Remote Sensing of Environment 164 (2015) 66–76

Contents lists available at ScienceDirect

Remote Sensing of Environment

journal homepage: www.elsevier.com/locate/rse

A linear physically-based model for remote sensing of soil moisture using

short wave infrared bands
Morteza Sadeghi a,⁎, Scott B. Jones a, William D. Philpot b
a
Dept. Plants, Soils and Climate, Utah State University, Logan, UT 84322-4820, USA
b
School of Civil & Environmental Engineering, Cornell University, Ithaca, NY, USA

a r t i c l e i n f o a b s t r a c t

Article history: Technological advances in satellite remote sensing have offered a variety of techniques for estimating surface soil
Received 23 November 2014 water content as a key variable in numerous environmental studies. Optical methods are particularly valuable for
Received in revised form 18 March 2015 remote sensing of soil moisture since reflected solar radiation is the strongest passive signal available to satellites
Accepted 3 April 2015
and thus observations at optical wavelengths are capable of providing high spatial resolution data. Since remote
Available online 28 April 2015
sensors do not measure soil water content directly, mathematical algorithms that describe the connection be-
Keywords:
tween the measured signal and surface water content must be derived. Here, we present a physically-based
Remote sensing soil moisture retrieval model in the solar domain (350–2500 nm) that is based on the Kubelka–Munk two-flux
Soil moisture radiative transfer theory. The model is designed to describe diffuse reflectance from a uniform, optically thick, ab-
Reflectance sorbing and scattering medium. The theory suggests a linear relationship between a transformed reflectance and
Absorption soil water content in the short wave infrared bands (e.g. band 7 of Landsat and MODIS satellites) providing an
Scattering easy-to-use algorithm in these bands. Accuracy of this model was tested and preliminarily verified using
laboratory-measured spectral reflectance data of different soils. Further studies on potentials and challenges of
this model for large-scale application using optical satellites data remain a topic of ongoing research.
© 2015 Elsevier Inc. All rights reserved.

1. Introduction Landrum, 2013; Ojha, Morbidelli, Saltalippi, Flammini, & Govindaraju,

Surface soil moisture is a fundamental state variable controlling a
wide range of processes occurring at the land–atmosphere interface in-
cluding water infiltration and runoff, evaporation, heat and gas ex-
change, solute infiltration, erosion, etc. The importance of measuring
and monitoring surface soil moisture at various spatial scales has been
highlighted by many authors (e.g. Ochsner et al., 2013; Robinson et al.,
2008; Vereecken et al., 2008; Wang & Qu, 2009).
Despite recent advances in electromagnetic (EM) sensing capability,
whether ground-based (Jones, Wraith, & Or, 2002; Jones et al., 2005;
Bogena, Huisman, Oberdorster, & Vereecken, 2007; Vaz, Jones,
Meding, & Tuller, 2013) or remotely deployed (de Jeu et al., 2008;
Huisman, Hubbard, Redman, & Annan, 2003; Kornelsen & Coulibaly,
2013; Njoku, Jackson, Lakshmi, Chan, & Nghiem, 2003), accurate assess-
ment of surface water content is still a challenging task due to complex
spatiotemporal variability of soil moisture in addition to scaling issues
(Brocca, Melone, Moramarco, & Morbidelli, 2010; Crow et al., 2012;
⁎ Corresponding author at: 4820 Old Main Hill, Logan, UT 84322-4820, USA. Tel.: +1
435 797 2175; fax: +1 435 797 3376.
E-mail addresses: morteza.sadeghi@usu.edu (M. Sadeghi), scott.jones@usu.edu
(S.B. Jones), philpot@cornell.edu (W.D. Philpot).
2014; Western, Grayson, & Blöschl, 2002; Western et al., 2004; Wu &
Li, 2009).
The interaction of EM radiation with soils at various wavelengths has
been shown to be significantly correlated with surface moisture con-
tent. Hence, various remote sensing (RS) methods have been developed
in different regions of the EM spectrum. These methods may be classi-
fied into three major groups (Verstraeten, Veroustraete, & Feyen,
2008; Wang & Qu, 2009):
i) Optical methods. Methods using optical (or solar domain) bands
(wavelengths between 0.35 and 2.5 μm) in which the reflected
radiation of the sun from the Earth's surface, known as reflec-
tance, is measured (Ben-Dor et al., 2009). Reflectance is com-
monly formulated as a function of soil moisture content using
empirical (regression) analysis (Wang & Qu, 2009).
ii) Thermal infrared methods. Methods in which the thermal
emission of the Earth (wavelengths between 3.5 and 14 μm) is
measured (Wang & Qu, 2009). The estimation of surface soil
moisture using remotely sensed thermal wavebands primarily
relies on the use of soil surface temperature measurements,
either singly by the thermal inertia method (Verstraeten,
Veroustraete, van der Sande, Grootaers, & Feyen, 2006) or in
combination with vegetation indices, for example using the so-

http://dx.doi.org/10.1016/j.rse.2015.04.007
0034-4257/© 2015 Elsevier Inc. All rights reserved.
 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76
called “triangle method” (Carlson, 2007; Carlson, Gillies, & Perry,
1994).
iii) Microwave methods. Methods in which the intensity of
microwave emission (wavelengths between 5 and 1000 mm) is
measured. The fundamental basis of microwave RS of soil
moisture is the large contrast between the dielectric properties
of water (~ 81) and other soil constituents (b4). As surface
soil moisture content increases, the dielectric constant of
the soil–water mixture increases significantly, and this change
is readily detectable by microwave sensors (Njoku & Kong,
1977).
Microwave RS techniques have demonstrated the most promising
ability for globally monitoring of soil moisture variations (e.g. Al-Yaari
et al., 2014). Unlike the thermal and optical sensing, there is significant
penetration into the soil at lower microwave frequencies. Measure-
ments at these frequencies are not impeded by clouds or darkness
(Njoku & Entekhabi, 1996). Most microwave models have a physical
basis, although they commonly include some empirical (calibration) pa-
rameters to account for surface roughness which causes perturbation of
the microwave signal (Shi et al., 2006).
Despite the benefits of microwave methods, optical and thermal
methods are also essential in remote sensing of soil moisture, because
of their capability for providing very high spatial resolution maps, espe-
cially when compared with the very coarse output of microwave sen-
sors. Therefore, optical/thermal satellites data provide favorable
means for downscaling (i.e. improving spatial resolution) microwave
satellite estimates of soil moisture (Merlin, Walker, Chehbouni, & Kerr,
2008; Merlin et al., 2013; Piles et al., 2011).
In this regard, thermal approaches have received more attention
than optical. Although thermal RS methods are powerful and have
been developed based on physical rationale, the methods are often
empirical and depend on local meteorological conditions, such as
wind speed, air temperature, and humidity (Nemani, Pierce, Running,
& Goward, 1993), and thus measurement results vary temporally and
with land cover types (Czajkowski, Goward, Stadler, & Waltz, 2000).
In other words, the relationship between soil moisture and surface tem-
perature is highly dependent on atmospheric conditions. Therefore, in
most cases these methods require ground measurements for calibration
of the empirical fitting parameters (Yang, Wu, Shi, & Yan, 2008).
This limitation motivated us to revisit the optical RS technique to po-
tentially develop a physically-based method with minimal calibration
requirements. We aimed to balance between physical significance and
practical application, where such a model is lacking among existing op-
tical RS methods.
In terms of physical significance, three classes of optical models may
be distinguished in the literature. The first class includes simple regres-
sion models for formulating soil moisture–reflectance relationship (Liu
et al., 2002; Liu et al., 2003; Lobell & Asner, 2002; Muller & De'camps,
2000; Nolet, Poortinga, Roosjen, Bartholomeus, & Ruessink, 2014;
Persson, 2005; Skidmore, Dickerson, & Schimmelpfennig, 1975; Zhu,
Wang, Shao, & Horton, 2011). Although such models are established
based on a physical understanding of the impact water has on the soil
spectrum from light absorption and refraction at the water–mineral in-
terfaces, they commonly contain physically meaningless empirical pa-
rameters and thus require soil information a priori in order to be
solved (i.e. calibration). Models such as that of Whiting, Li, and Ustin
(2004) may be distinguished as a second category having a stronger
physical basis than the first group. Such models are based on a fit to a
continuum of the soil water absorption bands, which are directly related
to the amount of water in the surface soil layer. In that sense, these
models relate the water content to a physical optical property rather
than the change in reflectance at one wavelength and the observation
is potentially independent of the soil type (Haubrock, Chabrillat,
Lemmnitz, & Kaufmann, 2008). Models introduced by Twomey,
Bohren, and Mergenthaler (1986), Bach and Mauser (1994) and
67
Philpot (2010) are among a few models in the third class which were
fully formulated based on physics of radiative transfer. These models
usually suffer from difficult-to-determine input information.
We could also consider an additional class of models being devel-
oped nowadays based on machine learning techniques such as artificial
neural networks and support vector machine (Hassan Esfahani,
Torres-Rua, Jensen, & McKee, 2015; Zaman, McKee, & Neale, 2012).
Commonly considering both optical and thermal bands, these models
provide powerful tools for inferring surface soil moisture in complex/
heterogeneous media (e.g. rough vegetated surfaces). However, these
models are fully black-box with no physical origin and hence require a
vast database for calibration and even then are limited to site-specific
applications.
In the following, we present a simple model with physically defin-
able parameters based on the Kubelka and Munk (1931) theory of re-
flectance. This model combines the benefits of both the first and third
classes simultaneously, as it is in the form of a simple regression
model, but is derived directly from a radiative transfer model and its pa-
rameters can be obtained either from fitting or from direct measure-
ment. As an added advantage, and supported by both theory and
experimental data, the proposed model reduces to a linear model at
short wave infrared (SWIR) wavelengths offering a straightforward
method for RS of soil moisture in these bands.
2. Theoretical considerations
2.1. Kubelka–Munk theory
Kubelka and Munk (1931) (KM hereinafter) developed a simple the-
ory describing radiative transfer in an absorbing and scattering layer,
considering a downward and an upward light propagation flux (I and
J, respectively), perpendicular to the layer (Fig. 1). The diffuse (volume)
reflectance could then be modeled incorporating both absorption by soil
particles and water films and light scattering due to differences in the
refractive index between soil particles and the surrounding water and
air. The KM model can be derived directly from the radiative transfer
equation (Sandoval & Kim, 2014; Thennadil, 2008) and thus has a
good physical basis underlying its simplicity.
The KM model assumes that: (i) the layer exhibits infinite lateral
extension (so that edge effects can be neglected), (ii) the light absorbing
and scattering particles are uniformly distributed in the layer,
(iii) particle dimensions are much smaller than the layer thickness,
d, and (iv) the whole layer is homogeneously illuminated with a
monochromatic diffuse light source (Ciani, Goss, & Schwarzenbach,
2005).
Fig. 1. Visualization of the Kubelka–Munk model, where I and J are the two light fluxes in
opposite directions and k and s are the absorption and scattering coefficients, respectively
(after Ciani et al., 2005).
68 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76

The validity of these assumptions will certainly vary depending on Assuming that kair and sair are negligible, and denoting absorption
soil properties, however, because of the simple and macroscopically- and scattering coefficients of the dry soil (i.e. θ = 0) by kd and sd,
designed nature of the KM model, it has been successfully applied in Eqs. (5) and (6) can be written as:
soil remote sensing studies (Barron & Torrent, 1986; Ciani et al.,
2005). The KM model has also been used under non-isotropic condi- k ¼ kd þ kwater θ ð7Þ
tions with satisfactory results (van Gemert and Star, 1987), despite
the assumption of isotropic light scattering. s ¼ sd þ swater θ ð8Þ
The KM model consists of a pair of coupled differential equations
describing the radiance, I (W m− 2 sr− 1) and J (W m−2 sr− 1), at a where kwater and swater are the absorption and scattering coefficients of
depth in the layer, z (m), and at a given wavelength, λ (m), with a soil water.
light absorption coefficient, k (m−1), and a light scattering coefficient, Combining Eqs. (4), (7), and (8) yields:
s (m−1):
ð1−RÞ2 kd þ kwater θ
r¼ ¼ : ð9Þ
dI ðλ; zÞ 2R sd þ swater θ
¼ −ðk þ sÞIðλ; zÞ þ sJ ðλ; zÞ ð1Þ
dz
Eq. (9) relates soil surface reflectance to volumetric soil water con-
dJ ðλ; zÞ
¼ ðk þ sÞJ ðλ; zÞ−sI ðλ; zÞ ð2Þ tent and soil optical properties (i.e. absorption and scattering coeffi-
dz
cient) of dry soil and soil water. Although the optical properties of
where k and s are the functions of the porous medium properties and λ. pure water are known, those of soil water can be different as it contains
By analytically solving these equations, reflectance (R) and transmit- dissolved organic matter and ions in addition to suspended particles
tance can be obtained (solutions not presented here) as a function of k, s, and the water itself is partially bound to the soil (Philpot, 2010).
and d. With increasing layer thickness, d, transmittance tends to zero, Hence, a more consistent and practical form of Eq. (9) is introduced in
whereas reflectance reaches a plateau as follows (Ciani et al., 2005; the following replacing the optical properties of soil water with those
Thennadil, 2008): of a wet soil.
Based on Eqs. (7) and (8), absorption and scattering coefficients of a
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s  saturated soil (θ = θs) denoted as ks and ss can be written as:
k k 2 k
R¼1þ − þ2 : ð3Þ
s s s ks ¼ kd þ kwater θs ð10Þ

Eq. (3) can be rearranged to form a “transformed reflectance” denot- ss ¼ sd þ swater θs : ð11Þ
ed by r = k/s:
Combining Eqs. (9), (10), and (11) yields the following r(θ) relation-
k ð1−RÞ2
r¼ ¼ : ð4Þ ship incorporating optical properties of the dry and saturated soils:
s 2R
   
θ θ
Variables R and r can be written in terms of each other using kd 1− þ ks
θs θs
Eqs. (3) and (4) and thus in the following model development they r¼    : ð12Þ
θ θ
will be equally considered as remotely sensible variables for estimating sd 1− þ ss
θs θs
soil water content. Eq. (4) has also been described as one form of the ef-
fective single-particle absorption thickness, or espat function (Hapke,
The absorption and scattering coefficients of a soil sample, whether
2012).
dry or wet, are directly measurable, for example, by the dilution method
as illustrated by Ciani et al. (2005) in which a mixture of soil and various
2.2. Proposed model
proportions of a white standard (e.g. barium sulfate) with known opti-
cal properties is analyzed. Nonetheless, a more convenient and practical
When applying the Kubelka–Munk theory to soil, the medium is
algorithm can be obtained when writing Eq. (12) based on the k/s ratio
thick enough (assuming the light penetration depth is in the microme-
(rather than the absolute values) which is directly related to reflectance
ter scale for optical wavelengths), and Eq. (3) can be used to describe
data through Eq. (4). Thus Eq. (12) is rearranged as follows:
light reflectance from a uniformly illuminated soil surface. Based on
this equation, reflectance is affected by the absorption and scattering co-    
θ θ
efficients (k and s) of the soil which are themselves functions of the soil σr d 1− þ rs
θs θs
particle characteristics (i.e., mineral composition, organic matter, nutri- r¼     ð13Þ
θ θ
ents, etc.) and the soil water content. σ 1− þ
θs θs
Although not strictly true (e.g., Hapke, 2012) a commonly accepted
and frequently effective assumption is that the absorption and scatter-
with:
ing coefficients of a mixed medium can be treated as a simple additive
function of coefficients of absorption and scattering of its constituents,
kd ð1−Rd Þ2
weighted in accordance with their proportion (Barron & Torrent, rd ¼ ¼ ð14Þ
sd 2Rd
1986; Ciani et al., 2005; Hu & Johnston, 2009). Given this assumption,
we can describe k and s of the soil surface as:
ks ð1−Rs Þ2
rs ¼ ¼ ð15Þ
k ¼ ksolid ð1−θs Þ þ kwater θ þ kair ðθs −θÞ ð5Þ ss 2Rs

s ¼ ssolid ð1−θs Þ þ swater θ þ sair ðθs −θÞ ð6Þ sd sd

σ¼ ¼ ð16Þ
ss sd þ swater θs
where θ is the volumetric water content, θs is the saturated volumetric
water content (here assumed equal to the soil porosity) and subscripts where Rd and Rs are the reflectance of the dry and saturated soil,
solid, water and air represent the corresponding soil components. respectively.
 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76
The advantage of Eq. (13) over Eq. (12) is that the parameters rd and
rs can be determined solely from remotely sensed reflectance data using
Eqs. (14) and (15) (i.e., without measuring the absorption and scatter-
ing coefficients). As Eq. (16) states, the remaining parameter, σ, ranges
between 0 and 1, serving as a shape parameter describing the concavity
of the r(θ) function (see Fig. 2). The parameter σ provides a metric for
the relative strength of scattering in dry and saturated soils.
For RS applications, one needs to retrieve soil water content from re-
flectance data. For such an application, Eq. (13) can be solved explicitly
for soil water content or saturation degree (i.e. θ/θs) as:
θ σ ðr−r d Þ
¼ :
θs r s −r þ σ ðr−r d Þ
As can be seen in Eq. (16), σ approaches unity when scattering of soil
water is negligible compared to that of dry soil. This condition may be
met at strong water absorbing wavelengths such as SWIR (e.g. bands
5 and 7 of the Landsat TM and ETM+ satellites) where Eq. (17) is re-
duced to a simple linear r(θ) relationship written:
θ r−r d
¼ :
θs r s −r d
Therefore, using Eq. (18), soil water content or saturation degree can
be simply retrieved from remotely sensed reflectance data at appropri-
ate SWIR bands. Although the developed theory is based on volumetric
water content, transformation between volumetric and gravimetric
water content is through multiplication by the bulk density, which is
canceled in Eqs. (17) and (18), thus the estimated soil water content
can be either volumetric or gravimetric depending on the units of θs.
2.3. Fresnel reflectance consideration
It is worthwhile to note that the KM model describes volume reflec-
tance and does not take the so-called Fresnel reflectance into account.
Fresnel reflectance occurs at the air–soil interface, especially from sur-
face water films due to differences in the refractive indices of the soil
surface and surrounding air (Philpot, 2010). Hence, the Fresnel reflec-
tance shows a positive correlation with soil water content opposite to
volume reflectance (i.e. the decreasing trend shown in Fig. 2) and thus
can provide a new physical explanation for the reverse (increasing)
trend of total reflectance with water content increase commonly
Rd
Reflectance (R)
=
1
=
0.
5
=
0.
1
Rs
s s
Soil water content ( )
Fig. 2. Mathematical behavior of Eq. (13) describing reflectance, R(θ) (left), or transformed reflectance, r(θ) (right), relationship. Note that r and R can be written in terms of each other
using Eqs. (3) or (4). The terms Rd and Rs are the reflectance and rd and rs are the transformed reflectance of the dry and saturated soil.
69
exhibited at visible bands for high water content values (Liu et al.,
2002; Neema, Shah, & Patel, 1987).
Fresnel reflectance is typically only a small fraction of the total re-
flectance. An exception occurs at extreme saturation (or oversatura-
tion), when water at the surface forms a nearly continuous film over
the soil particles, in which case the direct (specular) component of
Fresnel reflectance can dominate and result in a brief increase in total
reflectance. The effect can also be seen on a beach on a sunny day imme-
diately after a wave recedes. In that instance reflectance can increase
briefly as soil water content increases. This, however, is an extreme
case. This phenomenon can be observed in Fig. 3 showing images of a
dark masonry sand when dry, wet and oversaturated. While the wet
ð17Þ
sand image looks darker than the dry sand image, the surface water
films partly brighten the oversaturated sand image.
Although the Fresnel reflectance is commonly negligible compared
to the volume reflectance (i.e. the total reflectance in various wave-
lengths usually decreases when soil moisture increases), Philpot's
(2010) formulation (his Eq. (1)) indicates that the Fresnel reflectance
may become relatively significant at high water contents when dry
soil reflectance is small such as for the blue and green bands. Thus a
total reflectance model is introduced here for such conditions by adding
ð18Þ the Fresnel reflectance to the volume reflectance from the KM model
predictions (from Eq. (3)) written:
 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
nw −na
R¼ θ þ 1 þ r ðθÞ− ½r ðθÞ2 þ 2r ðθÞ ð19Þ
nw þ na
where the first term in the right-hand side is an approximate estimation
of the Fresnel reflectance based on Philpot's (2010) suggestion in which
volumetric water content (θ) is considered as the fraction of surface
area from which light is reflected directly from water films. Terms nw
and na are refractive indices of water (≈1.33) and air (≈1), respective-
ly, and r(θ) is the transformed volume reflectance described by Eq. (13).
The significance of the Fresnel reflectance described by Eq. (19) is
graphically presented in Fig. 4 where the Fresnel reflectance is the
same for both bands and it modifies the general behavior of the R–θ re-
lationship in the blue band yielding non-unique reflectance values
(i.e., minimum value at mid-range water content), while the monoton-
ically decreasing behavior in SWIR region is dominated by the volume
reflectance.
Although Eq. (19) improves the modeling capability for visible
bands, it is less applicable for the following reasons: i) The R–r relation-
ship in Eq. (19) is no longer universal as in the case in Eqs. (3) and (4)
Transformed reflectance ( r)
rs
1
0.
=
5
0.
=
=1
rd
Soil water content ( )
70 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76
Fig. 3. Images of a dark masonry sand: a) wet sand with dry surface. b) Wet sand surface with no water apparent in pore spaces; reflectance is lower although there is some water at the
surface. c) Fully saturated sand with liquid water between particles and draping over the surface. Average reflectance increases because of specular reflection from sand grain-oriented
water facets.
thus the variable r cannot be considered as an independent remotely
sensed variable (i.e. its solution requires soil water content), ii) the
parameters of Eq. (19) can no longer be physically defined as in
Eqs. (14)–(16), and iii) the soil water content in Eq. (19) cannot be ex-
plicitly solved as a function of R. Hence, Eq. (19) is not in line with our
objective in this paper and has been introduced here only to provide a
general explanation for the R–θ relationship at various wavelengths
and also to illustrate some discrepancies in Eq. (13) for visible bands.
3. Results and discussions
Accuracy of the proposed non-linear (Eqs. (13) and (19)) and linear
(Eq. (18)) models was evaluated using two laboratory datasets of: i)
Lobell and Asner (2002) and ii) Whiting et al. (2004), presented and
discussed in the following.
Total reflectance
Fresnel reflectance
KM reflectance
Reflectance (R)
SWIR
Blue
Soil water content (θ )
Fig. 4. Hypothetical reflectance–water content relationship based on Eq. (19) describing
the Fresnel, volume (KM), and total reflectances in the blue and SWIR bands. Note the
Fresnel reflectance is the same for both bands and the minimum reflectance for the blue
band occurs at a mid-range water content whereas for SWIR it occurs at saturation.
3.1. Evaluations using Lobell and Asner (2002) measurements
The Lobell and Asner (2002) data presented in Fig. 5 includes solar
domain (i.e. wavelength between 350 and 2500 nm) spectral reflec-
tance from four soil samples of different soil taxonomy orders of
Aridisol, Andisol, Mollisol, and Entisol. The soils were specifically chosen
by Lobell and Asner (2002) to span a broad range of properties such as
porosity and organic carbon (shown in detail in their Table 2). The di-
versity of soil optical properties exhibited in Fig. 5 illustrates how the
proposed models work for a range of soils.
To validate the models, the experimental data of Fig. 5 were evaluated
at six wavelengths corresponding to various bands of Landsat TM and
ETM+ satellite, including band 1 (blue, 480 nm), band 2 (green,
560 nm), band 3 (red, 660 nm), band 4 (near infrared, 830 nm), band
5 (SWIR, 1650 nm), and band 7 (SWIR, 2210 nm). Eq. (19) provides a
good fit for all soils in all bands. As an illustration we present results for
the Andisol and Mollisol in Fig. 6, which were the most challenging
soils for fitting Eq. (13), especially in the blue and green bands. As Fig. 6
indicates and is also well recognized from previous studies (Lobell &
Asner, 2002; Whiting et al., 2004), the shorter wavelength visible bands
are less sensitive to changes in soil moisture. Therefore, the improved ac-
curacy of Eq. (19) in the visible bands does not add significant value to
the RS of soil moisture and due to the additional complexity as described
above, Eq. (19) is less attractive compared to Eqs. (13) and (18) which
provide similar accuracy in the near infrared and SWIR bands.
Fig. 7 shows the transformed reflectance (Eq. (4)) of the first four
bands together with Eq. (13) predictions for the four soils. Eq. (13)
was fit by applying measured values of θs and the other parameters ob-
tained via least-squares optimization. Fig. 7 demonstrates a generally
good performance by Eq. (13) especially for the Aridisol and Entisol.
The Andisol and Mollisol were not well described in the blue and
green bands using Eq. (13) especially at high water contents where
the reflectance exhibits the reverse trend with increasing water content.
As discussed earlier, this discrepancy may arise from neglecting the
Fresnel reflectance. To illustrate this possibility, Eq. (19) predictions
were also plotted in Fig. 7 for the cases discussed. Note that the modeled
reflectance using Eq. (19) was transformed in the same way as the mea-
sured data were transformed using Eq. (4). This transformation is only
to facilitate a comparable presentation in the r–θ space and may not
make sense physically since Eq. (4) was derived only for the KM
(volume) reflectance.
 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76 71

0.30 0.30

0.25 0.25
θ=0 θ=0
0.20 0.20
Reflectance
0.022 0.044
0.15 0.15

0.058 0.087
0.10 0.131 0.10 0.131
0.167 0.235
0.239 0.409
0.05 0.05
0.319 0.584
Aridisol Andisol
0.00 0.00
500 1000 1500 2000 2500 3000 500 1000 1500 2000 2500 3000
0.5 0.5

0.4 0.4
θ=0
θ=0
Reflectance

0.3 0.3
0.024
0.024
0.2 0.058 0.2
0.058
0.086
0.086
0.149
0.1 0.1 0.149
0.341
0.456 0.341
0.696 0.456
Mollisol Entisol
0.0 0.0
500 1000 1500 2000 2500 3000 500 1000 1500 2000 2500 3000
Wavelength (nm) Wavelength (nm)

Fig. 5. Spectral reflectance measured within the solar domain from four different soils having various levels of volumetric soil water content (Lobell & Asner, 2002).

The different behavior of the Andisol and Mollisol may be related to
their porosity and organic carbon content which are significantly higher
than those of the Aridisol and Entisol (See Table 2 of Lobell & Asner,
2002). The role of soil porosity in the Fresnel reflectance calculation is
seen in Eq. (19). Organic matter in soil can also affect spectral signals
and to some degree mask the effect of soil moisture content on reflec-
tance (Knadel et al., 2014).
As shown in Fig. 7, σ is a wavelength-dependent parameter, which is
positively correlated with wavelength. This variability indicates that soil
reflectance spectra undergo nonlinear change with increasing soil water

0.25
Andisol Mollisol
0.4
480 nm (band 1)
0.20 560 nm (band 2)
660 nm (band 3)
830 nm (band 4) 0.3
1650 nm (band 5)
Reflectance

0.15
2210 nm (band 7)

0.2
0.10

0.05 0.1

0.00 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Soil water content Soil water content

Fig. 6. Measured reflectance from the Andisol and Mollisol soils in six different bands (data points) in comparison with Eq. (19) fit to each set of data (solid lines).
72 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76

25
12 480 nm (band 1)
560 nm (band 2)
660 nm (band 3) 20
10 830 nm (band 4) = 0.057
Transformed reflectance = 0.157
8 15

= 0.088
6 = 0.217
10
= 0.119
4 = 0.295
= 0.371 = 0.164
5
2

Aridisol Andisol
0 0
0.0 0.1 0.2 0.3 0.4 0.0 0.2 0.4 0.6 0.8
14 14

= 0.109
12 12
= 0.100
Transformed reflectance

10 10
= 0.114

8
= 0.162
8
= 0.134
6 6

4 4 = 0.289
= 0.193
= 0.431
2 2

Mollisol Entisol
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Soil water content Soil water content

Fig. 7. Transformed reflectance measurements of Lobell and Asner (2002) in bands 1, 2, 3, and 4 (data points) in comparison with fitted predictions from Eqs. (13) (solid lines) and (19)
(dashed lines).

content and thus the spectral signatures of moist soils usually require at different moisture levels in Fig. 5, for example, in the Entisol soil
additional information beyond the dry reflectance values in order to where the dry and saturated ends have a substantially different spectral
be solved (Torrent & Barrón, 2008). This response is seen in the spectra shape, σ exhibits a large variation in different bands.

6 6

Aridisol
5 Andisol
σ = 0.528
5
Mollisol
Transformed reflectance

Entisol
4
σ = 0.331
4

3 3

2 2

1 1

1650 nm (band 5) 2210 nm (band 7)

0 0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
Soil water content Soil water content

Fig. 8. Transformed reflectance values from Lobell and Asner (2002) measured in bands 5 and 7 (data points) in comparison with the best fit of Eq. (18) (σ = 1.0, solid lines) and Eq. (13)
(σ b 1.0, dashed lines).
 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76 73

0.8
2002; Whiting et al., 2004). However, the proposed r–θ space illustrated
Aridisol, RMSE = 0.005
Andisol, RMSE = 0.036 in Fig. 8 exhibits the sensitivity more clearly than the R–θ space shown
Mollisol, RMSE = 0.030 in Fig. 6.
Entisol, RMSE = 0.012
Estimated water content

0.6
3.2. Evaluations using Whiting et al. (2004) measurements

Because the optical properties of soils depend on their constituent

0.4
0.2
0.0
0.0 0.2 0.4 0.6
Measured water content
Fig. 9. Lobell and Asner (2002) measured volumetric water content compared with soil
water content estimates from Eq. (18) using measurements from band 7 (SWIR,
2210 nm).
Fig. 8 presents the transformed reflectance data in the two most sen-
sitive SWIR bands. The linear r(θ) relationship verifies the general ade-
quacy of Eq. (18) for these soils. Nonetheless, the Andisol data especially
in band 5 still show a non-linear behavior and thus could be fit better
using Eq. (13). This suggests something peculiar about the reflectance
of Andisols, which are formed from volcanic ash and commonly contain
high proportions of glass and amorphous colloidal materials (Soil
Survey Staff, 1999). In terms of Eq. (16), the high porosity and low dry
soil scattering coefficient (sd) common for Andisols may be reasons for
the non-linear behavior resulting in σ b 1.
The capability of the linear model (Eq. (18)) for estimating water
content from reflectance data from band 7 (2210 nm) was also quanti-
tatively examined using root mean squared error (RMSE) estimates. The
RMSE values shown in Fig. 9 were all below 4% demonstrating the pre-
dictive power of the proposed model.
Comparison of Figs. 7 and 8 indicates a greater potential of SWIR
bands than the visible and near infrared bands for RS of soil moisture,
because reflectance values in the SWIR bands correspond to unique
levels of soil moisture. This sensitivity for the SWIR bands was also re-
ported in previous studies (Haubrock et al., 2008; Lobell & Asner,
material physical and chemical properties, it is expected that parame-
ters of the proposed models should also be spatially variable over a het-
erogeneous area. Because reduction of the calibration requirements was
a goal of this study, we examined to what extend a single calibration
equation could represent a relatively large heterogeneous area. This
analysis was performed using data from Whiting et al. (2004), including
laboratory measurements of reflectance from various soil samples col-
lected from two distinct Mediterranean regions (15 soil samples per re-
0.8 gion) including: i) a 25 km2 area near the city of Lemoore (CA, USA) and
ii) a 27 km2 area around the town of Tomelloso (Castilla-La Mancha,
Spain). In this study, we used the Whiting et al. (2004) calibration
(model) dataset, including a total of 60 soil samples (2 regions × 5
sites per region × 3 samples per site × 2 replicates per sample). These
two regions provided soils of varying texture and salt content as sum-
marized in Table 2 of the original paper. The samples in Lemoore cov-
ered three textural soil classes of clay loam, sandy clay loam and silty
clay loam. The Tomelloso soils were also from three different soil classes
of loam, sandy loam and silt loam. Organic matter contents were low in
both regions, being less than 2% for Tomelloso and 1% in Lemoore.
Measured data in the blue and red bands are presented in Fig. 10 to-
gether with Eq. (13) fit representing well the general shape of the re-
flectance–water content relationship in these soils. In the blue band,
there are some discrepancies due to neglecting the Fresnel reflectance
as discussed earlier. The observed scattering of data around the model
is obviously due to spatial variability of the soils included. While the
sampling area in the two regions was similar, the variability of the
Tomelloso soils is considerably higher than those of Lemoore. Whiting
et al. (2004) discovered that the broader range of particle sizes and larg-
er pore space in the lighter textures of the Tomelloso soils is responsible
for this larger variability.
Fig. 11 includes measured reflectance data of Whiting et al. (2004) in
the SWIR bands in comparison with Eq. (18) plotted as a function of
gravimetric water content. The linear relationship predicted by the pre-
sented theory also describes these soils but with more variation in the
Tomelloso soils. The data presented in Fig. 11 exhibits less scattering
compared to those of Fig. 10, especially in band 7. This result is in

8 10
480 nm (band 1)
660 nm (band 3)
8
6
Transformed reflectance

σ = 0.071
6 σ = 0.042
4

4
σ = 0.094
2 σ = 0.100
2

Lemoore Tomelloso
0 0
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
Soil water content Soil water content

Fig. 10. Transformed reflectance of Lemoore- and Tomelloso-area soils as a function of gravimetric soil water content from Whiting et al. (2004) in bands 1 and 3 (data points) in com-
parison with the best fit of Eq. (13) (solid lines).
74 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76

2.5 4
Lemoore
Tomelloso
2.0
3
Transformed reflectance

1.5

1.0

1
0.5

1650 nm (band 5) 2210 nm (band 7)

0.0 0
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
Soil water content Soil water content

Fig. 11. Transformed reflectance data of Lemoore- and Tomelloso-area soils plotted against gravimetric soil water content from Whiting et al. (2004) in bands 5 and 7 (data points) in
comparison with the best fit of Eq. (18) (σ = 1.0, solid lines).

agreement with previous studies (e.g. Lobell & Asner, 2002) which indi-
cated the reflectance variability as a consequence of soil diversity is re-
duced in the SWIR wavelengths, pointing to another major advantage of
these bands for RS of soil moisture. The smaller spatial variability of the
SWIR reflectance coupled with the linear model resulted in reasonable
estimates of soil moisture (see Fig. 12) even in these diverse soil tex-
tures using a single calibration equation. Estimation errors could be fur-
ther reduced by developing individual calibration equations for each
soil or sub-region.
4. Conclusions and future perspective
Results of this study indicate that visible wavelengths are less attrac-
tive for remote sensing of soil moisture due to transparency of water
and the consequent insensitivity of reflectance values to soil moisture
variation. Soil is generally darker when wet, but the reduction in reflec-
tance levels off quickly providing much less sensitivity in the wetter re-
gion of soil moisture. This behavior is especially the case in the blue and
0.8
Lemoore, RMSE = 0.067
Tomelloso, RMSE = 0.077
Estimated water content
green bands where soil initially darkens when moistened up to a critical
saturation level and brightens afterward. Eq. (19) provides a physical
explanation for this phenomenon based on relative strengths of the sur-
face (Fresnel) reflectance and volume reflectance. Estimation of soil
moisture from reflectance data under these conditions becomes even
more complicated, because any reflectance value in these bands may
correspond to two different (non-unique) water contents; one below
the critical saturation level and the other above.
This study further verifies previous findings that the SWIR wave-
lengths provide the optimal bands in the solar domain (i.e. wavelength
between 350 and 2500 nm) for RS of soil moisture. Optimal because the
SWIR wavelengths provide the highest sensitivity of reflectance to soil
moisture variation and the least soil-dependence (i.e. spatial variability)
in reflectance values among the solar domain. An additional advantage
discovered in this study deals with applicability of the proposed models
(i.e., Eqs. (17) and (18)), between the transformed reflectance and soil
water content in the SWIR bands. The simplicity and physical basis of
Eq. (18) together with its accuracy at the SWIR wavelengths, especially
around 2210 nm (i.e. band 7 of the Landsat and MODIS satellites) make
it a promising tool for RS of soil moisture using optical airborne/satellite
imaging with minimal calibration requirements.
Another significant advantage of the proposed model is that it is not
confined to a single band or wavelength. Properly calibrated, it is capa-
ble of describing the response of reflectance to water content over the
full optical range. As mentioned, the greatest sensitivity to soil moisture

0.6
is encountered around 2210 nm. This band is favorably away from the
strong atmospheric water vapor absorption bands centered near
1440 nm and 1930 nm at which the EM radiation is absorbed when
0.4 passing through the atmosphere.
An obvious constraint of the model is that it requires calibration
using reflectance acquired under dry and saturated conditions. This is
not an insurmountable requirement given that our primary intent in
0.2 modeling the soil moisture–reflectance relationship has been to re-
motely retrieve surface soil moisture for agricultural and environmental
applications. Using the dry reflectance as a reference would be inappro-
priate in other applications such as mineralogy, for which the dry reflec-
0.0
0.0 0.2 0.4 0.6 0.8
tance is the goal in mineral analysis. On the other hand, in order to
Measured water content remove/correct the spectral effect of the water on the reflectance spec-
trum, which is of interest in mineralogy, a model describing the full
Fig. 12. Eq. (18) estimates of gravimetric soil water content for the Lemoore- and spectrum is commonly required. In that case, the approach presented
Tomelloso-area soils from Whiting et al. (2004) at band 7 (SWIR, 2210 nm) versus mea- here may be useful when considering rearrangement of Eq. (9) to
sured soil water content. solve for the dry soil properties (kd and sd) at all optical wavelengths.
 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76
Parameters of the proposed linear model, Eq. (18), include trans-
formed reflectance for dry and saturated soil water contents, while the
transformed reflectance itself is physically related to soil properties
through soil absorption and scattering coefficients (see Eq. (4)). There-
fore, future investigations toward linking the model parameters to soil
basic information such as texture, color or taxonomic class would be
feasible and would enhance the model's applicability using existing
soil databases (e.g. USDA–NRCS soil maps). Furthermore, the linear na-
ture of Eq. (18) suggests another simple method for its application,
without the need for ground measurements, which could be to collect
time series of reflectance data at a given location, including the full
range of field saturation (i.e., from air-dry to a maximum moisture con-
tent occurring over a long period of time). With these data, transformed
using Eq. (4), parameters of Eq. (18) could then be resolved as the max-
imum and minimum values of the transformed reflectance.
This study was a first step toward focusing on the theoretical aspects
of the model and its testing under well-controlled laboratory condi-
tions. Future studies are underway to examine/extend the model for
field and large-scale applications when facing challenges such as high
degrees of heterogeneity, surface roughness, topographical features,
shadow effects, etc. All optical methods will suffer from a shallow pen-
etration depth of the optical bands, as well as limited applicability in
vegetated soils. Some of these limitations may be overcome if the meth-
od is applied in conjunction with microwave satellite data (e.g. NASA's
SMAP satellite) as might be done to improve the spatial resolution or
to provide an independent assessment of water content. Another poten-
tial application would be to use the estimated surface water content to
provide the upper boundary condition for a numerical model designed
to solve for the soil moisture profile.
Acknowledgments
Financial support was provided by Utah State University Office of Re-
search and Graduate Studies Research Catalyst (RC) Grant and by the
Utah Agricultural Experiment Station, Utah State University, Logan,
Utah 84322-4810, approved as journal paper no. 8746. We also express
our appreciation to Michael Whiting and David Lobell for sharing their
valuable data.
References
Al-Yaari, A., Wigneron, J. -P., Ducharne, A., Kerr, Y., de Rosnay, P., de Jeue, R., et al. (2014).
Global-scale evaluation of two satellite-based passive microwave soil moisture
datasets (SMOS and AMSR-E) with respect to Land Data Assimilation System esti-
mates. Remote Sensing of Environment, 149, 181–195http://dx.doi.org/10.1016/j.rse.
2014.04.006.
Bach, H., & Mauser, W. (1994). Modelling and model verification of the spectral reflec-
tance of soils under varying moisture conditions. Proceedings of IGARSS'94. Vol. 4.
(pp. 2354–2356). Pasadena: IEEE.
Barron, V., & Torrent, J. (1986). Use of the Kubelka–Munk theory to study the influence of
iron oxides on soil colour. Journal of Soil Science, 37, 499–510.
Ben-Dor, E., Chabrillat, S., Demattê, J.A.M., Taylor, G.R., Hill, J., Whiting, M.L., et al. (2009).
Using imaging spectroscopy to study soil properties. Remote Sensing of Environment,
113(Supplement 1), S38–S55.
Bogena, H.R., Huisman, J.A., Oberdorster, C., & Vereecken, H. (2007). Evaluation of a low-
cost soil water content sensor for wireless network applications. Journal of Hydrology,
344, 32–42.
Brocca, L., Melone, F., Moramarco, T., & Morbidelli, R. (2010). Spatial–temporal variability
of soil moisture and its estimation across scales. Water Resources Research, 46,
W02516http://dx.doi.org/10.1029/2009WR008016.
Carlson, T. (2007). An overview of the “triangle method” for estimating surface evapo-
transpiration and soil moisture from satellite imagery. Sensors, 7, 1612–1629.
Carlson, T.N., Gillies, R.R., & Perry, E.M. (1994). A method to make use of thermal infrared
temperature and NDVI measurements to infer surface soil water content and frac-
tional vegetation cover. Remote Sensing Reviews, 9, 161–173.
Ciani, A., Goss, K.U., & Schwarzenbach, R.P. (2005). Light penetration in soil and particu-
late minerals. European Journal of Soil Science, 56, 561–574.
Crow, W.T., Berg, A.A., Cosh, M.H., Loew, A., Mohanty, B.P., Panciera, R., et al. (2012).
Upscaling sparse ground-based soil moisture observations for the validation of
coarse-resolution satellite soil moisture products. Reviews of Geophysics, 50,
RG2002http://dx.doi.org/10.1029/2011RG000372.
75
Czajkowski, K., Goward, S.N., Stadler, S.J., & Waltz, A. (2000). Thermal remote sensing of
near surface environmental variables: application over the Oklahoma Mesonet. The
Professional Geographer, 52, 345–357.
de Jeu, R., Wagner, W., Homes, T.R.H., Dolman, A.J., van de Giesen, N.C., & Friesen, J.
(2008). Global soil moisture patterns observed by space borne microwave radiome-
ters and scatterometers. Surveys in Geophysics, 29, 399–420.
Hapke, B.W. (2012). Theory of Reflectance and Emittance Spectroscopy (2nd ed.). Cam-
bridge, England: Cambridge University Press.
Hassan Esfahani, L., Torres-Rua, A., Jensen, A., & McKee, M. (2015). Assessment of surface
soil moisture using high-resolution multi-spectral imagery and artificial neural net-
works. Remote Sensing, 7, 2627–2646.
Haubrock, S. -N., Chabrillat, S., Lemmnitz, C., & Kaufmann, H. (2008). Surface soil moisture
quantification models from reflectance data under field conditions. International
Journal of Remote Sensing, 29(1), 3–29.
Hu, X., & Johnston, W.M. (2009). Concentration additivity of coefficients for maxillofacial
elastomer pigmented to skin colors. Dental Materials, 25, 1468–1473.
Huisman, J.A., Hubbard, S.S., Redman, J.D., & Annan, A.P. (2003). Measuring soil water con-
tent with ground penetrating radar: A review. Vadose Zone Journal, 2, 476–491.
Jones, S. B., Blonquist, J. M., Jr., Robinson, D. A., Rasmussen, V. P., & Or, D. (2005). Standard-
izing characterization of electromagnetic water content sensors: Part I. methodology.
Vadose Zone Jornal, 4, 1048–1058.
Jones, S.B., Wraith, J.M., & Or, D. (2002). Time domain reflectometry measurement princi-
ples and applications. Hydrological Processes, 16, 141–153.
Knadel, M., Deng, F., Alinejadian, A., de Jonge, L.W., Moldrup, P., & Greve, M.H. (2014). The
effects of moisture conditions – from wet to hyper dry – on visible near-infrared
spectra of Danish reference soils. Soil Science Society of America Journal, 78, 422–433.
Kornelsen, K.C., & Coulibaly, P. (2013). Advances in soil moisture retrieval from synthetic
aperture radar and hydrological applications. Journal of Hydrology, 476, 460–489.
Kubelka, P., & Munk, F. (1931). Ein Beitrag zur Optik der Farbanstriche. Zeitschrift für
Technische Physik, 12, 593–601.
Landrum, C. (2013). Mapping and decomposing scale-dependent soil moisture variability
within an inner bluegrass landscape. Theses and Dissertations—Plant and Soil Sciences
(Paper 34. http://uknowledge.uky.edu/pss_etds/34).
Liu, W.D., Baret, F., Gu, X.F., Tong, Q.X., Zheng, L.F., & Zhang, B. (2002). Relating soil surface
moisture to reflectance. Remote Sensing of Environment, 81, 247–252.
Liu, W.D., Baret, F., Gu, X.F., Zhang, B., Tong, Q.X., & Zheng, L.F. (2003). Evaluation of
methods for soil surface moisture estimation from reflectance data. International
Journal of Remote Sensing, 24, 2069–2083.
Lobell, D.B., & Asner, G.P. (2002). Moisture effects on soil reflectance. Soil Science Society of
America Journal, 66, 722–727.
Merlin, O., Escorihuela, M.J., Mayoral, M.A., Hagolle, O., Al Bitar, A., & Kerr, Y. (2013). Self-
calibrated evaporation-based disaggregation of SMOS soil moisture: an evaluation
study at 3 km and 100 m resolution in Catalunya, Spain. Remote Sensing of
Environment, 130, 25–38.
Merlin, O., Walker, J.P., Chehbouni, A., & Kerr, Y. (2008). Towards deterministic downscal-
ing of SMOS soil moisture using MODIS derived soil evaporative efficiency. Remote
Sensing of Environment, 112, 3935–3946.
Muller, E., & De'camps, H. (2000). Modeling soil moisture–reflectance. Remote Sensing of
Environment, 76, 173–180.
Neema, D.L., Shah, A., & Patel, A.N. (1987). A statistical optical model for light reflection
and penetration through sand. International Journal of Remote Sensing, 8(8),
1209–1217.
Nemani, R., Pierce, L., Running, S.N., & Goward, S.N. (1993). Developing satellite-derived
estimates of surface moisture status. Journal of Applied Meteorology, 32, 548–557.
Njoku, E.G., & Entekhabi, D. (1996). Passive microwave remote sensing of soil moisture.
Journal of Hydrology, 184, 101–129.
Njoku, E.G., Jackson, T.L., Lakshmi, V., Chan, T., & Nghiem, S.V. (2003). Soil moisture retriev-
al from AMSR-E. IEEE Transactions on Geoscience and Remote Sensing, 41(2), 215–229.
Njoku, E.G., & Kong, J.A. (1977). Theory for passive microwave remote sensing of near-
surface soil moisture. Journal of Geophysical Research, 82(20), 3108–3118.
Nolet, C., Poortinga, A., Roosjen, P., Bartholomeus, H., & Ruessink, G. (2014). Measuring
and modeling the effect of surface moisture on the spectral reflectance of coastal
beach sand. Plos One, 9(11), e112151http://dx.doi.org/10.1371/journal.pone.0112151.
Ochsner, T.E., Cosh, M., Cuenca, R., Dorigo, W., Draper, C., Hagimoto, Y., et al. (2013). State
of the art in large-scale soil moisture monitoring. Soil Science Society of America
Journal, 77, 1888–1919http://dx.doi.org/10.2136/sssaj2013.03.0093.
Ojha, R., Morbidelli, R., Saltalippi, C., Flammini, A., & Govindaraju, R.S. (2014). Scaling of
surface soil moisture over heterogeneous fields subjected to a single rainfall event.
Journal of Hydrology, 516, 21–36.
Persson, M. (2005). Estimating surface soil moisture from soil color using image analysis.
Vadose Zone Journal, 4, 1119–1122.
Philpot, W.D. (2010). Spectral reflectance of wetted soils. Symposium: “Art, Science and Ap-
plications of Reflectance Spectroscopy (ASARS)”, Analytical Spectral Devices, Inc.,
February 23–25, 2010http://dx.doi.org/10.13140/2.1.2306.0169.
Piles, M., Camps, A., Vall-llossera, M., Corbella, I., Panciera, R., Rudiger, C., et al. (2011).
Downscaling SMOS-derived soil moisture using MODIS visible/infrared data. IEEE
Transactions on Geoscience and Remote Sensing, 49(9), 3156–3166.
Robinson, D.A., Campbell, C.S., Hopmans, J.W., Hornbuckle, B.K., Jones, S.B., Knight, R., et al.
(2008). Soil moisture measurements for ecological and hydrological watershed scale
observatories: A review. Vadose Zone Journal, 7, 358–389http://dx.doi.org/10.2136/
vzj2007.0143.
Sandoval, C., & Kim, A.D. (2014). Deriving Kubelka–Munk theory from radiative transport.
Journal of the Optical Society of America A, 31, 628–636.
Shi, J., Jiang, L., Zhang, L., Chen, K.S., Wigneron, J.P., Chanzy, A., et al. (2006). Physically
based estimation of bare-surface soil moisture with the passive radiometers. IEEE
Transactions on Geoscience and Remote Sensing, 44(11), 3145–3152.
76 M. Sadeghi et al. / Remote Sensing of Environment 164 (2015) 66–76
Skidmore, E.L., Dickerson, J.D., & Schimmelpfennig, H. (1975). Evaluating surface soil
water content by measuring reflectance. Soil Science Society of America Proceedings,
39, 238–242.
Soil Survey Staff (1999). Soil taxonomy: A basic system of soil classification for making and
interpreting soil surveys Natural Resources Conservation Service (2nd ed.). U.S. De-
partment of Agriculture Handbook, 436.
Thennadil, S.N. (2008). Relationship between the Kubelka–Munk scattering and radiative
transfer coefficients. Journal of the Optical Society of America A, 25, 1480–1485.
Torrent, J., V. Barrón. 2008. Diffuse Reflectance Spectroscopy. pp. 367–385. In: Methods of
Soil Analysis. Part 5. Mineralogical Methods. SSSA Book Series no. 5. Soil Sci. Soc. Am.
Madison, WI.
Twomey, S.A., Bohren, C.F., & Mergenthaler, J.L. (1986). Reflectance and albedo differences
between wet and dry surfaces. Applied Optics, 25, 431–437.
Van Gemert, M. J., & Star, W. M. (1987). Relation between the Kubelka–Munk and the
transport equation models for anisotropic scattering. Lasers in Life Sciences, 1,
287–298.
Vaz, C.M.P., Jones, S., Meding, M., & Tuller, M. (2013). Evaluation of standard calibration
functions for eight electromagnetic soil moisture sensors. Vadose Zone Journal,
12http://dx.doi.org/10.2136/vzj2012.0160.
Vereecken, H., Huisman, J.A., Bogena, H., Vanderborght, J., Vrugt, J.A., & Hopmans, J.W.
(2008). On the value of soil moisture measurements in vadose zone hydrology:
A review. Water Resources Research, 44, W00D06http://dx.doi.org/10.1029/
2008WR006829.
Verstraeten, W.W., Veroustraete, F., & Feyen, J. (2008). Assessment of evapotranspiration
and soil moisture content across different scales of observation. Sensors, 8(1), 70–117.
Verstraeten, W.W., Veroustraete, F., van der Sande, C.J., Grootaers, I., & Feyen, J. (2006).
Soil moisture retrieval using thermal inertia, determined with visible and thermal
spaceborne data, validated for European forests. Remote Sensing of Environment,
101, 299–314.
Wang, L., & Qu, J.J. (2009). Satellite remote sensing applications for surface soil moisture
monitoring: A review. Frontiers of Earth Science in China, 3(2), 237–247.
Western, A.W., Grayson, R.B., & Blöschl, G. (2002). Scaling of soil moisture—A hydrologic
perspective. Annual Review of Earth and Planetary Sciences, 30, 149–180.
Western, A.W., Zhou, S. -L., Grayson, R.B., McMahon, T.A., Blöschl, G., & Wilson, D.J. (2004).
Spatial correlation of soil moisture in small catchments and its relationship to domi-
nant spatial hydrological processes. Journal of Hydrology, 286, 113–134.
Whiting, M.L., Li, L., & Ustin, S.L. (2004). Predicting water content using Gaussian model
on soil spectra. Remote Sensing of Environment, 89, 535–552.
Wu, H., & Li, Z.L. (2009). Scale issues in remote sensing: A review on analysis, processing
and modeling. Sensors, 9(3), 1768–1793http://dx.doi.org/10.3390/s90301768.
Yang, X., Wu, J.J., Shi, P.J., & Yan, F. (2008). Modified triangle method to estimate soil mois-
ture status with MODerate resolution Imaging Spectroradiometer (MODIS) products.
Proc. Int. Archives Photogramm., Remote Sens. Spatial Inf. Sci., Beijing, China. Vol. 37,
(Part B8).
Zaman, Z., McKee, M., & Neale, C.M.U. (2012). Fusion of remotely sensed data for soil
moisture estimation using relevance vector and support vector machines.
International Journal of Remote Sensing, 33, 6516–6552.
Zhu, Y., Wang, Y., Shao, M., & Horton, R. (2011). Estimating soil water content from
surface digital image gray level measurements under visible spectrum. Canadian
Journal of Soil Science, 91, 69–76.