• Monitoring Drylands - Problems
• The Vegetation Problem
• Vegetation and Soil Signatures
• Extracting Information
• Vegetation Indices
EARTH OBSERVATION CENTRE, UKM 1
Land Degradation Monitoring in Drylands
• Land degradation is a complex ensemble of surface processes (e.g.
wind erosion, water erosion, soil compaction, salinisation, and soil
• These can ultimately lead to "desertification".
• As the increasing world population places more demands on land for
food production etc., many marginal arid and semiarid lands will be
at risk of degradation.
• The need to maintain sustainable use of these lands requires that
they be monitored for the onset of land degradation so that the
problem may be addressed in its early stages.
• Monitoring will also be required to assess the effectiveness of
measures to control land degradation
EARTH OBSERVATION CENTRE, UKM 2
Problems with Monitoring Dryland Vegetation
• Remote sensing of arid regions is difficult and necessitates
• Desert plants typically manifest long periods of dormancy
interspersed with brief "greenings" associated with storms
or seasonal rainfall.
• During these relatively short productive periods, the
characteristic spectral features of desert plants change, as
does total vegetation cover
• Current long repeat times of Landsat and other present
satellite sensors provide insufficient temporal resolution to
reliably capture the short, but critical, greening.
EARTH OBSERVATION CENTRE, UKM 3
Specific Challenges for Land Degradation Monitoring
• Arid region vegetation is intrinsically difficult to study remotely
(1) vegetative cover usually is sparse compared to soil background,
(2) soil and plant spectral signatures tend to mix non-linearly, and
(3) arid plants tend to lack the strong red edge found in plants of
humid regions due to ecological adaptations to harsh desert
• A very important result of these studies is that conventional
vegetative red indices can be unreliable measures of arid region
plant cover with potential for over- or underestimation of the
actual vegetative cover.
EARTH OBSERVATION CENTRE, UKM 4
Satellite Remote Sensing as a Monitoring Tool – Pro’s
• The operational costs of satellite systems are significantly lower than for
other platform types (e.g. aircraft).
• Satellite systems provide automatically repeating coverage along
predictable flight paths with little variance compared to aircraft flight
lines. This provides the ability to track seasonal changes and, over a
longer time scale, changes related to climatic variability.
• This capability may also enable differentiation between anthropogenic
land degradation and natural variations.
• A satellite system also provides automatic coverage of much of the entire
globe, and therefore, potentially, may enable some degree of global
• Lastly, a satellite system monitoring drylands on a global scale has a
greater potential for producing data useful for currently unanticipated
needs than does dedicated airborne data collection.
EARTH OBSERVATION CENTRE, UKM 5
Airborne Remote Sensing ?
• Airborne remote sensing is not an efficient tool suitable for such
monitoring for many reasons.
• First, airborne sensors can only provide a relatively local view.
• Each acquisition of data using an airborne system requires an
active decision to fly the instrument over the target area.
• It is extremely difficult to accurately reproduce flight lines,
which dramatically increases the difficulty of analysing and
interpreting the monitoring data.
• Airborne instruments suffer through flight stresses each time that
the instrument is flown, which can compound the difficulty of
comparing data acquired at different times.
• The operating expenses for an airborne instrument are very high
EARTH OBSERVATION CENTRE, UKM 6
• The most vital single parameter for dryland monitoring is the
signature of vegetation cover.
• Vegetation provides protection against degradation processes
such as wind erosion, and subtle changes in vegetation are
likely to be a precursor of wind erosion.
• Decreasing vegetation cover, and changes in the population
of the vegetation cover, (e.g., from creosote bush to
bursage), are sensitive indicators of land degradation.
• Vegetation reflects the hydrological aspects of arid regions,
and provides an indicator of current and recent hydrological
EARTH OBSERVATION CENTRE, UKM 8
• The 0.4-1.0 µm part of the electromagnetic
spectrum contains the red edge feature of the
green vegetation reflectance spectrum which
is exploited by standard vegetation indices.
• Laboratory and field spectra of some desert
plants indicates that there are also interesting
features in the 2.0--2.5 µm range related to
leaf coatings, but the visible wavelength
pigment features are more easy to sense.
EARTH OBSERVATION CENTRE, UKM 9
Extracting a vegetation signal
• Current techniques of remotely measuring vegetation cover are
based on the characteristics of humid vegetation with large leaf
area, fairly continuous canopies, high chlorophyll content, and
thin, translucent leaves.
• Arid vegetation has special adaptations to the water and thermal
stresses which occur in these regions.
• The inability of arid region vegetation to regulate temperature
through transpiration leads to small leaves and open canopies to
improve the efficiency of cooling the leaves by moving air.
• The small leaves reduce the amount of leaf area in arid
vegetation, and the open canopies mean that a great deal of soil
is visible through most arid vegetation canopies.
EARTH OBSERVATION CENTRE, UKM 11
Extracting a vegetation signal (cont.)
• Further compounding the problem is the fact that arid
plants tend to have vertically oriented leaves to avoid
direct sunlight during midday, which is when remote
sensing observations are generally made in order to have
the brightest lighting and the fewest shadows.
• The edge-on view of these leaves means that little of the
small amount of leaf area present in arid plants can be seen
with remote sensing.
• Other plants change the orientation of their leaves by
rolling and unrolling or steering the leaves which has the
same effect of reducing the leaf area visible to remote
EARTH OBSERVATION CENTRE, UKM 12
Extracting a vegetation signal (cont.)
• Many arid region plants have leaf hairs and coatings
which alter the spectral properties of the leaves, and they
often have less chlorophyll concentration than humid
• On a larger scale, desert shrubs, which are the dominant
plant type in the vast majority of deserts around the
world, are sparsely distributed.
• This sparse distribution of shrubs, coupled with the open
canopies of the shrubs means that variability of the soil
background will be very significant in the reflected
spectrum in arid regions.
EARTH OBSERVATION CENTRE, UKM 13
Extracting a vegetation signal (cont.)
• The nature of the soil ``noise,'' which is partially due to
non-linear spectral mixing, will be different than that
observed in humid regions because very little light
physically passes through the leaves in arid plants, while
significant amounts pass through humid plant leaves
(Roberts et al., 1994).
• There is high variability in the nature, appearance, and
behaviour of arid vegetation with respect to recent rainfall.
• There are also significant variations in the appearance of
plants due to seasonal effects.
• Lastly, spectral characteristics differ significantly between
EARTH OBSERVATION CENTRE, UKM 14
EARTH OBSERVATION CENTRE, UKM 15
• The lower right boundary of this sort of plot is taken to be formed by pixels
containing only bare soil, and this boundary is referred to as the soil line.
• The “tip” opposite the soil line, which has high NIR reflectance and low red
reflectance, is taken to be where pixels completely covered with
vegetation plot on this diagram.
• All pixels covered by a mixture of bare soil and vegetation will plot
between these two extremes. This sort of figure is sometimes called a
tasselled-cap, because of its shape.
Points to Note: Soils
• Soil components that affect spectral reflectance can be
grouped into three components:
3. Water content
• Roughness also has the effect of decreasing reflectance
because of an increase in multiple scattering and
• Analysis has shown that for a given type of soil
characteristic, variability in one wavelength is often
functionally related to the reflectance in another
EARTH OBSERVATION CENTRE, UKM 17
Points to Note: Soils
• Thus, variation in any one soil parameter can give rise to a
line on a 2D scattergram.
• For RED-NIR scattergrams, this is termed the “soil line”,
and is used as a reference point in most vegetation studies.
• The problem is that real soil surfaces are not
homogeneous, and contain a composite of several types of
• However, Jasinki and Eagleson (1989) showed that when
experimentally varying three soil parameters together, the
composite line is generally linear, but can exhibit scatter.
EARTH OBSERVATION CENTRE, UKM 18
Points to Note: Vegetation Indices
• There are three types of vegetation
1. Simple, Intrinsic Indices
2. Indices which use a soil line
3. Atmospherically Corrected Indices
EARTH OBSERVATION CENTRE, UKM 19
Points to Note: Vegetation Indices
• Within these, there have been four general approaches
taken, based on the characteristics of the tasselled-cap.
1. The first approach is to measure the distance between
where the pixel plots in the tasselled cap plot from the
soil line. (The soil line is used because it is generally
easier to find than the 100% vegetation point).
2. The second approach is to assume that the isovegetation
lines all intersect at a single point.
3. The third approach is to recognise that lines do not
intersect at a single point.
4. The final possibility is to assume that the isovegetation
lines are non-linear.
EARTH OBSERVATION CENTRE, UKM 20
Simple Vegetation Indices
• As the first approximation, Jordan (1969) developed the
ratio vegetation index:
RVI = NIR
• RVI itself is no longer generally used in remote sensing.
Instead a index known as the normalized difference
vegetation index (NDVI) is used.
NIR-RED RVI +1
NIR+RED RVI - 1
Both RVI and NDVI basically measure the slope of the line
between the origin of red-NIR space and the red-NIR value
of the image pixel.
• The only difference between RVI and NDVI is the range of values that the
two indices take one. The range from -1.0-1.0 for NDVI is easier to deal with
than the infinite range of the RVI.
• NDVI can also be considered to be an improvement of DVI which eliminates
effects of broad-band red-NIR albedo through the normalization.
• Crippen (1990) recognized that the red radiance subtraction in the numerator
of NDVI was irrelevant, and he formulated the infrared percentage vegetation
IPVI = NIR = ½ (NDVI+1)
NIR + RED
• IPVI is functionally equivalent to NDVI and RVI, but it only ranges in value
• It also eliminates one mathematical operation per image pixel which is
important for the rapid processing of large amounts of data.
Soil Line ??
• The soil line will be different for different areas (soil types) and the soil line will
vary for different NIR and red band passes.
• Table 9 gives the slope and intercept for the soil line calculated from AVIRIS
data for different bandpasses.
• The clear implication is that the only truly valid way of making use of a
vegetation index which uses a soil line is to compute the soil line for each image.
• If a good calibration is available, calculating the soil line for each target for each
instrument once might suffice.
• Of course, even the assumption that all of the bare soil spectra in a single image
form a line may also be inaccurate.
• Elvidge and Chen (1995) found that SAVI and PVI consistently provided better
estimates of LAI and percent green cover than did NDVI or RVI.
• They also found that there was a steady improvement in all of these vegetation
indices as narrower and narrower bands were used for the near-infrared and red
reflectances, with SAVI being the best index at the very narrowest bandwidth.
• The advantage of narrow bands for use with vegetation indices provides
additional arguments for the use of high spectral resolution remote sensing.
Table 9: Red-NIR Soil Line Parameters for AVIRIS Data Sampled at
Instrument Red Band-pass NIR Band-pass
Simulated (m) (m)
MSS 0.6-0.7 0.8-1.1 0.9034 52.95
TM 0.63-0.68 0.8-0.9 0.7939 71.39
AVIRIS 0.674 0.755 0.8863 55.00
Indices Using the Soil Line
NIR Soil line
• The perpendicular vegetation index (PVI) of Richardson and Wiegand (1977)
assumes that the perpendicular distance of the pixel from the soil line is linearly
related to the vegetation cover. This index is calculated as follows:
PVI NIR red = - sin a (NIR) cos a (red)
• where (NIR) is the near-infrared reflectance, (red) is the red reflectance and (a) is
the angle between the soil line and the near-infrared axis. This means that the
isovegetation lines (lines of equal vegetation) would all be parallel to the soil line.
Soil Adjusted VI
• Huete (1988) suggested a new vegetation index which was designed to
minimize the effect of the soil background, which he called the soil-adjusted
vegetation index (SAVI). This vegetation index takes the form:
SAVI = NIR-RED (1+L)
• Huete showed evidence that the isovegetation lines do not converge at a single
point, and he selected the L-factor in SAVI based where lines of a specified
vegetation density intersect the soil line.
• The net result is an NDVI with an origin not at the point of zero red and near-
• For high vegetation cover, the value of L is 0.0, and L is 1.0 for low
• For intermediate vegetation cover L=0.5, and that is the values which is
most widely used. The appearance of L in the multiplier causes SAVI to
have a range identical to the of NDVI (-1.0 - 1.0).
• Huete (1988) suggested that SAVI takes on both the aspects of NDVI
• A further development of this concept is the transformed SAVI (TSAVI)
Baret and Guyot, 1991), defined as
TSAVI = a(NIR-aR-b)/[R=a(NIR-b) + 0.08(1+a2)]
• Where a and b are, respectively, the slope and intercept of the soil line
(NIRsoil = aRsoil +b), and the coefficient value 0.08 has been adjusted to
minimise soil effects
• Qi et al. (1994a) further developed a vegetation index which is
basically a version of SAVI where the L-factor is dynamically adjusted
using the image data.
• They referred to this index as the Modified Soil Adjusted Vegetation
Index or MSAVI. The factor L is given by the following expression:
L= 1 - (2 x slope x NDVI x WDVI)
• where WDVI is the Weighted Difference Vegetation of Clevers (1988)
which is functionally equivalent to PVI and calculated as follows
WDVI = NIR - (slope x RED)
• Qi et al. (1994a) also created an iterated version of this vegetation
which is called MSAVI2:
MSAVI2 = 1/2 * ((2*(NIR+1)) - (((2*NIR)+1)2 - 8(NIR-red))1/2).
Atmospherically Corrected Indices
• In order to reduce the dependence of the NDVI on the atmospheric
properties, Kaufman and Tanere (1992) proposed a modification to the
formulation of the index, introducing the atmospheric information contained
in the BLUE channel, defining
ARVI = (NIR – RB) / (NIR+RB)
• Where RB is a combination of the reflectances in the Blue (B) and Red (R)
RB = R – (B-R)
• And depends on the aerosol type (a good value is = 1 when the aerosol
model is not available)
• The authors emphasise the fact that this concept can be applied to other
indices. SAVI can be changed to SARVI by changing R to RB.
• However, Myneni and Asrar (1994) noted that although SAVI and ARVI
correct for soil and atmospheric effects independently, they fail to do so when
Atmospherically Corrected Indices
• Pinty and Verstraete, (1992) proposed a new index to
account for soil and atmospheric effects simultaneously.
• This is a non-linear index called GEMI:
GEMI = n(1-0.25n) – (R-0.125)/(1-R)
Where n =
[2(NIR2-R2) + 1.5NIR + 0.5R] / (NIR + R + 0.5)
• This index is seemingly transparent to the atmosphere, and
represents plant information at least as well as NDVI – but
is complicated, and difficult to use and interpret.
Which One to Use ?
• In a simulation study, Rondaux et al., (1996) found that an
optimised SAVI (OSAVI), where the value of X was tuned
to 0.16 easily out-performed all other indices for
application to agricultural surfaces.
• They found that a locally tuned SAVI (MSAVI) was more
appropriate for all other applications.
• However, in Niger, Leprieur et al (1996) found GEMI to
be less sensitive to the atmosphere – however, they found
it incapable of dealing with variations in soil reflectance.
• They suggest that the use of MSAVI with an accurate
atmospheric correction is essential or perhaps using a
combination of GEMI and MSAVI.
• One important difficulty which has been encountered in using the
vegetation indices which attempt to minimize the effect of a changing
soil background is an increase in the sensitivity to variations in the
atmosphere (Leprieur et al., 1994; Qi et al., 1994b).
• There have been several approaches in the development of vegetation
indices which are less sensitive to the atmosphere, such as the
Atmospherically Resistant Vegetation Index (ARVI) of Kaufman and
Tanré (1992) and the Global Environmental Monitoring Index (GEMI)
of Pinty and Verstraete (1991).
• Chehbouni has data demonstrating that GEMI is highly sensitive to soil
• Qi et al. (1994b) demonstrated that soil noise caused GEMI to violently
break down at low vegetation covers, and that all of the vegetation
indices designed to minimize the effect of the atmosphere have
increased sensitivity to the soil, which makes these indices completely
unsuitable for arid regions.