, ering, 1993). However, as a result of weathering, these two minerals
tetrahedral, and interlayer cations in the clay minerals structure could be transformed into smectite (Khormali and Abtahi, 2003). In
(Demattê, 2002; Stenberg et al., 2010). Besides, it is related to the other words, smectite can be originated from the transformation of illite
interaction of OH functional groups with iron, aluminum, and magne and chlorite, neoformation, or inheritance from the smectite-rich parent
sium cations (Demattê, 2002). material. Viscarra Rossel (2011) reported that soils developed on low
Pedons 1, 2, 4, and 6 with lower clay contents showed higher relief, flat or gently undulating landscapes with poor drainage and
reflectance in their spectrum with ascending morphology between alkaline pH, are rich in smectite. Smectite is commonly an authigenic
approximately 1000 to 2100 nm and with a continuous descending mineral formed from base-rich parent rocks, and can also form on al
trend. On the other hand, pedons 3 (Btkz and Bkz horizons), 5, 7, 8, and luvial deposits, in sedimentary lowlands, bare plains and calcareous
9 with higher clay contents showed low albedo and almost a straight and materials (Norish and Pickering, 1993; Rossel, 2011). The study area is
flat morphology between 800 and 2100 nm (Fig. 2) as was also sup dominated by evaporites including gypsiferous and saline sediments and
ported by Demattê and Garcia (1999). Analyzing the correlations be has been affected by Bakhtegan saline-alkali lake, which in turn, could
tween soil textural fractions and different wavelengths helps to better provide favorable conditions for palygorskite formation. The presence of
understand the interactions between soil particle size distribution and palygorskite in soils with calcic and gypsic horizons is implied on its
soil spectral behavior. Negative correlations were found among different pedogenic source. However, both pedogenic and inherited sources of
clay contents and different wavelengths, while sand and especially silt palygorskite are reported in Iran (Khormali and Abtahi, 2003).
contents showed positive correlations with the reflectance spectrum The absorption features of 2:1 and 1:1 soil clay minerals could be
wavelengths (Fig. 3). Although soil particle size distribution influences identified between 1400 and 1900 nm as a result of vibrations caused by
the whole reflectance spectrum, its effect increases from VIS to NIR and the interaction of OH and H2O molecules found between mineral layers
SWIR regions. The highest impact of clay (r = -0.64) occurs in the 1910 (Ben-Dor et al., 2008). In addition, an absorption feature was identified
nm wavelength, whereas silt (r = 0.57) impacts on a wider range of the around 2200 nm which was related to interactions of the AlOH group of
NIR-SWIR region in agreement with Viscarra Rossel et al. (2006b) and clay minerals (Stenberg et al., 2010) (Fig. 2). The small absorption bands
Davari et al., (2021). Besides, the range from 1900 to 2500 nm is around 2355 and 2448 nm in some spectra were also attributed to the
important for clay prediction. Regarding the positive significant rela AlOH group of 2:1 clay minerals (Demattê et al., 2004). In a study in
tionship between clay content and CEC (r = 0.73), the latter also showed tropical soils of southern China, Liu et al. (2021), applied VIS-NIR-SWIR
similar behavior and correlation coefficients with reflectance wave (350–2500 nm) through the second-derivative analyses to identify and
lengths in the soil spectrum. These results are consistent with Stenberg characterize clay minerals and reported a strong consistency between
et al. (2010) and Davari et al. (2021) (Fig. 3, Table 3). Because of the mineralogical interpretations obtained from the reflectance spectros
high influence of quartz in the range of 2000 to 2500 nm, Marques et al. copy and the XRD results.
(2019) related the pattern in this region of the spectrum to soil texture. In order to better detect and elucidate the spectral features of clay
They attributed the low reflectance of their studied soils to the lower minerals and iron oxides, the reflectance spectra were compared with
quartz content, and expected textural classes with greater clay content the spectral library (USGS; https://www.usgs.gov/labs/spec-lab) and
(White et al., 1997; Marques et al., 2019). Also, in their studies in the second-derivative analyses + Savitsky-Golay filter were carried out
tropical soils, Franceschini et al. (2013) and Lacerda et al. (2016), ob (Fig. 5a and b). Textural classification of soils provides a better under
tained a strong correlation between sand content and NIR spectral standing of the effect of particle size distribution on the absorption
features. bands of soil minerals. As Fig. 5 shows, the higher clay content leads to
The X-ray diffractograms (XRD) of the selected soil samples are greater spectral amplitude. The absorption features of illite usually
displayed in Fig. 4. Chlorite, illite, palygorskite, smectite, and kaolinite occur at 1400, 1900, and specifically around 2345 nm wavelengths
were identified as dominant clay minerals in the studied soils (Fig. 4). (Viscarra Rossel et al., 2009; Viscarra Rossel et al., 2011; Coblinski et al.,
Regarding the formation conditions of kaolinite which needs a warm 2021). However, the significant amounts of CaCO3 in the studied soils
and humid climate with suitable drainage, its presence in the studied and its spectral interference limited the detection of illite absorption
soils with an arid and semi-arid climate could probably be attributed to features in the raw spectra. Khayamim et al. (2018) reported that by
inheritance from parent material and its formation in the past climate increasing CaCO3, the absorption band of illite would be deviated from
(Farpoor and Irannejad, 2013; Nadimi and Farpoor, 2013). Besides, the the carbonate, making it difficult to detect. On the other hand, eluci
significant amounts of illite and chlorite minerals in the clay fraction of dation of absorption bands in the second-derivative spectrum indicated
the studied soils seem to be with an inherited origin. Illite can be that the range of 2344–2372 nm can be considered for better detection
5
Table 2
Selected physicochemical properties of studied pedons.
Horizon Depth Dry soil pH EC Sand Silt Clay Textural CCE OC Gypsum CEC Fed Feo Feo/ Fed-Feo ᵡlf
-1 - - - - -1 - - -1
Cm color dS g kg g kg g kg class g kg g kg g kg Cmol+ g kg g kg Fed g kg × 10-8
m− 1 1 1 1 1
kg− 1 1 1
m3
kg− 1
P1 (Flood plain) Soil Taxonomy: Gypsic Haplosalids WRB: Gypsic Solonchaks

A 0–40 10YR6/4 7.8 13.4 544 388 68 Sandy 290 4.1 71 12.9 3.11 0.38 0.12 2.732 47.4
loam
Bz 40–75 10YR6/4 7.4 87.4 616 256 128 Sandy 335 2.3 38 8 3.69 0.31 0.08 3.386 68
loam
Byz1 75–110 10YR7/4 7.2 456 384 508 108 Silt loam 245 7 63 10.6 3.35 0.64 0.19 2.701 51.3
Byz2 110–140 10YR5/4 7.6 408 676 216 108 Sandy 325 4.2 62 6.2 2.9 0.18 0.06 2.715 55.5
loam
Cz 140–190 10YR7/4 7.2 530 416 456 128 Loam 305 4.2 30 7 3.73 0.4 0.1 3.327 43.8
P2 (Alluvial fan) Soil Taxonomy: Typic Haplocalcids WRB: Haplic Calcisols

A 0–30 10YR6/4 7 16.5 536 312 152 Sandy 290 1.5 – 20.4 4.96 1.02 0.2 3.935 64.1
loam
Bk1 30–70 10YR6/6 7.2 12.8 468 416 116 Loam 350 0.3 – 18.6 5.03 0.35 0.07 4.671 68.6
Bk2 70–85 10YR6/4 7.3 9.6 416 468 116 Loam 470 0.4 – 15 4.47 0.34 0.07 4.127 65.1
Ck 85–130 10YR6/4 7.3 10.6 176 512 312 Silty clay 365 3 – 18.6 5.51 0.41 0.07 5.098 59.5
loam
P3 (Piedmont plain) Soil Taxonomy: Chromic Salitorrerts WRB: Salic Vertisols

A 0–50 10YR6/4 7.2 11.9 284 568 148 Silt loam 340 12.4 – 19 4.53 0.82 0.18 3.71 89.7
Btkz 50–70 10YR6/4 6.9 56.8 76 436 488 Silty clay 350 4.3 – 21.6 4.08 0.37 0.09 3.71 86.6
Bkz 70–110 10YR5/4 6.9 55.5 76 396 528 Clay 290 6.6 – 20.75 3.87 0.38 0.09 3.49 88.6
Ckz1 110–150 10YR6/4 6.5 130 192 640 168 Silt loam 366 6.2 – 15 4.46 0.3 0.08 4.16 52.4
Ckz2 150–170 10YR7/6 6.8 62.5 182 630 188 Silt loam 385 1.5 – 20 2.81 0.5 0.18 2.31 46.9
P4 (Lowland) Soil Taxonomy: Typic Aquisalids WRB: Hypersalic Solonchaks

Azg 0–30 5Y8/2 7 470 288 536 176 Silt loam 400 1.9 31 16 0.33 1.42 4.3 − 1.09 3.5
Bzg 30–60 5Y8/2 7.3 298 288 586 126 Silt loam 455 1.2 22 16.8 0.2 1.53 7.66 − 1.33 5.8
Bz 60–100 7.5YR7/ 7.5 176 308 576 116 Silt loam 355 3.5 13 15.1 2.4 0.48 0.19 1.94 36.4
4
P5 (Flood plain) Soil Taxonomy: Calcic Haploxeralfs WRB: Calcic Luvisols

Ap 0–40 10YR4/3 7.3 20 20 388 592 Clay 305 9.7 – 32.8 2.24 0.76 0.34 1.48 29.2
Btk 40–70 10YR6/4 7.2 10.8 96 232 672 Clay 330 7 – 31 3.18 1.08 0.34 2.1 28.4
Bk 70–110 10YR6/4 7.3 10.7 16 416 568 Silty clay 300 4 – 25.7 4.29 0.59 0.13 3.7 31.2
P6 (Piedmont plain) Soil Taxonomy: Typic Calcixerepts WRB: Haplic Calcisols

Az 0–30 10YR6/4 7.3 33.9 208 636 156 Silt loam 500 9 – 17.3 2.19 0.51 0.23 1.68 40.2
Bk1 30–55 10YR7/4 7.5 25.3 184 680 136 Silt loam 445 7.4 5 18.2 2.21 0.52 0.23 1.69 40.7
Bk2 55–80 10YR7/4 7.4 22.2 184 680 136 Silt loam 465 4.6 8 19 2.21 0.5 0.22 1.71 37.8
Bk3 80–110 10YR6/4 7.4 21.9 208 656 136 Silt loam 460 3.1 10.5 19 2.03 0.78 0.38 1.25 37.5
Ck 110–150 10YR8/4 7.4 14.5 228 636 136 Silt loam 475 – – 15 2.4 0.58 0.24 1.82 41
P7 (Piedmont plain) Soil Taxonomy: Typic Calcixerolls WRB: Calcic Kastanozems

Ap 0–30 10YR5/3 7.5 4.9 88 496 416 Silty clay 460 14.5 – 23.9 2.43 0.87 0.35 1.56 40.2
Bk1 30–50 10YR6/3 7.5 4.2 88 516 396 Silty clay 395 7.8 – 23.9 2.95 0.59 0.2 2.36 42.6
loam
Bk2 50–85 10YR6/4 7.5 4.8 108 496 396 Silty clay 425 2.3 – 22.2 3.37 0.65 0.19 2.72 52.4
loam
Bk3 85–135 10YR6/4 7.6 4.7 108 496 396 Silty clay 400 2 – 22.2 4.67 0.77 0.16 3.9 50.9
loam
P8 (Piedmont plain) Soil Taxonomy: Typic Haploxerepts WRB: Haplic Cambisols

Ap 0–35 10YR6/4 7.4 8.2 208 496 296 Clay loam 300 10.9 – 29.5 5.86 1.49 0.25 4.37 53.8
Bw1 35–65 10YR5/4 7.4 8.5 208 476 316 Clay loam 310 6.2 – 32.8 5.76 2.54 0.44 3.22 46
Bw2 65–100 10YR5/4 7.5 6.7 268 436 296 Clay loam 315 3.5 – 29.5 5.4 1.62 0.3 3.77 48.4
P9 (Piedmont plain) Soil Taxonomy: Calcic haploxeralfs WRB: Calcic Luvisols

Ap 0–30 10YR4/4 7.4 6.77 256 472 272 Clay loam 295 11.7 – 23.9 5 0.66 0.13 4.33 46.9
Btk 30–60 7.5YR6/ 7.6 6.27 176 432 392 Silty clay 300 4.6 – 22.2 4.46 0.58 0.13 3.88 53.7
4 loam
Bk1 60–80 10YR4/4 7.6 4.95 216 352 432 Clay 240 5 – 25.7 5.24 0.79 0.15 4.45 94.5
Bk2 80–120 10YR6/4 7.6 4.3 176 392 432 Clay 315 3.9 – 24.8 6.05 0.73 0.12 5.32 49.8
EC: electrical conductivity, OC: organic carbon, CCE: calcium carbonate equivalent, CEC: cation exchange capacity, Fed: dithionate – citrate – bicarbonate extractable
iron, Feo: oxalate extractable iron, ᵡlf: low frequency magnetic susceptibility.
6
Fig. 2. Spectral behavior of soil horizons in studied pedons.
Table 3
Correlation matrix values of soil properties.
CCE OC Gypsum Sand Silt Clay pH EC CEC ᵡlf Fed Feo Feo/Fed
CCE 1 0.03 − 0.23 − 0.16 0.64* − 0.28 − 0.01 − 0.12 − 0.17 − 0.38* − 0.55** − 0.09 0.26
OC 1 0.07 − 0.23 0.15 0.11 − 0.1 0.26 0.19 − 0.36* − 0.37* 0.38* 0.48**
Gypsum 1 0.53** − 0.16 − 0.43** 0.18 0.69** − 0.59** − 0.17 − 0.3 − 0.15 0.13
Sand 1 − 0.33* − 0.73** 0.09 0.43** − 0.68** 0.15 0.02 − 0.15 0.03
Silt 1 − 0.38* − 0.18 − 0.02 − 0.08 − 0.3 − 0.36* 0.01 0.18
Clay 1 0.01 − 0.4* 0.73** 0.07 0.25 0.14 − 0.16
pH 1 − 0.22 0.09 − 0.1 0.06 0.07 − 0.1
EC 1 − 0.62** − 0.3 − 0.4* − 0.03 0.4*
CEC 1 − 0.04 0.33* 0.57** − 0.08
ᵡlf 1 0.63** − 0.3 − 0.55**
Fed 1 0.09 − 0.56**
Feo 1 0.42*
Feo/Fed 1
*Significant (p < 0.05), ** Significant (p < 0.01).
of this mineral (Fig. 5a and b). Similar ranges are also reported by palygorskite. In situations in which both of the minerals equally exist in
Coblinski et al. (2021) for illite identification. The chlorite spectral li the soil, the absorption feature will appear around 1912 nm (Khayamim
brary shows poor absorption features at 1400 and 2300 nm. However, et al., 2018). In the present study, through elucidation of absorption
chlorite can be specifically detected around 2380 nm (Coblinski et al., bands, the 1385–1415 and 1885–1914 nm ranges were used to identify
2021). Similarly, the second-derivative spectrum in the present study these clay minerals.
proved the 2372–2395 nm range for better identification of chlorite. In the case of kaolinite, the most important absorption feature occurs
Smectite is another 2:1 clay mineral with strong absorption features around 2207 nm with a shoulder status (Viscarra Rossel et al., 2009;
around 1400, 1906, and 2206 nm (Viscarra Rossel et al., 2009). The Demattê and da Silva Terra, 2014). Due to a combination of Al-OH
absorption features around 1400, 1917, and 2200 nm are also reported bending and O-H stretching, kaolinite shows two specific absorption
for palygorskite in the spectral library. The strong absorption features of features around 1415 nm and 2200 nm (2165 and 2207 nm) (Stenberg
these two minerals are related to OH– and H2O molecules in which their et al., 2010). The low amounts of this mineral in the studied soils, and
morphology, intensity, and weakness help to identify the presence of the overlapping of its absorption features with other minerals, made its
these minerals in the soil clay fraction (Viscarra Rossel et al., 2009; detection difficult. However, the second-derivative spectrum showed
Viscarra Rossel et al., 2006a). The absorption features of palygorskite the range of 2218–2236 nm in this region. A similar absorption range for
and smectite are similar to each other (Clark et al., 1999). Hence, they kaolinite has been reported by Coblinski et al. (2021). In a study on
can affect and spectrally interfere with each other in soil spectral Australian soils, Viscarra Rossel (2011) successfully measured the rela
reflectance. As the soil smectite content decreases, its absorption feature tive abundance of dominant clay minerals using spectroscopy technique
deviates from 1906 nm to 1917 nm which is the absorption feature of and reported the 2165 nm, 2345 and 1912 nm wavelengths for detection
7
Fig. 3. Pearson’s correlation coefficients between spectral reflectance at each wavelength and soil attributes.
Fig. 4. Spectral behavior of selected horizons and X-ray diffractograms of the clay fraction of soil samples. a) Cz horizon, P1; b) Btkz horizon, P3; c) Btk horizon, P5;
d) Bk1 horizon, P7. (Sm: smectite, Ch: chlorite, I: illite, Pa: palygorskite, Ka: kaolinite, Q: quartz, Fel: feldspar).
of kaolinite, illite and smectite, respectively. Kaolinite is the most pre the dominant clay mineral in their study areas using a sharp absorption
dominant clay mineral in Australian soils which forms under both wet band around 2200 nm. Nevertheless, in the present study, the negligible
tropical climates in the north, and cooler conditions in the south of amounts of kaolinite in the studied soils, made it difficult to detect using
Australia. The existence of this mineral in arid regions with poor degrees its absorption features. Meanwhile, all of the approaches which were
of weathering was attributed to the past climate and inheritance from employed to detect and differentiate the clay minerals may be limited by
parent materials (Norish and Pickering, 1993). In addition, Demattê and the presence of compounds such as gypsum, which mask the spectral
da Silva Terra (2014) and Demattê et al. (2017) identified kaolinite as features and prevent their recognition (Khayamim et al., 2015). Such
8
Fig. 5. a) Second-derivative spectral signature of soil minerals as classified according to texture. Gt: goethite, Hm: hematite, Kt: kaolinite, Il: illite, Ch: chlorite, CCE:
calcium carbonate equivalent. b) Spectral signature and amplitude for each mineral in the clay, clay loam and loam texture classes.
conditions were observed for sandy loams among the studied soils 1990 and 1870 nm for carbonates. In fact, it is expected that higher
(Fig. 5). In other words, in soils with high mineralogical diversity, the amounts of carbonates increase the overall reflectance by brightening
detection of each mineral is complicated by the presence of the others. soil color, however other soil properties and their interactions may
The amount of CaCO3 in the studied soils was relatively high with an represent different results (Stenberg et al., 2010). The negative signifi
average value of 360 g kg− 1 which could be related to the dominant cant correlation between CCE and Fed (r = -55) can be considered as an
calcareous parent material in the study area (Table 1). The maximum example (Table 3). The correlation coefficients (r) of CaCO3 are positive
CaCO3 content (500 g kg− 1) was observed in the Az horizon of pedon 6. at all of the wavelengths (Fig. 3) and vary in the range of 0.41 to 0.57.
However, all horizons in this pedon were alkaline, reflecting the pres The most efficient part of this parameter was observed in the SWIR re
ence of disseminated CaCO3 throughout the pedon. It seems that all of gion. Asgari et al. (2020) also attributed 714, 850, 870, 1796, 2150, and
the studied soils were affected by different amounts of carbonates 2510 cm− 1 wavenumbers to the CaCO3 absorption features and reported
(Table 2). The related absorption bands were found at 2340 nm in the that with increasing CaCO3 content, the reflectance intensity also in
raw spectra and around 2330–2350 nm in the second-derivative spec creases in the range of 1200 to 1280 cm− 1.
trum for the studied soils. Viscarra Rossel et al. (2009) attributed the Among the studied soils, pedon 1 was a gypsiferous soil (Table 2).
absorption features around 2338 and 2340 nm wavelengths to the car The absorption features of gypsum were identified around 1450, 1530,
bonate nodules and calcite, respectively. With increasing carbonates, 1750, and 1940 nm in the studied soils spectra (Fig. 2). In addition,
the absorption rate at these wavelengths was also increased, resulting in absorption features around 1000, 1200, 1400, 1600, 1740, 1900, and
a broad concavity and creating a visible peak in this region. There are 2200 nm were reported due to stretching of OH group and bending vi
different absorption features in the VIS-NIR-SWIR region of the CaCO3 brations of H2O molecules in gypsum structure (Hunt and Salisbury,
reflectance spectrum which seem to be created by internal fundamental 1971). It seems that the presence of two water molecules in the structure
and radical vibrations of carbonate lattices as concluded by Viscarra of gypsum mineral (CaSO4⋅2H2O) affects the soil absorption spectrum
Rossel et al. (2009). In addition to the strong absorption band around the especially in the range of 1400 and 1900 nm, which is related to water
2340 nm, Clark (1999) reported weaker absorption bands around 2160, absorption. In addition to the moisture existed in the soil surface, the
9
structural water in gypsiferous soils enriched by gypsum reduces the absorption features of organic carbon in the VIS-NIR-SWIR region are
reflectance and increases the absorbance in the related absorption bands often weak and not easily detectable, while they appear in the visible
(Khayamim et al., 2015). The absorption band around 1750 nm in the region as broad and clear peaks. Accordingly, a better relationship be
soil spectra is created as a result of gypsum existence in soils (Khayamim tween VIS-NIR-SWIR and soil organic carbon was reported compared to
et al., 2015) (Fig. 2). It is crucial to select unique absorption features for the NIR-SWIR alone (Viscarra Rossel et al., 2006a). This does not mean
the soil parameters. Liu et al. (2009) attributed the absorption band that organic carbon has no effect in the NIR-SWIR region, but the effect
around 1750 nm to combination modes of SO42- and H2O vibrations, of different amounts of organic carbon is just more recognizable in the
which makes this band unique compared to other minerals commonly visible region. Although several studies have used soil color to estimate
occur in the soil. Meanwhile, the relatively high amounts of gypsum in organic carbon (Viscarra Rossel et al., 2006a), it can be affected by soil
some of the soil samples have caused the absorption bands to deviate properties such as moisture content, soil texture, structure, and miner
from the normal position of the desired range (Khayamim et al., 2015). alogy. This implies that soil color would only be a useful discriminator
High salinity was observed in the study area especially in pedons 1, within a limited geological variation (Hummel et al., 2001). The ab
3, and 4 (Table 2). White salt crystals usually accumulate on the surface sorption bands around 410, 570, and 660 nm in the visible region could
of saline soils, leading to enhancement of spectral reflectance. Appear be used to estimate soil organic carbon (Viscarra Rossel et al., 2006b;
ance of absorption features around 1600 and 1950 nm wavelengths can Stenberg et al., 2010). The higher concavity of reflectance spectrum in
be attributed to the water absorbed by salt crystals and the stretching the range of 500 to 1300 nm indicates more organic carbon content and
vibrations of H2O molecules trapped in their structure (Seifi et al., vice versa (Viscarra Rossel et al., 2006a). There are several reports about
2020). In their study on saline soils, Seifi et al. (2020) reported the the absorption bands around 960 and 1100 nm (Daniel et al., 2004),
absorption features of 550, 750, 1400, 1900, 2200, and 2350 nm for the 1400, 1720, 1900, 2180, and 2309 nm (Hummel et al., 2001), and 1744,
identification of salts in the soil and proved that the spectral reflectance 1870, and 2052 nm (Stenberg et al., 2010) as being particularly useful
in the VIS-NIR-SWIR region increases with the increase in salinity. Most for soil organic carbon identification. In this study, a negative correla
of the studies on soil salinity have reported that higher salinity correlates tion was found between organic carbon and reflectance spectrum, in
with increased spectral reflectance (Brown et al., 2006; Ganjegunte which the highest correlation coefficient (r = -0.67) was observed in the
et al., 2014; Setia et al., 2013). Pessoa et al. (2016) investigated the range of 600–700 nm in the visible region (Fig. 3). Mathews et al. (1973)
spectral properties of saline and non-saline agricultural soils in Brazil, and Heller Pearlshtien and Ben-Dor (2020) also found SOM correlations
and reported that soils with the highest salinity levels presented higher with reflectance values in the 500–1200 nm range. Excessive amounts of
spectral reflectance. However, salinity presented a low correlation co organic carbon can mask the absorption features of some soil properties
efficient with spectral reflectance. On the other hand, ESP values and make them difficult to be identified across the spectrum (Demattê
correlated strongly with spectral reflectance (r = 0.92 in VIS, and r = et al., 2017).
0.74 in NIR-SWIR) (Pessoa et al., 2016). Wang et al. (2018) also reported
that saline soils have a high reflectance in the VIS-NIR-SWIR region due 3.3. Spectral variations along the transect
to existence of salt crusts on their surface. In the present study, the
highest correlation coefficient was also found in the visible region (r = Generally, the soil spectral behavior along a transect varies as a
0.47). It was concluded that high salt contents in the studied soils were result of physico-chemical and mineralogical interactions which in turn,
the plausible reason for the increase of the overall spectral reflectance are due to the variation of soil properties during pedogenetic processes.
observed (Fig. 2). Therefore, regarding the complex formation conditions, each soil has
The effect of iron oxides on the studied soils spectra was detected at specific spectral behavior. In other words, the soil absorption features
the range of 400 to 1100 nm (Fig. 2). Among the studied soils, the have resulted from the overlapping bands of different mineral and
highest amounts of iron oxides were observed in pedons 2, 8, and 9 organic components (Henderson et al., 1992). The qualitative variations
(Table 2). However, due to similar absorption features of iron oxides and of the spectral curves can be detected by their general morphology,
organic carbon, the lower amounts of organic carbon facilitated the reflectance intensity, and absorption features (Demattê, 2002). The
detection of iron oxides spectral features (Stenberg et al., 2010; Heller studied transect was located in an arid and semi-arid region with various
Pearlshtien and Ben-Dor, 2020). The absorption features of goethite and soil types developed on different geomorphic surfaces. Each geomorphic
hematite were exhibited around 460 and 530 nm in the raw reflectance position meets specific conditions in terms of drainage, runoff, different
spectrum, respectively, while in the second derivative spectrum, they salt concentrations, and gypsum and CaCO3 contents. The most impor
occur in the range of 430–450 nm for goethite and 540–575 nm for tant soil-forming factors in the study area were climate and topography,
hematite (Fig. 5a and b). Fernandes et al. (2004) also reported the ranges and dominant soil forming processes included as calcification, gypsifi
of 420–480 and 520–560 nm for goethite and hematite, respectively. cation, lessivage, weathering, salinization, gleyzation, and illuviation.
The spectral features in these wavelengths are a result of energy transfer Some of the studied pedons showed similar trends in their reflectance
to electrons of orbitals in the goethite and hematite structure that then spectra, but their slight variations could not be neglected.
returns to the lower energy levels, which were absorbed by electron Overall, pedons 7 and 9 and A, Btkz and Bkz horizons of Pedon 3
transitions (Fernandes et al., 2004; Richter et al., 2009). The concave followed a similar geochemical and spectral morphology and trend, but
morphology appearing in the range of 800 to 1000 nm with a shoulder there were distinguished minor differences between them (Fig. 2). These
around 750 nm is an indication of iron oxide crystallinity, where soils pedons were located on a stable piedmont plain with a slow slope and
with more crystalline minerals show a distinct shoulder and a deeper poor degrees of erodibility. Their small reflectance intensities can be
and symmetric feature in this range (Richter et al., 2009). Additionally, attributed to the high clay content, dark soil color as a result of high
the degree of crystallinity can be determined based on the correlations organic carbon content especially in the surface horizons, iron oxides,
among reflectance intensity and the Fed-Feo and Feo/Fed ratios. This was weathering and more evolution degrees of the soils (Table 3). Some
confirmed by the highest correlation coefficients (r) of 0.74, 0.87, and other morphological characteristics of soils such as structural units,
0.70 obtained for Fed, Feo/Fed, and ᵡlf with average reflectance wave cracks, slickensides, etc. are different and cannot be detected by soil
lengths in the visible region, respectively (Fig. 3). Furthermore, the high spectroscopy (Demattê and da Silva Terra, 2014).
Feo/Fed ratios, which prove the dominance of amorphous goethite, On the other hand, pedon 8 is located on the same geomorphic
represented low reflectances around 940 nm (Demattê and da Silva surface with less clay content and less evolution degree. An ascending
Terra, 2014). morphology from VIS toward NIR and SWIR can be observed in the
Among the studied soils, the highest amounts of organic carbon were reflectance intensity of this pedon. This pedon shows a strong similarity
observed in the surface horizons of pedons 3, 7, 8, and 9 (Table 2). The with pedon 6 in the general morphology of the spectra, the homogeneity
10
of the spectra at different horizons, the ascending shape, and the in saline soils and reported high correlation coefficients (r) between
tensities and weaknesses of the absorption bands (Fig. 2). However, the salinity factors (SO42-= 0.9, TDS = 0.79 and EC = 0.93) and the raw
overall reflectance of Pedon 6 has increased, which can be related to the spectral data.
lower clay and iron oxides contents, more soluble salts, and more CaCO3 Pedon 2 (Typic Haplocalcids) is located on an alluvial fan. The
content which were observed in different forms (powder, concretion, highest concavity around 2340 nm wavelength was related to the Bk2
and nodules) in the studied soils. Similar spectral trends of Ckz1 and horizon with the highest amount of CaCO3. Although Bk1 and Bk2 ho
Ckz2 horizons in Pedon 3 with Pedon 6 were found. This similarity was rizons have almost similar characteristics, a spectral shift between ho
due to similar textural properties, decreasing trend of iron oxides, and rizons was observed around 1000 nm. This could be attributed to CaCO3
large amounts of soluble salts in the two pedons. increment in Bk2 horizon and increase of reflectance intensity in the
The existence of gypsum in pedons 1 and 4 has led the absorption NIR-SWIR region (Fig. 2). The degree of concavity in the range of 1200
bands to deviate from their normal positions and create new and unique to 1800 nm together with the absence of organic carbon facilitates the
ones (Fig. 2). In addition, salt crystals and low clay content have detection of crystalline iron (Ladoni et al., 2010; Heller Pearlshtien and
increased the reflectance intensity and created spectral similarities be Ben-Dor, 2020). Considering the Fed and Fed-Feo values (Table 2), the
tween these pedons. However, pedon 4 was located in the lowlands higher amounts of crystalline iron and the presence of iron containing
around the Bakhtegan lake with poor drainage conditions and prevailing opaque minerals have reduced the reflectance and resulted in more
aquic moisture regime, resulting in low amounts of OC and iron oxides concavity in the spectral reflectance of different horizons (Marques
and low soil evolution degrees. There are differences in the general et al., 2019) (Fig. 2). Zhao et al. (2020) pointed out that magnetic
shape of the reflectance spectra, especially in the initial parts in the properties and clay minerals are two important factors influencing soil
range of 350–1200 nm with Pedon 1, which distinguishes the spectrum spectral behavior.
of pedon 4. Pedon 3 (Chromic Salitorrerts) is located on the piedmont plain. The
Pedon 2 which was located on an alluvial fan, has clearly shown the spectral reflectance in this pedon can be interpreted in the two following
presence of iron oxides. The overall reflectance of this pedon shows an sections. First, the organic carbon content in the A, Btkz, and Bkz ho
increasing trend with an ascending morphology from VIS to NIR and rizons decreased with depth. This was confirmed by the reduction of
SWIR regions. The coarse soil texture and low contents of organic carbon concavity in the visible region. The higher amounts of organic carbon,
have caused better detection of the iron oxides. The amounts of iron clay content (argillic horizon), magnetic susceptibility values, and iron
oxides in this pedon are almost equal to pedons 8 and 9. However, as oxides in these horizons have created a set of homogeneous spectra with
Fig. 2 shows, the higher amounts of clay and organic carbon and the constant slopes in the range of 800 to 2000 nm (Fig. 2). However, with
darker color of the soils of pedons 8 and 9, have influenced the ab respect to the mentioned reasons, the overall reflectance intensity of the
sorption bands and reflectance intensities, making them difficult to be spectra is low. Second, as the shape of spectra and characteristics of Ckz1
distinguished by their absorption features. Marques et al. (2019) and Ckz2 horizons (Fig. 2) show, the soil textural classes have become
expressed that pedons with similar characteristics exhibit similar spec coarser following the significant increase in sand and silt contents.
tral features, but they may have originated from different distinct soil Generally, the soil textural variations along different layers of the pedon
classes. can be detected from spectral changes. As the spectra show, there is an
The presence of (2:1) clay minerals including illite, chlorite and ascending trend from VIS to NIR and SWIR regions and a higher
smectite has been significantly detected using XRD analyses, but the reflectance intensity has occurred in these horizons. In addition, the
strong absorption bands around 1400 and 1900 nm, such as reflectance effect of organic carbon on the visible region decreases by reducing the
in Padon 5, can be considered as the strong evidence for the presence of amounts of organic carbon with depth. This in turn leads to a clear
these minerals in the soil. These conditions in the reflectance curves of diagnosis of iron compounds in agreement with Demattê et al. (2017)
other studied pedons with different intensities and weaknesses could and Marques et al. (2019). It is to be noted that, due to the reduction of
also be observed. Overall, the obtained spectra of the studied pedons can organic carbon with depth, the analysis and identification of soil prop
be categorized into three general patterns: 1) pedons with a homoge erties in deeper horizons that are less affected by this parameter, could
neous spectral pattern, flat and low overall reflectance, which were be more easily performed. On the other hand, Demattê et al. (2014)
generally located on the piedmont plain, 2) pedons with a homogeneous reported that the morphological characteristics of Vertisols such as
spectral pattern, with ascending morphology and high overall reflec slickensides and cracks are not detectable by reflectance spectroscopy.
tance, which were located on the piedmont plain and alluvial fan, and 3) Aquic soils have special conditions regarding their pedohydrological
pedons with a heterogeneous spectral pattern, ascending morphology environments. Pedon 4 (Typic Aquisalids) was located on the lowlands
and high overall reflectance, which were mostly located on the lowlands around Bakhtegan Lake. Anaerobic and reductive conditions in these
and floodplains. soils lead to the significant dissolution of iron-containing minerals, and
consequently, the Fe3+ transforms to Fe2+, which is a more soluble form
3.4. Vertical spectral characterization as affected by pedogenesis of iron (Grimley et al., 2004). Under these conditions, the crystallization
of iron oxides decreases and iron compounds are mostly in organic and
Pedon 1 (Gypsic Haplosalids) is located on a floodplain and as a amorphous non-pedogenic forms (Owliaie and Ghiri, 2018). As can be
gypsiferous soil, contains different amounts of gypsum with a distinct seen in Fig. 2, despite the low amounts of organic carbon, no crystalline
gypsic horizon (Table 2). As mentioned earlier, the presence of gypsum iron oxides were observed in the surface layers. In addition, the differ
leads to a change in the appearance, intensity, and weakness of ab ences in the absorption bands, the concavity from surface to depth (Bz),
sorption features, as well as the development of new peaks (Khayamim the increase in the Fed content, the evolution rate, and the magnetic
et al., 2015). However, the low amount of organic carbon in this pedon properties are quitely obvious in the soil spectra (Fig. 2). The Azg and
provides an opportunity to better detection of iron oxides and their Bzg horizons have higher reflectance intensities, which can be attributed
variations (Marques et al., 2019). The relatively high reflectance in to low organic carbon, amorphous iron oxides, soil reductive color
tensity in the spectrum of this pedon can be attributed to the lower (greenish-gray), high EC and CaCO3 contents, and the presence of gyp
amounts of organic carbon and clay, together with higher amounts of sum. Besides, the soil texture was not changed in all horizons, resulting
sand, silt, gypsum, and EC in the soil layers (Fig. 2). The absorption in similar effects of sand, silt, and clay particles in the studied soil
intensity was increased with increasing EC, and the morphology of some spectra.
bands was changed around 950, 2200, 2380, and 2460 nm, especially Pedon 5 with a fine texture and high CEC values was located on a
for the EC affected wavelengths such as 1420 and 1950 nm (Mahajan floodplain and classified as Calcic Haploxeralfs (Table 2). A very sharp
et al., 2021). Zhang et al. (2013) investigated the spectral features of absorption band around 1910 nm was observed in different horizons of
11
this pedon due to the presence of (2:1) clay minerals including illite, behavior of A horizon around and after 1100 and 1200 nm could occur
chlorite, and especially smectite and palygorskite (Fig. 2). Generally, due to the reduction of organic carbon impact and interactions of other
soils rich in smectite have a large CEC and a greater ability to retain soil parameters.
nutrients compared to kaolinite and illite (Viscarra Rossel, 2011). The Pedon 8 (Typic Haploxerepts) was also located on a piedmont plain
sharp and strong absorption bands around 1910 nm which appeared in but, with less soil evolution compared to Pedon 6. Calcium carbonate
all soil spectra can be related to the dominance of H2O and OH– mole and soil particles were homogenously distributed along with the soil
cules between structural layers of minerals such as smectite (Ben-Dor layers. All three reflectance spectra showed similar trends in this pedon,
et al., 2008; Whiting et al., 2004). The effect of decreasing organic especially in the range of 1000 to 2500 nm (Fig. 2). The greater depth of
carbon with depth was clearly detected in the visible region, while absorption properties at 1410 and 1910 nm and the reverse trend at
CaCO3 showed a homogeneous trend along with different layers (Fig. 2). 2210 nm indicate the higher amounts of (2:1) clay minerals and lower
In addition, the increase of iron oxides with depth was detected by amounts of kaolinite, respectively (Demattê and da Silva Terra, 2014).
increasing the concavity around 800 to 1000 nm (Heller Pearlshtien and Regarding the low evolution degree and soil properties under dominated
Ben-Dor, 2020). The overall reflectance intensity was increased with environmental conditions, higher amounts of weatherable minerals
depth, which could be attributed to the reduction of soil darkness which were expected. The iron oxide content in this pedon was higher than
was resulted due to a decrease in organic carbon and clay contents pedon 6. This was better detected in the Bw2 horizon due to lower
(Marques et al., 2019). amounts of organic carbon. Overall, the reflectance intensity in this
Pedon 6 was located on a piedmont plain with high amounts of pedon was lower than pedon 6 because of higher amounts of clay con
CaCO3 throughout the pedon and was classified as Typic Calcixerepts tent, iron oxides, and soil darkness.
(Table 2). The formation of a Calcic horizon and the presence of CaCO3 Pedon 9 (Calcic Haploxeralfs) was also located on a piedmont plain.
as concretions and nodules are the main characteristics of this pedon. Alfisols usually contain various clay minerals such as illite, chlorite,
Different soil horizons showed completely homogeneous spectra with smectite, palygorskite, and kaolinite, but none of them are usually
similar trends (Fig. 2). This was confirmed by the same trends in the soil predominant. Absorption bands around 1410, 1910, and 2210 nm
properties such as particle size distribution, and CaCO3 and iron oxide indicated the presence of (2:1) and (1:1) clay minerals (Fig. 2). In this
contents throughout the soil layers. The absorption feature around 2340 pedon, several factors such as high clay content, organic carbon and
nm wavelength was related to the carbonate nodules and calcite and crystalline iron oxides, soil color, magnetic susceptibility, and soil evo
distinguished as a sharp absorption band (Viscarra Rossel et al., 2009). lution degree (Feo/Fed) have reduced the overall reflectance of soil ho
Organic carbon was the only variable parameter along the pedon, rizons (Fig. 2, Table 2). The changes in CaCO3 content in different
affecting the reflectance intensity of the soil spectra. As mentioned horizons were detected based on the concavity of the absorption band
earlier, iron oxides and organic carbon show more influences on VIS-NIR around 2340 nm. While the lowest reflectance intensity was expected to
region of the spectrum. Clay, sand, and silt, in turn, are the most be related to Ap, it was replaced in the visible region by the Bk1 horizon.
important factors affecting the reflectance intensity and absorption This, in turn, suggests the possibility of magnetic properties and iron
bands, especially in the SWIR region (Demattê et al., 2014; Demattê and oxides to reduce the reflectance intensity and favor further spectral
da Silva Terra, 2014; Marques et al., 2019) (Fig. 2). Sand and silt par concavity in this region (Fig. 2).
ticles mainly contain quartz, feldspar, and mica minerals. Regarding
high amounts of silt and sand particles in different soil horizons of this
pedon, the soil spectra represented a high reflectance intensity (Demattê 3.5. Quantitative assessment of selected soil properties
et al., 2004). An ascending slope was observed from the visible to the
NIR and SWIR regions and the reflectance intensity was increased, Table 4 presents the estimated soil properties using PLSR, correlation
because quartz has the greatest effect on the reflectance intensity in the coefficients, RMSE, RPD and RPIQ values of the cross-validation. The
SWIR region. Supporting the observed trend, silt particles showed a data modeling was performed to investigate the feasibility of the PLSR
positive correlation with the reflectance spectrum at different wave model in the prediction of key soil properties by minimum data and
lengths indicating that it was more effective than sand (Fig. 3). quantitative study of the relationships between different parameters and
As expected, pedon 7 (Typic Calcixerolls) on the piedmont plain spectral data. The best PLSR fitted model among different pre-processing
geomorphic position showed a low reflectance intensity as a conse methods was found between key soil properties and output spectra with
quence of the effects of organic carbon content, fine textural class, and the highest R2, and lowest RMSE, RPD and RPIQ values (Table 4). The
iron oxides (Fig. 2). Mollisols contain a variety of 2:1 clay minerals, but RPD values demonstrated that the PLSR prediction accuracy was precise
their predominant minerals are generally smectite and illite. The studied for CEC (RPD > 2), but reliable for clay, Fed, ᵡlf, and OC (1.4 ≤ RPD ≤ 2).
soils contained low amounts of kaolinite. Since favorable conditions for Pinheiro et al. (2017) reported good predictions for sand (R2 = 0.62 and
kaolinite formation have not existed at the present time, kaolinite could RPD = 1.6) and poor predictions for silt (R2 = 0.36 and RPD = 1.25) in
only be inherited from the parent material. A relatively weak absorption their study. In addition, Mousavi et al. (2020) predicted the Atterberg
band around 2207 nm (Fig. 2) proves this finding. Due to the homoge limits of 60 soil samples with reliable results. In their study on nine soil
nous particle size distribution in all horizons, the observed changes were pedons in Brazil, Demattê et al. (2017) predicted some soil properties
related to other soil properties. One of the most important characteris including clay, sand, OC, and CEC using the principle component
tics of Mollisols is the accumulation of organic carbon in the surface analysis (PCA) method and reported that the OC prediction showed a
horizon which masks the spectral reflectance of other parameters such higher correlation coefficient (R2 = 0.9) and lower RMSE values
as iron oxides. As expected, the Ap horizon in pedon 7, showed the compared to clay (R2 = 0.5) and sand (R2 = 0.55). They also stated that
highest concavity in the visible region due to its higher amounts of due to the masking effects of other soil properties such as OC, gypsum,
organic carbon in consisted with Stenberg et al. (2010) and Demattê
et al. (2017). However, spectral shifts between horizons have occurred Table 4
in two continued sections (Fig. 2). The first section was around 1200 nm, Statistical results of PLSR for selected soil properties.
which could be related to higher amounts of (Feo/Fed) ratio and lower Soil Properties PLSR Factor R2 RMSE RPD RPIQ
amounts of iron oxides (Table 2). The reflectance intensity of the Ap -1)
Clay (g kg 9 0.65 99 1.68 2.68
horizon gradually increased toward the NIR-SWIR region, as the second Fed (g kg-1) 3 0.56 0.97 1.5 2.53
shift. This could be due to the higher impacts of CaCO3 and lower im ᵡlf (×10-8 m3 kg− 1) 3 0.65 12.07 1.7 1.52
pacts of iron oxides and organic carbon in this spectral range compared CEC (Cmol+ kg− 1) 9 0.91 1.99 3.36 4.34
OC (g kg-1) 10 0.6 2.7 1.6 1.74
to other ranges. Demattê et al. (2017) expressed that the changes in the
12
and CaCO3, the estimation of soil texture (sand, silt, and clay) could only Ben-Dor, E., Inbar, Y., Chen, Y., 1997. The reflectance spectra of organic matter in the
visible near-infrared and short wave infrared region (400–2500 nm) during a
be performed with lower R2 and higher RMSE values. Despite the results
controlled decomposition process. Remote Sens. Environ. 61 (1), 1–15.
of these researchers, the model prediction for CEC was quite appropriate Brown, D.J., Shepherd, K.D., Walsh, M.G., Mays, M.D., Reinsch, T.G., 2006. Global soil
in the present study. Guerrero et al. (2016) also reported the R2 = 0.95 characterization with VNIR diffuse reflectance spectroscopy. Geoderma 132 (3–4),
for OC prediction with minimum data. These results clearly proved the 273–290.
Chang, C.-W., Laird, D.A., Mausbach, M.J., Hurburgh, C.R., 2001. Near-infrared
possibility of prediction using minimum data. Besides, these results reflectance spectroscopy–principal components regression analyses of soil
highlighted the necessity of qualitative spectroscopic studies together properties. Soil Sci. Soc. Am. J. 65 (2), 480–490.
with quantitative investigations. Chapman, H., 1965. Cation-exchange capacity. Methods of Soil Analysis. Part 2 Chemical
and Microbiological Properties. 9, 891–901.
Clark, R.N., 1999. Spectroscopy of rocks and minerals, and principles of spectroscopy.
4. Conclusions Manual of remote sensing 3 (3–58), 2.
Coblinski, J.A., Giasson, É., Demattê, J.A., Dotto, A.C., Costa, J.J.F., Vašát, R., 2020.
Prediction of soil texture classes through different wavelength regions of reflectance
The efficiency of the VIS-NIR-SWIR method to be used in soil genesis spectroscopy at various soil depths. CATENA 189, 104485.
and evolution studies was investigated using qualitative analyses of soil Coblinski, J.A., Inda, A.V., Demattê, J.A., Dotto, A.C., Gholizadeh, A., Giasson, É., 2021.
Identification of minerals in subtropical soils with different textural classes by
spectral behavior. The obtained general spectra of the studied pedons
VIS–NIR–SWIR reflectance spectroscopy. CATENA 203, 105334.
followed from three different patterns including: 1) homogeneous, flat Cohen, M., Mylavarapu, R.S., Bogrekci, I., Lee, W.S., Clark, M.W., 2007. Reflectance
and low overall reflectance, 2) homogeneous, with ascending spectroscopy for routine agronomic soil analyses. Soil Sci. 172, 469–485.
morphology and high overall reflectance, and 3) heterogeneous, with Daniel, K., Tripathi, N., Honda, K., Apisit, E., 2004. Analysis of VNIR (400–1100 nm)
spectral signatures for estimation of soil organic matter in tropical soils of Thailand.
ascending morphology and high overall reflectance. Soil constituents Int. J. Remote Sens. 25 (3), 643–652.
such as texture, mineralogy, iron oxides, organic carbon, gypsum, car Davari, M., Karimi, S.A., Bahrami, H.A., Hossaini, S.M.T., Fahmideh, S., 2021.
bonate, salinity as well as evolution degree influenced the spectral sig Simultaneous prediction of several soil properties related to engineering uses based
on laboratory Vis-NIR reflectance spectroscopy. CATENA 197, 104987.
natures and behavior of the studied soils. These results were approved de Souza, S. F. F., Araújo, M., Demattê, J. A., de Araujo Filho, J. C., 2020. Spectral
by high correlation coefficients of the studied soil properties with patterns in the mid-ir of soils in the northeast of Brazil and their relation to the
reflectance wavelengths in different VIS-NIR-SWIR ranges. Additionally, taxonomic classification. Embrapa Solos-Artigo em periódico indexado (ALICE).
Demattê, J.A.M., 2002. Characterization and discrimination of soils by their reflected
the results of qualitative identification of clay minerals using XRD electromagnetic energy. Pesqui. Agropecu. Bras. 37 (10), 1445–1458.
method were specifically in accordance with the spectral absorption Demattê, J.A.M., Bellinaso, H., Romero, D.J., Fongaro, C.T., 2014. Morphological
bands. This indicates that minor changes in the morphology of absorp Interpretation of Reflectance Spectrum (MIRS) using libraries looking towards soil
classification. Sci. Agric. 71 (6), 509–520.
tion bands refer to their dominance in the soil. The predominance of clay Demattê, J.A.M., Garcia, G.J., 1999. Alteration of soil properties through a weathering
minerals such as illite, chlorite, kaolinite, palygorskite, and especially sequence as evaluated by spectral reflectance. Soil Sci. Soc. Am. J. 63 (2), 327–342.
smectite in the studied soils was confirmed through investigation of the Demattê, J.A.M., Horák-Terra, I., Beirigo, R.M., da Silva Terra, F., Marques, K.P.P.,
Fongaro, C.T., Vidal-Torrado, P., 2017. Genesis and properties of wetland soils by
absorption features in comparison with the spectral library of clay
VIS-NIR-SWIR as a technique for environmental monitoring. J. Environ. Manage.
minerals and second derivative spectrum analyses. However, the VIS- 197, 50–62.
NIR-SWIR method was unable to recognize some soil morphological Demattê, J.A., Campos, R.C., Alves, M.C., Fiorio, P.R., Nanni, M.R., 2004. Visible–NIR
and structural characteristics which are necessary to distinguish soil reflectance: a new approach on soil evaluation. Geoderma 121 (1–2), 95–112.
Demattê, J.A., da Silva Terra, F., 2014. Spectral pedology: a new perspective on
classes. The PLSR model indicated an excellent capability to estimate evaluation of soils along pedogenetic alterations. Geoderma 217, 190–200.
CEC and presented reliable results for OC, clay, Fed, and ᵡlf. The results Farifteh, J., Farshad, A., George, R., 2006. Assessing salt-affected soils using remote
of this study revealed the high potential of VIS-NIR-SWIR method to be sensing, solute modeling, and geophysics. Geoderma 130 (3–4), 191–206.
Farpoor, M.H., Irannejad, M., 2013. Soil genesis and clay mineralogy on Aliabbas River
used in soil characterization and monitoring of soil variations. However, alluvial fan. Kerman Province. Arab. J. Geosci. 6 (3), 921–928.
recognizing the effects of soil properties separately encounters various Fernandes, R., Barrón, V., Torrent, J., Fontes, M., 2004. Quantification of iron oxides in
challenges due to interactions of soil properties with spectral behavior. Brazilian latosols by diffuse reflectance spectroscopy. Rev. Bras. Cienc. Solo. 28 (2),
245–257.
Finally, it can be concluded from the findings of this study that spectral Franceschini, M.H.D., Demattê, J.A.M., Sato, M.V., et al., 2013. Abordagens
pedology can be successfully applied to understand and interpret soil semiquantitativa e quantitativa na avaliac¸ãoda textura do solo por espectroscopia
characteristics and evolution pathways with a new perspective and de reflectaˆncia bidirecional no VIS-NIR-SWIR. Pesqui Agropecuá ria Bras 48,
1569–1582.
reasonable capability. Ganjegunte, G.K., Sheng, Z., Clark, J.A., 2014. Soil salinity and sodicity appraisal by
eletromagnetic induction in soils irrigated to grow cotton. Land Degrad Dev. 25 (3),
228–235.
Funding
Gee, G. W., Bauder, J. W., 1986. Particle-size analysis. Methods of soil analysis: Part 1
Physical and mineralogical methods. 5, 383-411.
This research did not receive any specific grant from funding Grimley, D.A., Arruda, N.K., Bramstedt, M.W., 2004. Using magnetic susceptibility to
agencies in the public, commercial, or not-for-profit sectors. facilitate more rapid, reproducible and precise delineation of hydric soils in the
midwestern USA. CATENA 58 (2), 183–213.
Guerrero, C., Wetterlind, J., Stenberg, B., Mouazen, A.M., Gabarrón-Galeote, M.A., Ruiz-
Sinoga, J.D., Rossel, R.A.V., 2016. Do we really need large spectral libraries for local
Declaration of Competing Interest scale SOC assessment with NIR spectroscopy? Soil Tillage Res. 155, 501–509.
Hanesch, M., Rantitsch, G., Hemetsberger, S., Scholger, R., 2007. Lithological and
pedological influences on the magnetic susceptibility of soil: Their consideration in
The authors declare that they have no known competing financial magnetic pollution mapping. Sci Total Environ. 382 (2–3), 351–363.
interests or personal relationships that could have appeared to influence Heller Pearlshtien, D., Ben-Dor, E., 2020. Effect of organic matter content on the spectral
the work reported in this paper. signature of iron oxides across the VIS–NIR spectral region in artificial mixtures: An
example from a red soil from Israel. Remote Sens 12 (12), 1960.
Henderson, T., Baumgardner, M., Franzmeier, D., Stott, D., Coster, D., 1992. High
References dimensional reflectance analysis of soil organic matter. Soil Sci. Soc. Am. J. 56 (3),
865–872.
Hummel, J.W., Sudduth, K.A., Hollinger, S.E., 2001. Soil moisture and organic matter
Araújo, S.R., Söderström, M., Eriksson, J., Isendahl, C., Stenborg, P., Demattê, J.M.,
prediction of surface and subsurface soils using a NIR soil sensor. Comput. Electron.
2015. Determining soil properties in Amazonian Dark Earths by reflectance
Agric. 32 (2), 149–165.
spectroscopy. Geoderma 237, 308–317.
Hunt, G.R., 1977. Spectral signatures of particulate minerals in the visible and near
Asgari, N., Ayoubi, S., Demattê, J.A.M., Dotto, A.C., 2020. Carbonates and organic matter
infrared. Geophysics 42 (3), 501–513.
in soils characterized by reflected energy from 350–25000 nm wavelength. J. Mt.
Hunt, G.R., Salisbury, J.W., 1971. Visible and near infrared spectra of minerals and
Sci. 17 (7), 1636–1651.
rocks. II. Carbonates. Modern Geology 2, 23–30.
Banaie, M. H., 2001. Map of Iran soils moisture and temperature regimes. Soil and Water
Iran Geology Organization, 1995. Neyriz and Estahban map 1:250000. Tehran map
research institute. Tehran. Iran.
publication.
Ben-Dor, E., Heller, D., Chudnovsky, A., 2008. A novel method of classifying soil profiles
in the field using optical means. Soil Sci. Soc. Am. J. 72 (4), 1113–1123.
13
Islam, K., Singh, B., McBratney, A., 2003. Simultaneous estimation of several soil CaCO3 using laboratory Vis-NIR spectra: A case study in a semi-arid region of Iran.
properties by ultra-violet, visible, and near-infrared reflectance spectroscopy. Soil Geoderma 314, 102–112.
Res. 41 (6), 1101–1114. Owliaie, H., Ghiri, M.N., 2018. The Magnetic Susceptibility and Iron Oxides of Aquic
IUSS Working Group WRB., 2015. World reference base for soil resources. In: World Soil Soils in Southern Iran. Eurasian. Soil. Sci. 51 (10), 1252–1265.
Resources Reports No. 106. FAO, Rome. Page, A., Miller, R., Keeney, D., 1982. Methods of soil analysis, Part 2: Chemical and
Jackson, M.L., 1975. Soil Chemical Analysis: Advanced Course. Department of Soils, microbiological properties, 2nd ed. Madison, Wisconsin, USA.
College of Agriculture, University of Wisconsin, Madison, Wisconsin, USA. Pessoa, L.G.M., Freire, M.B.G.D.S., Wilcox, B.P., Green, C.H.M., De Araújo, R.J.T., De
Jensen, J.R., 2009. Remote sensing of the environment: An earth resource perspective 2/ Araújo Filho, J.C., 2016. Spectral reflectance characteristics of soils in northeastern
e. Pearson Education India. Brazil as influenced by salinity levels. Environ Monit Assess 188 (11), 1–11.
Khayamim, F., Khademi, H., Ayoubi, S., 2018. Capability and Limitations of Clay Pinheiro, É.F., Ceddia, M.B., Clingensmith, C.M., Grunwald, S., Vasques, G.M., 2017.
Minerals Estimation in Surface Soils of the Isfahan Province by Vis-NIR Prediction of soil physical and chemical properties by visible and near-infrared
Spectroscopy. IJSR. 32 (1), 129–140. In persian. diffuse reflectance spectroscopy in the central Amazon. Remote Sensing 9 (4), 293.
Khayamim, F., Wetterlind, J., Khademi, H., Robertson, A.J., Cano, A.F., Stenberg, B., Ramos, P.V., Inda, A.V., Barrón, V., Siqueira, D.S., Júnior, J.M., Teixeira, D.D.B., 2020.
2015. Using visible and near infrared spectroscopy to estimate carbonates and Color in subtropical brazilian soils as determined with a Munsell chart and by diffuse
gypsum in soils in arid and subhumid regions of Isfahan. Iran. J. Near Infrared reflectance spectroscopy. CATENA 193, 104609.
Spectrosc. 23 (3), 155–165. Richter, N., Jarmer, T., Chabrillat, S., Oyonarte, C., Hostert, P., Kaufmann, H., 2009. Free
Khormali, F., Abtahi, A., 2003. Origin and distribution of clay minerals in calcareous arid iron oxide determination in Mediterranean soils using diffuse reflectance
and semi-arid soils of Fars Province, southern Iran. Clay Miner. 38 (4), 511–527. spectroscopy. Soil Sci. Soc. Am. J. 73 (1), 72–81.
Kim, I., Pullanagari, R., Deurer, M., Singh, R., Huh, K., Clothier, B., 2014. The use of Sacko, B.D., Sanogo, S., Konare, H., Ba, A., Diakite, T., 2018. Capability of visible-nea
visible and near-infrared spectroscopy for the analysis of soil water repellency. Eur. infrared spectroscopy in estimating soils carbon, potassium and phosphorus. Opt.
J. Soil. Sci. 65 (3), 360–368. Photonics J. 08 (5), 123–134.
Kittrick, J., Hope, E., 1963. A procedure for the particle-size separation of soils for X-ray Schwertmann, U., 1973. Use of oxalate for Fe extraction from soils. Can. J. Soil Sci. 53
diffraction analysis. Soil Sci. 96 (5), 319–325. (2), 244–246.
Lacerda, M.P., Demattê, J.A., Sato, M.V., Fongaro, C.T., Gallo, B.C., Souza, A.B., 2016. Seifi, M., Ahmadi, A., Neyshabouri, M.-R., Taghizadeh-Mehrjardi, R., Bahrami, H.-A.,
Tropical texture determination by proximal sensing using a regional spectral library 2020. Remote and Vis-NIR spectra sensing potential for soil salinization estimation
and its relationship with soil classification. Remote Sence. 8 (9), 701. in the eastern coast of Urmia hyper saline lake. Iran. Remote Sensing Applications:
Ladoni, M., Bahrami, H.A., Alavipanah, S.K., Norouzi, A.A., 2010. Estimating soil organic Society and Environment 20, 100398.
carbon from soil reflectance: A review. Precis. Agric. 11 (1), 82–99. Setia, R., Lewis, M., Marschner, P., Raja Segaran, R., Summers, D., Chittleborough, D.,
Liu, J.C., He, H.P., Michalski, J., Cuadros, J., Yao, Y.Z., Tan, W., Wei, G.J., 2021. 2013. Severity of salinity accurately detected and classified on a paddok scale with
Reflectance spectroscopy applied to clay mineralogy and alteration intensity of a high resolution multispectral satellite imagery. Land Degrad Dev. 24 (4), 375–384.
thick basaltic weathering sequence in Hainan Island. South China. Appl Clay Sci. Shahrayini, E., Noroozi, A., Eghbal, M.K., 2020. Prediction of Soil Properties by Visible
201, 105923. and Near-Infrared Reflectance Spectroscopy. Eurasian. Soil. Sci. 53 (12), 1760–1772.
Liu, Y., Wang, A., Freeman, J.J., 2009. “Raman, MIR and NIR spectroscopic study of Silva, L.S., Marques Júnior, J., Barrón, V., Gomes, R.P., Teixeira, D.D.B., Siqueira, D.S.,
calcium sulfates: gypsum, bassanite and anhydrite”, in Proceedings of the 40th Lunar Vasconcelos, V., 2020. Spatial variability of iron oxides in soils from Brazilian
and Planetary Science Conference, 2128. sandstone and basalt. CATENA 185, 104258.
Mahajan, G.R., Das, B., Gaikwad, B., Murgaonkar, D., Desai, A., Morajkar, S., Kulkarni, R. Staff, S.S., 2014. Keys to Soil Taxonomy, 12th edn. USDA-Natural Resources
M., 2021. Monitoring properties of the salt-affected soils by multivariate analysis of Conservation Service, Washington, DC.
the visible and near-infrared hyperspectral data. CATENA 198, 105041. Soriano-Disla, J.M., Janik, L.J., Viscarra Rossel, R.A., Macdonald, L.M., McLaughlin, M.
Marques, K.P., Rizzo, R., Carnieletto Dotto, A., Souza, A.B.E., Mello, F.A., Neto, L.G., J., 2014. The performance of visible, near-, and mid-infrared reflectance
Demattê, J.A., 2019. How qualitative spectral information can improve soil profile spectroscopy for prediction of soil physical, chemical, and biological properties.
classification? J. Near Infrared Spectrosc. 27 (2), 156–174. Appl. Spectrosc. Rev. 49 (2), 139–186.
Mathews, H.L., Cunningham, R.L., Peterson, G.W., 1973. Spectral reflectance of selected Stenberg, B., Rossel, R.A.V., Mouazen, A.M., Wetterlind, J., 2010. Visible and near
Pennsylvania soils. Soil Sci. Soc. Am. J. 37 (3), 421–424. infrared spectroscopy in soil science. Adv. Agron. 107, 163–215.
Mehra, O., Jackson, P., 1958. Iron oxide removal from soils and clays in a dithionite- Stoner, E. R., Baumgardner, M. F., 1981. Characteristics variations in reflectance of
citrate-bicarbonate system buffered with sodium. Clay Clay Miner. 7, 317–327. surface soils. Soil Sci. Soc. Am. J. 45.
Miller, C. E., 2001. Chemical principles of near-infrared technology. In ‘‘Near-Infrared Summers, D., Lewis, M., Ostendorf, B., Chittleborough, D., 2011. Visible near-infrared
Technology in the Agricultural and Food Industries’’ (P. Williams and K. Norris, reflectance spectroscopy as a predictive indicator of soil properties. Ecol. Indic. 11
Eds.), The American Association of Cereal Chemists Inc., St. Paul, MN. pp. 19–37. (1), 123–131.
Mina, M., Rezaei, M., Sameni, A., Moosavi, A.A., Ritsema, C., 2021. Vis-NIR spectroscopy Terra, F.D.S., 2012. Espectroscopia de reflectância do visível ao infravermelho médio
predicts threshold velocity of wind erosion in calcareous soils. Geoderma 401, aplicada aos estudos qualitativos e quantitativos de solos (Doctoral dissertation,
115163. Universidade de São Paulo).
Minasny, B., McBratney, A., Tranter, G., Murphy, B., 2008. Using soil knowledge for the Terra, F.S., Demattê, J.A., Rossel, R.A.V., 2018. Proximal spectral sensing in pedological
evaluation of mid-infrared diffuse reflectance spectroscopy for predicting soil assessments: vis–NIR spectra for soil classification based on weathering and
physical and mechanical properties. Eur. J. Soil. Sci. 59 (5), 960–971. pedogenesis. Geoderma 318, 123–136.
Mortimore, J.L., Marshall, L.-J.-R., Almond, M.J., Hollins, P., Matthews, W., 2004. Viscarra Rossel, R.A., 2011. Fine-resolution multiscale mapping of clay minerals in
Analysis of red and yellow ochre samples from Clearwell Caves and Çatalhöyük by Australian soils measured with near infrared spectra. J Geophys Res-Earth 116 (F4).
vibrational spectroscopy and other techniques. Spectrochimica Acta Part A: Viscarra Rossel, R.A., Behrens, T., Ben-Dor, E., Brown, D., Demattê, J., Shepherd, K.D.,
Molecular and Biomolecular Spectroscopy 60 (5), 1179–1188. Adamchuk, V., 2016. A global spectral library to characterize the world’s soil. Earth-
Mousavi, F., Abdi, E., Ghalandarzadeh, A., Bahrami, H.A., Majnounian, B., Ziadi, N., Science Reviews 155, 198–230.
2020. Diffuse reflectance spectroscopy for rapid estimation of soil Atterberg limits. Viscarra Rossel, R.A., Cattle, S.R., Ortega, A., Fouad, Y., 2009. In situ measurements of
Geoderma 361, 114083. soil color, mineral composition and clay content by vis–NIR spectroscopy. Geoderma
Nadimi, M., Farpoor, M.H., 2013. Genesis and clay mineralogy of soils on different 150 (3–4), 253–266.
geomorphic surfaces in Mahan-Joupar area, central Iran. Arab. J. Geosci. 6 (3), Viscarra Rossel, R.A., McGlynn, R., McBratney, A., 2006a. Determining the composition
825–833. of mineral-organic mixes using UV–vis–NIR diffuse reflectance spectroscopy.
Naimi, S., Ayoubi, S., Demattê, J.A., Zeraatpisheh, M., Amorim, M.T.A., Mello, F.A.D.O., Geoderma 137 (1–2), 70–82.
2021. Spatial prediction of soil surface properties in an arid region using synthetic Viscarra Rossel, R.A., Webster, R., 2012. Predicting soil properties from the Australian
soil image and machine learning. Geocarto Int 1–24. soil visible–near infrared spectroscopic database. Eur. J. Soil. Sci. 63 (6), 848–860.
Nanni, M.R., Demattê, J.A.M., 2006. Espectral reflectance methodology in comparison to Viscarra Rossel, R.A., Walvoort, D.J.J., McBratney, A.B., Janik, L.J., Skjemstad, J.O.,
soil analysis. Soil Sci. Soc. Am. J. Madison 70 (2), 393–407. 2006b. Visible, near infrared, mid infrared or combined diffuse reflectance
Nawar, S., Mouazen, A.M., 2017. Predictive performance of mobile vis-near infrared spectroscopy for simultaneous assessment of various soil properties. Geoderma 131,
spectroscopy for key soil properties at different geographical scales by using spiking 59–75.
and data mining techniques. CATENA 151, 118–129. Walkley, A., Black, I.A., 1934. An examination of the Degtjareff method for determining
Nelson, R.E., 1982. Carbonate and Gypsum. In: Page, A.L. (Ed.), Methods of Soil Analysis. soil organic matter, and a proposed modification of the chromic acid titration
American Society of Agronomy Inc, Madison, Wisconsin, USA, pp. 181–197. method. Soil Sci. 37 (1), 29–38.
Norish, K., Pickering, J.G., 1993. Mineralogy. In: Soils: An AustralianViewpoint. Wang, J., Ding, J., Abulimiti, A., Cai, L., 2018. Quantitative estimation of soil salinity by
Academic, London, pp. 281–308. means of different modeling methods and visible-near infrared (VIS–NIR)
Ogen, Y., Goldshleger, N., Ben-Dor, E., 2017. 3D spectral analysis in the VNIR-SWIR spectroscopy, Ebinur Lake Wetland. Northwest China. PeerJ 6, e4703.
spectral region as a tool for soil classification. Geoderma 302, 100–110. Weidong, L., Baret, F., Xingfa, G., Qingxi, T., Lanfen, Z., Bing, Z., 2002. Relating soil
Ostovari, Y., Ghorbani-Dashtaki, S., Bahrami, H.-A., Abbasi, M., Dematte, J.A.M., surface moisture to reflectance. Remote Sens. Environ. 81 (2–3), 238–246.
Arthur, E., Panagos, P., 2018. Towards prediction of soil erodibility, SOM and
14
Weindorf, D.C., Chakraborty, S., Herrero, J., Li, B., Castañeda, C., Choudhury, A., 2016. Zeng, R., Zhang, G.L., Li, D.C., Rossiter, D.G., Zhao, Y.G., 2016. How well can VNIR
Simultaneous assessment of key properties of arid soil by combined PXRF and spectroscopy distinguish soil classes? Biosyst Eng. 152, 117–125.
Vis–NIR data. Eur J Soil Sci. 67 (2), 173–183. Zhang, F., Tiyip, T., Ding, J., Kung, H., Johnson, V.C., Sawut, M., Gui, D., 2013. Studies
White, K., Walden, J., Drake, N., Eckardt, F., Settlell, J., 1997. Mapping the iron oxide on the reflectance spectral features of saline soil along the middle reaches of Tarim
content of dune sands, Namib Sand Sea, Namibia, using Landsat Thematic Mapper River: a case study in Xinjiang Autonomous Region, China. Environ Earth Sci. 69 (8),
data. Remote Sens Environ. 62 (1), 30–39. 2743–2761.
Whiting, M.L., Li, L., Ustin, S.L., 2004. Predicting water content using Gaussian model on Zhang, Z., Ding, J., Wang, J., Ge, X., 2020. Prediction of soil organic matter in
soil spectra. Remote Sens. Environ. 89 (4), 535–552. northwestern China using fractional-order derivative spectroscopy and modified
Williams, P., Dardenne, P., Flinn, P., 2017. Tutorial: Items to be included in a report on a normalized difference indices. CATENA 185, 104257.
near infrared spectroscopy project. J. Near Infrared Spectrosc. 25 (2), 85–90. Zhao, L., Hong, H., Fang, Q., Algeo, T.J., Wang, C., Li, M., Yin, K., 2020. Potential of
Xuemei, L., Jianshe, L., 2013. Measurement of soil properties using visible and short VNIR spectroscopy for prediction of clay mineralogy and magnetic properties, and
wave-near infrared spectroscopy and multivariate calibration. Measurement. 46 its paleoclimatic application to two contrasting Quaternary soil deposits. CATENA
(10), 3808–3814. 184, 104239.
15

1 s2.0 S0341816222002740 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0341816222002740 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0341816222002740 Main

Uploaded by

Copyright:

Available Formats

Catena 214 (2022) 106288

Contents lists available at ScienceDirect

Pedological assessments along an arid and semi-arid transect using soil

2.2. Soil laboratory analyses

Fig. 1. Geographical location of the study area and studied pedons.

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ correlations between soil properties and the corresponding absorption

P1 (Flood plain) Soil Taxonomy: Gypsic Haplosalids WRB: Gypsic Solonchaks

P2 (Alluvial fan) Soil Taxonomy: Typic Haplocalcids WRB: Haplic Calcisols

P3 (Piedmont plain) Soil Taxonomy: Chromic Salitorrerts WRB: Salic Vertisols

P4 (Lowland) Soil Taxonomy: Typic Aquisalids WRB: Hypersalic Solonchaks

P5 (Flood plain) Soil Taxonomy: Calcic Haploxeralfs WRB: Calcic Luvisols

P6 (Piedmont plain) Soil Taxonomy: Typic Calcixerepts WRB: Haplic Calcisols

P7 (Piedmont plain) Soil Taxonomy: Typic Calcixerolls WRB: Calcic Kastanozems

P8 (Piedmont plain) Soil Taxonomy: Typic Haploxerepts WRB: Haplic Cambisols

P9 (Piedmont plain) Soil Taxonomy: Calcic haploxeralfs WRB: Calcic Luvisols

Fig. 2. Spectral behavior of soil horizons in studied pedons.

*Significant (p < 0.05), ** Significant (p < 0.01).

You might also like