Remote Sensing of Environment 140 (2014) 257–266

Contents lists available at ScienceDirect

Remote Sensing of Environment

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

Capability of C-band backscattering coefficients from high-resolution

satellite SAR sensors to assess biophysical variables in paddy rice
Yoshio Inoue a,⁎, Eiji Sakaiya b, Cuizhen Wang c
a
National Institute for Agro-Environmental Sciences, Tsukuba, Ibaraki 305-8604, Japan
b
Aomori-ITC Agricultural Research Institute, Kuroishi 036-0522, Japan
c
University of Missouri, Columbia, MO 65211, USA

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

Article history: High-resolution (ca. 1 m) synthetic aperture radar (SAR) sensors have great potential for all-weather monitoring
Received 22 June 2013 of crop biophysical variables in small and mosaic crop fields in Asia. Rice is the most important staple crop in
Received in revised form 30 August 2013 monsoon Asia, and the timely monitoring of rice growth is critical for precision farming and the assessment of
Accepted 1 September 2013
productivity. The objective of this study was to determine the potential capability of backscattering coefficients
Available online 28 September 2013
(σ0) from satellite C-band SAR sensors for the assessment of biophysical variables in rice. SAR images were ac-
Keywords:
quired by a Radarsat-2 sensor in spotlight mode during the critical growth stages over 4 years in one of the
C-band major rice-producing areas of Japan. Detailed plant biophysical measurements were made concurrently with
fAPAR the SAR observations. The seasonal consistency of C-band σ0 was clearly demonstrated. The baseline σ0 values
LAI (minimum σ0 for zero-biomass paddy fields) were determined to be −28.5 dB in VH and −21.1 dB in HH
Microwave and VV, respectively. The dynamic change in σ0 during the full range of rice growth was similar (ca. 12 dB) in
Paddy rice all polarizations. A comprehensive analysis revealed the response of C-band σ0 to biophysical canopy variables.
Radarsat-2 High or moderate sensitivity of σ0 to canopy height, water content, or chlorophyll content was superficial and
SAR
was attributable to the change in leaf biomass and structure. Both the leaf area index (LAI) and leaf biomass
Spotlight mode
were significantly and consistently correlated with σ0 throughout all growth stages. These relationships were
expressed by exponential curves with high coefficients of determination, although σ0 saturates at around a LAI
of 3 and a leaf biomass of 180 gDW m−2. The response of σ0 to total biomass was expressed by an exponential
function with a high coefficient of determination, but the sensitivity was clear only within the lower 20% range
of the seasonal maximum biomass. The C-band σ0 had the highest correlation with fAPAR, and the σ0–fAPAR re-
lationship was linear throughout the growth stages. The results suggest the suitability of C-band σ0 for the assess-
ment of LAI or fAPAR and show promise for the timely monitoring of rice growth by C-band SAR and/or through
its constellation with optical sensors.
© 2013 Elsevier Inc. All rights reserved.

1. Introduction
A wide range of satellite sensors are available in the optical, thermal,
and microwave spectral domains for the observation of terrestrial eco-
systems. However, agricultural applications are highly demanding for
advanced specifications in spatial, spectral, and temporal resolutions
(Inoue, 2003; Moran, Inoue, & Barnes, 1997). In crop monitoring for
precision farming and yield forecasting, the timely observation of
plant biophysical and ecophysiological status (e.g., leaf area, biomass,
and chlorophyll content) is critical (Doraiswamy et al., 2004; Inoue,
2003; Moran, Inoue, et al., 1997). High spatial resolution is required
when observing regions with small and mosaic crop fields, as in many
countries in Asia.
⁎ Corresponding author. Tel.: +81 29 838 8220; fax: +81 29 838 8199.
E-mail address: yinoue@affrc.go.jp (Y. Inoue).
Rice (Oryza sativa L.) is an important stable crop in monsoon Asia, but
rice productivity is strongly affected by water, temperature, and solar
radiation. Therefore, timely and high-resolution observation is essential
for monitoring rice crops in Asia. Accordingly, SAR sensors have great
potential for the timely assessment of biophysical and ecophysiological
variables of rice in Asia (e.g., Le Toan et al., 1997; Ribbes & Le Toan,
1999). The extraction of rice fields is relatively robust due to the specular
features under flooded surface conditions (e.g., Choudhury, Chakraborty,
Santra, & Parihar, 2012; Kurosu, Fujita, & Chiba, 1997; Ribbes & Le Toan,
1999). The combined use of multiple polarizations can effectively classify
agricultural fields (e.g., Bouvet, Le Toan, & Lam-Dao, 2009; McNairn,
Champagne, Shang, Holmstrom, & Reichert, 2009). Cropping systems or
agricultural management practices in rice-growing regions can be identi-
fied successfully from SAR observations (e.g., Bouvet & Le Toan 2011;
Lopez-Sanchez, Ballester-Berman, & Hajnsek, 2011). However, quantita-
tive assessments of biophysical and ecophysiological plant variables
using SAR signatures remain challenging.

0034-4257/$ – see front matter © 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.rse.2013.09.001
258 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266
Lopez-Sanchez and Ballester-Berman (2009) highlighted the
great potential of microwave remote sensing for monitoring crop
biophysical variables by reviewing various experimental studies using
ground-based scatterometers (e.g., Bouman, 1991; Inoue et al., 2002;
Prévot, Campion, & Guyot, 1993). For example, Inoue et al. (2002)
indicated the most suitable combination of frequency band, polariza-
tion, and incidence angle for the assessment of each biophysical variable
of rice by analyzing a comprehensive dataset of backscattering coeffi-
cients in five frequency bands (Ka, Ku, X, C, and L), full polarizations
(VV, VH, HV, HH), and four incidence angles (25°, 35°, 45°, 55°) during
a full growing-period of paddy rice. Among the major results, the C-
band σ0 was suggested to be the most useful for estimating the leaf
area index (LAI) and whole-canopy variables such as biomass.
Most previous studies using satellite C-band SAR sensors have sug-
gested that C-band σ0 may be able to detect the seasonal changes in
crop growth based on the similarity of the seasonal trends in σ0 and
plant height (e.g., Baghdadi, Boyer, Todoroff, El Hajj, & Bégué, 2009;
Chakraborty, Manjunath, Panigrahy, Kundu, & Parihar, 2005; Fieuzal,
Baup, & Marais-Sicre, 2012; Ribbes & Le Toan, 1999; Shao et al.,
2001). However, due to the limited biophysical and ecophysiological
data and limited spatial resolutions of SAR sensors, their results re-
main ambiguous, and the accuracy and consistency seem insufficient
for operational applications. Additionally, the apparent similarity of
seasonal trends in SAR signatures and plant height does not imply a
real relationship but an indirect or superficial relationship, consider-
ing that the C-band σ0 was similar before and after the harvesting
of a rice canopy of full height (1 m). More importantly, indications
of the similarity of seasonal trends in both σ0 and plant height are
too preliminary to evaluate the potential of SAR sensors for crop
diagnostics at critical growth stages and the assessment of crop
productivity.
One reason for such uncertainty lies in the insufficient ground-based
data relating to canopy biophysical or ecophysiological status. Some
sophisticated parameters such as entropy and alpha derived from
multi-polarization images have sometimes been used to derive more ac-
curate and/or consistent information on canopy and surface conditions
(Lopez-Sanchez, Hajnsek, & Ballester-Berman, 2012, Lopez-Sanchez
et al., 2011), but further investigations based on detailed ground-based
data are needed for a clear interpretation of the behavior of such param-
eters. Backscattering process models are useful for the interpretation of
measured SAR signatures (e.g., Le Toan et al., 1997; Wang et al., 2009),
but their ability is still very limited because of difficulties in the modeling
of scattering processes, the determination of key parameters, and the
acquisition of various input data (e.g., 3-D canopy structure, plant water
content, and soil roughness). Considering the state-of-the-art in these re-
search fields, new investigations need to be based on a detailed sensitivity
analysis based on accurate canopy biophysical and ecophysiological
measurements.
Another reason for the ambiguity of earlier studies using satellite
SAR sensors is the low spatial resolution. Spatial resolution on the
order of 1 m has been suggested for advanced investigations to provide
more consistent, reliable, and innovative results (Inoue, 2003, Moran,
Inoue, et al., 1997). Such high spatial resolution is critical, especially in
the small and mosaic crop fields of monsoon Asia, but such spatial
resolution has not been available for long. The latest SAR sensors in
the C-band (Radarsat-2) and X-band (COSMO-SkyMed and TerraSAR-X)
have achieved the required high spatial resolution in spotlight mode,
but in-depth studies of the capability of such high-resolution data are
still limited (e.g., Fieuzal et al., 2012; Inoue & Sakaiya, 2013; Santi et al.,
2012).
Thus, the objectives of this study were to examine the consistency
of a high-resolution C-band SAR sensor (i.e., Radarsat-2, spotlight
mode) for rice crop monitoring, to investigate the comprehensive rela-
tionship of σ0 with canopy biophysical variables, and to explore the new
capabilities of the C-band sensor for assessment of crop growth and
yield.
2. Materials and methods
2.1. Test site
One of the major rice-growing regions in northeast Japan (Tsugaru
Plain, Aomori Prefecture) was selected as a study area (center: 40°36′
20.74″N, 140°33′36.02″E). The land is flat, and rice varieties and crop
management practices are relatively uniform in the area. In the region,
rice is grown once a year during May–September. The mean air temper-
ature and total precipitation for the period are 18.6 °C and 513 mm,
respectively.
In general, paddy fields are flooded and smoothed several days be-
fore transplanting, and rice seedlings are transplanted in late May
using a rice-planting machine. The normal dates for the panicle initia-
tion stage, heading stage, and maturity stage (harvesting) are mid-
July, early August, and mid-September, respectively (Fig. 1). A single
rice variety (O. sativa L. japonica, variety: Tsugaru Roman) is grown in
the study area. In general, a bundle of 3–5 seedlings (hill) of about
15 cm long are machine planted at a spacing of 30 × 15 cm under
flooded conditions. After transplanting, paddy fields are irrigated con-
tinuously until the mid-maturing stage, and the soil surface of the
paddy fields is under flooded conditions during most growing periods.
Normally, plant growth within each field is highly uniform (coefficient
of variance, CV b 10%), but between-field variability in growth and
yield is significant because of differences in the soil condition and farm-
ing management practices. The majority of paddy fields in the region
are 30 × 100 m in size, but the orientation of fields (i.e., row direction)
is not identical. The field size is much smaller than that in the United
States or European countries but is typical for Japan and most Asian
countries.
2.2. Acquisition and processing of C-band SAR images
High-resolution SAR images in the C-band (5.4 GHz) were acquired
using the Radarsat-2 sensor in the spotlight mode. As discussed in
Section 1, the spotlight mode was considered most suitable for investigat-
ing the relationship of SAR signatures with rice biophysical variables be-
cause degradation of spatial resolution due to noise-reduction processing
is unavoidable. All images were taken in spotlight mode, i.e., at a spatial
resolution of 1 × 1 m on the ground.
Seven images were obtained over the vegetative to maturing growth
periods in the growing seasons of 2009–2012: on 06 September 2009,
15 July 2010, 16 July 2010, 10 July 2011, 30 September 2011, 27 June
2012, and 24 September 2012. These periods cover the critical growth
stages such as panicle initiation, heading, and maturing, when quantita-
tive information on biophysical variables is most useful for growth
diagnosis and yield prediction. Stem density, LAI, biomass, fraction of
photosynthetically active radiation (fAPAR), and water and chlorophyll
contents were the key canopy variables. The time of measurement was
around 17:30 LST for all seven images. The size of the observation area
was approximately 8 × 20 km.
Measurement configurations such as polarization and incidence
angles were carefully selected from the possible configurations during
each targeted time window based on the results of our previous study
(Inoue et al., 2002). Because only a single-polarization mode was avail-
able due to the inherent limitation of the spotlight mode, VH polariza-
tion was selected, with some additional polarizations (VV and HH)
also used for comparative analysis. Accordingly, intermediate incidence
angles (25–35°) were selected from those available.
All image data were converted to σ0 signatures based on the
radiometric parameters provided for each dataset using the Next ESA
SAR Toolbox (NEST) 4B (European Space Agency). All images were
georeferenced using a high-resolution airborne image (1 m) obtained
by the CASI hyperspectral sensor. The boundaries of individual paddy
fields accurately matched the GIS polygon data from GIS for the whole
region. A 3 × 3 enhanced Lee filter was applied to each image to reduce
 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266
Fig. 1. Examples of rice canopies in the study area; nadir and side view of typical rice canopies at (1) transplanting, (2) early-vegetative, (3) late-vegetative, and (4) mid-maturing stages,
respectively.
speckle noise. The high spatial resolution of the spotlight mode allowed
the extraction of signatures for the exact area of interest in each paddy
field. We compared the average σ0 values for the area of interest
(around 300 pixels for uniform surfaces) using the raw σ0 data and
the processed data by enhanced Lee filters (with 3 × 3 and 5 × 5), but
their differences were negligible (b 0.02 dB). From these results, we
assumed that the speckle reduction filters may not have significant
impacts on the spatial average values in case the surfaces are uniform
and sufficient numbers of pixels are averaged. The σ0 values were
extracted from the individual areas of interest in paddy fields, and
plant samples were taken as well as from nearby water surfaces (rivers
and ponds). Such nearby water surfaces were assumed to be usable as a
reference throughout a growing season because there was no difference
between the calm water surfaces and flooded paddy fields just before
transplanting.
2.3. Biophysical measurements of rice canopies
Detailed biophysical and ecophysiological measurements were
made concurrently with the SAR observations at 41, 52, 12, and 24
paddy fields in 2009, 2010, 2011, and 2012, respectively. Fig. 1 shows
the nadir and side view of typical rice canopies at the transplanting,
early-vegetative, late-vegetative, and mid-maturing stages. Hill density,
plant height, and water depth were recorded for each field.
Rice plants in each field were very uniform (e.g., the maximum coef-
ficient of variance was 10%). Therefore, five representative hills (a bun-
dle of plants) were sampled from each paddy field to characterize the
biophysical structure of the canopy. The wet biomass and dry biomass
of leaves, stems, panicles, and whole plants were determined by de-
structive measurements of the sampled plants based on normal gravi-
metric processes. Canopy structural variables such as stem density,
LAI, number of leaves per stem, and the vertical position of panicles
and leaves, and major morphological variables such as panicle size,
leaf size (length, width, and thickness), and stem diameter were mea-
sured for each sample. The chlorophyll content of three fully expanded
leaves in the upper canopy layer was measured using a SPAD 502 chloro-
phyll meter (Minolta) and then averaged for each canopy. The fAPAR was
derived from the ascending and descending values of photosynthetic
photon flux density (PPFD) measured at the top and bottom of a canopy
using a line PPFD sensor (LI-191, Li-Cor).
259
These structural, morphological, and physiological variables were used
to investigate the sensitivity of σ0 to rice canopies. They can be used as in-
puts to physically based scattering models or for the interpretation of var-
ious responses of σ0 to rice-canopy conditions in a range of studies.
2.4. Physically-based backscattering models for a rice canopy
In support of the analysis of the dataset obtained by the satellite SAR
sensor and ground-based measurements, we used two types of canopy
backscattering models; i.e., 1) a simple process-based model (the water
cloud model; Attema & Ulaby, 1978; Prévot et al., 1993), and 2) a structur-
al canopy backscattering model (Karam et al., 1995; Wang et al., 2009).
2.4.1. Water cloud model
In the water cloud (WC) model, a crop canopy is represented by two
layers, i.e., vegetation layer and soil layer. The backscattering coefficient
σ0 for the whole canopy is expressed by the following equations.
0 0 2 0
σ ¼σ veg þτ σ soil ð1Þ


σ
0
¼ AV1 cosθ 1−τ
2
ð2Þ
veg
2
τ ¼ exp½−2BV2 = cosθ ð3Þ
where, σ0veg and σ0soil are the backscattering coefficients in power units
(m2 m−2) for vegetation and soil, respectively; τ2 is the two-way atten-
uation through the canopy; θ is the incident angle; V1 and V2 are the de-
scriptors of the canopy; and A and B are coefficients that depend on
canopy type. σ0soil is usually expressed as a function of soil moisture.
However, in paddy rice, the σ0soil can be simplified to be constant
(σ0BG) due to the flooded surface conditions (Inoue et al., 2002). By as-
suming the V1 = V2 = V = LAI or TFW (Inoue et al., 2002; Prévot et al.,
1993), the backscatter for the rice paddy can be expressed (in dB) as
follows:
n o
0 0
σ ¼ 10 log AV cosθð1− exp½−2BV= cosθÞ þ exp½−2BV= cosθÞσ BG : ð4Þ
In this study, we applied iterative parameterization to determine the
coefficients A and B of the WC model using LAI as a key vegetation
260 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266

variable (V = LAI). The σ0BG was derived from the image data. In general,
the degree of fitness of measured data (LAI and σ0 in this case) to theoret-
ical models can be a useful indicator to infer the physical backscattering
processes in a canopy.
2.4.2. Structural canopy scattering model
This model is based on a radiative transfer model for forests (Karam
et al., 1995), so modified to express the scattering processes in a paddy
rice canopy with several assumptions; 1) the ground surface is a smooth
surface with dielectric constant of water since paddy fields are flooded
during the growing season, 2) a rice canopy consists of three layers
(i.e., panicle-, leaf-, and stem-layers), 3) panicles and stems are expressed
as short cylinders, 4) leaves are expressed as narrow and long ellipses,
and 5) the leaf angle distribution is expressed by a specific probability
distribution function with a few parameters. Accordingly, the total back-
scattering coefficient from a canopy (σtotal) is expressed as a linear com-
bination of volume scattering from each component, its double bounce
with ground, and ground surface scattering (in power unit);
σ total ¼ σ leaf þ σ leaf–ground þ σ stem þ σ stem–ground þ σ panicle
þ σ panicle–ground þ σ ground
where the σleaf, σstem, and σpanicle are volume scattering of leaves, stems,
and panicles, respectively. The σleaf–ground, σstem–ground, and σpanicle–ground
are double bounce between each component and ground. The σground
is ground surface scattering. Details of the model structure are given
in Wang et al. (2009) and Karam et al. (1995).
Consequently, the model enables us to simulate the σtotal in all polar-
izations at any incidence angles for given frequency bands. The neces-
sary inputs and parameters were derived from the ground-based
biophysical and morphological measurements (Section 2.3) and default
values in Wang et al. (2009).
3. Results and discussion
3.1. Characteristics of C-band σ0 over paddy fields in VH, HH, and
VV polarizations
Fig. 2 shows the average σ0 values extracted for rice canopies and
water surfaces in VH, HH, and VV images. The σ0 values for water
surfaces were very consistent in cross (VH) and like (HH & VV)
0
1)
polarizations, irrespective of the rice growing stage. The average σ0
value was −28.5 dB in VH, whereas it was −21.1 dB in HH and VV;
each value indicates the baseline for zero biomass (i.e., no scattering
elements in paddy fields). Accordingly, the σ0 value for rice canopies
changes along the arrow lines in Fig. 2 in response to the scattering
characteristics of the rice canopy. Note that the scattering characteris-
tics are determined by the structural, morphological, and dielectric
conditions of the canopy, i.e., they are not always related to the growth
stage of rice. Even in a specific growth stage, rice canopies in a region
display large variability in biophysical variables, and this variability
between fields has to be assessed for crop diagnosis.
Overall, the range of seasonal change in σ0 (i.e., between the seed-
ling and hull growth stages) was 12 dB in all polarizations. The dotted
line connecting the VH (20110716) and HH (20110715) points was
nearly parallel to 1:1. Because these two images of the same canopy
were taken only 1 day apart, the response of σ0 in VH and HH to the
rice canopy at the vegetative stage was considered almost equivalent, al-
though HH appeared to be slightly more sensitive than VH. The seasonal
consistency of the C-band σ0 from the Radarsat-2 spotlight mode was
clearly confirmed by the results over 4 years. The seasonal baseline
ð5Þ values (minimum σ0 for zero-biomass paddy fields) were determined
for cross and like polarizations. The results provide much more definitive
information for the monitoring of rice growth compared with preceding
studies (e.g., Chakraborty et al., 2005; Choudhury & Chakraborty, 2006;
Ribbes & Le Toan, 1999; Shao et al., 2001).
3.2. A comprehensive analysis of the relationship between C-band σ0 and
biophysical variables
Because the response of the C-band σ0 in VH was similar to that in HH
and VV, we assumed that an intensive analysis of the relationship be-
tween the C-band σ0 in VH would provide useful insights on the overall
capability of C-band SAR sensors in other configurations. Accordingly,
we focused on the VH images because the multi-polarization measure-
ment was not yet possible in the spotlight model. Fig. 3 summarizes the
correlation coefficients between C-band σ0 in VH and canopy biophysical
and ecophysiological variables. Figs. 4–8 show scatter plots for the major
canopy variables.
Overall, the whole-canopy structural variables such as canopy height,
biomass, and LAI were strongly correlated with σ0, although stem densi-
ty, leaf density, and hill density were only moderately correlated. These
results agreed well with the results of a ground-based study by Inoue
et al. (2002). Results included some weak or negligible relationships as

well as strong ones. However, both negative and positive results are
:
(1

-5 presented here to clearly show the inherent limitations and potential

C-band σ0 for rice canopy (dB)

20100715HH for accurate assessment of biophysical variables. In addition, we should

-10 20120924HH note that high correlation coefficients do not always imply direct mecha-
20110710VV nistic relationships. Indirect or superficial relationships are essential
20090906VH
20110930VH
constraints for higher accuracy and wider applicability. Therefore,
-15
ecophysiological insights into such close relationships would be useful.
20100716VH
-20
HH, VV 3.2.1. Canopy height
Canopy height was strongly correlated with σ0, but the relationship
-25 20120627VH
was presumed to be indirect and superficial because canopy height is
VH
not directly related to the scattering processes. Fig. 5a clearly indicates
-30 that seasonal trends in C-band σ0 and canopy height are similar, which
has often been reported (e.g., Chakraborty et al., 2005; Choudhury &
-28.5 -21.1
-35 Chakraborty, 2006; Ribbes & Le Toan, 1999; Shao et al., 2001), but it is
-35 -30 -25 -20 -15 -10 -5 0 also obvious that the canopy height has no relationship to σ0 at each
C-band σ0 for water surface (dB) growth stage. Therefore, the difference in canopy height at similar growth
stages cannot be detected by σ0. The superficial sensitivity of σ0 to canopy
Fig. 2. Comparison of C-band σ0 for water surfaces and rice canopies. Each point indicates height is attributable to changes in leaf biomass and structure and
the average values for water surfaces and rice canopies in each SAR image. A label for each
data point (yyyymmddPP) indicates the year (yyyy), month (mm), date (dd), and polari-
is highly dependent on the relationship between plant height and canopy
zation (PP), respectively. The numbers −28.5 and −21.1 along the x-axis are average σ0 structure. For example, LAI and leaf biomass were moderately related
values for cross (VH) and like (HH & VV) polarizations, respectively. to canopy height for the entire season (r = 0.89 and r = 0.90,
 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266 261

Correlation coefficient (r) -5

-1.0 -0.5 0.0 0.5 1.0 a)
Height (m) -10
Hill density (m-2)
n = 127 Stem density (m-2 )
-15

C-VH σ0 (dB)
LAI((m2 m-22 )
fAPAR
SPAD (digital number) -20

Leaf FW (g m-2) -25

Stem FW (g m-2)
Panicle FW (g m-22) 20090906VH
Total FW (g m-2) -30 20100716VH
20120627VH
Leaf DW (g m-2)
Stem DW (g m-2) -35
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Panicle DW (g m-2) Height (m)
Total DW (g m-2)
-5
Leaf W (%)
Stem W (%) b)
Panicle W (%) -10
TotalW(%)
Total
-15

C-VH σ0 (dB)
Leaf density (m-2)
Leaf length (m)
Leaf width (m) -20
Leaf thickness (m)
3-D Leafdensity(m-3)
-25
Layer-1 thickness (m)
Layer-2 thickness (m) 20090906VH
Layer-3 thickness (m) -30 20100716VH
Stem radius (m) 20120627VH
P=0.01 P=0.01
-35
0
0 10 20 30 40 50
Fig. 3. Correlation of C-band σ with biophysical variables in rice canopies over a wide
range of growth stages. SPAD: chlorophyll index by SPAD 502; FW: fresh biomass; DW: SPAD (DN)
dry biomass; W: water content; Layer-1, -2, and -3: thickness of layers for panicles (1),
0
leaves (2), and stems (3), respectively. The correlation coefficient (r) at significance levels Fig. 4. Relationship of C-band σ with a) canopy height and b) SPAD.
of P = 0.01 and P = 0.05 is 0.33 and 0.25, respectively.

respectively), but their relationships were not consistent for individual
growth stages (r b 0.58 and r b 0.76).
3.2.2. Water content and chlorophyll content
Interestingly, ecophysiological variables such as water content and
SPAD (chlorophyll content) were negatively correlated with σ0 (Fig. 3).
These correlations were also indirect or superficial because these
variables are strongly affected by both phenology and canopy
structure. Fig. 5a clearly shows that stem water content is very stable
throughout the growth stage, suggesting that σ0 could change signif-
icantly without any substantial change in stem water content. Fig. 5b
also indicates no sensitivity of σ0 to leaf water content during the
vegetative stage. Additionally, σ0 displays little variation despite the
large variability of leaf water content, i.e., a clear decreasing trend with
senescence. Water area index (LAI × leaf water content; Dabrowska-
Zielinska, Inoue, Kowalik, & Gruszczynska, 2007) is much more closely re-
lated to σ0 (Fig. 5c), but the relationship is mainly determined by LAI; the
leaf water content has only a minor effect. Therefore, plant water status
may not be directly detected by C-band σ0 under the normal conditions
of paddy rice under flood irrigation. The difference in water content
would be detected only though significant changes in leaf biomass and
structure.
The highest negative correlation was found between SPAD and σ0
(Fig. 3) despite the absence of any direct mechanistic influence of chloro-
phyll content on the microwave scattering processes. This is also attribut-
able to the physiological relationship between chlorophyll content and
the amount of leaves; i.e., leaf area increase is associated with a decrease
in chlorophyll. In fact, SPAD was negatively correlated with increasing
biomass and LAI (r = −0.65 and −0.61, respectively). There is a weak
relationship between σ0 and the SPAD values for the data in each growth
stage (Fig. 4b). The apparent high correlation for the canopy structural
variables such as canopy layer thickness (Layers 2, 3) or morphological
variables (e.g., leaf length and leaf width) can also be explained by
synchronous changes in the amount of leaves with elongation of
internodes or an enlargement of leaf size, respectively (Fig. 3). For exam-
ple, the thickness of Layer 2 (leaf layer) and the thickness of Layer 3 (stem
layer) were best correlated with leaf biomass (r = 0.88) and stem bio-
mass (r = 0.96), respectively. Leaf length was also best correlated to can-
opy leaf biomass (r = 0.89).
3.2.3. Leaf area and leaf biomass
Both LAI and leaf biomass produce significantly high correlations
with σ0 (Fig. 3). The scatter plots between σ0 and LAI (Fig. 6a) and
leaf biomass (Fig. 6b) indicate consistent relationships throughout
all growth stages. Overall, even within a single stage (i.e., single
image), data points are plotted along the similar fitted lines. The C-band
σ0 increases exponentially with increasing LAI or leaf biomass. These
relationships produce exponential curves with high coefficients of deter-
mination (r2 = 0.84–0.85).
0 2
σ ¼ −15:17ð1 þ 0:879 exp½−0:637 LAIÞ r ¼ 0:844 ð6Þ
262 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266

-5 -5
a) a)
-10 -10
σ0 = -15.17 (1 + 0.879 exp [-0.637 LAI])
r2 = 0.844*
-15
C-VH σ0 (dB)

-15

C-VH σ0 (dB)
-20
-20

-25
20090906VH -25

-30 20100716VH
20090906VH
20120627VH -30 20100716VH
-35 20120627VH
0 20 40 60 80 100
-35
Stem W (%) 0 1 2 3 4 5 6
LAI (m2m2)
-5 -5
b) b)
-10 σ0 = -15.15 (1+0.886 exp [-0.0117 LDW])
-10
r2 = 0.850*
-15
C-VH σ0 (dB)

-15

-20 C-VH σ0 (dB)

-20

-25
20090906VH -25
-30 20100716VH 20090906VH
20120627VH -30 20100716VH
-35 20120627VH
0 20 40 60 80 100
Leaf W (%) -35
0 100 200 300
Leaf DW (g m-2)
-5
Fig. 6. Relationship of C-band σ0 with a) leaf area index (LAI) and b) leaf biomass (leaf
c) DW). * indicates the statistical significance at P = 0.001.
-10

These exponential curves are similar to the NDVI–LAI curve

-15
C-VH σ0 (dB)

(e.g., Asrar, Myneni, & Kanemasu, 1989; Huete, 1988). The normal-
ized difference vegetation index (NDVI) using reflectance values in
-20 the red and near-infrared has been used extensively in various applica-
tions. Both σ0 and NDVI are saturated at an LAI of around 3, and they
-25
20090906VH
20100716VH -5
-30
20120627VH σ0 = -16.51(1 + 0.737 exp[-0.00588 TDW] )
-10 r2 = 0.869*
-35
0 1 2 3 4
C-VH σ0 (dB)

Water Area Index -15

Fig. 5. Relationship of C-band σ0 with a) stem water content, b) leaf water content, and -20
c) water area index (water content × LAI).

-25 20090906VH
20100716VH
0
σ ¼ −15:15ð1 þ 0:886 exp½−0:0117 LDWÞ
2
r ¼ 0:850 ð7Þ -30 20120627VH
20110930VH
-35
The σ0 value saturates at a LAI of around 3 and a leaf biomass of 0 500 1000 1500 2000 2500
180 gDW m−2. The asymptotic values for the equations, i.e., −15.17 dB Total DW (g m-2)
for LAI and −15.15 dB, suggest that the maximum σ0 for rice canopies
is approximately −15.2 dB in VH polarization of the Radarsat-2 spotlight Fig. 7. Relationship between C-band σ0 and total biomass (total DW). * indicates the sta-
mode. tistical significance at P = 0.001.
 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266
0 -10
σ 0 = 13.81 fAPAR -27.85
-5
r2 = 0.814*
-10
C-VH σ0 (dB)
-15
-20
-25
-30
20090906VH
0090906V
-35 20100716VH
20120627VH
-40
0.0 0.2 0.4 0.6 0.8 1.0
fAPAR
Fig. 8. Relationship between C-band σ0 and fAPAR. * indicates the statistical significance at
P = 0.001.
lose sensitivity at higher LAI values. This similarity implies a similar
scattering process in vegetation canopies for optical and microwave sig-
natures. Because the normal range of LAI changes in rice is 0–6, both σ0
and NDVI can be used to estimate LAI up to half of the range. The results
suggest the applicability of C-band σ0 for the assessment of LAI,
especially in constellation with optical sensors.
However, the accuracy may not be high enough to detect the small
variability between fields, although this is still the case for NDVI. Because
SAR observations are always oblique to the canopy surface, the effects of
incidence angle and relative azimuth angle relative to the row direction
(i.e., field orientation) would be additional error sources. However, the
flexibility of SAR sensors in the choice of angular configurations may pro-
vide additional opportunities to achieve higher accuracy. The process
modeling approach may be useful to normalize the effects of such sensor
configurations. Fig. 9 indicates the degree of fitness of the WC model to
the measured σ0 under a wide range of LAI conditions. High statistical in-
dicators (r2 = 0.818, RMSE = 1.52) suggest the theoretical soundness of
the above interpretation on the relationship between LAI and C-band σ0.
This result also suggests the applicability of the WC model to some simple
analysis or operational assessment, even though a canopy is expressed by
only two layers (vegetation and soil) and the vegetation is expressed by
only one canopy descriptor such as LAI. The current physically based
models are useful for the interpretation of measured data but are not
fully applicable to inversion purposes (e.g., Prévot et al., 1993; Wang
et al., 2009). Therefore, simple semi-empirical approaches such as
normalized indices for polarization and incidence angles would play
some useful roles in various applications (Inoue et al., 2002).
These results and considerations all suggest that the amount and
structure of leaves are the dominant factors determining the C-band σ0
of a rice canopy, even if some other variables may produce superficially
high correlations. This is supported by the results from ground-based
scatterometry experiments (e.g., Bouman, 1991; Inoue et al., 2002;
Prévot et al., 1993). Accordingly, based on our results, various phenome-
na observed in previous studies (e.g., Baghdadi et al., 2009; Bouvet & Le
Toan 2011; Chakraborty et al., 2005; Fieuzal et al., 2012; Ribbes & Le
Toan, 1999; Shao et al., 2001) can be reasonably interpreted, and their
generality or applicability is also inferred.
3.2.4. Total biomass
The total biomass was moderately correlated with C-band σ0
(Fig. 3). However, Fig. 7 clearly shows that the σ0 saturates at small
values of total biomass (320 gDW m−2), and the asymptotic maximum
(−16.5 dB) does not change during the maturing stage irrespective of
263
C-VH σ0 estimated by WC model (dB)
r2 = 0.818*
RMSE=1.52
-15
-20
-25
-30
-35
-35 -30 -25 -20 -15 -10
Measured C-VH σ0 (dB)
Fig. 9. Comparison of measured C-VH σ0 values with those estimated by a parameterized
water cloud (WC) model. LAI was used as a key canopy variable for parameterization of
the model. *indicates the statistical significance at P = 0.001.
any large change in biomass (up to 1600 gDW m−2). The rice growth
stage with a biomass of 320 gDW corresponds approximately to the
end of leaf expansion. A simple exponential curve can be fitted with a
high coefficient of determination (r2 = 0.87) for the entire stage, but
σ0 has sensitivity to the total biomass only within the range of the
lower 20% of the full biomass (320 gDW m−2/1600 gDW m−2).
0
σ ¼ −16:51ð1 þ 0:737 exp½−0:00588 TDWÞ
2
r ¼ 0:869 ð8Þ
Additionally, the correlation coefficient for the lower 20% range
(r = 0.52) was not high enough for operational applications. These
results indicate that the overall response to total biomass was attribut-
able to the structural change in the canopy due to leaf growth and that
C-band σ0 has relatively poor predictive power for total biomass.
3.2.5. fAPAR
Fig. 3 indicates that of all the biophysical and ecophysiological vari-
ables examined, C-band σ0 was most strongly correlated with fAPAR
(r = 0.91). The scatter plot (Fig. 8) suggests that the σ0–fAPAR relation-
ship is linear throughout the growth stages as expressed by the follow-
ing equation:
0
σ ¼ 13:81 fAPAR−27:85
2
r ¼ 0:814: ð9Þ
fAPAR has often been used as a key variable in light (or radiation)
use efficiency models to estimate the biomass productivity of plant
canopies (Monteith, 1977; Shibles & Weber, 1966). Originally, these
studies showed a simple linear relationship between crop biomass
and integrated solar radiation absorbed by a canopy (APAR) for an en-
tire growing season. The slope (i.e., the proportionality coefficient,
often referred to as LUE, RUE, or ε) was assumed to be constant, and
the fAPAR was estimated by instrumental radiation measurements or
expressed as a function of LAI.
However, these measurements can be avoided by using remotely
sensed fAPAR. A number of experimental and theoretical studies have
shown that fAPAR can be estimated using NDVI or improved forms of
NDVI (e.g., Asrar et al., 1989; Baret & Guyot, 1991; Huete, 1988;
Myneni & Williams, 1994; Qi, Chehbouni, Huete, Kerr, & Sorooshian,
1994). Accordingly, NDVI has been used as the most common index,
but a discrepancy in the NDVI–fAPAR relationship between the vegeta-
tive and reproductive (senescent) stages has also been recognized
(e.g., Asrar et al., 1989; Choudhury, 2000; Inoue, Moran, & Horie,
264 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266

1998). Therefore, more consistent spectral indices have been consid- 0

ered based on hyperspectral measurement (e.g., NDSI (R550, R410)
and NDSI (R720, R420) by Inoue, Peñuelas, Miyata, & Mano, 2008).
a)
Another key variable, LUE (light use efficiency), proved to be variable -5

Simulated C-VH25 σ0 (dB)

that is affected by growth stage and environmental stresses, espe-
cially when the algorithm is applied at short temporal resolutions
(e.g., day or week). Accordingly, several spectral indices such as PRI -10 r² = 0.5 10*
(Gamon, Peñuelas, & Field, 1992; Garbulsky, Peñuelas, Gamon,
Inoue, & Filella, 2010; Peñuelas & Filella, 1998) and NDSI (R730,
R520) have been proposed for remote sensing of LUE (Inoue et al., -15
2008).
From an ecophysiological viewpoint, both variables are related to
optical remote-sensing signatures because the absorption of photosyn-
-20
thetic light and the synthesis of carbohydrates are both based on light
reactions in pigments such as chlorophyll. Because microwave signa-
tures are never sensitive to pigments, LUE in particular would never
-25
be assessed by microwave remote sensing. However, our results sug- 0.0 0.2 0.4 0.6 0.8 1.0
gest that the fAPAR could be estimated by C-band σ0. This is reasonable
0
because fAPAR, i.e., the light absorbing capacity of a canopy, is strongly
determined by the amount and structure of leaf elements in a canopy b)
rather than by pigment concentration. As a result, the C-band σ0
would be able to provide the equivalent information to that provided -5

Simulated C-VH25 σ0 (dB)

by NDVI. This is a promising basis for the timely monitoring of fAPAR
by C-band SAR and/or through a constellation of SAR and optical
sensors. -10
r² = 0.400*
In support of these analytical results between measured fAPAR and
C-band σ0, we compared the relationship between fAPAR and simulated
σ0 using a physically-based scattering model. In this study, the simula- -15
tion was used just to infer the theoretical soundness of the experimental
results derived from the satellite SAR data and ground-based data dur-
ing the vegetative stages, since the absolute values of SAR σ0 are highly -20
dependent on sensor calibrations (e.g., Inoue & Sakaiya, 2013), and the
scattering processes in a rice canopy especially with panicles are not
fully modeled in the simple model. Fig. 10 shows the relationships be- -25
tween fAPAR and simulated σ0 values in VH polarization at incidence 0.0 0.2 0.4 0.6 0.8 1.0
angles of 25°, 35° and 45° during early vegetative stages (i.e., without 0
panicles). We can see obvious linear relationships in all polarizations.
Comparisons using the simulation results at various incidence angles
c)
(25–55°) indicated that the observations at around 25–35° would be
-5
Simulated C-VH25 σ0 (dB)

most suitable to detect the difference in fAPAR. These relationships

agree quite well to that in Fig. 8. These results suggest the soundness
of the experimental results in Fig. 8, although the absolute values of σ0
were not identical. Similar comparisons between different polarizations -10
r² = 0.397*
suggested that the VH-σ0 would be most sensitive to fAPAR.
Accordingly, these results also suggest that the selection of both
polarizations (VH) and incidence angles (around 25–35°) based on -15
the scatterometer experiment (Inoue et al., 2002) was optimal for the
satellite SAR observations in this filed experiment.
-20
3.2.6. Opportunities to improve the capability of C-band σ0 for the
assessment of biophysical variables
Our analysis using the high-resolution satellite images taken by -25
Radasat-2 over 4 years has determined clear and consistent relationships 0.0 0.2 0.4 0.6 0.8 1.0
between C-band σ0 and canopy biophysical variables. Although a correla- fAPAR
tion coefficient can be affected by the amount and range of data as well as
Fig. 10. Relationship between fAPAR and simulated C-band σ0 values by a canopy scatter-
by the underlying mechanisms, the reasonable interpretation of our con-
ing model during early vegetative stages. The C-VHdd σ0 indicates simulated σ0 values in
sistent results strongly suggests the potential capability of C-band SAR for VH polarization at an incidence angle of dd degrees. * indicates the statistical significance
the timely monitoring of rice growth. These results are well supported at P = 0.01.
by the comprehensive results obtained using a scatterometer under a
wide range of configurations (Inoue et al., 2002). However, several
approaches would be useful to further enhance the potential for into the capability of the other configurations such as HH and VV,
operational applications. but a more sophisticated use of multi-polarization images might be
At present, high-resolution SAR images can be acquired only in a useful to improve accuracy or to obtain other useful information
single polarization mode due to the limited capability of the current for a canopy. Several previous studies have shown that the use of
sensor/mode. Accordingly, our analysis was mainly based on the multi-polarization data is effective for improving the accuracy of classifi-
dataset at VH polarization. The present results provide clear insight cation (e.g., Bouvet et al., 2009; Lopez-Sanchez & Ballester-Berman, 2009;
 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266
Ribbes & Le Toan, 1999). When full-polarimetric images are available in a
high-resolution mode in the near future, more sensitive components of
the signatures may be extracted, allowing greater accuracy in the assess-
ment of biophysical variables.
The accuracy of assessment may be further improved by taking
account of some supplemental information related to field conditions
such as row orientation and soil roughness, which affect the backscatter-
ing processes. For example, the shape, orientation, and fractional vegeta-
tion cover of a field can be derived from the segmentation of high-
resolution multispectral images (e.g., Inoue et al., 2010; Liu & Xia,
2010). Such supplemental information regarding the confounding factors
from other types of sensors is useful to reduce the variability or uncer-
tainty in the relationships between C-band σ0 and biophysical variables.
Several studies have demonstrated the potential of the synergistic use
of SAR and optical sensors (e.g., McNairn et al., 2009; Moran, Vidal,
et al., 1997; Qi, Wang, Inoue, & Zhang, 2003).
Mechanistic models may be useful to improve the accuracy of assess-
ment. Backscattering signatures from land surfaces are affected by many
factors, including plant biomass, canopy structure (e.g., LAI), soil mois-
ture, and roughness. Interactions with sensor configurations, such as fre-
quency, polarization, and incidence angle, strongly affect backscattering
coefficients (Inoue et al., 2002; Lopez-Sanchez & Ballester-Berman,
2009; Qi et al., 2003). Therefore, a simple water-cloud model or more
detailed models are useful for the interpretation or rough prediction of
microwave backscattering processes in plant canopies (e.g., Karam
et al., 1995; Prévot et al., 1993; Ulaby, Allen, Eger, & Kanemasu, 1984;
Wang et al., 2009). However, the retrieval of biophysical variables by
the inversion of physically based models is still difficult because of
three reasons; 1) backscattering processes within a canopy, especially
during the panicle growth stage, are not sufficiently modeled, 2) a large
number of parameters and inputs related to the three-dimensional struc-
ture, electromagnetic characteristics, and biophysical variables have to be
provided to run the models, and 3) inherent limitations in the inversion
of complex multi-variable models. A synergistic linkage of a backscatter-
ing model with plant growth and 3-D morphogenetic models (e.g., Godin
& Sinoquet, 2005) would be one of the useful approaches.
4. Conclusions
Our analytical results revealed clear and consistent relationships
between C-band σ0 and canopy biophysical variables based on a dataset
consisting of high-resolution C-band SAR images from Radasat-2 and
concurrent measurements of detailed canopy biophysical variables
over 4 years.
The baseline σ0 values (minimum σ0 for zero-biomass paddy fields)
were determined to be −28.5 dB in VH and −21.1 dB in HH and VV.
However, the dynamic range of σ0 during rice growth was similar
under all polarizations (approximately 12 dB). The results provide
much more definitive information for monitoring rice growth com-
pared with previous studies.
The comprehensive analysis of the relationship of C-band σ0 in VH
with canopy biophysical and ecophysiological variables has revealed
the response of SAR signatures to paddy rice canopies. The superficial
high sensitivity of σ0 to canopy height is attributable to the change in
leaf biomass and structure. Ecophysiological variables such as water
content and SPAD (chlorophyll content) were negatively correlated
with σ0, but these correlations also proved to be indirect or superficial.
The plant water status of rice canopies under flood irrigation may be
poorly detected directly by C-band σ0.
Both LAI and leaf biomass had significantly high correlations with σ0,
and these relationships were consistent throughout all the growth
stages. These relationships are expressed by exponential curves with
high coefficients of determination. However, the σ0 value saturates at
a LAI of around 3 and a leaf biomass of 180 gDW m−2. The asymptotic
equation suggests that the maximum σ0 for rice canopies is approxi-
mately −15.2 dB in VH polarization. The response of σ0 was similar to
265
that of NDVI, implying a similar scattering process in vegetation cano-
pies for optical and microwave signatures. The response of σ0 to total
biomass was expressed by an exponential equation with a high coeffi-
cient of determination, but the sensitivity was clear only within the
lower 20% of the full biomass. This sensitivity was attributable to the
structural changes due to leaf growth, and C-band σ0 proved to have a
relatively poor predictive ability for estimating total biomass.
Of all the biophysical and ecophysiological variables, the C-band σ0
had the highest correlation with fAPAR. A scatter plot suggested that
the σ0–fAPAR relationship is linear throughout the growth stages.
These results suggest the capability of C-band σ0 for the assessment of
LAI or fAPAR. This is a promising basis for the timely monitoring of
fAPAR by C-band SAR and/or through a constellation of SAR and optical
sensors.
Not only the positive results but also some of the negative results
would be useful for basic studies on backscattering processes and for
operational applications of SAR sensors in the future.
Acknowledgments
This work was supported financially by the JSPS and MEXT, Japan. The
authors are grateful for the efficient help from the technical staff of the
National Institute for Agro-Environmental Sciences. We would like to
thank Dr. Thuy Le Toan (CESBIO, France) and Prof. Juan M. Lopez-
Sanchez (University of Alicante, Spain) for their helpful comments.
References
Asrar, G., Myneni, R. B., & Kanemasu, E. T. (1989). Estimation of plant-canopy attributes
from spectral reflectance measurements. In G. Asrar (Ed.), Theory and application of
optical remote sensing (pp. 252–296). New York: Wiley Interscience.
Attema, E. P. W., & Ulaby, F. T. (1978). Vegetation modeled as a water cloud. Radio Science,
13, 357–364.
Baghdadi, N., Boyer, N., Todoroff, P., El Hajj, M., & Bégué, A. (2009). Potential of SAR sen-
sors TerraSAR-X, ASAR/ENVISAT and PALSAR/ALOS for monitoring sugarcane crops
on Reunion Island. Remote Sensing of Environment, 113, 1724–1738.
Baret, F., & Guyot, G. (1991). Potentials and limits of vegetation indices for LAI and APAR
assessment. Remote Sensing of Environment, 35, 161–173.
Bouman, B.A.M. (1991). Crop parameter estimation from ground-based X-band (3-cm
wave) radar backscattering data. Remote Sensing of Environment, 37, 193–205.
Bouvet, A., & Le Toan, T. (2011). Use of ENVISAT/ASAR wide-swath data for timely rice fields
mapping in the Mekong River Delta. Remote Sensing of Environment, 115, 1090–1101.
Bouvet, A., Le Toan, T., & Lam-Dao, N. (2009). Monitoring of the rice cropping system in
the Mekong Delta using ENVISAT/ASAR dual polarization data. IEEE Transactions on
Geoscience and Remote Sensing, 47, 517–526.
Chakraborty, M., Manjunath, K. R., Panigrahy, S., Kundu, N., & Parihar, J. S. (2005). Rice
crop parameter retrieval using multi-temporal, multi-incidence angle RADARSAT
SAR data. ISPRS Journal of Photogrammetry and Remote Sensing, 59, 310–322.
Choudhury, B. J. (2000). A sensitivity analysis of the radiation use efficiency for gross
photosynthesis and net carbon accumulation by wheat. Agricultural and Forest
Meteorology, 101, 217–234.
Choudhury, I., & Chakraborty, M. (2006). SAR signature investigation of rice crop using
RADARSAT data. International Journal of Remote Sensing, 27, 519–534.
Choudhury, I., Chakraborty, M., Santra, S.C., & Parihar, J. S. (2012). Methodology to classify
rice cultural types based on water regimes using multi-temporal RADARSAT-1 data.
International Journal of Remote Sensing, 33, 4135–4160.
Dabrowska-Zielinska, K., Inoue, Y., Kowalik, W., & Gruszczynska, M. (2007). Inferring the
effect of plant and soil variables on C- and L-band SAR backscatter over agriculture
fields based on model analysis. Advances in Space Research, 39, 139–148.
Doraiswamy, P. C., Hatfield, J. L., Jackson, T. J., Akhmedov, B., Prueger, J., & Stern, A. (2004).
Crop condition and yield simulation using Landsat and MODIS. Remote Sensing of
Environment, 92, 548–559.
Fieuzal, R., Baup, F., & Marais-Sicre, C. (2012). Sensitivity of TERRASAR-X, RADARSAT-2
and ALOS satellite radar data to crop variables. Proceeding of the IEEE International
Geoscience and Remote Sensing Symposium (IGARSS) 2012, 22–27 July 2012, Munich
(pp. 3740–3743).
Gamon, J. A., Peñuelas, J., & Field, C. B. (1992). A narrow-waveband spectral index that
tracks diurnal changes in photosynthetic efficiency. Remote Sensing of Environment,
41, 35–44.
Garbulsky, M. F., Peñuelas, J., Gamon, J., Inoue, Y., & Filella, I. (2010). The photochemical
reflectance index (PRI) and the remote sensing of leaf, canopy and ecosystem radia-
tion use efficiencies. Remote Sensing of Environment, 115, 281–297.
Godin, C., & Sinoquet, H. (2005). Functional–structural plant modelling. New Phytologist,
166, 705–708.
Huete, A.R. (1988). A soil vegetation adjusted index (SAVI). Remote Sensing of Environment,
25, 295–309.
Inoue, Y. (2003). Synergy of remote sensing and modeling for estimating ecophysiological
processes in plant production. Plant Production Science, 6, 3–16.
266 Y. Inoue et al. / Remote Sensing of Environment 140 (2014) 257–266
Inoue, Y., Kiyono, Y., Asai, H., Ochiai, Y., Qi, J., Olioso, A., et al. (2010). Assessing land use
and carbon stock in slash-and-burn ecosystems in tropical mountain of Laos based
on time-series satellite images. International Journal of Applied Earth Observation and
Geoinformation, 12, 287–297.
Inoue, Y., Kurosu, T., Maeno, H., Uratsuka, S., Kozu, T., Dabrowska-Zielinska, K., et al.
(2002). Season-long daily measurements of multifrequency (Ka, Ku, X, C, and L)
and full-polarization backscatter signatures over paddy rice field and their relation-
ship with biological variables. Remote Sensing of Environment, 81, 194–204.
Inoue, Y., Moran, M. S., & Horie, T. (1998). Analysis of spectral measurements in paddy
field for predicting rice growth and yield based on simple crop simulation model.
Plant Production Science, 1, 269–279.
Inoue, Y., Peñuelas, J., Miyata, A., & Mano, M. (2008). Normalized difference spectral indi-
ces for estimating photosynthetic efficiency and capacity at a canopy scale derived
from hyperspectral and CO2 flux measurements in rice. Remote Sensing of
Environment, 112, 156–172.
Inoue, Y., & Sakaiya, E. (2013). Relationship between X-band backscattering coefficients
from high-resolution satellite SAR and biophysical variables in paddy rice. Remote
Sensing Letters, 4, 288–295.
Karam, M.A., Amar, F., Fung, A. K., Mougin, E., Lopes, A., Le Vine, D.M., et al. (1995). A mi-
crowave polarimetric scattering model for forest canopies based on vector radiative
transfer theory. Remote Sensing of Environment, 53, 16–30.
Kurosu, T., Fujita, M., & Chiba, K. (1997). The identification of rice fields using multi-temporal
ERS-1 C band SAR data. International Journal of Remote Sensing, 18, 2953–2965.
Le Toan, T., Ribbes, F., Wang, L. F., Nicolas, F., Ding, K. H., Kong, J. A., et al. (1997). Rice crop
mapping and monitoring using ERS-1 data based on experiment and modeling
results. IEEE Transactions on Geoscience and Remote Sensing, 35, 41–56.
Liu, D., & Xia, F. (2010). Assessing object-based classification: Advantages and limitations.
Remote Sensing Letters, 1, 187–194.
Lopez-Sanchez, J. M., & Ballester-Berman, J.D. (2009). Potentials of polarimetric
SAR interferometry for agriculture monitoring. Radio Science, 44, RS2010.
http://dx.doi.org/10.1029/2008RS004078.
Lopez-Sanchez, J. M., Ballester-Berman, J.D., & Hajnsek, I. (2011). First results of rice mon-
itoring practices in Spain by means of time series of TerraSAR-X dual-pol images. IEEE
Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 4, 412–422.
Lopez-Sanchez, J. M., Hajnsek, I., & Ballester-Berman, J.D. (2012). First demonstration of
agriculture height retrieval with PolInSAR airborne data. Geoscience and Remote
Sensing Letters, 9, 242–246.
McNairn, H., Champagne, C., Shang, J., Holmstrom, D., & Reichert, G. (2009). Integra-
tion of optical and synthetic aperture radar (SAR) imagery for delivering
operational annual crop inventories. ISPRS Journal of Photogrammetry and
Remote Sensing, 64, 434–449.
Monteith, J. L. (1977). Climate and the efficiency of crop production in Britain.
Philosophical Transactions of the Royal Society B, 281, 277–294.
Moran, M. S., Inoue, Y., & Barnes, E. M. (1997). Opportunities and limitations for
image-based remote sensing in precision crop management. Remote Sensing of
Environment, 61, 319–346.
Moran, M. S., Vidal, A., Troufleau, D., Qi, J., Clarke, T. R., Pinter, P. J., Jr., et al. (1997). Com-
bining multifrequency microwave and optical data for crop management. Remote
Sensing of Environment, 61, 96–109.
Myneni, R. B., & Williams, D. L. (1994). On the relationship between FAPAR and NDVI.
Remote Sensing of Environment, 49, 200–211.
Peñuelas, J., & Filella, I. (1998). Visible and near-infrared reflectance techniques for diag-
nosing plant physiological status. Trends in Plant Science, 3, 151–156.
Prévot, L., Campion, I., & Guyot, G. (1993). Estimating surface soil moisture and leaf area
index of a wheat canopy using a dual-frequency (C and X bands) scatterometer.
Remote Sensing of Environment, 46, 331–339.
Qi, J., Chehbouni, A., Huete, A.R., Kerr, Y. H., & Sorooshian, S. (1994). A modified soil ad-
justed vegetation index. Remote Sensing of Environment, 48, 119–126.
Qi, J., Wang, C., Inoue, Y., & Zhang, R. (2003). Synergy of optical and radar remote sensing
in agricultural applications. Proceedings of the SPIE International Symposium on Optical
Science and Technology, 5153, 153–158.
Ribbes, F., & Le Toan, T. (1999). Rice field mapping and monitoring with RADARSAT data.
International Journal of Remote Sensing, 20, 745–765.
Santi, E., Fontanelli, G., Montomoli, F., Brogioni, M., Macelloni, G., Paloscia, S., et al. (2012).
The retrieval and monitoring of vegetation parameters from COSMO-SkyMed images.
Proceeding of the IEEE International Geoscience and Remote Sensing Symposium
(IGARSS) 2012, 22–27 July 2012, Munich (pp. 7031–7034).
Shao, Y., Fan, X., Liu, H., Xiao, J., Ross, S., Brisco, B., et al. (2001). Rice monitoring and pro-
duction estimation using multitemporal RADARSAT. Remote Sensing of Environment,
76, 310–325.
Shibles, R. M., & Weber, R. C. (1966). Interception of solar radiation and dry matter pro-
duction by various soybean planting patterns. Crop Science, 6, 55–59.
Ulaby, F. T., Allen, C. T., Eger, G., & Kanemasu, E. T. (1984). Relating the microwave back-
scattering coefficient to leaf area index. Remote Sensing of Environment, 14, 113–133.
Wang, C., Wu, J., Zhang, Y., Pan, G., Qi, J., & Salas, W. A. (2009). Characterizing L-band scat-
tering of paddy rice in southeast China with radiative transfer model and
multitemporal ALOS/PALSAR imagery. IEEE Transactions on Geoscience and Remote
Sensing, 47, 988–998.