LAI, APARcanopy, and FPARcanopy have all been a focus of both the ecology and the remote sensing communities over the past few decades� A number of remote sensing studies have been conducted to develop quantitative relationships between the NDVI and LAI, and between the NDVI and FPARcanopy (Prince and Goward 1995; Ruimy et al� 1999)� Approaches based on the NDVI-LAI and NDVI-FPARcanopy relationships have been the dominant para- digm at the crossroads of the fields of remote sensing science and ecology, for example, satellite-based PEMs (Figure 15�2)�
A number of satellite-based PEMs use the concept of FPARcanopy to estimate the GPP and NPP (Potter et al� 1993; Field et al� 1995; Prince and Goward 1995; Running et al� 2004)� GPP is calculated as follows:
GPP=εg×FPARcanopy×PAR (15�3)
where εg is the LUE for photosynthesis or GPP� Brief descriptions of two models are provided in Sections 15�2�1 and 15�2�2�
15.2.1 global Production efficiency Model
The global production efficiency model (GLO-PEM) estimates both the GPP and NPP based on the production efficiency approach (see Equation 15�3)� It has several linked components that describe the processes of canopy radiation absorption, utilization, autotrophic respira- tion, and the regulation of these processes by environmental factors (Prince and Goward 1995; Goetz et al� 2000)� The GLO-PEM uses NDVI to estimate FPARcanopy (see Goward and Huemmrich 1992 for more details):
FPARcanopy =1 08� ×NDVI−0 08� (15�4)
EVI
Partition of chlorophyll and NPV within a leaf and canopy
NDVI
=(ρNIR–ρred)/(ρNIR+ρred) Leaf and canopy water contentLSWI=(ρNIR–ρSWIR)/(ρNIR+ρSWIR) LAI-centered algorithms
GPP= εg×FPARcanopy×PAR FPARcanopy=a×(1 – e-k×LAI) FPARcanopy=f(NDVI) NDVI=f(LAI)
Chlorophyll-centered algorithms
FPARcanopy=FPARChl+FPARNPV GPP= εg×FPARchl×PAR Canopy=Chl+NPV
FPARchl=f(EVI)
Leaf area index (LAI)
EVI=2.5× ρNIR–ρred
ρNIR+6× ρred– 7.5× ρblue+1
FIgure 15.2
A simple comparison between two paradigms of production efficiency models� GPP = gross primary produc- tion; NDVI = normalized difference vegetation index; FPAR = fraction of photosynthetically active radiation;
PAR = photosynthetically active radiation; EVI = enhanced vegetation index; LSWI = land surface water index;
Chl = chlorophyll; NPV = nonphotosynthetic vegetation; NIR = near-infrared; and SWIR = shortwave infrared�
In the GLO-PEM, εg is estimated through a modeling approach based on plant physio- logical principles (Prince and Goward 1995)� Plant photosynthesis depends on both the capacity of the photosynthetic enzymes to assimilate CO2 (Collatz et al� 1991; Farquhar et al� 1980) and the stomatal conductance of CO2 from the atmosphere into the intercellu- lar spaces (Harley et al� 1992)� These two processes are affected by environmental factors, such as air temperature, water vapor pressure deficit, soil moisture, and atmospheric CO2 concentration� Detailed descriptions of approaches for modeling εg have been provided in many earlier publications (Prince and Goward 1995; Collatz et al� 1991; Goetz and Prince 1998; Collatz et al� 1992; Goetz and Prince 1999)�
15.2.2 MODIS Daily Photosynthesis Model
The photosynthesis (PSN) model uses Equation 15�3 to estimate GPP, but εg and FPARcanopy are derived using different methods (Running et al� 2004; Running et al� 1999; Running et al� 2000)� FPARcanopy is produced as a part of the MOD15 (LAI and FPAR) product suite�
In MOD17, a set of biome-specific maximum LUE parameters is extracted from the biome properties lookup table (Running et al� 2000)�
εg =ε0×Tscalar×Wscalar (15�5)
where ε0 is the maximum LUE, Tscalar is estimated as a function of daily minimum tempera- ture, and Wscalar is estimated as a function of daylight average water vapor pressure deficit�
In this approach, biome is defined according to the MODIS land-cover product (MOD12) (Running et al� 2004; Running et al� 2000; Friedl et al� 2002)�
15.3 Chlorophyll, Light Absorption by Chlorophyll, and FPARchl
From the biochemical perspective, vegetation canopies are composed of chlorophyll (chl) and nonphotosynthetic vegetation (NPV)� The latter includes both canopy-level (e�g�, stem, senescent leaves) and leaf-level (e�g�, cell walls, vein, and other pigments) materi- als� Therefore, FPARcanopy should be partitioned into FPARchl and FPAR absorbed by NPV (FPARNPV) (Xiao et al� 2004a,b; Xiao et al� 2005a)�
Canopy chlorophyll NPV= + (15�6)
FPARcanopy =FPARchl+FPARNPV (15�7)
How much difference is there between FPARcanopy and FPARchl in a vegetation canopy?
Does the difference between FPARcanopy and FPARchl change over the plant growing sea- son? Using a radiative transfer model (PROSAIL2) and daily MODIS data, results from tem- perate deciduous forests (Zhang et al� 2005; Zhang et al� 2006) have shown that FPARcanopy is significantly larger than FPARchl, and the difference between FPARcanopy and FPARchl changes as much as 30%–40% over the plant growing season (Figure 15�3)�
As shown in Figure 15�1, photosynthesis starts with light absorption by leaf chlorophyll�
Only the PAR absorbed by chlorophyll (product of PAR × FPARchl) is responsible for pho- tosynthesis or GPP� Based on the conceptual partitioning of chlorophyll and NPV within a leaf and canopy, the VPM was developed for estimating GPP over the photosynthetically active period of vegetation (Xiao et al� 2004a)� The VPM is briefly described as follows:
GPP=εg×FPARchl×PAR (15�8)
This biochemical approach, based on the chlorophyll–FPARchl relationship, is currently an emerging paradigm in the field of remote-sensing–based PEMs, and other additional models have been developed using the FPARchl concept (Sims et al� 2008; Sims et al� 2006;
Mahadevan et al� 2008)� Figure 15�2 summarizes the major differences between FPARcanopy and FPARchl approaches in estimating light absorption and GPP�