and FPAR chl ) and structural (LAI and FPAR canopy ) variables across leaf, canopy,
and landscape levels and develope novel algorithms of chlorophyll content over
terrestrial ecosystems. Since a large portion of leaf nitrogen is contained within
leaf chloroplasts, more attention should also be paid to develop quantitative
relationships among chlorophyll, FPAR chl , and nitrogen content.
Secondly, the uncertainty concerning the scaling-up of light-use efficiency from
chloroplast and leaf levels to canopy, ecosystem, and landscape levels is also still
significant. The estimation methods and values of maximum light-use efficiency
for various types of terrestrial ecosystems differ substantially and need more
integrative studies across various levels: chloroplast, leaf, plant, canopy, ecosystem and landscape. LUE is the primary controlling parameter of satellite-based
PEMs. For most PEMs, a species- or biome-specific maximum LUE value is
predefined, and it is then down-regulated by the scalars representing various
environmental stresses. Different definitions and choices of maximum LUE and
environmental scalars are the main sources of uncertainty about PEMs. Daily NEE
and PAR data are used to examine the continuous short-time changes of LUE.
Recently, increasing effort has been reported for estimating LUE with remote
sensing techniques (Garbulsky et al. 2011). Remote sensing can indirectly estimate
LUE by adjusting maximum LUE values through environmental factors, including
soil water content, temperature, nitrogen content, and so forth. Remote sensing can
also directly estimate LUE by detecting the photoprotective mechanism from the
leaf spectral reflectance change due to the epoxidation of xanthophyll cycle pigments (Barton and North 2001). The PRI (Photochemical Reflectance Index),
which demonstrates the characteristics of spectrum absorption around 505 nm and
531 nm (Gamon et al. 1992), is found to have a strong correlation with LUE
(Garbulsky et al. 2011; Hilker et al. 2009, 2010, 2012; Wu et al. 2010). Even
though the temporal, spatial, and spectral resolutions of remote sensing, and the
semiempirical feature of this approach, limit its application, the PRI can represent
the integral effect of various environmental factors, and it has great potential to
estimate LUE on regional and global scales in the future (Drolet et al. 2008).
Considerations of the temporal and spatial dynamics of LUE will greatly improve
the simulation accuracy of PEMs (Garbulsky et al. 2011).
Thirdly, there is still a great deal of uncertainty in delineating vegetation
growing seasons (starting date, ending date, and the length of the vegetation
growing season) from remote sensing data, which substantially affects the estimates of total GPP over the vegetation growing season (Falge et al. 2002; Richardson et al. 2010). There is a need for better understanding and quantification of
vegetation phenology through analyses of both remote sensing data and CO 2 flux
data (ecosystem physiology approach). Vegetation phenology varies over years,
driven by interannual climate variability and climate change (e.g., temperature and
precipitation) (Piao et al. 2006, 2011), and is also affected by topography (Doktor
et al. 2009; Hwang et al. 2011; Piao et al. 2011). In addition, there are also
significant differences among various algorithms that use remotely sensed data to
retrieve vegetation phenology (Cong et al. 2012). Therefore, it is very important to
continue evaluation and development of satellite-based algorithms to retrieve
5 Gross Primary Production of Terrestrial Vegetation
143
and landscape levels and develope novel algorithms of chlorophyll content over
terrestrial ecosystems. Since a large portion of leaf nitrogen is contained within
leaf chloroplasts, more attention should also be paid to develop quantitative
relationships among chlorophyll, FPAR chl , and nitrogen content.
Secondly, the uncertainty concerning the scaling-up of light-use efficiency from
chloroplast and leaf levels to canopy, ecosystem, and landscape levels is also still
significant. The estimation methods and values of maximum light-use efficiency
for various types of terrestrial ecosystems differ substantially and need more
integrative studies across various levels: chloroplast, leaf, plant, canopy, ecosystem and landscape. LUE is the primary controlling parameter of satellite-based
PEMs. For most PEMs, a species- or biome-specific maximum LUE value is
predefined, and it is then down-regulated by the scalars representing various
environmental stresses. Different definitions and choices of maximum LUE and
environmental scalars are the main sources of uncertainty about PEMs. Daily NEE
and PAR data are used to examine the continuous short-time changes of LUE.
Recently, increasing effort has been reported for estimating LUE with remote
sensing techniques (Garbulsky et al. 2011). Remote sensing can indirectly estimate
LUE by adjusting maximum LUE values through environmental factors, including
soil water content, temperature, nitrogen content, and so forth. Remote sensing can
also directly estimate LUE by detecting the photoprotective mechanism from the
leaf spectral reflectance change due to the epoxidation of xanthophyll cycle pigments (Barton and North 2001). The PRI (Photochemical Reflectance Index),
which demonstrates the characteristics of spectrum absorption around 505 nm and
531 nm (Gamon et al. 1992), is found to have a strong correlation with LUE
(Garbulsky et al. 2011; Hilker et al. 2009, 2010, 2012; Wu et al. 2010). Even
though the temporal, spatial, and spectral resolutions of remote sensing, and the
semiempirical feature of this approach, limit its application, the PRI can represent
the integral effect of various environmental factors, and it has great potential to
estimate LUE on regional and global scales in the future (Drolet et al. 2008).
Considerations of the temporal and spatial dynamics of LUE will greatly improve
the simulation accuracy of PEMs (Garbulsky et al. 2011).
Thirdly, there is still a great deal of uncertainty in delineating vegetation
growing seasons (starting date, ending date, and the length of the vegetation
growing season) from remote sensing data, which substantially affects the estimates of total GPP over the vegetation growing season (Falge et al. 2002; Richardson et al. 2010). There is a need for better understanding and quantification of
vegetation phenology through analyses of both remote sensing data and CO 2 flux
data (ecosystem physiology approach). Vegetation phenology varies over years,
driven by interannual climate variability and climate change (e.g., temperature and
precipitation) (Piao et al. 2006, 2011), and is also affected by topography (Doktor
et al. 2009; Hwang et al. 2011; Piao et al. 2011). In addition, there are also
significant differences among various algorithms that use remotely sensed data to
retrieve vegetation phenology (Cong et al. 2012). Therefore, it is very important to
continue evaluation and development of satellite-based algorithms to retrieve
5 Gross Primary Production of Terrestrial Vegetation
143