Figure 10. Relationship graphs between estimated and
measured values of optimized spectral indices.
suitable for the study area, and through the range of sensitive
band the spectral index of we constructed work much better
and purified for the study area.
Discussion
Water stress generally results in changes in plant biomass and
leaf structure. It has been long recognized that the lack of wa-
ter availability has an adverse effect on the plant growth pro-
cess. In arid and semi-arid area, water conditions determine
the growth of
halophyte
, it has great theoretical and practical
significance to understand and grasp the variation of spectral
curve and sensitive band of
halophyte
. Therefore, this study
demonstrated the potential of hyperspectral spectroscopy
data in monitoring plant water stress response (Maimaitiyim-
ing
et al
., 2013).
In this research, hyperspectral remote sensing technology
was used to investigate the differences between
VWC
spectra
and water content inversion of wetland vegetation. Com-
parison of
VWC
measured with a hyperspectral radiometer
to which obtained using traditional methods revealed that
the results of hyperspectral remote sensing was acceptable
indicate it has the potential to reduce costs and increase effi-
ciency (Lin
et al
., 2011). Inversion of
VWC
using hyperspectral
data has been widely applied to crops, such as corn, rice and
soy (Lin
et al
., 2015). Moreover, some researchers use geodetic
GPS receivers to measure
VWC
(Wei
et al
., 2015), while others
use the
AMSR-E VWC
to assess crop water stress (Chakraborty
et
al
., 2016).
A large number of spectral indices have been proposed for
VWC
estimation and widely used in remote sensing. Penuelas
et al
. (1993) found that the water index (
WI
) had potential
applicability in estimating
VWC
. Gao (1996) discovered that
usage of the normalized difference water index (
NDWI
) can
be a good estimate factor of the equivalent water thickness
(
EWT
). Wu et al. (2009) combined the measured spectral data
with the
PROSAIL
model and constructed NDWI
1240
, NDWI
1450
,
and NDWI
1940
, by using the water absorption bands of 1240
nm, 1450 nm, 1940 nm, and 860 nm. This led to establish-
ment of a
VWC
estimation model of canopy and leaf level good
results. Ceccato
et al
. (2002) established the global vegetation
moisture index (
GVMI
) based on the radiative transfer model,
using a shortwave infrared band (
SWIR
), instead of a near
infrared band (
NIR
) of
NDVI
. Their model could be used for
water retrieval of different types of vegetation cover. Wang
et
al
. (2003) recently developed a new normalized multiband
drought index (
NMDI
).
Considering the influence of moisture absorption and other
factors, most vegetation indices are composed of the ration
of two bands; the selection of wavelength is significant to
the sensitivity of the index to changes in vegetation water
status (Eitel
et al
., 2006). 2 bands of information,
DVI
and
RSI
,
enhances the radiation difference between the vegetation and
soil background. The normalization of
NDSI
is beneficial to
eliminate the proportional variation of the reflection spec-
trum, enhance the spectral response of the observed target,
and reduce the influence of the instrument (Qi
et al
., 1994;
Stagakis
et al
., 2010). Thus, in this paper, the researchers
constructed three new vegetation indices (
DVI
,
NDSI
, and
RSI
)
to estimate
VWC
of
Halophyte.
In this paper,
VWC
was considered as vegetation growth’s
the main stress restricting factor. However, vegetation stress
is multifarious in other aspects, such as fertilizer deficiency,
insect pests, multiple heavy metals, and disease. Therefore,
investigating the spectral response of plants in different
vegetation stresses is important. Further research is needed to
improve the
VWC
models, which using new vegetation indices
under different combined vegetation stress conditions (Shi
et
al
., 2016).
Removal of high spectral reflectance background and
scattered light noise is an important step in the process of
spectral data. Moreover, during the crop growth period, the
leaf area was small, while factors in the surrounding environ-
ments, such as soil reflectance resulted in a high diffuse ef-
fect; accordingly, future studies should investigate methods of
effectively removing noise and accurately predict crop water
content (Lin
et al
., 2015).
The present study determined the spectrum of vegeta-
tion leaves directly by using a spectrometer probe. Using the
spectrometer probe setup, the spectra of the vegetation leaves
are only for single leaf blade; it is bound to be affected by soil,
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
September 2018
545
