15.4.1.1 Satellite Data
The satellite-based VPM uses two vegetation indices as input data: the enhanced vegetation index (EVI) and the land surface water index (LSWI)� These vegetation indices differ from the widely used NDVI (Equation 15�2)� NDVI is often applied in PEMs to estimate the
0.4140 0.5 0.6 0.7
FPARcanopy, FPARleaf, or FPARchl
0.8 0.9 1.0
160 180
Day of year (DOY)200 220 240
FPARcanopy in 2001 FPARcanopy in 2002 FPARcanopy in 2003 FPARleaf in 2001 FPARleaf in 2002 FPARleaf in 2003 FPARchl in 2001 FPARchl in 2002 FPARchl in 2003
FIgure 15.3
A comparison of the fractions of photosynthetically active radiation absorbed by canopy, leaf, and chlorophyll (FPARcanopy, FPARleaf, and FPARchl respectively), as illustrated in a deciduous broadleaf forest at the Harvard forest site, Massachusetts (see Zhang et al� 2005 for more details)�
vegetation productivity of terrestrial ecosystems (Field et al� 1995; Prince and Goward 1995; Nemani et al� 2003)� It is known that NDVI suffers several limitations, including sensitivity to atmospheric conditions, sensitivity to soil background (e�g�, soil moisture), and saturation of index values in multilayered and closed canopies (Xiao et al� 2004)� EVI directly adjusts the reflectance in the red band as a function of the reflectance in the blue band, accounting for residual atmospheric contamination (e�g�, aerosols), variable soil, and canopy background reflectance (Huete et al� 1997):
EVI NIR red
NIR red blue
= −
+ − +
G
C C L
( )
( )
ρ ρ
ρ 1ρ 2ρ
(15�9)
where G is 2�5, C1 is 6, C2 is 7�5, and L is 1, and ρNIR, ρred, and ρblue are land surface reflec- tances of the NIR, red, and blue bands, respectively�
Because the shortwave infrared (SWIR) spectral band is sensitive to vegetation water content and soil moisture, a combination of NIR and SWIR bands has been used to derive water-sensitive vegetation indices (Xiao et al� 2004b; Ceccato et al� 2002a,b; Ceccato et al�
2001)� LSWI is calculated as the normalized difference between NIR and SWIR spectral bands (Xiao et al� 2002):
LSWI NIR SWIR
NIR SWIR
= −
+
ρ ρ
ρ ρ (15�10)
where ρNIR and ρSWIR are reflectances of the NIR and the SWIR band, respectively�
Satellite images from two advanced optical sensors (vegetation onboard SPOT-4 satellite and MODIS onboard Terra satellite) have blue, red, NIR, and SWIR bands, which allow the calculation of EVI and LSWI indices� EVI and LSWI have now been used widely to char- acterize the growing conditions of vegetation (Zhang et al� 2003; Boles et al� 2004)�
15.4.1.2 Climate Data
The climate input data sets for the VPM include daily minimum temperature (°C), daily maximum temperature (°C), and the daily sum of PAR (mol/day)� The daily climate data come from either in situ measurements (e�g�, CO2 flux tower sites, weather stations) or cli- mate model simulations (e�g�, NCEP Reanalysis climate data), depending upon the avail- ability of climate data (Zhang et al� 2009; Zhao et al� 2005; Raich et al� 1991)�
15.4.2 estimation of Vegetation Photosynthesis Model Parameters 15.4.2.1 Light Absorption by Chlorophyll
In the VPM, FPARchl within the photosynthetically active period of vegetation is estimated as a linear function of EVI, and the coefficient, a, is set to be 1�0 (Xiao et al� 2004a,b):
FPARchl= ×α EVI (15�11)
15.4.2.2 Effect of Temperature on Gross Primary Production
Temperature affects photosynthesis; there are a number of ways to estimate the effect of temperature on photosynthesis (Tscalar)� In the VPM, Tscalar is estimated at each time step using the equation developed for the terrestrial ecosystem model (Raich et al� 1991):
T T T T T
T T T T
scalar = − −
− − −
( )( )
[( )( )]
min max
min max (T T− opt)2 (15�12)
where Tmin, Tmax, and Topt are the minimum, maximum, and optimum temperatures for photosynthetic activities, respectively� If air temperature falls below Tmin, Tscalar is set to be zero� The values of the Tmin, Tmax, and Topt parameters vary with vegetation types�
15.4.2.3 Effect of Water on Gross Primary Production
Wscalar, the effect of water on plant photosynthesis, has been estimated as a function of soil moisture and/or water vapor pressure deficit in a number of PEMs (Field et al� 1995;
Prince and Goward 1995; Running et al� 2000)� For instance, in the Carnegie Ames Stanford Approach (CASA) model, soil moisture was estimated using a one-layer bucket model (Malmstrom et al� 1997)� Soil moisture represents water supply to the leaves and canopy, and water vapor pressure deficit represents evaporative demand in the atmosphere� The leaf and canopy water content is largely determined by dynamic changes of both the soil moisture and water vapor pressure deficit�
The availability of time-series data of NIR and SWIR bands from the new generation of advanced optical sensors (e�g�, variable geometry turbocharger, MODIS) offers opportuni- ties for quantifying the canopy water content at large spatial scales through both the vege- tation indices approach (Ceccato et al� 2002a) and the radiative transfer modeling approach (Zarco-Tejada et al� 2003)� Vegetation indices that are based on NIR and SWIR bands are sensitive to changes in equivalent water thickness (g/cm2) at the leaf and canopy levels (Ceccato et al� 2002a,b; Ceccato et al� 2001; Hunt and Rock 1989)� As a first-order approxi- mation, the VPM uses a satellite-derived water index to estimate the seasonal dynamics of Wscalar:
Wscalar LSWI
= +LSWI + 1
1 max
(15�13)
where LSWImax is the maximum LSWI value within the plant growing season for indi- vidual pixels�
15.4.2.4 Effect of Leaf Age and Phenology on Gross Primary Production
The leaf age affects the seasonal patterns of photosynthetic capacity and NEE in decidu- ous forest (Wilson et al� 2001)� Turner et al� (Turner et al� 2003) compared daily LUE from four CO2 flux tower sites: an agriculture field, a tall grass prairie, a deciduous broadleaf forest, and a boreal forest� Their results suggested that parameters on cloudiness and the
phenological status of vegetation should be included in modeling vegetation primary pro- duction� In the VPM, Pscalar is used to account for the effect of the leaf age on photosynthesis at the canopy level� Calculation of Pscalar is dependent upon leaf longevity (deciduous ver- sus evergreen)� For a canopy that is dominated by leaves with a life expectancy of 1 year (one growing season, e�g�, deciduous trees and shrubs), Pscalar is calculated at two different phases:
Pscalar LSWI
(From bud-burst to complete
= +1
2 leaf expansion) (15�14)
Pscalar=1(After complete leaf expansion) (15�15)
Evergreen trees and shrubs have a green canopy throughout the year, because foliage is retained for several years� The canopy of evergreen forests is thus composed of green leaves of various ages� To deal with different age classes in evergreen forest canopies, fixed turnover rates of foliage of evergreen forests at the canopy level have been used in some process-based ecosystem models (Aber and Federer 1992; Law et al� 2000)� For evergreen forests, we simply assume Pscalar = 1 (Xiao et al� 2004b); we also assume this for tundra, grassland, and cropland (e�g�, wheat) vegetation, which have new leaves emerging through most of the plant growing season (Li et al� 2007)�
15.4.2.5 Maximum LUE
LUE (εg) is affected by temperature, water, and leaf phenology:
εg=ε0×Tscalar×Wscalar×Pscalar (15�16)
where ε0 is the apparent quantum yield or the maximum LUE (μmol CO2/μmol PPFD), and Tscalar, Wscalar, and Pscalar are the scalars for the effects of temperature, water, and leaf phenology on the LUE of vegetation, respectively� A full description of the VPM is given elsewhere (Xiao et al� 2004b; Xiao et al� 2005a)�
The maximum LUE (ε0) for individual vegetation types can be estimated from the non- linear analysis of the observed half-hourly NEE and incident PAR data from eddy cova- riance flux tower sites� In the VPM, the ecosystem-level ε0 values vary with vegetation types� The Michaelis–Menten function (Equation 15�17) is used to estimate the ε0 values of individual vegetation types; half-hourly NEE and PAR data for weekly to 10-day periods within the peak period of the plant growing season (e�g�, from July to August) are used:
NEE PPFD GPP
PPFD GPP e
= × ×
× + −
α
α maxmax
R (15�17)
where α is the maximum LUE or apparent quantum yield (as photosynthetic photon flux density [PPFD] approaches 0), GPPmax is the maximum gross ecosystem exchange, and Re
is the ecosystem respiration� The estimated α value is used as an estimate of the ε0 param- eter in the VPM�
15.4.3 Model evaluation
Evaluating GPP estimates at the canopy level is a challenging task� Recent progress in partitioning the observed NEE data into GPP and Re makes it possible to directly evaluate GPP estimates from various models� Daily GPP and Re flux data at individual flux sites are generated from half-hourly NEE flux data by the CO2 eddy covariance flux community (Baldocchi et al� 2001; Mizoguchi et al� 2009)�