← Back
Nucleic Acid Binding Behavior and Cytotoxicity Properties of a Ruthenium(II) Polypyridyl Complex
Int. J. Disaster Risk Sci. 2013, 4 (2): 68–76
doi:10.1007/s13753-013-0008-8
ARTICLE
Evaluation of the Visible and Shortwave Infrared Drought
Index in China
Ning Zhang1,2, Yang Hong2, Qiming Qin1,*, and Lin Zhu3
1
Institute of Remote Sensing and GIS, Peking University, Beijing 100871, China
School of Civil Engineering and Environmental Sciences, University of Oklahoma, OK 73072, USA
3
National Satellite Meteorological Center of China Meteorological Administration, Beijing 100081, China
2
Abstract In this article, the performance of the Visible and
Shortwave infrared Drought Index (VSDI), a drought index
recently developed and validated in Oklahoma, United States,
is further explored and validated in China. The in-situ measured soil moisture from 585 weather stations across China
are used as ground-truth data, and five commonly used
drought indices are compared with VSDI for surface drought
monitoring. The results reveal that VSDI is robust and reliable
in the estimation of surface dryness—it has the highest
correlation with soil moisture among the six indices when
computed using both the original and cloud removed data. All
six indices show the highest correlation with soil moisture
at the 10 cm layer and the averaged 10–50 cm layer. The
spatiotemporal patterns of surface moisture indicated by
the MODIS-based VSDI are further compared with the
precipitation-based drought maps and the Global Land Data
Assimilation System (GLDAS) simulated surface soil moisture
maps over five provinces located in the Middle-Lower Yangtze
Plain of China. The results indicate that despite the difference
between the spatial and temporal resolutions of the three
products, the VSDI maps still show good agreement with the
other two drought products through the rapidly alternating
drought and flood events in 2011 in this region. Therefore,
VSDI can be used as an effective surface wetness indicator at
both the provincial and the national scales in China.
Keywords China, drought map, drought monitoring, optical
remote sensing, soil moisture, VSDI
1
Introduction
Drought is a slow developing phenomenon that accumulates
over a period across a vast area, and its effects may last for
years after it ends (Tannehill 1947). Droughts impact both
surface and groundwater resources and can lead to reduced
water supply, deteriorated water quality, crop failure, reduced
range productivity, diminished power generation, and great
economic and social damages (Riebsame, Changnon, and
Karl 1991; Wilhite 2000; Mishra and Singh 2010). Bryant
* Corresponding author. E-mail: qmqinpku@163.com
(1991) ranked natural hazards based on various key characteristics including severity, duration, areal extent, loss of life,
economic loss, social effect, long-term impact and so on, and
found that drought ranks first among all natural hazards
(Mishra and Singh 2010). Monitoring drought conditions and
surface moisture status using satellite remote sensing is of
great interest for drought disaster management and for the
sustainable development of eco-environments.
In optical remote sensing, various studies have been conducted to investigate the effectiveness of existing drought
indices in surface moisture monitoring. The Normalized
Difference Vegetation Index (NDVI) is commonly used in
vegetation drought monitoring (Brown et al. 2008). Jimmy
and Andrew (2002) analyzed the sensitivity of NDVI to soil
moisture in the U.S. Corn Belt and found weak correlation
between them. Specifically, NDVI was found to lag 8 weeks
behind the soil moisture variation. The research of Chen,
Huang, and Jackson (2005) revealed that NDVI and the
Normalized Difference Water Index (NDWI) were both good
candidates for vegetation moisture monitoring, and NDWI
performed better than NDVI, which is consistent with the
findings of Jackson et al. (2004). Gu et al. (2007) also compared NDVI and NDWI using MODIS data for grassland
drought assessment in the central United States. The results
indicate strong relationships among NDVI, NDWI, and
drought conditions, and NDWI had a quicker response to
drought conditions than NDVI. However, the experiment
conducted in Oklahoma, United States (Gu et al. 2008)
indicated that NDVI and NDWI had comparable sensitivities
to soil moisture variation and no additional benefit was gained
by using NDWI. Yilmaz, Hunt, and Jackson (2008) analyzed
the relationship between the Normalized Difference Infrared
Index (NDII) (Hardinsky, Lemas, and Smart 1983) and
Vegetation Water Content (VWC) through the Soil Moisture
Experiment 2002 and 2005 (SMEX02 and SMEX05) in Iowa,
United States, and concluded that NDII was related to canopy
Equivalent Water Thickness (EWT) and indirectly related to
VWC. Zhao et al. (2009) validated the shortwave infrared
water stress index (SIWSI) developed by Fensholt and
Sandholt (2003) in northwestern China and recommended
© The Author(s) 2013. This article is published with open access at Springerlink.com
www.ijdrs.org www.springer.com/13753
Zhang et al. Evaluation of the Visible and Shortwave Infrared Drought Index in China
using MODIS band 6 as shortwave infrared band in SIWSI
calculation. Wang and Qu (2009) comprehensively reviewed
progress in soil moisture monitoring using optical, thermal,
passive microwave, and active microwave remote sensing
techniques. Zhang et al. (2010) also reviewed advances in
research of vegetation water content retrieval using optical
remote sensing, including various vegetation moisture
indices and the radiative transfer model (RTM) methods.
Recently, a simple method for the estimation of surface
dryness, the Visible and Shortwave infrared Drought Index
(VSDI), has been developed (Zhang et al. 2013). The VSDI is
based on the combination of optical spectral bands located in
Blue, Red, and Shortwave infrared (SWIR) regions. It shows
potential advantages for monitoring both soil and vegetation
moisture and for drought monitoring throughout plant growing seasons, which distinguish it from other drought indices
that were either designed for vegetation water content estimation or confined to soil moisture monitoring. This index has
been proven effective in monitoring the drought development
over Oklahoma in the United States (Zhang et al. 2013), but
its responses to moisture dynamics in other regions and different phenological conditions requires further examination.
The main objective of this study is to further explore the
performance of VSDI in China as a surface drought index
at the regional and national scales. To achieve this goal, two
experiments were carried out: one to compare VSDI with five
commonly used drought indices over China’s 585 weather
stations; and the other to compare VSDI drought maps with
other drought products covering the Middle-Lower Yangtze
Plain in China. Based on the above validation, VSDI can be
deemed an efficient drought index applicable for drought
monitoring in China at different scales. Our work on VSDI is
to provide a new solution for future surface dryness monitoring and VSDI is expected to be used as a generalized drought
index in various areas and ecosystems.
2
Methods
Five typical drought indices were selected as the candidate
drought indices to compare with VSDI, including the Land
Surface Water Index (LSWI) (Xiao et al. 2004), the Moisture
Stress Index (MSI) (Hunt and Rock 1989), the Surface Water
Capacity Index (SWCI) (Du et al. 2007), the Shortwave
Infrared Soil Moisture Index (SIMI) (Yao et al. 2011), and
the Normalized Difference Vegetation Index (NDVI)
(Deering 1978). These indices are commonly used for surface
dryness monitoring and only require optical spectral bands
for calculation.
2.1
Land Surface Water Index (LSWI)
The LSWI is a popular drought index for vegetation moisture
monitoring by using the normalized difference between
Near-infrared (NIR) and shortwave infrared (SWIR) bands.
It was proposed by Xiao et al. (2004) based on the band 2
69
(NIR, 841–876 nm) and band 6 (SWIR, 1628–1652 nm) of
MODIS data:
LSWI = (RNIR – RSWIR)/(RNIR + RSWIR)
Eq. 1
This index has also been studied and referred to under
other names, such as NDII (Hardinsky, Lemas, and Smart
1983), NDWI (Gao 1996), and SIWSI (Fensholt and Sandholt
2003), which varies with the specific wavelengths or sensordependent bands used. Despite the different names, one thing
these indices have in common is that the NIR spectral region
serves as a moisture reference band and the SWIR spectral
domain is used as the moisture measuring band. This index
has been proven effective in monitoring vegetation water
content in a variety of studies (Zarco-Tejada, Rueda, and
Ustin 2003; Maki, Ishiahra, and Tamura 2004; Xiao et al.
2005; Gu et al. 2008).
2.2
Moisture Stress Index (MSI)
The MSI is a simple water ratio index for the estimation of
leaf relative water content (%) and equivalent water thickness
(EWT, g cm–2) of different plant species (Hunt and Rock
1989). It is calculated as R1600 nm/R820 nm. In this study,
the SWIR band (band 6 of MODIS data) is used instead
of the MIR band in equation 2 considering that the strong
water absorption bands at SWIR spectrum are more sensitive
to moisture variation than other optical spectral regions
(Dawson et al. 1999; Ceccato et al. 2001; Chuvieco et al.
2002). In addition, the weak water absorption at NIR band
makes it less sensitive to water variation (Gao 1996; Ghulam
et al. 2008), thus the ratio between SWIR and NIR bands can
effectively reduce the scattering effect of the single band and
highlight the water variation in vegetation leaves.
MSI = RSWIR6/RNIR
2.3
Eq. 2
Surface Water Capacity Index (SWCI)
The SWCI is a surface moisture index constructed by using
the normalized difference between two SWIR bands (bands 6
and 7) of MODIS data (Du et al. 2007):
SWCI = (RSWIR6 – RSWIR7)/(RSWIR6 + RSWIR7)
Eq. 3
Since MODIS band 6 (1628–1652 nm) and band 7 (2105–
2155 nm) correspond to the valley and the peak of the water
absorption curve respectively and are both sensitive to
moisture variation, the difference between the two bands
(RSWIR6 – RSWIR7) has the potential of indicating the surface
dryness conditions and reducing the atmospheric effect to
some extent considering the similar response of the two bands
to atmospheric influences. The term RSWIR6 + RSWIR7 is used to
limit the index value within –1 and 1. The validation results
indicate that SWCI has strong correlation with soil water
content in Inner Mongolia and Liaoning Province of China
(Du et al. 2007). Compared with NDVI, SWCI also presents
higher correlation with the average soil moisture for the
0–50 cm layer in Henan Province of China (Zhang et al.
2008).
70 Int. J. Disaster Risk Sci. Vol. 4, No. 2, 2013
2.4
Shortwave Infrared Soil Moisture Index (SIMI)
The SIMI (Yao et al. 2011) is a soil moisture index developed
based on the SWIR spectral space using MODIS band 6
and band 7 data. It can also be applied to remote sensor
with two shortwave infrared bands centered on 1650 nm and
1950 nm.
SIMI =
(R
2
SWIR6
+ R SWIR7
2
)2
Eq. 4
The combined use of MODIS band 6 and band 7 in the
SWIR spectral space can highlight the moisture information
and reduce the disturbance from the complex land surface
components at the same time. This index has been proven
more efficient than the Temperature-Vegetation index (TVX),
which is the ratio between Land Surface Temperature
(LST) and NDVI, in soil moisture retrieval in Ningxia Hui
Autonomous Region of China (Yao et al. 2011).
2.5
Normalized Difference Vegetation Index (NDVI)
NDVI = (RNIR – RRED)/(RNIR + RRED)
Eq. 5
The NDVI is a well-known vegetation greenness index based
on the normalized difference between NIR and Red reflectance. It has been widely applied in drought monitoring under
the assumption that water stress is the most important factor
that interferes with the plant growing process (McVicar and
Bierwirth 2001; Ji and Peters 2003; Wan, Wang, and Li 2004;
Wang et al. 2007; Gu et al. 2007, 2008; Brown et al. 2008).
2.6 Visible and Shortwave infrared Drought Index
(VSDI)
The VSDI is a newly developed drought index for drought
monitoring of both soil and vegetation surfaces. This index is
a combination of MODIS Blue (band 3), Red (band 1), and
SWIR (band 6) bands, and is defined in equation 6:
VSDI = 1 – [(RSWIR6 – RBlue) + (RRed – RBlue)]
Eq. 6
By analyzing the spectral response to water stress of plants
and soils, SWIR and Red bands are found sensitive to moisture variation for both types of surfaces, thus they are used as
the moisture measuring bands. The Blue band is less sensitive
to water changes and can serve as the moisture benchmark
(Zhang et al. 2013). In this way, VSDI is constructed based
on the difference between moisture sensitive bands (SWIR
and Red) and reference band (Blue). The combination of
(RSWIR – RBlue) and (RRed – RBlue) may maximize the moisture
variation and give the potential to estimate surface water
independent of land cover types. Finally, “(RSWIR – RBlue) +
(RRed – RBlue)” is subtracted from 1 to make VSDI positively
correlated to moisture variation. The theoretical VSDI range
is defined in Table 1 with brief explanation.
The validation of VSDI was carried out in Oklahoma,
United States using the soil moisture measurements from 49
Mesonet stations.i The results show that VSDI presented high
correlation with soil moisture and was efficient for drought
monitoring over different land cover types and was applicable
throughout the plant growing seasons in Oklahoma (Zhang
et al. 2013).
3
Test Site and Data Processing
Our study area is located in China (Figure 1) and the evaluation of the performance of VSDI was conducted at two scales.
First, the soil moisture measurements from 585 meteorological and environmental observation stations across China
(indicated by black dots in Figure 1) were used as groundtruth data to compare VSDI with other drought indices. At
these stations soil moisture is measured for every 10 days
and recorded as relative soil moisture at three depths—10 cm,
20 cm, and 50 cm. The mean 10–20 cm soil moisture (later
referred to as 20_ave) and the mean 10–50 cm moisture (later
referred to as 50_ave) were also calculated respectively
by average soil moisture of the first two layers and all three
layers. The data for the 585 stations used in this research are
from March to October in 2011.
Second, VSDI is investigated at a smaller scale in the
Middle-Lower Yangtze Plain in southeast China (the enlarged
area in Figure 1). Five provinces are included in this region:
Hubei, Hunan, Auhui, Jiangxi, and Zhejiang Provinces.
In 2011, this region experienced frequent severe drought
and flood events; thus it represents an interesting case for
examining the performance of VSDI under extreme weather
conditions.
For the Yangtze River region, the monthly precipitation
anomaly data were collected from 84 local weather stations
(denoted by the plus signs in Figure 1). The station-based
precipitation anomalies are interpolated using Kriging
interpolation (Oliver 1990) embedded in ArcGIS software to
produce a continuous precipitation map. With reference to
the four precipitation anomaly-based drought categories
(indicated by asterisks in Table 2) defined in the Chinese
Classification of Meteorological Drought (GB/T 20481-2006)
(China Meteorological Administration 2006), five more categories were introduced and finally nine drought categories
were adopted to describe the moisture condition of this region
(Table 2).
The remote sensing data used in this study are the 8-day
MODIS reflectance products (MOD09A1). The spatial
resolution of MOD09A1 is 500 m for bands 1–7 covering the
Table 1. Definition of the theoretical range of the Visible and
Shortwave Infrared Drought Index (VSDI)
0 < VSDI ≤ 1
1 < VSDI ≤ 2
The smaller the value is, the drier the condition it
indicates (for the surface of farmland or any surface that
can be simply classified as soil, vegetation and the
combination of the two)
Water or Snow Water Equivalent (including water bodies,
snow, and ice cover)
Source: Zhang et al. (2013).
Zhang et al. Evaluation of the Visible and Shortwave Infrared Drought Index in China
Figure 1. Map of the study area in China. The black dots
(upper panel) represent the 585 weather stations across
China where the relative soil moisture data were obtained.
The plus signs (lower panel) denote the 84 weather stations
within the five provinces in the Middle-Lower Yangtze Plain
where precipitation anomalies were recorded
visible, near-infrared, and shortwave-infrared spectral
domains. The MODIS data were obtained from two sources.
In accordance with the station-based soil moisture measurements, the MODIS reflectance data that geographically
correspond to the 585 weather stations were downloaded
Table 2. Drought categories based on precipitation anomalies (Pa). The asterisks indicate the drought categories
defined in the Chinese Classification of Meteorological
Drought (GB/T 20481-2006) (China Meteorological Administration 2006)
Moisture Condition Pa Range
Normal
Abnormally dry*
Moderate drought*
Severe drought*
Extreme drought*
Moisture Condition Pa Range
−40 < Pa ≤ 40
Abnormally wet
−60 < Pa ≤ −40* Moderately wet
−80 < Pa ≤ −60* Severely wet
−95 < Pa ≤ −80* Extremely wet
Pa ≤ −95*
40 < Pa ≤ 60
60 < Pa ≤ 80
80 < Pa ≤ 95
95 < Pa
71
from the Earth Observation and Modeling website of the
University of Oklahoma (http://www.eomf.ou.edu/), which
provides the retrieved time series MODIS data for multiple
sites. Considering the different time intervals between
MODIS data (8-day period) and the soil moisture measurements (10-day period), the MODIS images that have the longest overlap with the 10-day ground-truth data were selected
for correlation analysis. Considering that cloud contamination is a major problem in optical remote sensing application,
the reflectance data before and after the quality control and
exclusion of cloud contaminated pixels are both used in this
study for computing and comparing the six drought indices
and to evaluate the cloud influences on these indices in large
scale.
For the investigation of the alternating drought and flood
events in the Middle-Lower Yangtze region, the MODIS
images covering the five provinces were downloaded from
NASA’s LAADS Web (Level 1 and Atmosphere Archive and
Distribution System, http://ladsweb.nascom.nasa.gov/data)
and mosaicked. A quality control process was applied to filter
the “cloud” pixels by using the MODIS quality assurance
(QA) data product.
Another set of data used in this study comes from a
web-based application developed by Goddard Earth Sciences
Data and Information Services Center (GES DISC), called
Giovanni. It provides a simple and intuitive way to visualize,
analyze, and access vast amounts of earth science remote
sensing data without having to download the data (http://disc.
sci.gsfc.nasa.gov/giovanni/overview/). In this study, the
monthly data set of average soil moisture at the 0–10 cm layer
simulated by the GLDAS-1 NOAH Model were selected from
the Global Land Data Assimilation System (GLDAS) data
portal in Giovanni. This data set covers the period from
January 1979 to May 2012 and has a spatial resolution of
1 degree. The modeled monthly soil moisture of the five provinces was produced by Giovanni and used for the following
analysis.
4
Results and Discussion
Using the above data sets, VSDI was validated by comparing
with the other five drought indices and drought maps. The
results are discussed in this section.
4.1
Comparison among Different Drought Indices
In this study, the ground-truth data (relative soil moisture)
are assumed to have a normal distribution. The Correlation
Coefficient (R), which is a measure of the strength of linear
dependence between two variables, is calculated between
the six drought indices (VSDI, LSWI, MSI, SWCI, SIMI,
and NDVI) and the relative soil moisture respectively. The
Correlation Coefficient R is computed at five depths (10 cm,
20 cm, 50 cm, 20_ave, and 50_ave) over 585 stations across
China, and two atmospheric conditions are also considered—
one represented by the original images and the other by
72 Int. J. Disaster Risk Sci. Vol. 4, No. 2, 2013
Table 3. The Correlation Coefficient (R) calculated between each of the six drought indices† and relative soil moisture at five
depths from 585 weather stations across China. A maximum number of 17,550 samples (30 periods x 585 stations) are included in the statistical analysis. Fisher (F) Test is conducted to test this linear regression and the Rs for all indices have
passed the significance test (p_value<0.05). The original condition stands for the results without quality control (cloud pixels
included in R computation) and the cloud removed condition denotes that the cloud pixels have been filtered before R calculation.
Condition Indices
Relative Soil Moisture
10 cm
R
20 cm
F_value p_value
R
50 cm
F_value p_value
R
20_ave
F_value p_value
R
50_ave
F_value p_value
R
F_value p_value
Original
VSDI
LSWI
MSI
SWCI
SIMI
NDVI
0.39 2142.8
0.23 643.6
−0.23 651.8
0.27 923.2
−0.24 747.5
0.15 262.3
0.00
0.00
0.00
0.00
0.00
0.00
0.37
0.19
−0.19
0.23
−0.23
0.12
1863.8
455.4
460.0
690.1
691.9
162.9
0.00
0.00
0.00
0.00
0.00
0.00
0.37
0.23
−0.24
0.26
−0.26
0.17
1794.2
641.3
685.1
840.9
859.0
352.2
0.00
0.00
0.00
0.00
0.00
0.00
0.39
0.21
−0.22
0.26
−0.24
0.13
2121.1
573.0
579.2
851.1
755.5
220.3
0.00
0.00
0.00
0.00
0.00
0.00
0.41
0.25
−0.26
0.29
−0.26
0.18
2358.1
768.9
826.9
1042.9
842.9
372.3
0.00
0.00
0.00
0.00
0.00
0.00
Cloud
removed
VSDI
LSWI
MSI
SWCI
SIMI
NDVI
0.35 1474.6
0.20 436.0
−0.20 443.2
0.26 783.9
−0.35 1473.7
0.22 548.5
0.00
0.00
0.00
0.00
0.00
0.00
0.34
0.17
−0.17
0.23
−0.33
0.18
1345.1
296.2
299.2
576.3
1319.3
361.6
0.00
0.00
0.00
0.00
0.00
0.00
0.35
0.21
−0.22
0.25
−0.36
0.23
1411.5
479.0
512.6
699.1
1516.3
553.2
0.00
0.00
0.00
0.00
0.00
0.00
0.35
0.19
−0.19
0.25
−0.35
0.21
1489.3
380.5
385.3
717.7
1475.3
474.7
0.00
0.00
0.00
0.00
0.00
0.00
0.38
0.23
−0.24
0.28
−0.38
0.24
1742.4
561.0
606.7
869.9
1704.6
657.9
0.00
0.00
0.00
0.00
0.00
0.00
Note: †The six drought indices are Visible and Shortwave Infrared Drought Index (VSDI), Land Surface Water Index (LSWI), Moisture Stress Index (MSI), Surface
Water Capacity Index (SWCI), Shortwave Infrared Soil Moisture Index (SIMI), and Normalized Difference Vegetation Index (NDVI).
Relative Soil Moisture
Relative Soil Moisture
images with “cloudy” pixels removed. The results are listed
in Table 3. Figure 2 shows the scatter plots and the linear
fitting between the six drought indices and the 10 cm relative
soil moisture. The cloud pixels have been eliminated from the
analysis in Figure 2.
From Table 3 we can see that among the six indices, MSI
and SIMI are negatively correlated to soil moisture variation,
while other indices have positive correlation with surface
dryness. The absolute value of correlation coefficients (|R|)
between the six drought indices and soil moisture at all depths
ranges from 0.13 to 0.41. It is worth explaining that although
all R values have passed the significance test, they are not
significantly high. The coefficient of determination values
(R2) of the six indices, which describes the proportion of variance in a data set that is accounted for by the statistical model
(Steel and Torrie 1960), are relatively low with the highest
1
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2 y=1.2x-0.2
R=0.35
0.2 y=0.3x+0.7
R=0.20
0.2 y=−0.2x+0.9
0
0
0.5
1
1.5
0
−1
0
1
0
1
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2 y=0.5x+0.6
0
−1
0
SWCI
y=−1.1x+1.0
0.2
R=0.26
1
0
R=−0.35
0
0.5
SIMI
R=−0.20
0
1
2
0
NDVI
1
0.2 y=0.2x+0.6
R=0.22
1
0
−1
Figure 2. Scatter plots between each drought index (VSDI, LSWI, MSI, SWCI, SIMI, and NDVI) and the 10 cm relative soil
moisture over 585 stations across China. The cloud contaminated pixels have been removed from this correlation analysis
Zhang et al. Evaluation of the Visible and Shortwave Infrared Drought Index in China
value of 0.17 found in VSDI. Several reasons may be ascribed
to this phenomenon, such as the different spatial resolution
between MODIS data (500 m) and the ground-truth data
(station-based point measurements), the mismatch of the
temporal resolution of the two data sets, and the positioning
and image registration errors, which may all add great
uncertainties to the final results and reduce the proportion of
variance that may be explained by the indices themselves.
However, given that all six drought indices are calculated and
processed in the same way and may be affected by the same
factors, and the five other indices are representative moisture
indices and have been proved efficient in moisture estimation
by a variety of studies, the correlation results can be considered reliable and the relative superior performance of VSDI
(with higher R values) compared with other indices can
justify the capability of VSDI for moisture monitoring.
For both conditions (original and cloud removed), VSDI
possesses the highest R values among the six indices, followed by the soil moisture drought indices (SWCI and SIM),
which are slightly higher than that of the vegetation drought
indices (LSWI and MSI). It is also worth noting that, after
removing the cloud contaminated pixels (Table 3 and
Figure 2), both SIMI and NDVI show enhanced correlation
with the observed relative soil moisture: the performance
of SIMI is almost as good as VSDI and NDVI has higher R
values than the vegetation drought indices (LSWI and MSI).
In contrast, other indices show slightly subdued correlation
with the observed relative soil moisture after excluding the
cloud contamination. The results suggest that VSDI is an
efficient and reliable drought index among the six indices
and the atmospheric conditions may have a stronger effect on
SIMI and NDVI compared with other indices. In this sense,
the cloud pixels should be removed or cloud-free conditions
should be selected when applying SIMI and NDVI for surface
moisture monitoring with high accuracy.
A further examination of Table 3 also reveals that for both
conditions (original and cloud removed), all indices seem to
have higher correlation with surface soil moisture (10 cm
layer) than soil moisture at deeper layers (20 cm and 50 cm).
This can be explained by the limited penetrability of optical
remote sensing signals. With the attenuation of optical signals, less information can be gathered from the deeper layers,
and therefore a decreased correlation with soil moisture at
these layers can be expected. Besides, stronger correlations
are also observed between averaged soil moisture and drought
indices, especially for the 10–50 cm layer. This may be due
to the averaging effect that reduces the measuring noises by
averaging soil moisture from different layers. Therefore,
for operational applications the optical drought indices are
recommended to estimate the surface dryness or the average
soil moisture content.
4.2
Comparison with Other Drought Maps
In this study, the VSDI color maps for five provinces (Hubei,
Hunan, Auhui, Jiangxi, and Zhejiang) in the Middle-Lower
Yangtze Plain (Figure 1) were also produced and compared
73
with two other drought products: the monthly drought classification maps based on precipitation anomalies; and the
monthly average soil moisture in the 0–10 cm layer simulated
by the GLDAS-1 NOAH Model from Giovanni (http://disc.
sci.gsfc.nasa.gov/giovanni/overview/index.html). The acquisition and processing of the two drought products were
introduced in Section 3. Finally, the three drought products
are plotted and compared in Figure 3. The first column shows
the precipitation-based drought maps with 9 moisture categories (as listed in Table 2). The middle column shows the
GLDAS simulated monthly surface soil moisture. The last
column shows the VSDI color maps calculated from the
8-day MODIS reflectance products. Considering the difference in temporal resolution between the first two products
(monthly) and VSDI color maps (8-day), the cloud-free VSDI
images covering the middle or the last period of each month
are selected for comparison.
From Figure 3, both agreement and some differences can
be observed among the three products. In April, all drought
maps indicate serious water-stressed conditions with warm
and red color in the northern and southern parts of the study
area, especially in Anhui and Jiangxi Provinces. This is
consistent with the drought events that started early in 2011
and were still persistent at this time in this area. In May, the
drought condition seems relieved to some extent in the south
of the study area (Hunan and Jiangxi Provinces) with light
and blue color presented in all three products. However, for
the northern part, both the soil moisture product and the VSDI
product indicate sustained water stress, while the precipitation-based drought map shows a normal to abnormally dry
condition. In June, the first two products show an apparent
moderate to severe wet condition (blue color) in the middle of
the study area along the Yangtze River. This is consistent with
the flood events that struck this area from early June. Since
the cloud pixels have been filtered from the VSDI maps, the
middle part of the VSDI color map on June 10 is almost blank,
which means that heavy cloud was persistent in this area. This
can be viewed as an indirect indication of the flood events
considering that heavy rains usually coincide with heavy
clouds. In July, drought conditions can be observed from all
three products in the western and northern parts of the study
area, but relatively wet conditions can be observed in the last
two products in the eastern and southern parts. In August, a
wet condition can be observed in the northeastern part of
the study area in the first two products and corresponds to
the blank area in the VSDI map after removing the cloud
contamination. In August the drought condition in the VSDI
map is more prominent in the southwestern part than in the
other two products.
In summary, there are satisfactory agreements among the
three products, especially between the GLDAS modeled
surface soil moisture and the VSDI maps. This is reasonable
because for these two products drought is measured by surface soil moisture, but for the precipitation maps drought is
measured by precipitation anomalies. The different temporal
resolution (monthly vs. 8-day) may also have contributed to
74 Int. J. Disaster Risk Sci. Vol. 4, No. 2, 2013
Percentage of Precipitation Anomalies
GLDAS_NOAH10_M.001 Average layer 1 soil moisture [kg/m^2]
Normal
Extreme Drought
Abnormally Wet
Severe Drought
Moderate Wet
Moderate Drought
Severe Wet
Abnormally Dry
Extreme Wet
Percentage of Precipitation Anomalies
GLDAS_NOAH10_M.001 Average layer 1 soil moisture [kg/m^2]
Normal
Extreme Drought
Abnormally Wet
Severe Drought
Moderate Wet
Moderate Drought
Severe Wet
Abnormally Dry
Extreme Wet
Percentage of Precipitation Anomalies
GLDAS_NOAH10_M.001 Average layer 1 soil moisture [kg/m^2]
Normal
Extreme Drought
Abnormally Wet
Severe Drought
Moderate Wet
Moderate Drought
Severe Wet
Abnormally Dry
Extreme Wet
Percentage of Precipitation Anomalies
GLDAS_NOAH10_M.001 Average layer 1 soil moisture [kg/m^2]
Normal
Extreme Drought
Abnormally Wet
Severe Drought
Moderate Wet
Moderate Drought
Severe Wet
Abnormally Dry
Extreme Wet
Percentage of Precipitation Anomalies
GLDAS_NOAH10_M.001 Average layer 1 soil moisture [kg/m^2]
Normal
Extreme Drought
Abnormally Wet
Severe Drought
Moderate Wet
Moderate Drought
Severe Wet
Abnormally Dry
Extreme Wet
Figure 3. Comparison between different drought maps of the five provinces located in the Middle-Lower Yangtze Plain from
April to August, 2011. The first column shows the drought maps produced by interpolating the monthly precipitation anomalies from the 84 weather stations. The middle column shows the GLDAS modeled monthly surface soil moisture (10 cm) from
a web-based visualization application, Giovanni (http://disc.sci.gsfc.nasa.gov/giovanni/overview/index.html). The last column
shows the VSDI color maps calculated from the 8-day MODIS reflectance products
Zhang et al. Evaluation of the Visible and Shortwave Infrared Drought Index in China
75
the differences among the three products. In general, VSDI
can be viewed as an effective tool for monitoring surface
moisture conditions. Compared with the precipitation-based
drought maps interpolated from the 84 weather stations and
the 1-degree GLDAS simulated soil moisture maps, the VSDI
maps have a spatial resolution of 500 m, which may facilitate
a finer-scale interpretation of surface moisture distribution.
Although not elaborated in this study, the relatively high
temporal resolution of VSDI maps (8-day) may also enable a
closer and timely monitoring of drought development.
thank Dr. Zhenghong Tang, Dr. Jianjun Wu, and the anonymous reviewers for their critical and helpful comments and
suggestions. The authors appreciate the kind financial support
of the National Natural Science Foundation of China
(41230747, 41071221, 41201331) and the National Key
Technology R&D Program in the 12th Five-Year Plan of
China (2012BAH29B03). The first author also likes to thank
the HyDROS Lab (http://hydro.ou.edu) at the National
Weather Center, Norman, Oklahoma, United States, for the
helpful and valuable suggestions for this article.
5
Note
Conclusion
In this article, VSDI, a drought index recently developed, is
further explored and validated in China. We come to the
following conclusions:
The six optical drought indices applied in this study all
have the ability of monitoring surface dryness and higher
correlation is observed both with the top 10 cm layer and the
10–50 cm averaged soil moisture than with the single deeper
layer (that is, 20 cm and 50 cm) and the 10–20 cm averaged
moisture. Therefore, it is suggested that the optical drought
indices can be used to better monitor the surface soil moisture
(10 cm) or multilayer averaged soil moisture than in other
cases.
Among the six drought indices, VSDI shows the highest
correlation with soil moisture at various depths using both the
original and cloud removed data. SIMI has similar performance as VSDI after removing the cloud pixels, and both
SIMI and NDVI show enhanced correlation, while the
correlations of other indices are slightly decreased after
removing the cloud contamination. This suggests that VSDI
is robust and reliable in the estimation of surface dryness
among the six indices. SIMI and NDVI are more sensitive to
atmospheric influences compared with other indices, therefore either images under clear weather conditions should
be selected or atmospheric correction should be conducted
before applying these two indices for drought monitoring.
Although differences exist among the three drought products for the Yangtze River region due to the different spatial
and temporal resolutions, VSDI maps show a satisfactory
agreement with the other two moisture products, even in the
case of alternating drought and flood events in the study area.
The MODIS-based VSDI maps have higher temporal and
spatial resolutions than the other two products, and therefore
may serve as an effective tool in real-time regional surface
drought monitoring at field scale.
Acknowledgments
The authors would like to thank the National Satellite Meteorological Center of China Meteorological Administration
for providing the ground measurement data. The authors also
i The Mesonet stations are an extensive environmental observation
network located in Oklahoma, United States, which can provide
quality-controlled measurements of meteorological and land surface
parameters such as precipitation, temperature, and soil moisture
every five minutes. More detailed information can be found at http://
www.mesonet.org/.
References
Brown, J. F., B. D. Wardlow, T. Tadesse, M. J. Hayes, and B. C. Reed.
2008. The Vegetation Drought Response Index (VegDRI): A New
Integrated Approach for Monitoring Drought Stress in Vegetation.
GIScience and Remote Sensing 45 (1): 16–46.
Bryant, E. A. 1991. Natural Hazards. Cambridge, UK: Cambridge
University Press.
Ceccato, P., S. Flasse, S. Tarantola, S. Jacquemoud, and J. Gregoire.
2001. Detecting Vegetation Leaf Water Content Using Reflectance in
the Optical Domain. Remote Sensing of Environment 77 (1): 22–33.
Chen, D., J. Huang, and T. J. Jackson. 2005. Vegetation Water Content
Estimation for Corn and Soybeans Using Spectral Indices Derived
from MODIS Near- and Short-Wave Infrared Bands. Remote Sensing
of Environment 98 (2–3): 225–236.
China Meteorological Administration. 2006. Chinese Classification of
Meteorological Drought. GB/T 20481-2006. http://www.tsinfo.js.cn/
inquiry/gbtdetails.aspx?A100=GB/T%2020481-2006.
Chuvieco, E., D. Riãno, I. Aguado, and D. Cocero. 2002. Estimation
of Fuel Moisture Content from Multitemporal Analysis of Landsat
Thematic Mapper Reflectance Data: Applications in Fire Danger
Assessment. International Journal of Remote Sensing 23 (11): 2145–
2162.
Dawson, T. P., P. J. Curran, P. R. J. North, and S. E. Plummer. 1999. The
Propagation of Foliar Biochemical Absorption Features in Forest
Canopy Reflectance: A Theoretical Analysis. Remote Sensing of
Environment 67 (2): 147–159.
Deering, D. W. 1978. Rangeland Reflectance Characteristics Measured
by Aircraft and Spacecraft Sensors. Ph.D. Dissertation, Texas A & M
University, College Station, TX.
Du, X., S. Wang, Y. Zhou, and H. Wei. 2007. Construction and Validation of a New Model for Unified Surface Water Capacity Based
on MODIS Data. Geomatics and Information Science of Wuhan
University 32 (3): 205–207 (in Chinese).
Fensholt, R., and I. Sandholt. 2003. Derivation of a Shortwave Infrared
Stress Index from MODIS Near- and Shortwave Infrared Data in
a Semiarid Environment. Remote Sensing of Environment 87 (1):
111−121.
Gao, B. C. 1996. NDWI—A Normalized Difference Water Index for
Remote Sensing of Vegetation Liquid Water from Space. Remote
Sensing of Environment 58 (3): 257−266.
76 Int. J. Disaster Risk Sci. Vol. 4, No. 2, 2013
Ghulam, A., Z. L. Li, Q. M. Qin, H. Yimit, and J. H. Wang. 2008.
Estimating Crop Water Stress with ETM+NIR and SWIR Data.
Agricultural and Forest Meteorology 148 (11): 1679–1695.
Gu, Y., J. F. Brown, J. P. Verdin, and B. Wardlow. 2007. A Five-Year
Analysis of MODIS NDVI and NDWI for Grassland Drought Assessment Over the Central Great Plains of the United States. Geophysical
Research Letters 34 (6): L06407.
Gu, Y., E. Hunt, B. Wardlow, J. B. Basara, J. F. Brown, and J. P. Verdin.
2008. Evaluation of MODIS NDVI and NDWI for Vegetation Drought
Monitoring Using Oklahoma Mesonet Soil Moisture Data. Geophysical Research Letters 35 (5): L22401.
Hardinsky, M. A., V. Lemas, and R. M. Smart. 1983. The Influence
of Soil Salinity, Growth Form, and Leaf Moisture on the Spectral
Reflectance of Spartina Alternifolia Canopies. Photogrammetric
Engineering and Remote Sensing 49 (1): 77–83.
Hunt, Jr. E. R., and B. N. Rock. 1989. Detection of Changes in Leaf
Water Content Using Near- and Middle-Infrared Reflectance. Remote
Sensing of Environment 30 (1): 43– 54.
Jackson, T. J., D. Chen, M. Cosh, F. Li, M. Anderson, C. Walthall, P.
Doriaswamy, and E. R. Hunt. 2004. Vegetation Water Content
Mapping Using Landsat Data Derived Normalized Difference Water
Index for Corn and Soybeans. Remote Sensing of Environment 92 (4):
475−482.
Ji, L., and A. J. Peters. 2003. Assessing Vegetation Response to Drought
in the Northern Great Plains Using Vegetation and Drought Indices.
Remote Sensing of Environment 87 (1): 85–98.
Jimmy, O. A., and M. C. Andrew. 2002. Relations between Soil Moisture
and Satellite Vegetation Indices in the U.S. Corn Belt. Journal of
Hydrometeorology 3 (4): 395–405.
Maki, M., M. Ishiahra, and M. Tamura. 2004. Estimation of Leaf Water
Status to Monitor the Risk of Forest Fires by Using Remotely Sensed
Imagery. Remote Sensing of Environment 90 (4): 441−450.
McVicar, T. R., and P. N. Bierwirth. 2001. Rapidly Assessing the 1997
Drought in Papua New Guinea Using Composite AVHRR Imagery.
International Journal of Remote Sensing 22 (11): 2109–2128.
Mishra, A. K., and V. P. Singh. 2010. A Review of Drought Concepts.
Journal of Hydrology 391 (1–2): 202–216.
Oliver, M. A. 1990. Kriging: A Method of Interpolation for Geographical
Information Systems. International Journal of Geographic Information Systems 4 (3): 313–332.
Riebsame, W. E., S. A. Changnon, and T. R. Karl. 1991. Drought and
Natural Resource Management in the United States: Impacts and
Implications of the 1987–1989 Drought. Boulder, CO: Westview
Press.
Steel, R. G. D., and J. H. Torrie. 1960. Principles and Procedures of
Statistics. New York: McGraw-Hill.
Tannehill, I. R. 1947. Drought: Its Causes and Effects. Princeton, NJ:
Princeton University Press.
Wan, Z., P. Wang, and X. Li. 2004. Using MODIS Land Surface
Temperature and Normalized Difference Vegetation Index for Monitoring Drought in the Southern Great Plains, USA. International
Journal of Remote Sensing 25 (1): 61–72.
Wang, L., and J. J. Qu. 2009. Satellite Remote Sensing Applications
for Surface Soil Moisture Monitoring: A Review. Frontiers of Earth
Science in China 3 (2): 237–247.
Wang, X., H. Xie, H. Guan, and X. Zhou. 2007. Different Responses of
MODIS-Derived NDVI to Root-Zone Soil Moisture in Semi-Arid
and Humid Regions. Journal of Hydrology 340 (1 –2): 12–24.
Wilhite, D. A. 2000. Drought as a Natural Hazard: Concepts and
Definitions. In Drought: A Global Assessment, Vol. I, edited by D. A.
Wilhite, 3–18. New York: Routledge.
Xiao, X., S. Boles, J. Liu, D. Zhuang, S. Frolking, C. Li, W. Salas, and
B. Moore III. 2005. Mapping Paddy Rice Agriculture in Southern
China Using Multi-Temporal MODIS Images. Remote Sensing of
Environment 95 (4): 480–492.
Xiao, X., Q. Zhang, B. Braswell, S. Urbanski, S. Boles, S. Wofsy, B.
Moore, III, and D. Ojima. 2004. Modeling Gross Primary Production
of Temperate Deciduous Broadleaf Forest Using Satellite Images and
Climate Data. Remote Sensing of Environment 91 (2): 256–270.
Yao, Y., Q. Qin, S. Zhao, and W. Yuan. 2011. Retrieval of Soil Moisture
Based on MODIS Shortwave Infrared Spectral Feature. Journal of
Infrared and Millimeter Waves 30 (1): 61–67.
Yilmaz, M. T., E. R. Hunt Jr., and T. J. Jackson. 2008. Remote Sensing
of Vegetation Water Content from Equivalent Water Thickness Using
Satellite Imagery. Remote Sensing of Environment 112 (5): 2514–
2522.
Zarco-Tejada, P. J., C. A. Rueda, and S. L. Ustin. 2003. Water Content
Estimation in Vegetation with MODIS Reflectance Data and Model
Inversion Methods. Remote Sensing of Environment 85 (1): 109–
124.
Zhang, H., H. Chen, S. Shen, G. Zhou, and W. Yu. 2008. Drought
Remote Sensing Monitoring Based on the Surface Water Content
Index (SWCI) Method. Remote Sensing Technology and Application
23 (6): 624–628 (in Chinese).
Zhang, N., Y. Hong, Q. Qin, and L. Liu. 2013. VSDI: A Visible and
Shortwave Infrared Drought Index for Monitoring Soil and
Vegetation Moisture Based on Optical Remote Sensing. International
Journal of Remote sensing 34 (13): 4585–4609.
Zhang, J., Y. Xu, F. Yao, P. Wang, W. Guo, L. Li and L. Yang. 2010.
Advances in Estimation Methods of Vegetation Water Content
Based on Optical Remote Sensing Techniques. Science ChinaTechnological Sciences 53 (5): 1159−1167.
Zhao, S., Q. Qin, L. You, Y. Yao, N. Yang, and J. Li. 2009. Application
of Two Shortwave Infrared Water Stress Indices to Drought Monitoring Over Northwestern China. Geoscience and Remote Sensing
Symposium (IGARSS 2009), III-530 – III-533.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and
reproduction in any medium, provided the original author(s) and source are credited.