agronomy

Article
Ground Hyper-Spectral Remote-Sensing Monitoring of Wheat
Water Stress during Different Growing Stages
Qiang Li, Maofang Gao * and Zhao-Liang Li

Key Laboratory of Agricultural Remote Sensing, Ministry of Agriculture and Rural Affairs/Institute of
Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences,
Ministry of Agriculture and Rural Affairs, Beijing 100081, China
* Correspondence: [email protected]

Abstract: Monitoring agricultural drought via ground hyper-spectral remote sensing has always
been a hot topic in the fields of agriculture and meteorology. In this study, a greenhouse experiment
was conducted on wheat subjected to water stress during its different growth stages, namely tillering,
jointing, and milk maturity. An instrument (HOBO ware PRO) used to continuously measure soil
moisture was employed to measure the soil water content (SWC). An analytical spectral device (ASD)
was utilized to obtain the spectral curve of wheat subject to different water treatment methods. The
canopy temperature was obtained using thermal infrared sensors (METER SI-400). The relationships
between the SWC, wheat drought stage, canopy temperature, and spectral response characteristics
were elucidated. The results showed that the significant differences in spectral characteristics were
due to water stress during the different growth stages of wheat. Red-edge parameters of red-valley
position (RVP) and red-edge position (REP) both changed by 21 nm for the tillering-stage drought
and the jointing-stage drought; however, the RVP and REP values for the milk maturity stage drought
and the treatment under no water stress changed by 2 nm. Further, it was proved that the red-edge
blue-shift phenomenon was affected not only by the different wheat growth processes, but also by
Citation: Li, Q.; Gao, M.; Li, Z.-L.
the water stress at different growth stages. Red-edge reflectance clearly reflects wheat water stress
Ground Hyper-Spectral
at different growth stages. From SWC and canopy temperature analysis results, SWC and canopy
Remote-Sensing Monitoring of
Wheat Water Stress during Different
temperature had a significant difference between wheat drought at different growth stages, and
Growing Stages. Agronomy 2022, 12, the canopy temperature at the jointing stage drought had the strongest change. The water index
2267. https://doi.org/10.3390/ (WI) based on eleven vegetation water indexes exhibited a good performance for distinguishing
agronomy12102267 wheat water stress at different growth stages. In conclusion, ground-based hyperspectral remote
sensing can provide a large amount of high temporal and spectral resolution data on vegetation and
Academic Editor: Enrico Corrado
its surrounding environment, making it an important technical tool for wheat drought monitoring,
Borgogno
which has a great significance on the monitoring and early warning of wheat drought, reducing
Received: 6 August 2022 drought-related yield losses, and ensuring food security.
Accepted: 19 September 2022
Published: 22 September 2022 Keywords: hyper-spectral; greenhouse experiment; water stress; red edge; blue shift
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations. 1. Introduction
Crop canopy hyper-spectral analysis is a remote-sensing technology used for non-
destructive testing, which is effective for monitoring the growth, yield, and stress of
crops [1–3]. Using hyper-spectral imaging to estimate wheat water stress is significantly
Copyright: © 2022 by the authors.
important for large-scale wheat drought monitoring [4]. A massive information source
Licensee MDPI, Basel, Switzerland.
for spectral curves can be obtained from the reflectivity, reflection peak, and absorption
This article is an open access article
distributed under the terms and
valley positions; the absorption width and depth; the spectral derivative; the absorption
conditions of the Creative Commons
area; and the symmetry, slope, and characteristic index [5–7]. The hyper-spectral character-
Attribution (CC BY) license (https:// istics for wheat water stress mainly focus on the perspective of the spectrum, the selection
creativecommons.org/licenses/by/ of characteristic bands using the differential or ratio method, red-edge information, and
4.0/). parameter extraction from spectral characteristic [8,9]. The red edge of the crop is defined

Agronomy 2022, 12, 2267. https://doi.org/10.3390/agronomy12102267 https://www.mdpi.com/journal/agronomy

Agronomy 2022, 12, 2267 2 of 15

by the sharp change in the canopy reflectance of the leaves between 680 and 750 nm. The
red edge is formed because of the multiple scattering of the near-infrared band inside the
leaves and the absorption of visible light by chlorophyll [10,11]. The position of the red
edge is considered to be closely related to the moisture and nitrogen contents in the leaf,
the leaf area index (LAI), and the chlorophyll content in the vegetation [12–14]. Liu et al.
indicated that the amount of red and blue shifts from the red edge can be used to monitor
crop water stress and phenological characteristics [8]. He et al. proved that the red edge of
a crop is an indicator of crops suffering from water stress [13].
The vegetation water index (VWI) is an effective method used for the parameter
extraction of spectral characteristics for describing the level of vegetation drought; it is not
only widely used in leaf-scale measurements and field-scale experiments, but also in long-
term large-scale crop water monitoring combined with remote-sensing technology [15,16].
These indexes are usually calculated using the spectral values of the vegetation water
absorption band, such as the visible light and near-infrared bands with different scales [17].
VWIs for field-scale and leaf-scale experiments can be measured with a spectrometer, and
VWIs for satellite-scale experiments can be calculated via satellite imaging. Some related
VWIs have been proposed since the 1990s. For example, field-scale VWIs include the
WI [18], WI/NDVI [19], and RDI1450 [20]; leaf-scale VWIs include the MSI [21], R975 [22],
and RDI [23]; and satellite-scale VWIs include the NDWI [19], SRWI [24], NDII [25], and
GVMI [26]. These indexes differ with regard to the data obtained, equipment used, and
the resulting wave widths; their accuracy, sensitivity, and uncertainty should be analyzed
according to different application scenarios.
Drought is a natural disaster caused by short-term climatic anomalies and is character-
ized by its high frequency, impact, duration, and damage compared to other agricultural
disasters [27]. Agricultural droughts are considered to be the most complex natural hazards,
affecting the largest number of people but with little understanding of the mechanisms [28].
Many scholars suggest that the main drivers affecting drought are meteorology, physical
properties of the soil, and agricultural management, etc. The development of ground-based
hyperspectral sensors has become an effective means of acquiring parameters of crops and
their surroundings, and is important for monitoring crop drought. Judging from existing
research, the jointing, booting, and heading stages are the key growth stages of wheat,
and these three growth stages afford the most significant differences with regard to wheat
phenology characteristics [29–31]. This study is novel as we designed a greenhouse exper-
iment for these three growth stages of the wheat exposed to drought and subsequently
obtained useful crop/environment information using the ground-based hyperspectral
monitor method, aiming to reveal the mechanism of drought in different wheat growth
stages. The primary objectives of this study were: (1) designing the jointing, booting, and
heading stages of the wheat exposed to drought, which was followed by collecting and
analyzing related data; (2) comparing the spectral characteristics (water absorption bands
and red-edge parameters), soil moisture, and canopy temperatures for each treatment,
which were collected using ground-based remote-sensing sensors; and (3) calculating and
comparing the eleven VWIs and finding the best VWIs to characterize wheat drought based
on wheat canopy hyper-spectral. The great significance of this study is to reveal the re-
sponses of the wheat exposed to drought in different wheat growth stages, and accumulate
technical experience to advancing the development of ground-based hyperspectral remote
sensing on wheat drought monitoring.

2. Materials and Methods

2.1. Experiment Design
The experiment was conducted in the greenhouse of the Chinese Academy of Agri-
cultural Sciences, Beijing (116.33◦ E, 39.97◦ N) from October 15 2019 to March 18 2020.
The soil was sourced from the Langfang Experimental Station of the Chinese Academy
of Agricultural Sciences (116.76◦ E, 39.55◦ N); it exhibited a sandy loam texture and the
clay fraction was 5.43%. For the soil organic carbon content of 3.91 ± 0.78 g/kg, the initial
Agronomy 2022, 12, 2267 3 of 15

total nitrogen content was 0.45 ± 0.07 g/kg, the initial ammonium nitrogen content was
1.44 ± 0.18 mg/L, and the initial soil pH was 7.4. The wheat variety was Jimai No. 38.
Four controlled trials were set up, including the jointing, booting, and heading stages under
water stress, along with a trial with no water stress for comparison. Subsequently, 30 kg/ha
of nitrogen fertilizer was applied on each group; however, no top dressing was applied to
ensure that the wheat growth was affected by nitrogen stress. Insecticides and herbicides
were sprayed evenly on Nov 30. Each set followed different water management strategies
and irrigation details, as listed in Table 1. Information on the wheat phenology of this
trial is as follows. The wheat was planted on 15 October 2019; the wheat germinated on
6 November and in the tillering phase from 6 to 24 November 2019; the jointing stage lasted
from 25 November 2019 to 30 January 2020; and the wheat flowered and filled in the last
days of the jointing stage, and in milk maturity stage from 30 January to 20 March 2020.

Table 1. The amount of irrigation once a week per time for different sets (unit: mL). Date format
(yyyy.mm.dd).

Before Tillering Jointing Stage Milk Maturity

Tillering Stage
Stage (2019.11.25– Stage Remarks
(2019.11.06–11.24)
(2019.10.15–11.06) 2020.01.30) (2020.01.30–03.20)
S1 2000 500 2000 2000 Tillering stage drought
S2 2000 500 2000 2000 Tillering stage drought
S3 2000 2000 500 2000 Jointing stage drought
S4 2000 2000 500 2000 Jointing stage drought
S5 2000 2000 2000 500 Milk maturity stage drought
S6 2000 2000 2000 500 Milk maturity stage drought
S7 2000 2000 2000 2000 No water stress
S8 2000 2000 2000 2000 No water stress

An ASD FieldSpec Pro spectrometer (Analytical Spectral Devices, Inc., Boulder, CO,
USA) was used to measure the wheat canopy spectral. The spectrometer had a spectral
range of 350~2500 nm, with a 3 nm sample footprint size in the 350~1000 nm region and a
10 nm sample footprint size in the 1000~2500 nm region. The spectrometer automatically
interpolated the final result to 1 nm. Wheat spectra were measured twice for each group
during the experimental period, namely on Jan 9 and Mar 9. Furthermore, spectral data
for different periods of drought were obtained. We measured wheat spectrums on sunny
days, preferably between 10:00 and 14:00 a.m. when the light conditions were the best
before irrigation [32]. When we measured the spectral, the distance between the sensor and
the wheat canopy was kept at about 15 cm. Whenever one spectral acquisition operation
was performed, the ASD acquired five continuous spectral profiles for one sample. Five
sets of acquisition operation were obtained for one sample, and 25 spectrum profiles were
averaged as the final spectrum of the canopy. Then, the average spectra of the two groups
were used as the spectrum observation data of one group. A continuous observation
instrument (HOBO ware PRO) was used to obtain soil water content, and the sampling
interval was set to 5 min. The canopy temperature was obtained using the thermal infrared
sensors (METER SI-400), which can continuously record the canopy temperature and
ambient temperature every 1 min. Observations of soil moisture and canopy temperature
were taken after the wheat entered the tillering stage (November 6, 2019). To facilitate
analysis and mapping, we averaged the soil moisture and canopy temperature observations
to 1 day. The data quality and instrument error of the above instruments have been verified
in a previous experiment [32].

2.2. Red-edge Parameters

The red edge is a steep and nearly straight bevel between 660 nm and 770 nm and
occurs due to the strong absorption of vegetation chlorophyll in the red band and the
multiple scattering of the radiation in the near-infrared band inside the blade [33,34]. The
Agronomy 2022, 12, 2267 4 of 15

1st derivative of reflectance is commonly used to describe the red-edge parameters and
estimate the position of the red edge (Equation (1)).

R ( λ i ) − R ( λ i −1 )
R 0 ( λi ) = (1)
λ i − λ i −1

where R represents the reflectance, R0 represents the 1st derivative for reflectance, λ repre-
sents the wavelength, and i represents the spectral channel.
Six red-edge parameters were used in this study to study the response of spectral
characteristics of wheat during the different growing stages of drought. The seven red-
edge parameters included the red-valley position (RVP), the red-valley reflectance (RVR),
the red-edge position (REP), the red-edge reflectance (RER), the red-edge width (REW),
and the red-edge amplitude (REA) [35]. The definitions and equations for the 6 red-edge
parameters are listed in Table 2.

Table 2. Definitions and equations for 6 red-edge parameters.

Items Abbreviation Definition Equation Unit

The wavelength value corresponding to the
Red-Valley Position RVP λMin( R(λ)) nm
minimum reflectance within the red band.
Red-Valley Reflectance RVR The minimum reflectance within the red band. Min(R(λ)) –
The wavelength value corresponding to the
Red-Edge Position REP λMax( R0 (λ)) nm
maximum value of the first derivative spectrum.
The reflectance value at the band of the maximum  
Red-Edge Reflectance RER R λmax( R0 ) –
value of the 1st derivative.
The distance from the red-valley position to the
Red-Edge Width REW REP-RVP nm
red-edge position.
The maximum value of the first derivative of
Red-Edge Amplitude REA Max(R’(λ)) –
reflectance with red band.

2.3. Vegetation Water Indexes (VWI)

Eleven types of vegetation water indexes (VWIs) were applied in this study to quantify
drought change for different drought stages. The equations and references for the eleven
VWIs are listed in Table 3. Most of the VWIs were calculated via two hyper-spectral bands
that exhibited sensitivity for the water content in crops. We collected eleven VWIs, which
not only included the leaf-scale and field-scale VWIs, but also the satellite-scale VWIs; this
was aimed at comparing and verifying the applicability of different scales of VWIs in a
greenhouse experiment. Although various vegetation water indexes were scale-limited, we
tried to investigate if these indexes have good performances at the canopy scale, so we did
not stick to the index-scale limitations when we calculated indexes.
Agronomy 2022, 12, 2267 5 of 15

Table 3. Equations and references of the vegetation water indexes.

Index Equation Scale Reference

ρ900
1 WI ρ970 Field [18]
ρ900 /ρ970
2 WI/NDVI Field [19]
(ρ800 −ρ680 )/(ρ800 +ρ680 )
2 i =960
3 R975 31 ∑990 ρi Leaf [22]
1 i =920 1 i =1090
21 ∑ 940 ρ i + 11 ∑1100 ρi
ρ1600
4 MSI ρ820 Leaf [21]
ρ816 −ρ2218
5 DRI ρ816 +ρ2218
Leaf [23]
ρmax −ρmin
6 RDI1450 ρmax
Field [20]
ρ858 −ρ1240
7 NDWI ρ858 +ρ1240
Satellite [19]
ρ858
8 SRWI ρ1240 Satellite [24]
ρ820 −ρ1650
9 NDII ρ820 +ρ1650
Satellite [25]
ρ1650
10 R5/R7 ρ2218 Leaf [36]
11 GVMI (ρ820 +0.1)−(ρ1600 +0.02) Satellite [26]
(ρ820 +0.1)+(ρ1600 +0.02)

3. Results
3.1. Wheat Spectral Characteristics
The spectrum curve and the first derivative for the reflectance curve for different growth
stages of drought are shown in Figures 1 and 2. The oscillations centered at 1350, 1850, and
2400 nm were due to the influence of the water vapor in air, which were related to the moisture
content of the air on the day of the spectral measurement. Owing to the varying moisture
content in the air for different spectrum measurement periods, noise intensities were also
observed to be different. As shown in Figure 1a, the curve representing the milk maturity
stage drought and the treatment under no water stress was almost overlaid because the wheat
had not reached the milk maturity stage by Jan 9, and the watering gradient was the same as
that of the control group without water stress. The spectrum characteristics for the tillering
stage drought and jointing stage drought were quite different, as shown in Figure 1a. The
reflectance of the tillering stage drought was smaller than the reflectance values of the jointing
stage drought in the blue band of 350–450 nm and in the near-infrared band of 800–2300 nm.
The reflectance under no water stress was observed to be the smallest of the three groups. The
differences in spectrum characteristics for the four comparison groups are shown in Figure 1b.
In the near-infrared band, the decreasing order of reflectivity for different stages was the
tillering stage drought, the jointing stage drought, the milk maturity stage drought, and no
water stress. As these results indicate that crop reflectance in the near-infrared band was
closely related to the structure of chloroplasts, the upper limit of the water content of wheat
leaves may be proportional to the time of drought. However, in the blue and green bands, the
reflectivity of the four control groups had no obvious regularity.
Agronomy
Agronomy 2022,
2022, 12,
12, x2267
x FOR
FOR PEER
PEER REVIEW
REVIEW 66 6of
of15
of 15

Figure1.
Figure
Figure 1. Wheatspectral
1.Wheat
Wheat spectralcharacteristics
spectral characteristicsfor
characteristics fordifferent
for differentgrowing
different growing stages.
growing stages. (a)
stages. (a)Spectra
(a) Spectrafor
Spectra forwheat
for wheatmeasured
wheat measured
measured
on
on Jan 9 and (b) Mar 5. (c) The 1st derivative curve for (a). (d) The 1st derivative curve
on Jan 9 and (b) Mar 5. (c) The 1st derivative curve for (a). (d) The 1st derivative curve for (b).
Jan 9 and (b) Mar 5. (c) The 1st derivative curve for (a). (d) The 1st derivative curve for
for (b).
(b).

Figure
Figure 2.
Figure Soilwater
2. Soil
2. Soil watercontent
water content variation
content variation for
variation for continuous
for continuous observation
continuous observation (Date
observation (Date format:
(Date format: yyyymmdd).
format: yyyymmdd).
yyyymmdd).

The first
The
The first derivative
first derivative of
derivative of reflectivity
of reflectivity was
reflectivity was of
was of great
of great significance
great significancefor
significance forconfirming
for confirmingthe
confirming thered-
the red-
red-
edge
edge position,
edge position, reflecting
position, reflecting the
reflecting the wheat
the wheat spectral
wheat spectral characteristics
spectral characteristics and
characteristics and red-edge
and red-edge parameters.
red-edge parameters.
parameters. As As
As
shown
shown in Figure 1c,d, the first derivative of the spectrum fluctuated in the following ranges:
shown in in Figure
Figure 1c,d,
1c,d, the
the first
first derivative
derivative ofof the
the spectrum
spectrum fluctuated
fluctuated in in the
the following
following
450–750450–750
ranges: nm, 950–1150 nm, 1350–1550 nm, 1850–1950 nm, and 2300–2350 nm. The shape
ranges: 450–750 nm, 950–1150 nm, 1350–1550 nm, 1850–1950 nm, and 2300–2350 nm.
nm, 950–1150 nm, 1350–1550 nm, 1850–1950 nm, and 2300–2350 nm. The
The
of the
shape first derivative was mainly determined by the spectral characteristics of the vege-
shape of
of the
the first
first derivative
derivative waswas mainly
mainly determined
determined by by the
the spectral
spectral characteristics
characteristics of
of the
the
tation reflectance and and
vegetation observation conditions. When the first derivative waswas positive in a
vegetation reflectance
reflectance and observation
observation conditions.
conditions. When
When thethe first
first derivative
derivative was positive
positive
Agronomy 2022, 12, 2267 7 of 15

certain band, the reflectivity gradually increased in this band; otherwise, the reflectance
gradually decreased.

3.2. Soil Water Content Tendency

The variation in the soil water content for each group is shown in Figure 2. The blue
points in the top of the curves represent the irrigation time, which led to these variations.
There were three irrigation cycles during the wheat tillering stage, ten during the wheat
jointing stage, and six during the milk maturity jointing stage. After irrigation, the soil
water content immediately increased and then decreased gradually over the next 3–5 days.
The water stress test group also received a small amount of irrigation to prevent wheat
annihilation due to drought. The soil water contents of the experimental group subjected
to drought treatment in different growth periods were extremely significant. The soil
water content of the experimental group with the least irrigation in each growth period
maintained the lowest soil water content. However, the soil water content in the no-water-
stress group maintained a high level of soil water content.

3.3. Wheat Spectral Characteristics

To observe the variations in the red edge and the red-edge parameters for different
wheat treatments, the red-edge band at 660–770 nm was discussed separately (Figure 3).
The calculation results and numerical changes in the six red-edge parameters are shown
in Tables 4 and 5. The red-valley position (RVP) and red-edge position (REP) decided
the red-edge starting position. From the tillering stage to the milk maturity stage, the
RVP continued to decrease and the REP continued to increase with time, indicating that
a blue shift phenomenon appeared for the red edge. There were two main reasons for
the occurrence of this phenomenon: the phenology of wheat and the water stress during
the growth stages [13]. In addition, the values for REP and red-edge width (REW) for
the tillering stage drought and jointing stage drought were larger than those for the milk
maturity stage and no-water-stress groups, which indicates that the blue shift of red edge
was more obvious during the tillering stage drought and jointing stage drought. Red-valley
reflectance (RVR) reflected the minimum value of red-edge reflectivity. RVR decreased
for each group during the wheat growth periods and did not show obvious differences
between the four groups, which means that RVR was affected by the influence of wheat
phenology, and RVR was not sensitive to water stress. The red-edge reflectance (RER) was
the most obvious index among the four comparison groups, indicating that RER could
better reflect the spectral response characteristics of drought in different growth periods.
The change in the red-edge amplitude (REA) value was small during the wheat growth
periods, and there were small differences between the four comparison groups.
VP −1 −9 −3 −5
VR −0.039 −0.028 −0.04 −0.054
EP 21 21 2 2
ER 0.3 0.236 −0.008 −0.068
EW 22 12, 2267
Agronomy 2022, 30 5 7 8 of 15
EA 0.004 0.003 0.001 0.001

3. Red-edge
Figurespectra
Figure 3. Red-edge spectrawheat
for different for different
growing wheat growing
stages. stages. the
(S represents (S represents the tillering stage
tillering stage
drought; JS represents the jointing stage drought; MS represents the milk maturity stage drought; stage drought;
drought; JS represents the jointing stage drought; MS represents the milk maturity
and no
and NS represents NSwater
represents
stress.no water stress.0109
Furthermore, Furthermore,
represents0109 represents
the spectra the spectra
obtained on Janobtained
9, whileon Jan 9, while
0305 represents0305
the represents
spectra obtained on Mar 5.
the spectra obtained on Mar 5.

3.4. Canopy Temperature Characteristics

Table 4. Calculation of the Analysis
six red-edge parameters for wheat exposed to drought during differ-
In order ent stages. the influence of ambient temperature on canopy temperature, we
to reduce
introduced the concept of the difference between canopy temperature and ambient tem-
Milk
Tillering Jointing to analyze canopy temperature results.
perature (DCTAT) TilleringDue to Jointing
the transpiration of the No Water
Milk Maturity No Water Maturity
Stage Stage Stage Stage
crop,
Drought
the canopy
Drought
temperature of
Stage Droughtwheat was
Stressusually lower than the ambient temperature
Drought Drought
Stage Stress
(Jan. 9) (Jan. 9) Drought (Mar. 5)
and
(Jan. 9) the DCTAT
(Jan. 9)value was usually negative. During wheat
(Mar. 5) tillering stage,
(Mar. 5) the DCTAT
(Mar. 5)
value ranged between −7.8 and −1.3 °C (Figure 4a), and the canopy temperature in the
RVP 669 668 669 669 670 672 674 677
RVR water-stress 0.079
0.073 experimental group0.054 (green line)
0.055 was, on 0.112
average, 0.720.094 ± 0.56 °C higher 0.109 than 0.107
REP in
720the no-water-stress
716 experimental
698 group.697During the699 wheat jointing 696 stage, the 695DCTAT 695
RER 0.531
value ranged0.465 between −9.92 0.193
and 0.76 °C 0.18
(Figure 4b), and0.231 the canopy 0.201temperature 0.248in the 0.229
REW 51 48 29 28 29 24 21 18
REA water-stress experimental
0.017 0.016 group
0.016 (blue line)0.016was, on average,
0.013 1.36 ± 0.015
0.44 °C higher than in
0.015 0.013
the no-water-stress experimental group. During the wheat milk and maturity stage, the
DCTAT valueTable ranged between
5. Changes in−13.49
the six and 0.80 °C
red-edge (Figure 4c),
parameters (the and
valuethewascanopy temperature
calculated by subtracting red-edge
in the water-stress experimental group (grey line) was, on average, 0.67 ±
parameters obtained on Jan 9 with the red-edge parameters on Mar 5; + represents 0.78 °C higher an increase in the
than in the no-water-stress experimental
red-edge parameters, group. Aahigher
while—represents decreaseDCTAT represents
in the red-edge a higher can-
parameters).
opy temperature, a higher canopy temperature indicates weaker transpiration in wheat,
and weaker
Tillering transpiration
Stage Drought reflects
Jointing a higher
Stage intensity
Drought of water
Milk Maturity stress
StageonDrought
wheat. Comparing No Water Stress
RVP
water − stress
1
treatment DCTAT results
−9
with those of no-water-stress
−3
treatment during −5
RVR wheat different growth periods−above,
−0.039 0.028 we found that the largest −0.04changes in canopy tem- −0.054
REP perature 21 of wheat under water stress 21 occurred at the jointing stage. 2 2
RER 0.3 0.236 −0.008 −0.068
REW 22 30 5 7
REA 0.004 0.003 0.001 0.001

3.4. Canopy Temperature Characteristics Analysis

In order to reduce the influence of ambient temperature on canopy temperature,
we introduced the concept of the difference between canopy temperature and ambient
temperature (DCTAT) to analyze canopy temperature results. Due to the transpiration of
the crop, the canopy temperature of wheat was usually lower than the ambient temperature
and the DCTAT value was usually negative. During wheat tillering stage, the DCTAT
value ranged between −7.8 and −1.3 ◦ C (Figure 4a), and the canopy temperature in the
water-stress experimental group (green line) was, on average, 0.72 ± 0.56 ◦ C higher than
Agronomy 2022, 12, 2267 9 of 15

in the no-water-stress experimental group. During the wheat jointing stage, the DCTAT
value ranged between −9.92 and 0.76 ◦ C (Figure 4b), and the canopy temperature in the
water-stress experimental group (blue line) was, on average, 1.36 ± 0.44 ◦ C higher than in
the no-water-stress experimental group. During the wheat milk and maturity stage, the
DCTAT value ranged between −13.49 and 0.80 ◦ C (Figure 4c), and the canopy temperature
in the water-stress experimental group (grey line) was, on average, 0.67 ± 0.78 ◦ C higher
than in the no-water-stress experimental group. A higher DCTAT represents a higher
canopy temperature, a higher canopy temperature indicates weaker transpiration in wheat,
and weaker transpiration reflects a higher intensity of water stress on wheat. Comparing
water stress treatment DCTAT results with those of no-water-stress treatment during wheat
different growth periods above, we found that the largest changes in canopy temperature
of wheat under water stress occurred at the jointing stage.

3.5. Vegetation Water Index Analysis

The VWI used in this study was widely shown to perform well in characterizing the
crop water content, and the hyperspectral data of wheat from ground-based observations
are a reliable source for calculating the VWI. Using two spectral observation data with
200 spectral curves from Jan. 9 and Mar. 5 for four different experiment groups, six vegeta-
tion water indexes were calculated to study variations in VWIs (Figure 5 and Table 6). The
5main basis for determining whether VWI can characterize different stages of drought in
wheat is whether there are significant differences in VWI values calculated at different times
for each control, and whether such differences have regularity (including a comparison
between samples exposed to drought at different growth stages, and between drought
samples and no-water-stress samples). Compared with the no-water-stress test group, the
differences in the WI curve slopes for different water sets showed significant differences
and regularity, which implied that the WI might be able to classify different growth stages
of drought for wheat. Water stress in different stages tends to cause permanent damage to
the wheat, which could be distinguished by the WI at the leaf scale.

Table 6. Results with each VWI for different drought stages.

Tillering Stage Jointing Stage Milk Maturity

Date No Water Stress
Drought Drought Stage Drought
Jan. 9 1.06 1.09 1.04 1.04
WI
Mar. 5 1.04 1.04 1.03 1.09
Jan. 9 1.37 1.49 2.06 2.07
WI/NDVI
Mar. 5 2.16 2.41 3.24 4.16
Jan. 9 0.96 0.94 0.97 0.97
R975
Mar. 5 0.98 0.96 0.97 0.90
Jan. 9 0.58 0.52 0.61 0.61
MSI
Mar. 5 0.54 0.55 0.59 0.43
Jan. 9 0.43 0.47 0.43 0.43
DRI
Mar. 5 0.51 0.52 0.48 0.62
Jan. 9 0.94 0.95 0.98 0.96
RDI1450
Mar. 5 0.92 0.94 0.95 0.97
Jan. 9 0.10 0.13 0.06 0.06
NDWI
Mar. 5 0.08 0.09 0.07 0.06
Agronomy 2022, 12, 2267 10 of 15

Table 6. Cont.

Tillering Stage Jointing Stage Milk Maturity

Date No Water Stress
Drought Drought Stage Drought
Jan. 9 1.22 1.31 1.12 1.13
SRWI
Mar. 5 1.18 1.21 1.16 1.12
Jan. 9 0.24 0.28 0.20 0.20
NDII
Mar. 5 0.26 0.35 0.26 0.22
Jan. 9 1.52 1.56 1.66 1.65
R5/R7
Mar. 5 1.81 2.08 1.87 1.81
Jan. 9
Agronomy 2022, 12, x FOR PEER REVIEW 0.30 0.35 0.28 90.28
of 15
GVMI
Mar. 5 0.33 0.43 0.33 0.30

Figure 4. Difference
Figure between
4. Difference canopy
between temperature
canopy andand
temperature ambient temperature
ambient (DCTAT).
temperature (a–c)(a–c)
(DCTAT). repre-
represent
sent the tillering stage, jointing stage, and milk maturity stage, respectively. No data were attributed
the tillering stage, jointing stage, and milk maturity stage, respectively. No data were attributed to
to sensor damage or equipment overhaul.(Date format: yyyymmdd).
sensor damage or equipment overhaul.(Date format: yyyymmdd).

3.5. Vegetation Water Index Analysis

The VWI used in this study was widely shown to perform well in characterizing the
crop water content, and the hyperspectral data of wheat from ground-based observations
are a reliable source for calculating the VWI. Using two spectral observation data with 200
 drought samples and no-water-stress samples). Compared with the no-water-stress test
group, the differences in the WI curve slopes for different water sets showed significant
differences and regularity, which implied that the WI might be able to classify different
Agronomy 2022, 12, 2267 11 of 15
growth stages of drought for wheat. Water stress in different stages tends to cause perma-
nent damage to the wheat, which could be distinguished by the WI at the leaf scale.

Figure 5. The slope

slope change in eleven VWIs on two observation dates (Jan. 9 and Mar. 5) for different
growth stages drought for wheat.

Except for WI/NDVI and R5/R7, the other five leaf/field-scale VWIs (WI, R975, MSI,
RDI, and RDI1450) can better distinguish wheat water stress and the absence of it via slope
variability. For RDI1450 and the WI, when the slope was greater than 0, it represented that
the wheat had not suffered water stress during the growth period, and when the slope
was less than 0, it represented that the wheat had suffered water stress. For R975 and MSI,
when the slope was less than 0, it represented that the wheat had not suffered water stress
during the growth period, and when the slope was greater than 0, it represented that the
wheat had suffered water stress. For RDI, the increase in slope decreased, indicating that
the wheat suffered water stress.
However, WI/NDVI and R5/R7, which were both leaf/field-scale VWIs, showed
no significant differences between the water-stress groups and the no-water-stress group.
All the satellite-scale VWIs (including NDWI, SRWI, NDII, and GVMI) did not exhibit
Agronomy 2022, 12, 2267 12 of 15

good performance in distinguishing water-stressed and non-water-stressed wheat. This

result proved that VWIs had certain requirements for researching different scales; this is
reinstated by the fact that satellite-scale VWIs could not be applied at the leaf/field scale.
However, leaf-scale and field-scale VWIs exhibited a certain degree of applicability for each
of the scales.

4. Discussion
4.1. Advantages of Ground-Based Hyperspectral Monitor Wheat Drought
With the development of ground-based remote-sensing sensors, higher temporal and
spectral resolution sensors emerged, covering various wavelength bands, such as visible
light, near-infrared, mid-infrared, far-infrared, and microwave [37]. A variety of vegetation
parameters monitored tends to diversify, such as the leaf area index, chlorophyll, chloro-
phyll fluorescence, plant water content, etc. A wide range of environmental parameters can
be monitored to assist in drought monitoring, such as temperature, canopy temperature,
soil moisture, water vapor pressure, wind speed, air pressure, etc. [38,39]. With this large
number of precise multi-parameter observations, we can better clarify the mechanisms of
drought in winter wheat by analyzing the interrelationships between the various vegetation
and environmental parameters [40]. In this study, for example, by designing comparative
experiments in the greenhouse with wheat subjected to drought at different growth stages,
we collected a large amount of raw data on canopy temperature, soil moisture, and canopy
spectra. We observed differences in soil moisture and canopy temperature in different
comparison samples, and observed the interrelationship between different irrigation quan-
tities, canopy temperature, and soil moisture. The spectral measurements of the wheat
canopy were used to observe the differences in spectra between the different samples,
and the changes in the red-edge parameters. Vegetation moisture indexes were calculated
from the vegetation hyper spectrum to select the most suitable index for evaluating the
wheat drought at different growth stages. These results demonstrate the importance of
ground-based hyperspectral remote sensing in the study of drought mechanisms in wheat.
This refined access to data and the ability to analyze qualitatively and quantitatively is
irreplaceable to other research methods.

4.2. Enlightenment for Large-Scale Drought Monitoring

Observing the spectral characteristics of drought and analyzing the water stress
spectrum changes at different growth stages for wheat is important for clarifying the
drought mechanism of wheat, which has great significance for guiding scientific irrigation,
realizing crop growth monitoring, improving crop yield, and ensuring food security [41,42].
Research on wheat spectrum based on the greenhouse-controlled variable experiment has
accumulated rich data and valuable methodology experience for large-scale wheat drought
monitoring. With the rapid developments in remote-sensing technology in recent years, the
spatial resolution, temporal resolution, and spectral resolution of remote-sensing sensors
have greatly improved, resulting in the continuous development of large-scale, timely, and
accurate wheat cropping monitoring. In particular, the use of hyper-spectral remote-sensing
sensors is increasingly popular and can easily obtain remote-sensing images with high
resolutions for wheat in each growth stage. Based on the spectral response characteristics
of wheat, wheat growth with different drought periods can be classified. Combined with
accurate soil moisture monitoring technology, meteorology, and field wheat observation
data, a deeper understanding of drought mechanisms and a better arrangement of irrigation
events can be achieved while minimizing yield losses. However, solving the applicability
of hyper-spectral research at the leaf/field scale and the satellite scale is a difficult problem
in the field of hyper-spectral drought monitoring.

5. Conclusions
In this study, we designed a greenhouse experiment with three different growth stages
(tillering, jointing, and milk maturity stage) to study drought in wheat for mechanisms
Agronomy 2022, 12, 2267 13 of 15

of wheat drought. A variety of ground-based continuous-record remote-sensing sensors

were used to monitor key vegetation and environmental parameters during wheat growth
periods, which was an effective way to collect precise data automatically, continuously
towards wheat drought. The monitored parameters included ambient temperature, canopy
temperature, soil moisture, and wheat canopy spectral. The collected observations were
statistically analyzed to understand the different changing conditions of soil moisture and
canopy temperature for different water strategies samples, as well as using canopy spectral
to calculate red-edge parameters and some commonly used vegetation moisture indices to
characterize wheat drought at different growth stages from a quantitative and qualitative
perspective. This study is of great significance for understanding the drought mechanism
of wheat and the spectral response characteristics of drought at different growth stages.
It also provided a valuable experience for large-scale remote sensing for the monitoring
of wheat under drought conditions to increase wheat yield and ensure food security. The
main conclusions drawn from this study are as follows:
(1) SWC and canopy temperature had a significant difference between wheat drought
at different growth stages, and canopy temperature at the jointing stage drought had the
strongest change (−9.92 to 0.76 ◦ C).
(2) The wheat at different growth stages showed significant differences in spectral re-
sponse characteristics during the drought period. The red-edge parameters and vegetation
water indexes can better show the spectral characteristic differences.
(3) Owing to the wheat phenology and water stress during growth periods of wheat
with different water stress treatments, the changes in RVP and REP can constitute a red-
edge blue shift. The value of blue shift was 21 nm for the tillering stage drought and
jointing stage drought, which were significantly larger than those of the milk maturity
stage drought (2 nm) and the treatment with no water stress (2 nm).
(4) Among the red-edge parameters, RER was the most optimum to reveal the red-edge
characteristics for wheat at different growth stages.
(5) Based on eleven vegetation water indexes (VWIs), the WI exhibited a good perfor-
mance for distinguishing wheat water stress at different growth stages.

Author Contributions: Conceptualization, Q.L.; methodology, Q.L.; software, Q.L.; validation, M.G.
and Z.-L.L.; formal analysis, Q.L., M.G. and Z.-L.L.; resources, Q.L.; data curation, Q.L.; writing—
original draft preparation, Q.L.; writing—review and editing, Q.L., M.G. and Z.-L.L.; visualization,
Q.L.; project administration, M.G.; funding acquisition, M.G. All authors have read and agreed to the
published version of the manuscript.
Funding: Funding for this research was provided by the National Natural Science Foundation of
China (grant number 41921001, 41871282).
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. González, A.; Bermejo, V.; Gimeno, B.S. Effect of different physiological traits on grain yield in barley grown under irrigated and
terminal water deficit conditions. J. Agric. Sci. 2010, 148, 319–328. [CrossRef]
2. Dwivedi, S.K.; Arora, A.; Kumar, S. Paclobutrazol-induced alleviation of water-deficit damage in relation to photosynthetic
characteristics and expression of stress markers in contrasting wheat genotypes. Photosynthetica 2016, 55, 351–359. [CrossRef]
3. Du, L.; Yang, J.; Chen, B.; Sun, J.; Chen, B.; Shi, S.; Song, S.; Gong, W. Novel Combined Spectral Indices Derived from
Hyperspectral and Laser-Induced Fluorescence LiDAR Spectra for Leaf Nitrogen Contents Estimation of Rice. Remote Sens. 2020,
12, 185. [CrossRef]
4. Wang, H.; Vicente-Serrano, S.M.; Tao, F.; Zhang, X.; Wang, P.; Zhang, C.; Chen, Y.; Zhu, D.; El Kenawy, A. Monitoring winter
wheat drought threat in Northern China using multiple climate-based drought indices and soil moisture during 2000–2013. Agric.
For. Meteorol. 2016, 228–229, 1–12. [CrossRef]
5. Li, H.; Zhang, Y.; Lei, Y.; Antoniuk, V.; Hu, C. Evaluating Different Non-Destructive Estimation Methods for Winter Wheat
(Triticum aestivum L.) Nitrogen Status Based on Canopy Spectrum. Remote Sens. 2019, 12, 95. [CrossRef]
6. Fu, Y.; Yang, G.; Li, Z.; Li, H.; Li, Z.; Xu, X.; Song, X.; Zhang, Y.; Duan, D.; Zhao, C.; et al. Progress of hyperspectral data processing
and modelling for cereal crop nitrogen monitoring. Comput. Electron. Agric. 2020, 172, 105321. [CrossRef]
Agronomy 2022, 12, 2267 14 of 15

7. Wang, M.; Yang, K.-M.; Zhang, W. Hyperspectral monitoring of maize leaves under copper stress at different growth stages.
Remote Sens. Lett. 2020, 11, 343–352. [CrossRef]
8. Liu, L.; Wang, J.; Huang, W.; Zhao, C.; Zhang, B.; Tong, Q. Estimating winter wheat plant water content using red edge parameters.
Int. J. Remote Sens. 2004, 25, 3331–3342. [CrossRef]
9. Kamal, N.M.; Gorafi, Y.S.A.; Abdelrahman, M.; Abdellatef, E.; Tsujimoto, H. Stay-Green Trait: A Prospective Approach for Yield
Potential, and Drought and Heat Stress Adaptation in Globally Important Cereals. Int. J. Mol. Sci. 2019, 20, 5837. [CrossRef]
10. Meng, J.; Wu, B.; Niu, L.; Du, X.; Zhang, F. Validation of HJ-1 HSI Red Edge Parameters Based on Spectral Reflectance of Hyperion.
Spectrosc. Spectr. Anal. 2010, 30, 2205–2210.
11. Wang, J.; Liu, L.; Li, C.; Xiang, T.; Zheng, S.; Ding, S.; Liu, P. The Study for Formation Mechanism of Spectral Red Edge of Green
Vegetation. Spectrosc. Spectr. Anal. 2017, 37, 3940–3946.
12. Ju, C.-H.; Tian, Y.-C.; Yao, X.; Cao, W.-X.; Zhu, Y.; Hannaway, D. Estimating Leaf Chlorophyll Content Using Red Edge Parameters.
Pedosphere 2010, 20, 633–644. [CrossRef]
13. He, K.; Zhao, S.; Lai, J.; Luo, Y.; Qin, Z. Effects of Water Stress on Red-Edge Parameters and Yield in Wheat Cropping. Spectrosc.
Spectr. Anal. 2013, 33, 2143–2147.
14. Li, Y.; Chen, X.; Luo, X.; Li, B.; Zhang, L. Effects of Cuprum Stress on Position of Red Edge of Maize Leaf Reflection Hyperspectra
and Relations to Chlorophyll Content. Spectrosc. Spectr. Anal. 2018, 38, 546–551.
15. Rasheed, F.; Delagrange, S.; Lorenzetti, F. Detection of plant water stress using leaf spectral responses in three poplar hybrids
prior to the onset of physiological effects. Int. J. Remote Sens. 2020, 41, 5127–5146. [CrossRef]
16. Claudio, H.C.; Cheng, Y.; Fuentes, D.A.; Gamon, J.A.; Luo, H.; Oechel, W.; Qiu, H.L.; Rahman, A.F.; Sims, D.A. Monitoring
drought effects on vegetation water content and fluxes in chaparral with the 970 nm water band index. Remote Sens. Environ.
2006, 103, 304–311. [CrossRef]
17. Tao, F.; Chen, Y.; Fu, B. Impacts of climate and vegetation leaf area index changes on global terrestrial water storage from 2002 to
2016. Sci. Total Environ. 2020, 724, 138298. [CrossRef]
18. Penuelas, J.; Filella, I.; Biel, C.; Serrano, L.; Savé, R. The reflectance at the 950–970 nm region as an indicator of plant water status.
Int. J. Remote Sens. 1993, 14, 1887–1905. [CrossRef]
19. Peñuelas, J.; Pinol, J.; Ogaya, R.; Filella, I. Estimation of plant water concentration by the reflectance water index WI (R900/R970).
Int. J. Remote Sens. 1997, 18, 2869–2875. [CrossRef]
20. Zhao, C.-J.; Zhou, Q.; Wang, J.; Huang, W.-J. Band selection for analysing wheat water status under field conditions using relative
depth indices (RDI). Int. J. Remote Sens. 2004, 25, 2575–2584. [CrossRef]
21. Ceccato, P.; Flasse, S.; Tarantola, S.; Jacquemoud, S.; Grégoire, J. Detecting vegetation leaf water content using reflectance in the
optical domain. Remote Sens. Environ. 2001, 77, 22–33. [CrossRef]
22. Pu, R.; Ge, S.; Kelly, N.M.; Gong, P. Spectral absorption features as indicators of water status in coast live oak ( Quercus agrifolia )
leaves. Int. J. Remote Sens. 2003, 24, 1799–1810. [CrossRef]
23. Datt, B. Remote Sensing of Water Content in Eucalyptus Leaves. Aust. J. Bot. 1999, 47, 909–923. [CrossRef]
24. Zarco-Tejada, P.J.; Ustin, S.L. Modeling canopy water content for carbon estimates from MODIS data at land EOS validation sites.
IEEE Int. Geosci. Remote Sens. Symp. 2001, 2, 976152.
25. Hardisky, M.S.; Klemas, V.; Smart, R.M. The influence of soil salinity, growth form, and leaf moisture on the spectral radiance of
Spartina Alterniflora canopies. Photogramm. Eng Remote Sens 1983, 49, 77–84.
26. Ceccato, P.; Gobron, N.; Flasse, S.; Pinty, B.; Tarantola, S. Designing a spectral index to estimate vegetation water content from
remote sensing data: Part 1. Remote Sens. Environ. 2002, 82, 188–197. [CrossRef]
27. Gao, M.; Li, Z.L.; Liu, S.; Gao, Y.; Leng, P.; Duan, S. Drought loss assessment combining remote sensing and a crop growth model
for maize in Yunnan Province, China. Int. J. Remote Sens. 2019, 5, 2151–2165. [CrossRef]
28. Wang, W.; Ertsen, M.W.; Svoboda, M.D.; Hafeez, M. Propagation of Drought: From Meteorological Drought to Agricultural and
Hydrological Drought. Adv. Meteorol. 2016, 2016, 6547209. [CrossRef]
29. Wu, S.; Yang, P.; Ren, J.; Chen, Z.; Liu, C.; Li, H. Winter wheat LAI inversion considering morphological characteristics at different
growth stages coupled with microwave scattering model and canopy simulation model. Remote Sens. Environ. 2020, 240, 111681.
[CrossRef]
30. Li, B.; Wang, T.; Meng, F. Double Quantitative Analysis of the Effects of Meteorological Factors on Winter Wheat Yield at Different
Growth Stages Based on Grey Relational Analysis. J. Grey Syst. 2018, 30, 31–43.
31. Tao, F.; Shuai, Z.; Zhang, Z. Spatiotemporal changes of wheat phenology in China under the effects of temperature, day length
and cultivar thermal characteristics. European Journal of Agronomy. 2012, 43, 201–212. [CrossRef]
32. Shen, L.; Gao, M.; Yan, J.; Li, Z.; Duan, S. Hyperspectral Estimation of Soil Organic Matter Content using Different Spectral
Preprocessing Techniques and PLSR Method. Remote Sens. 2020, 12, 1206. [CrossRef]
33. Horler, D.N.H.; Dockray, M.; Barber, J. The red edge of plant leaf reflectance. Int. J. Remote Sens. 1983, 4, 273–288. [CrossRef]
34. Jiang, S.; Wang, F.; Shen, L.; Liao, G. Local detrended fluctuation analysis for spectral red-edge parameters extraction. Nonlinear
Dyn. 2018, 93, 995–1008. [CrossRef]
35. Curran, P.J.; Dungan, J.L.; A Macler, B.; E Plummer, S. The effect of a red leaf pigment on the relationship between red edge and
chlorophyll concentration. Remote Sens. Environ. 1991, 35, 69–76. [CrossRef]
Agronomy 2022, 12, 2267 15 of 15

36. Elvidge, C.D.; Lyon, R.J. Influence of rock-soil spectral variation on the assessment of green biomass. Remote Sens. Environ. 1985,
17, 265–279. [CrossRef]
37. Picard, R.H.; Ben-David, A.; Davidson, C.E.; SchFer, K.; Comeron, A.; Weele, M.V. Infrared radiative transfer model for aerosol
clouds: Implications to remote sensing by ground-based and airborne sensors. SPIE Remote Sens. 2009, 7475, 108–117.
38. Marzano, F.S.; Cimini, D.; Montopoli, M. Investigating precipitation microphysics using ground-based microwave remote sensors
and disdrometer data. Atmos. Res. 2010, 97, 583–600. [CrossRef]
39. Sempreviva, A.M. Study of the vertical structure of the coastal boundary layer integrating surface measurements and ground-
based remote sensing. Sensors 2020, 20, 6516.
40. Zhang, H.; Lan, Y.; Suh, C.P.-C.; Westbrook, J.; Hoffmann, W.C.; Yang, C.; Huang, Y. Fusion of remotely sensed data from airborne
and ground-based sensors to enhance detection of cotton plants. Comput. Electron. Agric. 2013, 93, 55–59. [CrossRef]
41. Su, B.; Zhao, G.; Dong, C. Spatiotemporal variability of soil nutrients and the responses of growth during growth stages of winter
wheat in northern China. PLoS ONE 2018, 13, e0203509. [CrossRef] [PubMed]
42. Franz, T.E.; Pokal, S.; Gibson, J.P.; Zhou, Y.; Gholizadeh, H.; Tenorio, F.A.; Rudnick, D.; Heeren, D.; McCabe, M.; Ziliani, M.;
et al. The role of topography, soil, and remotely sensed vegetation condition towards predicting crop yield. Field Crop. Res. 2020,
252, 107788. [CrossRef]