Van Leeuwen et al. 2006). Multi-decadal, validated, consistent global and regional
data sets of LAI/FPAR from the AVHRR, MODIS, and the SPOT-VGT sensors
are now available at resolutions of 1 km to 1° in service of several national and
international initiatives (Chen 2002; Fernandes and Butson 2003; Ganguly et al.
2008a; Myneni et al. 2002). Long-term records of LAI and FPAR are required by
various terrestrial biosphere models, like the Terrestrial Ecosystem Model (TEM)
(Melillo et al. 1993), Biome-BGC (Running and Gower 1991), Simple Biospheric
Model (SiB) (Sellers et al. 1986), Integrated Biosphere Simulated Model (IBIS)
(Foley et al. 1996), Lund-Potsdam-Jena (LPJ) dynamic global vegetation model in
Land Surface Model (LSM) (Bonan et al. 2003) and the Atmospheric-Vegetation
Interactive Model (AVIM) (Jinjun 1995), for the investigation of the response of
ecosystems to the changes in climate, carbon cycle, land cover and land use. The
Landsat series of sensors also provides a unique opportunity to characterize terrestrial ecosystem processes at a spatial scale at which most natural resources
management decisions are made. Although regional- to continental-scale multitemporal mosaics of Landsat data have been constructed for pilot studies of
national land use change monitoring and disturbance mapping (Chander et al.
2009; Hansen et al. 2008; Wulder et al. 2002), the Landsat archive has not yet been
exploited to derive long-term biophysical products. This chapter provides a brief
overview of the recent progresses in some of the key LAI/FPAR estimation
algorithms and resulting biophysical products from the AVHRR, MODIS, SPOT
and Landsat data at global to continental scales.
2.2 Algorithmic Theoretical Basis
There is considerable literature on the estimation of LAI from vegetation indices
like the Normalized Difference Vegetation Index (NDVI), Simple Ratio and
Reduced Simple Ratio (RSR) (Asrar et al. 1984; Chen and Cihlar 1996; Stenberg
et al. 2004; Brown et al. 2000). In particular, (Sellers et al. 1996) introduced an
empirical algorithm that calculated FPAR as a function of the simple ratio. Lu and
Shuttleworth (2002) used this definition of FPAR and approximated the relationship between LAI and FPAR to be exponential (Monteith and Unsworth 1990) for
evenly distributed vegetation. Strong positive correlations were found between
LAI and NDVI for various vegetation types (Myneni et al. 1997), as well as with
simple ratio in coniferous forests (Chen and Cihlar 1996). Site-specific NDVI/
RSR-LAI empirical relationships have been used in various ecosystems (Colombo
et al. 2003; Fassnacht et al. 1997; Stenberg et al. 2004), but with limited success
when applied across sites and vegetation classes.
The sensitivity of NDVI or RSR to LAI is controlled by the relationship
between NDVI/RSR and fractional vegetation cover when LAI is in the range of
about 2–4 (Carlson and Ripley 1997; Stenberg et al. 2004). Steltzer and Welker
(2006) incorporated fractional cover of photosynthetic vegetation for multiple
species into the exponential NDVI-LAI model for a regional scale analysis, and
2 Green Leaf Area and Fraction of Photosynthetically
45
data sets of LAI/FPAR from the AVHRR, MODIS, and the SPOT-VGT sensors
are now available at resolutions of 1 km to 1° in service of several national and
international initiatives (Chen 2002; Fernandes and Butson 2003; Ganguly et al.
2008a; Myneni et al. 2002). Long-term records of LAI and FPAR are required by
various terrestrial biosphere models, like the Terrestrial Ecosystem Model (TEM)
(Melillo et al. 1993), Biome-BGC (Running and Gower 1991), Simple Biospheric
Model (SiB) (Sellers et al. 1986), Integrated Biosphere Simulated Model (IBIS)
(Foley et al. 1996), Lund-Potsdam-Jena (LPJ) dynamic global vegetation model in
Land Surface Model (LSM) (Bonan et al. 2003) and the Atmospheric-Vegetation
Interactive Model (AVIM) (Jinjun 1995), for the investigation of the response of
ecosystems to the changes in climate, carbon cycle, land cover and land use. The
Landsat series of sensors also provides a unique opportunity to characterize terrestrial ecosystem processes at a spatial scale at which most natural resources
management decisions are made. Although regional- to continental-scale multitemporal mosaics of Landsat data have been constructed for pilot studies of
national land use change monitoring and disturbance mapping (Chander et al.
2009; Hansen et al. 2008; Wulder et al. 2002), the Landsat archive has not yet been
exploited to derive long-term biophysical products. This chapter provides a brief
overview of the recent progresses in some of the key LAI/FPAR estimation
algorithms and resulting biophysical products from the AVHRR, MODIS, SPOT
and Landsat data at global to continental scales.
2.2 Algorithmic Theoretical Basis
There is considerable literature on the estimation of LAI from vegetation indices
like the Normalized Difference Vegetation Index (NDVI), Simple Ratio and
Reduced Simple Ratio (RSR) (Asrar et al. 1984; Chen and Cihlar 1996; Stenberg
et al. 2004; Brown et al. 2000). In particular, (Sellers et al. 1996) introduced an
empirical algorithm that calculated FPAR as a function of the simple ratio. Lu and
Shuttleworth (2002) used this definition of FPAR and approximated the relationship between LAI and FPAR to be exponential (Monteith and Unsworth 1990) for
evenly distributed vegetation. Strong positive correlations were found between
LAI and NDVI for various vegetation types (Myneni et al. 1997), as well as with
simple ratio in coniferous forests (Chen and Cihlar 1996). Site-specific NDVI/
RSR-LAI empirical relationships have been used in various ecosystems (Colombo
et al. 2003; Fassnacht et al. 1997; Stenberg et al. 2004), but with limited success
when applied across sites and vegetation classes.
The sensitivity of NDVI or RSR to LAI is controlled by the relationship
between NDVI/RSR and fractional vegetation cover when LAI is in the range of
about 2–4 (Carlson and Ripley 1997; Stenberg et al. 2004). Steltzer and Welker
(2006) incorporated fractional cover of photosynthetic vegetation for multiple
species into the exponential NDVI-LAI model for a regional scale analysis, and
2 Green Leaf Area and Fraction of Photosynthetically
45