APAR canopy ¼ FPAR canopy  PAR
ð5:4Þ
APAR chl ¼ FPAR chl  PAR
ð5:5Þ
APAR NPV ¼ FPAR NPV Â PAR
ð5:6Þ
The framework of PEMs was first proposed in the early 1970s as a function of
APAR and LUE (Monteith 1972, 1977). Therefore, GPP and NPP can be estimated
as a product of (1) the total amount of absorbed PAR, and (2) the radiation (light)use efficiency (e g for GPP, and e n for NPP) with the following formulas:
GPP ¼ e g  APAR
ð5:7Þ
NPP ¼ e n  APAR
ð5:8Þ
Based on the above-mentioned conceptual partitioning between FPAR chl and
FPAR npv , the PEMs can be divided into two groups. One group of PEMs uses
FPAR canopy to estimate GPP:
GPP ¼ e g  FPAR canopy  PAR
ð5:9Þ
The other group of PEMs uses FPAR chl to estimate GPP:
GPP ¼ e g  FPAR chl  PAR
ð5:10Þ
5.3 Methods
The development histories, key parameters, and variables of several wellestablished PEMs are summarized in Table 5.1. Variations among the different
PEMs appear in the calculation of FPAR, the estimation of the maximum LUE
parameter, and the use of scalars or downregulation factors.
Generally, FPAR canopy can be estimated with two methods by using (1) the leaf
area index (LAI) from in situ measurements at the site scale, or (2) satellitederived vegetation indices at the large scale. For large scale GPP modeling with
PEMs, the Normalized Difference Vegetation Index (NDVI, Tucker et al. 1979) is
often used to estimate FPAR canopy (Potter et al. 1993; Prince and Goward 1995;
Ruimy et al. 1999; Running et al. 2004; Veroustraete et al. 2002; Yuan et al. 2007)
with a simple empirical or physical function, as shown in Eqs. 5.11 and 5.12.
About the estimation of FPAR chl , Xiao et al. (2004a) first proposed using the
Enhanced Vegetation Index (EVI) (Huete et al. 1997) to estimate FPAR chl (see
Eqs. 5.13 and 5.14). By simulating FPAR canopy and FPAR chl using the radiative
transfer model, it was reported that EVI time series data are closer to the dynamic
of FPAR chl (Zhang et al. 2006). Recently, EVI and other chlorophyll vegetation
indices have increasingly been adopted by new PEMs (Gitelson et al. 2006; Potter
130
X. Xiao et al.
ð5:4Þ
APAR chl ¼ FPAR chl  PAR
ð5:5Þ
APAR NPV ¼ FPAR NPV Â PAR
ð5:6Þ
The framework of PEMs was first proposed in the early 1970s as a function of
APAR and LUE (Monteith 1972, 1977). Therefore, GPP and NPP can be estimated
as a product of (1) the total amount of absorbed PAR, and (2) the radiation (light)use efficiency (e g for GPP, and e n for NPP) with the following formulas:
GPP ¼ e g  APAR
ð5:7Þ
NPP ¼ e n  APAR
ð5:8Þ
Based on the above-mentioned conceptual partitioning between FPAR chl and
FPAR npv , the PEMs can be divided into two groups. One group of PEMs uses
FPAR canopy to estimate GPP:
GPP ¼ e g  FPAR canopy  PAR
ð5:9Þ
The other group of PEMs uses FPAR chl to estimate GPP:
GPP ¼ e g  FPAR chl  PAR
ð5:10Þ
5.3 Methods
The development histories, key parameters, and variables of several wellestablished PEMs are summarized in Table 5.1. Variations among the different
PEMs appear in the calculation of FPAR, the estimation of the maximum LUE
parameter, and the use of scalars or downregulation factors.
Generally, FPAR canopy can be estimated with two methods by using (1) the leaf
area index (LAI) from in situ measurements at the site scale, or (2) satellitederived vegetation indices at the large scale. For large scale GPP modeling with
PEMs, the Normalized Difference Vegetation Index (NDVI, Tucker et al. 1979) is
often used to estimate FPAR canopy (Potter et al. 1993; Prince and Goward 1995;
Ruimy et al. 1999; Running et al. 2004; Veroustraete et al. 2002; Yuan et al. 2007)
with a simple empirical or physical function, as shown in Eqs. 5.11 and 5.12.
About the estimation of FPAR chl , Xiao et al. (2004a) first proposed using the
Enhanced Vegetation Index (EVI) (Huete et al. 1997) to estimate FPAR chl (see
Eqs. 5.13 and 5.14). By simulating FPAR canopy and FPAR chl using the radiative
transfer model, it was reported that EVI time series data are closer to the dynamic
of FPAR chl (Zhang et al. 2006). Recently, EVI and other chlorophyll vegetation
indices have increasingly been adopted by new PEMs (Gitelson et al. 2006; Potter
130
X. Xiao et al.