This article was downloaded by: [University of California, San Francisco]

On: 22 November 2014, At: 00:09

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Geocarto International
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tgei20

Vegetation Indices: Advances Made in Biomass

Estimation and Vegetation Monitoring in the Last 30
Years
a b c
Nikolaos G. Silleos , Thomas K. Alexandridis , Ioannis Z. Gitas & Konstantinos Perakis
d

a
Lab of Remote Sensing and GIS, Faculty of Agronomy , Aristotle University of
Thessaloniki , G.R. 54124, Thessaloniki, Greece E-mail:
b
Lab of Remote Sensing and GIS, Faculty of Agronomy , Aristotle University of
Thessaloniki , G.R. 54124, Thessaloniki, Greece E-mail:
c
Lab of Forest Management and Remote Sensing, Faculty of Forestry and Natural
Environment , Aristotle University of Thessaloniki , G.R. 54124, Thessaloniki, Greece E-
mail:
d
University of Thessaly, School of Engineering, Department of Planning and Regional
Development , G.R. 38334, Pedion Areos, Volos, Greece E-mail:
Published online: 04 Jan 2008.

To cite this article: Nikolaos G. Silleos , Thomas K. Alexandridis , Ioannis Z. Gitas & Konstantinos Perakis (2006)
Vegetation Indices: Advances Made in Biomass Estimation and Vegetation Monitoring in the Last 30 Years, Geocarto
International, 21:4, 21-28, DOI: 10.1080/10106040608542399

To link to this article: http://dx.doi.org/10.1080/10106040608542399

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of
the Content. Any opinions and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of information. Taylor and Francis shall
not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other
liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Vegetation Indices: Advances Made in Biomass Estimation and
Vegetation Monitoring in the Last 30 Years

Nikolaos G. Silleos
Lab of Remote Sensing and GIS
Faculty of Agronomy, Aristotle University of Thessaloniki
G.R. 54124, Thessaloniki, Greece
E-mail: silleos@agro.auth.gr

Thomas K. Alexandridis
Lab of Remote Sensing and GIS
Faculty of Agronomy, Aristotle University of Thessaloniki
Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

G.R. 54124, Thessaloniki, Greece

E-mail: thalex@agro.auth.gr

Ioannis Z. Gitas
Lab of Forest Management and Remote Sensing
Faculty of Forestry and Natural Environment, Aristotle University of Thessaloniki
G.R. 54124, Thessaloniki, Greece
E-mail: igitas@for.auth.gr

Konstantinos Perakis
University of Thessaly
School of Engineering, Department of Planning and Regional Development
G.R. 38334, Pedion Areos, Volos, Greece
e-mail: perakis@prd.uth.gr

Abstract

During the last 30 years Vegetation Indices (VI) have been extensively used for tracing and monitoring vegetation
conditions, such as health, growth levels, production, water and nutrients stress, etc. In this paper the characteristics of
over 20 VIs based on the VNIR spectrum are described in order to provide the reader with adequate material to form a
picture of their nature and purpose. It is not, though, a review article due to the fact that a huge volume of work exists all
over the world and a simple lining up of the related papers would not contribute to an understanding of the usefulness of
VIs. A limited number of review work is included, together with research results from various operational and research
applications of VI for wheat damage assessment in Northern Greece.

Introduction strongly reflected through the internal structure of the leaves.

It is this strong contrast, particularly between the reflected
One of the major applications of remote sensing in energy in the red and near-infrared regions of the
environmental resources management and decision making is electromagnetic spectrum that has been the focus of a large
the detection and quantitative assessment of green vegetation. variety of attempts to develop quantitative indices of vegetation
In this sense, vegetation analyses and detection of changes in condition using remotely sensed imagery.
vegetation patterns and structure are keys to natural resources The proposed vegetation indices (VI) are applicable to both
assessment and monitoring. low and high spatial resolution multispectral satellite sensors,
Healthy canopies of green vegetation have a very distinct such as NOAA AVHRR, Terra MODIS, Landsat TM and
interaction with certain portions of the electromagnetic MSS, SPOT HRV/XS, and all the others that acquire data in
spectrum. In the visible regions, chlorophyll causes strong the visible and near-infrared regions. They have been used in a
absorption of energy, primarily for use in photosynthesis. This variety of contexts to assess green biomass and have also been
absorption peaks in the red and blue areas of the visible used as a proxy to overall environmental change, especially in
spectrum, while the green area is reflected by chlorophyll, thus the context of drought and land degradation risk assessment
leading to the characteristic green appearance of most leaves. (Kogan, 1990; Tripathy et al., 1996; Liu and Kogan, 1996).
At the same time, the near-infrared region of the spectrum is Consequently, special interest has been focused on the

Geocarto International, Vol. 21, No. 4, December 2006 E-mail: geocarto@geocarto.com 21

Published by Geocarto International Centre, G.P.O. Box 4122, Hong Kong. Website: http://www.geocarto.com
 assessment of green biomass in arid environments where soil
background becomes a significant component of the signal
detected.
A wide review of VIs can be found in several textbooks
(Bastiaanssen, 1998; Jensen, 2000; Jensen, 2005). The aim of
this paper, however, is not to evaluate the extensive amount of
work carried out on VIs, but rather to provide the reader with a
clear understanding of their nature and usefulness. With this in
mind, a description is provided of the characteristics of over 20
VIs used on the VNIR spectrum. A limited amount of review
work, together with the research results of various operational
and research applications using VIs have been included in this
paper.

Classification of Vegetation Indices

Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

Jackson and Huete (1991) classify VIs into two groups: (1)
slope-based and (2) distance-based VIs. To appreciate this
distinction, it is necessary to consider the position of vegetation
pixels in a two-dimensional graph (or bi-spectral plot) of red
versus infrared reflectance (Figure 1). The slope-based VIs are
simple arithmetic combinations that focus on the contrast
between the spectral response patterns of vegetation in the red
and near-infrared portions of the electromagnetic spectrum. In
contrast to the slope-based group, the distance-based group Figure 1 Bi-spectral plot in the near-infrared and red domain of all pixels
measures the degree of vegetation present by gauging the in a scene. All pixels are found in the grey shaded area. Bare
difference of any pixel’s reflectance from the reflectance of soil fields are located along the soil line. The greater the
biomass and/or canopy cover, the greater the near-infrared
bare soil. A key concept here is that a plot of the positions of
bare soil pixels of varying moisture levels in the bi-spectral plot
will tend to form a line (known as a soil line). As vegetation index is constructed as a ratio, problems of variable
canopy cover increases, this soil background will become illumination as a result of topography are minimized. However,
progressively obscured, with vegetated pixels showing a the index is susceptible to division by zero errors and the
tendency towards increasing perpendicular distance from this resulting measurement scale is not linear. A study regarding
soil line. All of the members of this group (such as the its efficiency has been published by Vaiopoulos et al. (2004).
Perpendicular Vegetation Index-PVI) thus require that the slope The Normalized Difference Vegetation Index (NDVI)
and intercept of the soil line be defined for the image under was introduced by Rouse et al. (1974) in order to produce a
consideration. spectral VI that separates green vegetation from its background
To these two groups of vegetation indices, a third group can soil brightness using Landsat MSS digital data. It is expressed
be added, namely orthogonal transformation VIs. Orthogonal as the difference between the near infrared and red bands
indices undertake a transformation of the available spectral normalized by the sum of those bands. It is the most commonly
bands to form a new set of uncorrelated bands within which a used VI as it retains the ability to minimize topographic
green vegetation index band can be defined. The Tasseled Cap effects while producing a linear measurement scale. In
transformation is perhaps the most well-known of this group. addition, division by zero errors are significantly reduced.
Furthermore, the measurement scale has the desirable property
The Slope-Based VIs of ranging from -1 to 1, with 0 representing the approximate
value of no vegetation, and negative values non-vegetated
Slope-based VIs are combinations of the visible red and the surfaces.
near infrared bands and are widely used to generate vegetation The Soil-Adjusted Vegetation Index (SAVI) was
indices. Their values indicate both the state and abundance of proposed by Huete (1988). It is intended to minimize the
green vegetation cover and biomass. The slope-based VIs include effects of soil background on the vegetation signal by
the RATIO, NDVI, SAVI, RVI, NRVI, TVI, CTVI, TTVI, and incorporating a constant soil adjustment factor L into the
EVI (Table 1). denominator of the NDVI equation. L varies with the
The Ratio Vegetation Index (RATIO) was originally reflectance characteristics of the soil (e.g., colour and
described by Birth and McVey (1968). It is calculated by brightness). The author provides a graph from which the
simply dividing the reflectance values of the near infrared band values of L can be extracted. The L factor chosen depends on
by those of the red band. The result clearly captures the contrast the density of the vegetation one wishes to analyse. In cases
between the red and infrared bands for vegetated pixels, with of very low vegetation, the use of an L factor of 1.0 is
high index values being produced by combinations of low red suggested, for intermediate 0.5, and for high densities 0.25.
(because of absorption by chlorophyll) and high infrared (as a Eastman (2003) suggests that the best L value to select is
result of leaf structure) reflectance. In addition, because the where the difference between SAVI values for dark and light
22
 Table 1 Slope-based Vegetation Indices

SLOPE BASED VEGETATION INDICES Explanation of symbols Author

RATIO =— NIR
NIR = near infrared, Birth and McVey (1968)
R
R = red,
NIR-R
NDVI = –––– B = blue, Rouse et al. (1974)
NIR + R L = Soil
SAVI = ––NIR-R
––– (1+L) adjustment factor, C1 and C2 are
Huete (1988)
(NIR + R) constants, G is a gain factor
TVI = –(NIR-R)
––– + 0.5 Deering et al. (1975)
NIR + R
NDVI + 0.5
CTVI = ––––––––– × ABS (NDVI + 0.5) Perry and Lautenschlager (1984)
ABS (NDVI + 0.5)
TTVI = ABS (NDVI + 0.5) Thiam (1997)
Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

RVI =— R
Richardson and Wiegand (1977)
NIR
NRVI = –RVI
––-1
Baret and Guyot (1991)
RVI +1
EVI = G –––– NIR
–– - R––– (1+L)
Huete et al. (1999)
NIR + C1R - C2B+L

soil is minimal. For L = 0, SAVI equals NDVI. For L = 1, computationally being more simple. RVI is clearly the inverse
SAVI approximates PVI. of the standard Ratio Vegetation Index (RATIO).
The Transformed Vegetation Index (TVI) (Deering et The Normalized Ratio Vegetation Index (NRVI) is a
al., 1975) modifies NDVI by adding a constant of 0.5 to all its modification of the RVI by Baret and Guyot (1991) whereby
values and taking the square root of the results. The constant 0. the result of [RVI - 1] is normalized over [RVI + 1]. This
5 is introduced in order to avoid operating with negative NDVI normalization is similar in effect to that of the NDVI, i.e., it
values. The calculation of the square root is intended to correct reduces topographic, illumination and atmospheric effects and
NDVI values that approximate the Poisson distribution and creates a statistically desirable normal distribution.
introduce a normal distribution. Moreover, there is no technical The Enhanced Vegetation Index (EVI) was developed
difference between NDVI and TVI in terms of image output or by the MODIS Land Discipline Group for use with MODIS
active vegetation detection. data. It is a modified NDVI with a soil adjustment factor L and
The Corrected Transformed Vegetation Index (CTVI) two coefficients C1 and C2, which describe the use of the blue
was proposed by Perry and Lautenschlager (1984) to correct band in correction of the red band for atmospheric aerosol
the TVI. Clearly adding a constant of 0.5 to all NDVI values scattering. This VI has improved sensitivity to high biomass
does not always eliminate the negative values in the square regions and reduced atmospheric influence (Huete et al., 1999).
root, as NDVI can be as low as -1. Thus, the CTVI is intended
to resolve this situation by dividing the term [NDVI + 0.5] by The Distance-Based VIs
its absolute value [ABS(NDVI + 0.5)] and multiplying the
result by the square root of the absolute value [SQRT(ABS This group of vegetation indices is derived from the
(NDVI + 0.5))]. This suppresses the negative NDVI. Given Perpendicular Vegetation Index (PVI) discussed in detail below.
that the correction is applied in a uniform manner, the output The main objective of these VIs is to cancel the effect of soil
image using CTVI should have no difference with the initial brightness in cases where vegetation is sparse and pixels
NDVI image or the TVI whenever TVI properly carries out the contain a mixture of green vegetation and soil background.
square root operation. The correction is intended to eliminate This is particularly important in arid and semi-arid
negative values and generate a VI image that is similar to, if environments.
not better than, the NDVI. The procedure is based on the soil line concept as outlined
However, Thiam (1997) indicates that the resulting image earlier. The soil line represents a description of the typical
of the CTVI can be very “noisy” due to an overestimation of signatures of soils in a red/near-infrared bi-spectral plot (Figure
the greenness. He suggests ignoring the first term of the CTVI 1). It is obtained through linear regression of the near-infrared
equation in order to obtain better results. This is done by band against the red band for a sample of bare soil pixels.
simply taking the square root of the absolute values of the Pixels falling near the soil line are assumed to be soil, while
NDVI in the original TVI expression to have a new VI called those far away are assumed to be vegetation. Distance-based
Thiam’s Transformed Vegetation Index (TTVI). VIs (Table 2) using the soil line require the slope (b) and
The simple Ratio Vegetation Index (RVI) was suggested intercept (a) of the line as inputs to the calculation.
by Richardson and Wiegand (1977) as graphically having the Unfortunately, there has been a remarkable inconsistency
same strengths and weaknesses as the TVI (see above), while in the logic with which this soil line has been developed for

23
 Table 2 Distance-based Vegetation Indices

SLOPE BASED VEGETATION INDICES Explanation of symbols and author

2 3 Where Rp is the reflectance at a vegetation spot for

PVI = (Rgg5 - Rp5) + (Rgg7 - Rp7)
Landsat bands MSS5 and MSS7, and Rgg is the
reflectance of soil background (Richardson and
Wiegand, 1977)
NIR = reflectance in the near infrared band
(bNIR-R) +–
a R = reflectance in the visible red band
PVI1 = –– – –– – a = intercept of the soil line
b2 + 1
b = slope of the soil line (Perry and Lautenschlager,
1984)
(NIR-a)
PVI2 = ––– – – *–
(R
– +–
–b) Similar to PVI (Bannari et al., 1996)
2
1+a
Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

pNIR = reflectance in the near infrared band

pR = reflectance in the visible red band
PVI3 = apNIR - bpR a = intercept of the soil line
b = slope of the soil line (Qi et al., 1994)
g = the slope of the soil line
MSS7 = reflectance in the near infrared 2 band
DVI = gMSS7 - MSS5
MSS5 = reflectance in the visible red band
(Richardson and Wiegand, 1977)
AVI = 2.0 MSS7 - MSS5 Similar to DVI (Ashburn, 1978)
NIR = reflectance in the near infrared band
(expressed as reflectances)
TSAVI1 = –a(NIR
–– – –-–
a)–(R
– –-–
b) R = reflectance in the visible red band (expressed
R + a* NIR - a*b as reflectances)
a = slope of the soil line
b = intercept of the soil line (Baret et al., 1989)

TSAVI2 = –– – –a(NIR
– – ––-–aR–-–
b)
– –– – Similar to TSAVI1 (Baret and Guyot, 1991)
R + aNIR - ab + 0.08 (1+ a2)
NIR = reflectances in the near infrared band
MSAVI1 = ––NIR
––– -–
R– (1 + L)
R = reflectances in the visible red band
NIR + R + L
L = 1 -2 NDVI * WDVI (Qi et al., 1994)
pNIR = reflectance of the near infrared band
2pNIR + 1- (2pNIR + 1)2 - 8(pNIR - pR) pR = reflectance of the red band
MSAVI2 = ––– ––– –– –
– –
–– –– –
– – – –– –– –
– – ––
2 (Qi et al., 1994)
pNIR = reflectance of near infrared band
pR = reflectance of visible red band
WDVI = pNIR -γpR
γ = slope of the soil line (Richardson and Wiegand,
1977; Clever, 1988)

specific VIs. One group requires the red band be the independent who argued that the original PVI equation is computationally
variable and the other requires the near-infrared band be the intensive and does not discriminate between pixels that fall to
independent variable for the regression. the right or left side of the soil line (i.e., water from vegetation).
The Perpendicular Vegetation Index (PVI) suggested by Given the spectral response pattern of vegetation in which the
Richardson and Wiegand (1977) is the parent index from infrared reflectance is higher than the red reflectance, all
which this entire group is derived. The PVI uses the vegetation pixels will fall to the right of the soil line.
perpendicular distance from each pixel coordinate to the soil PVI2 Bannari et al., (1996) weights the red band with the
line (Figure 1). Attempts to improve the performance of the intercept of the soil line, similar to PVI3, presented by Qi et al.
PVI have yielded three other indices suggested by Perry and (1994).
Lautenschlager (1984), Bannari et al., (1996), and Qi et al. The Difference Vegetation Index (DVI) is also suggested
(1994). by Richardson and Wiegand (1977) as an easier vegetation
PVI1 was developed by Perry and Lautenschlager (1984) index calculation algorithm. The particularity of the DVI is

24
 that it weights the near-infrared band by the slope of the soil transformation of n-dimensional image data that produces a
line. Similar to the PVI1, zero values of DVI indicate bare soil, new set of images (components) that are uncorrelated with one
negative values indicate water, and positive values indicate another and ordered with respect to the amount of variation
vegetation. (information) they represent from the original image set.
The Ashburn Vegetation Index (AVI) (Ashburn, 1978) is PCA is typically used to uncover the underlying
presented as a measure of green growing vegetation. The dimensionality of multivariate data by removing redundancy
values in MSS7 are multiplied by 2 in order to scale the 6-bit (evident in inter-correlation of image pixel values), with specific
data values of this channel to match with the 8-bit values of applications in GIS and image processing ranging from data
MSS5. This scaling factor would not apply if both bands were compression to time series analysis. In the context of remotely
either 7-bit or 8-bit; in this case, the equation is rewritten as a sensed images, the first component typically represents albedo
simple subtraction. (in which the soil background is represented), while the second
The Transformed Soil-Adjusted Vegetation Index component most often represents variation in vegetative cover.
(TSAVI-1) was defined by Baret et al. (1989) who argued that For example, the second component generally has positive
the SAVI concept is exact only if the constants of the soil line loadings on the near-infrared bands and negative loadings on
are a=1 and b=0. Because this is not generally the case, the the visible bands. As a result, the green vegetation pattern is
authors transformed SAVI. By taking into consideration the highlighted in this component (Singh and Harrison, 1985;
Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

PVI concept, a first modification of TSAVI, designated as Fung and LeDrew, 1988; Thiam, 1997).
TSAVI-1, was proposed. With some resistance to high soil The Green Vegetation Index (GVI) of the Tasseled Cap
moisture, TSAVI-1 was specifically designed for semi-arid is the second of the four new bands that Kauth and Thomas
regions and does not perform well in areas with full vegetation (1976) extracted from raw MSS images. The GVI provides
cover. global coefficients that are used to weight the original MSS
TSAVI was readjusted a second time by Baret and Guyot digital counts to generate the new transformed bands. The
(1991) with an additional correction factor of 0.08 to minimize expression of the green vegetation index band, GVI, is written
the effects of the background soil brightness. The new version as follows for MSS or TM data:
was named TSAVI-2.
The Modified Soil-Adjusted Vegetation Indices (MSAVI- GVI = [(-0.386 MSS4) + (-0.562 MSS5) + (0.600 MSS6) + (0.
1 and MSAVI-2) suggested by Qi et al. (1994) are based on a 491 MSS7)]
modification of the L factor of the SAVI. Both are intended to GVI = [(-0.2848 TM1) + (-0.2435 TM2) + (-0.5436 TM3) +
better correct the soil background brightness in different (0.7243 TM4) + (0.0840 TM5) + (-0.1800 TM7)]
vegetation cover conditions. With MSAVI-1, L is selected as
an empirical function due to the fact that L decreases with a The negative weights of the GVI on the visible bands tend
decrease of vegetation cover, as is the case in semi-arid lands to minimize the effects of the background soil, while its
(Qi, et al., 1994). In order to cancel or minimize the effect of positive weights on the near infrared bands emphasize the
the soil brightness, L is set to be the product of NDVI and green vegetation signal.
WDVI (described below). Therefore, it displays the opposite
trends of NDVI and WDVI. Applications
The second modified SAVI, MSAVI-2, uses an inductive L Selected publications, which are evidence of the extensive
factor to: (i) remove the soil “noise” that was not cancelled out use of VIs in the last decades in various applications employing
by the product of NDVI by WDVI, and (ii) correct values a wide range of sensors and VIs, are explored in this section.
greater than 1 that MSAVI-1 may have due to the low negative Payero et al. (2004) compared 11 vegetation indices in
value of [NDVI*WDVI]. Thus, its use is limited for high order to estimate plant height and develop its quantitative
vegetation density areas. relationship with VIs. Among the VIs used were NDVI, IPVI
The Weighted Difference Vegetation Index (WDVI) has (Infrared Percentage Vegetation Index), TVI and RATIO.
been attributed to Richardson and Wiegand (1977) and Clevers Steven et al. (2003) used NDVI and SAVI from a range of
(1988). Although simple, WDVI is as efficient as most of the earth observation satellites currently in operation, such as
slope-based VIs. The effect of weighting the red band with the AVHRR, ATSR-2, Landsat MSS, TM and ETM+, SPOT-2
slope of the soil line is maximization of the vegetation signal and SPOT-4 HRV, IRS, IKONOS, SeaWiFS, MISR, MODIS,
in the near-infrared band and minimization of the effect of soil POLDER, QuickBird, and MERIS. Spectroradiometric
brightness. measurements were made over a range of crop canopy densities,
soil backgrounds and foliage colour. The reflected spectral
The Orthogonal Transformations radiances were convoluted with the spectral response functions
of the satellite instruments to simulate their responses. The
The derivation of vegetation indices has also been results indicated that vegetation indices could be interconverted
approached through orthogonal transformation techniques such to a precision of 1-2%.
as the PCA, the GVI of the Kauth-Thomas Tasseled Cap Ferreira and Huete (2004) used MQUALS (light aircraft-
Transformation and the MGVI of the Wheeler-Misra orthogonal based Modland Quick Airborne Looks package, consisting of
transformation. The link between these three techniques is that a spectroradiometer and digital camera), MODIS, AVHRR
they all express green vegetation through the development of and Landsat ETM+ in order to investigate VI’s ability to
their second component. differentiate the physiognomies in the Brazilian Cerrado and
Principal Components Analysis (PCA) is an orthogonal monitor their seasonal dynamics.
25
 Fensholt (2004) identified and validated net primary LAI and observed ETM+ reflectances were strongest in band
production (NPP) model variables from the MODIS sensor, 7.
focusing on the semi-arid ecosystem. Two MODIS VIs were Based on these examples, it is clear that remote sensing has
evaluated against the NOAA AVHRR NDVI to exploit the presented a significant opportunity to study and monitor
improvement of the MODIS sensor quality and also to examine vegetation and vegetation dynamics. The application of various
the possibility of establishing an empirical relation in order to VIs in a test site located in Northern Greece is shown in Figure
reap the full benefit of 20 years’ availability of NOAA AVHRR 2. Visual interpretation of the VI images can reveal differences
data. in their performances.
Price et al. (2002) evaluated the use of raw Thematic Silleos et al. (2002) carried out the first operational work to
Mapper (TM) band combinations and several derived VIs to assessing crop damage using space remote sensing techniques
determine optimal vegetation indices and band combinations (Figure 3). A linear regression model was used to compare
for discriminating among six grassland management practices remote sensing estimations with field observations. The results
in eastern Kansas. Tasselled Cap Brightness Index, GVI, of the model application for all the studied fields showed an
Wetness Index, the first three components from Principle agreement in 60% of cases, with a deviation of about 10%.
Component Analysis (PCA1, 2, 3), NDVI, GR (Green Ratio) In Figure 4, the orthogonal VIs, namely the Tasseled Cap,
and MR (MIR Ratio) were used as VIs. Principal Components Analysis, and the decorrelation stretch
Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

Huete et al. (2002) performed an initial analysis of the of the original bands, were produced in order to extract new
MODIS NDVI and EVI performances from both radiometric bands. In all these orthogonal transformations, a green band
and biophysical perspectives, using MODIS, airborne that was free of soil background effects was produced. This
radiometric measurements, Landsat ETM+ and in situ field was due to the fact that almost all soil characteristics were
biophysical data collected over four validation test sites. The ascribed to a brightness band. The results are displayed in a
results showed good correspondence between airborne- perspective view in order to facilitate visual interpretation.
measured, top-of-canopy reflectances and VI values with those An important issue to be considered when using VIs is the
from the MODIS sensor at four intensively measured test sites. pre-processing of the images. Radiometric calibration is usually
Gilabert et al. (2002) introduced a generalized soil-adjusted a standard procedure performed by the image distribution
vegetation index (GESAVI), theoretically based on a simple companies. Atmospheric correction is essential when
vegetation canopy model. biophysical parameters (e.g. biomass, LAI, percent vegetation
Peddle et al. (2001) compared ten vegetation indices for cover) are extracted from the VIs as final products. Erroneous
estimating boreal forest biophysical information from airborne VI estimations could result in misleading information, often
data. The authors used Landsat-TM images to test the hypothesis with severe consequences (Jensen, 2005). Also, atmospheric
that Spectral Mixture Analysis (SMA) is able to derive the correction is needed when VIs are to be compared with
percentage of sunlit crowns, background, and shadows within measurements obtained at different times, such as for multi-
a remote sensing image pixel. This sub pixel scale information temporal change detection (Lu et al., 2004). On the other hand,
has been shown to consistently provide significantly improved the demanding and complex algorithms used for atmospheric
estimates of forest biophysical information such as biomass, correction can be avoided in case of single date dataset (Song
leaf area index (LAI) and NPP compared to that provided by et al., 2001).
NDVI using airborne and satellite imagery. To accomplish the
work, ten different VIs were used to predict forest biophysical
parameters, the results of which were compared with those
obtained from SMA using airborne multispectral data from the
NASA COVER Project (Superior National Forest, Minnesota,
USA). In all cases, SMA shadow fraction provided significantly
better results than those of any VI, with improvements in the
order of 20% compared to the best VI result. In most cases, one
or more of the new vegetation indices provided a small to
moderate improvement compared to NDVI, with NDVI and
SAVI-1 performing best among the VIs, possibly due to the
inclusion of background reflectance.
Eklundhe et al. (2001) investigated the relationship between
Landsat ETM+ sensor data and leaf area index in a boreal
conifer forest. The authors used Landsat ETM+ and detailed
field data from a coniferous forest area, dominated by Norway
spruce and Scots pine. A forest canopy reflectance model was
used to simulate stand reflectances in the Landsat ETM+
wavelengths bands in order to investigate the theoretical
response to LAI changes. The analysis showed that the response
to changes in LAI was the strongest in the visible wavelength Figure 2 The following indices were estimated in a group of large fields in
bands, particularly in band 3, whereas only a weak response a test site in Northern Greece: NDVI (A), RVI (band 4/band3)
(B), SQRT(IR/R) (C), VI (band4 - band 3) (D), TNDVI (Sqrt
was noted in the near-infrared band and for some vegetation NDVI + 0.5) (E), and Iron Oxide (band3/band 1) (F). Bare land,
indices (RATIO and NDVI). Statistical relationships between corn and cotton were the main land cover types.

26
 Discussion

The use of any of these transformations depends on

the objective of the investigation and the general
geographic characteristics of the studied area. In theory,
any of these can be applied to any geographic area,
regardless of their sensitivity to various environmental
components that might limit their effectiveness. In this
respect, one might consider applying the slope-based
indices, as they are simple to use and yield numerical
results that are easy to interpret. However, it is important
to note that all slope-based indices, excluding SAVI,
have the major weakness of not being able to minimize
soil background effects. This means that a certain
proportion of their values, negative or positive, represent
Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

the background soil brightness. The effect of the

background soil is a major limiting factor in certain
statistical analyses geared towards the quantitative
assessment of above-ground green biomass.
Although there exist indices whose extremes may
be much lower and higher than those of the more
familiar NDVI, the distance-based VIs have the
advantage of being able to minimize the effects of
background soil brightness by combining the input
bands with the slope and intercept of the soil line. This
represents an important quantitative and qualitative
improvement in the significance of the indices for all
types of applications, particularly for those dealing
with arid and semi-arid environments. Despite the large
number of vegetation indices currently in use, it is clear
that much more needs to be carried out into how these
procedures can be applied in different environments.
Figure 3 Assessment of damage caused by frost to wheat crops in Northern Greece.
The Normalized Difference Vegetation Index (NDVI) was used as main Acknowledgements
input image in the estimation model. Class 0-25% (green) refers to very
low damage, Class 25-50% (yellow) refers to low damage, Class 50-70%
This study was funded by “Pythagoras”, a research
(pink) refers to moderate damage, and Class 70-100% (red) refers to
severe damage. Source: Silleos et al., 2002. grant awarded by the Managing Authority of the

Figure 4 Principal Components Analysis (upper left), Tasseled Cap (upper right), Decorrelation stretch (lower
left) and Natural colour (lower right) of a SPOT-5 image in 3D representation of Kerkini wetland in
North Greece.

27
 Operational Programme “Education and Initial Vocational Jensen, J.R., 2000. Remote sensing of the environment: an earth resources
Training” of Greece, which is partially funded by the European perspective. Prentice-Hall, Inc., 526p.
Social Fund - European Commission. The project for crop Jensen, J.R., 2005. Introductory digital image processing: A remote sensing
damage assessment was funded by the Hellenic Organization perspective. Prentice-Hall, Inc., 544p.
of Agricultural Insurance (HOAI) with the participation of Mr. Kauth, R.J., and G.S. Thomas, 1976. The Tasseled Cap - A Graphic
Description of the Spectral Temporal Development of Agricultural
George Petsanis, agronomist and stuff member of HOAI. Crops As Seen By Landsat. Proceedings of the Symposium on Machine
Processing of Remotely Sensed Data, Perdue University, West Lafayette,
References Indiana, 41-51.
Kogan, F.N., 1990. Remote Sensing of Weather Impacts on Vegetation in
Ashburn, P., 1978. The vegetative index number and crop identification, Nonhomogeneous Areas, International Journal of Remote Sensing, 11
The LACIE Symposium Proceedings of the Technical Session, 843- (8): 1405-1419.
850. Liu, W.T., and F.N. Kogan, 1996. Monitoring Regional Drought Using the
Bannari, A., A.R. Huete, D. Morin, and F. Zagolski, 1996. Effets de la Vegetation Condition Index, International Journal of Remote Sensing,
Couleur et de la Brillance du Sol Sur les Indices de Végétation, 17(14): 2761-2782.
International Journal of Remote Sensing, 17(10): 1885-1906. Lu, D., P. Mausel, E. Brondizio, and E. Moran, 2004. Change detection
Baret, F., and G. Guyot, 1991. Potentials and limits of vegetation indices techniques. International Journal of Remote Sensing, 25(12):2365-
Downloaded by [University of California, San Francisco] at 00:09 22 November 2014

for LAI and APAR assessment, Remote Sensing of Environment, 35(2- 2407.
3):161-173. Payero, J.O., C.M.U. Neale, and J.L. Wright, 2004. Comparison of eleven
Baret, F., G. Guyot, and D. Major, 1989. TSAVI: A Vegetation Index vegetation indices for estimating plant height of alfalfa and grass,
Which Minimizes Soil Brightness Effects on LAI and APAR Estimation, Applied Engineering in Agriculture, 20(3):385-393.
12th Canadian Symposium on Remote Sensing and IGARSS’90, Peddle, D.R., S.P. Brunke, and F.G. Hall, 2001. A comparison of spectral
Vancouver, Canada, 4 p. mixture analysis and ten vegetation indices for estimating boreal forest
Bastiaanssen, W.G.M., 1998. Remote sensing in water resources biophysical information from airborne data, Canadian Journal of
management: The state of the art, Colombo, Sri-Lanka, International Remote Sensing, 27(6): 627-635.
Water Management Institute, 118 p. Perry, C. Jr., and L.F. Lautenschlager, 1984. Functional Equivalence of
Birth, G.S., and G. McVey, 1968. Measuring the color of growing turf Spectral Vegetation Indices, Remote Sensing of Environment, 14(1-3):
with a reflectance spectroradiometer, Agronomy Journal, 60:640-643. 169-182.
Clevers, J.P.G.W., 1988. The derivation of a simplified reflectance model Price, K.P., X.L. Guo, and J.M. Stiles, 2002. Optimal Landsat TM band
for the estimation of Leaf Area Index, Remote Sensing of Environment, combinations and vegetation indices for discrimination of six grassland
25(1):53-69. types in eastern Kansas, International Journal of Remote Sensing, 23
Deering, D.W., J.W. Rouse, R.H. Haas, and J.A. Schell, 1975. Measuring (23):5031-5042.
“Forage Production” of Grazing Units From Landsat MSS Data, Qi, J., A. Chehbouni, A.R. Huete, Y.H. Kerr, and A. Sorooshian, 1994. A
Proceedings of the 10th International Symposium on Remote Sensing Modified Soil Adjusted Vegetation Index, Remote Sensing of
of Environment, II:1169-1178. Environment, 48(2):119-126.
Eastman, R.J., 2003. IDRISI Kilimanjaro, Guide to GIS and Image Richardson, A.J., and C.L. Wiegand, 1977. Distinguishing Vegetation
Processing, Clark Labs, Clark University, 950 Main Street, Worcester, From Soil Background Information, Photogramnetric Engineering
MA,01610-1477 USA. and Remote Sensing, 43(12):1541-1552.
Eklundh, L., L. Harrie, and A. Kuusk, 2001, Investigating relationships
Rouse, J.W. Jr., R.H. Haas, D.W. Deering, J.A. Schell, and J.C. Harlan,
between Landsat ETM plus sensor data and leaf area index in a boreal
1974. Monitoring the Vernal Advancement and Retrogradation (Green
conifer forest, Remote Sensing of Environment, 78(3): 239-251.
Wave Effect) of Natural Vegetation, NASA/GSFC Type III Final
Fensholt, R., 2004. Earth observation of vegetation status in the Sahelian Report, Greenbelt, MD., 371p.
and Sudanian West Africa: comparison of terra MODIS and NOAA
AVHRR satellite data, International Journal of Remote Sensing, 25 Silleos N., K. Perakis, and G. Petsanis, 2002. Assessment of crop damages
(9): 1641-1659. using Space remote sensing and GIS. International Journal of Remote
Sensing, 23(3):417-427.
Ferreira, L.G., and A.R. Huete, 2004. Assessing the seasonal dynamics of
the Brazilian Cerrado vegetation through the use of spectral vegetation Singh, A., and A. Harrison, 1985. Standardized Principal Components,
indices, International Journal of Remote Sensing, 25(10): 1837-1860. International Journal of Remote Sensing, 6(6):883-896.
Fung, T., and E. LeDrew, 1988. The Determination of Optimal Threshold Song, C., C.E. Woodcock, K.C. Seto, M. Pax Lenney, and S.A. Macomber,
Levels for Change Detection Using Various Accuracy Indices, 2001. Classification and Change Detection Using Landsat TM Data:
Photogrammetric Engineering and Remote Sensing, 54(10): 1449- When and How to Correct Atmospheric Effects? Remote Sensing of
1454. Environonment, 75(2):230-244.
Gilabert, M.A., J. Gonzalez-Piqueras, F.J. Garcia-Haro, and J. Melia, Steven, M.D., T.J. Malthus, F. Baret, H. Xu, and M.J. Chopping, 2003.
2002. A generalized soil-adjusted vegetation index, Remote Sensing of Intercalibration of vegetation indices from different sensor systems,
Environment, 82(2-3): 303-310. Remote Sensing of Environment, 88(4):412-422.
Huete, A.R., 1988. A Soil-Adjusted Vegetation Index (SAVI), Remote Thiam, A.K., 1997. Geographic Information Systems and Remote Sensing
Sensing of Environment, 25(3): 295-309. Methods for Assessing and Monitoring Land Degradation in the Sahel:
Huete, A.R., C. Justice, and W. van Leeuwen, 1999. MODIS Vegetation The Case of Southern Mauritania. Doctoral Dissertation, Clark
Index (MOD13) Algorithm Theoretical Basis Document, NASA University, Worcester Massachusetts.
Goddard Space Flight Center, http://modis.gsfc.nasa.gov/data/atbd/ Tripathy, G.K., T.K. Ghosh, and S.D. Shah, 1996. Monitoring of
atbd_mod13.pdf, 120p. Desertification Process in Karnataka State of India Using Multi-
Huete, A., K. Didan, T. Miura, E.P. Rodriguez, X. Gao, and L.G. Ferreira, Temporal Remote Sensing and Ancillary Information Using GIS,
2002. Overview of the radiometric and biophysical performance of the International Journal of Remote Sensing, 17(12):2243-2257.
MODIS vegetation indices, Remote Sensing of Environment, 80(1-2): Vaiopoulos D., G. Skianis, and K. Nikolakopoulos, 2004. The contribution
195-213. of probability theory in assessing the efficiency of two frequently used
Jackson, R.D., and A.R. Huete, 1991. Interpreting vegetation indices. vegetation indices. International Journal of Remote Sensing, 25(20):
Preventive Veterinary Medicine, 11:185-200. 4219-4236.

28