and energy exchanges between ecosystems and atmosphere. This provides valuable opportunities for evaluation of satellite VI measures of vegetation growth,
phenology, and seasonal dynamics. Monteith and Unsworth (1990) noted that VIs
can legitimately be used to estimate the rate of processes that depend on absorbed
light, such as gross primary production (GPP, photosynthesis) and transpiration.
Several studies have shown potential satellite data validation opportunities via
FLUXNET, e.g., strong, multiple-biome satellite EVI relationships have been
reported with tower GPP flux measurements across AmeriFlux tower sites and
tropical forests in the Amazon and Southeast Asia with MODIS and SPOT-VGT
satellite data (Rahman et al. 2005; Sims et al. 2006; Xiao et al. 2004, 2005; Huete
et al. 2006, 2008) (Fig. 1.10).
1.4.2 Biophysical Validation
Field-based vegetation sampling is fundamental for validating and assessing VI
performance in depicting vegetation dynamics and biophysical phenomena. A
good correspondence between VIs and field measurements lends confidence in
their use as biophysical surrogates for variables that are otherwise difficult to
sample in the field. VI relationships with biophysical properties are mostly derived
from empirical field measurements and canopy radiative transfer models, where
numerous and often ambiguous relationships have been reported (Sellers 1985).
Hence, although VIs have been validated within numerous environments, the
resulting biophysical relationships tend to be local-based and with limited spatial
extent that rarely extend to landscape-relevant temporal and spatial scales.
The biophysical validation of VIs is complicated by a lack of consensus on
what VIs explicitly measure about a canopy and how to interpret a VI value. VIs
respond to upper sunlit leaves to a greater extent than lower leaves, resulting in
strongly non-linear relationships with field-derived LAI values. NDVI sensitivity
to LAI variations is generally restricted to values below 2 or 3 with differing
correlations between broadleaf vs needle-leaf canopy stands (Fassnacht et al. 1997;
Chen et al. 2005). NDVI-LAI relationships may further vary across different
canopy phenophases, as was found in a beech deciduous forest in Europe (Wang
et al. 2005) (Fig. 1.11).
Linear combination and optimized indices provide extended LAI sensitivity and
are less prone to saturate in high-biomass areas (Fensholt et al. 2004). Houborg
and Soegaard (2004) found EVI from MODIS to accurately describe the variations
in green LAI up to 5 in agriculture areas in Denmark (r
2 = 0.91). However, in
comparison with relationships reported using fine resolution Landsat satellite data,
poorer relationships are commonly found when using coarser resolution VIs, such
as MODIS and SPOT-VEGETATION (VGT). In a multi-biome validation field
campaign known as Bigfoot, Cohen et al. (2003) found only weak correlations
between field measured LAI and several MODIS products, including LAI and VIs.
1 Indices of Vegetation Activity
17
phenology, and seasonal dynamics. Monteith and Unsworth (1990) noted that VIs
can legitimately be used to estimate the rate of processes that depend on absorbed
light, such as gross primary production (GPP, photosynthesis) and transpiration.
Several studies have shown potential satellite data validation opportunities via
FLUXNET, e.g., strong, multiple-biome satellite EVI relationships have been
reported with tower GPP flux measurements across AmeriFlux tower sites and
tropical forests in the Amazon and Southeast Asia with MODIS and SPOT-VGT
satellite data (Rahman et al. 2005; Sims et al. 2006; Xiao et al. 2004, 2005; Huete
et al. 2006, 2008) (Fig. 1.10).
1.4.2 Biophysical Validation
Field-based vegetation sampling is fundamental for validating and assessing VI
performance in depicting vegetation dynamics and biophysical phenomena. A
good correspondence between VIs and field measurements lends confidence in
their use as biophysical surrogates for variables that are otherwise difficult to
sample in the field. VI relationships with biophysical properties are mostly derived
from empirical field measurements and canopy radiative transfer models, where
numerous and often ambiguous relationships have been reported (Sellers 1985).
Hence, although VIs have been validated within numerous environments, the
resulting biophysical relationships tend to be local-based and with limited spatial
extent that rarely extend to landscape-relevant temporal and spatial scales.
The biophysical validation of VIs is complicated by a lack of consensus on
what VIs explicitly measure about a canopy and how to interpret a VI value. VIs
respond to upper sunlit leaves to a greater extent than lower leaves, resulting in
strongly non-linear relationships with field-derived LAI values. NDVI sensitivity
to LAI variations is generally restricted to values below 2 or 3 with differing
correlations between broadleaf vs needle-leaf canopy stands (Fassnacht et al. 1997;
Chen et al. 2005). NDVI-LAI relationships may further vary across different
canopy phenophases, as was found in a beech deciduous forest in Europe (Wang
et al. 2005) (Fig. 1.11).
Linear combination and optimized indices provide extended LAI sensitivity and
are less prone to saturate in high-biomass areas (Fensholt et al. 2004). Houborg
and Soegaard (2004) found EVI from MODIS to accurately describe the variations
in green LAI up to 5 in agriculture areas in Denmark (r
2 = 0.91). However, in
comparison with relationships reported using fine resolution Landsat satellite data,
poorer relationships are commonly found when using coarser resolution VIs, such
as MODIS and SPOT-VEGETATION (VGT). In a multi-biome validation field
campaign known as Bigfoot, Cohen et al. (2003) found only weak correlations
between field measured LAI and several MODIS products, including LAI and VIs.
1 Indices of Vegetation Activity
17