SEARCH

SEARCH BY CITATION

Keywords:

  • fraction of absorbed photosynthetically active radiation (FAPAR);
  • light-use efficiency;
  • net ecosystem exchange;
  • production efficiency models;
  • respiration;
  • vegetation indices.

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • The fraction of absorbed photosynthetically active radiation (FAPAR) is a key vegetation biophysical variable in most production efficiency models (PEMs). Operational FAPAR products derived from satellite data do not distinguish between the fraction of photosynthetically active radiation (PAR) absorbed by nonphotosynthetic and photosynthetic components of vegetation canopy, which would result in errors in representation of the exact absorbed PAR utilized in photosynthesis.
  • The possibility of deriving only the fraction of PAR absorbed by photosynthetic elements of the canopy (i.e. FAPARps) was investigated.
  • The approach adopted involved inversion of net ecosystem exchange data from eddy covariance measurements to calculate FAPARps. The derived FAPARps was then related to three vegetation indices (i.e. Normalized Difference Vegetation Index (NDVI), Medium Resolution Imaging Spectrometer (MERIS) Terrestrial Chlorophyll Index (MTCI) and Enhanced Vegetation Index (EVI)) in an attempt to determine their potential as surrogates for FAPARps. Finally, the FAPARps was evaluated against two operational satellite data-derived FAPAR products (i.e. MODIS and CYCLOPES products).
  • The maximum FAPARps from the inversion approach ranged between 0.6 and 0.8. The inversion approach also predicted site-specific Q10-modelled daytime respiration successfully (R2 > 0.8). The vegetation indices were positively correlated (R2 = 0.67–0.88) to the FAPARps. Finally, the two operational FAPAR products overestimated the FAPARps. This was attributed to the two products deriving FAPAR for the whole canopy rather than for only photosynthetic elements in the canopy.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The fraction of absorbed photosynthetically active radiation absorbed by plant canopies (FAPAR) is a critical biophysical variable for extrapolating ecophysiological measurements from leaf to landscape scale (Asner et al., 1998). The absorbed PAR represents the available light energy for plant productivity and is therefore the key variable influencing photosynthesis, transpiration and energy balance in most production efficiency models (PEMs; Prince & Goward, 1995; Ruimy et al., 1999; Myneni et al., 2002; Morisette et al., 2006; Gobron et al., 2006). In situ measurement of FAPAR requires simultaneous measurement of photosynthetically active radiation (PAR) above and below a canopy, as well as estimation of the canopy architecture information to account for the nonleaf absorptions (Gower et al., 1999; Huemmrich et al., 2005; Gobron & Verstraete, 2009). This approach is often difficult to implement as it is time-consuming and limited in spatial coverage. Therefore, spatially continuous FAPAR is often derived by either developing empirical relationships with satellite sensor-derived spectral vegetation indices (e.g. Normalized Difference Vegetation Index (NDVI); Prince & Goward, 1995; Fensholt et al., 2004) and leaf area index (LAI; Ruimy et al., 1999) or through inversion of physically based radiative transfer models using satellite remote sensing data as constraints (Knyazikhin et al., 1998; Myneni et al., 2002; Deng et al., 2006; Gobron et al., 2006; Baret et al., 2007).

A number of FAPAR products have been developed from spectral data acquired through satellite sensors (e.g. the Moderate Resolution Imaging Spectroradiometer (MODIS) FAPAR (Knyazikhin et al., 1998); Medium Resolution Imaging Spectrometer (MERIS)-MGVI (Gobron et al., 2002); and CYCLOPES FAPAR (Baret et al., 2007)). However, these FAPAR products often represent the fraction of photosynthetically active radiation absorbed by the whole canopy (i.e. FAPARcanopy), while the vegetation canopy is composed of both photosynthetically active vegetation (PAV) and nonphotosynthetic vegetation (NPV, e.g. branches, stem, senescent foliage; Asner et al., 1998; Zhang et al., 2005, 2009). As the fraction of PAR absorbed by NPV is not utilized in photosynthesis, calculating FAPARcanopy would result in overestimation of the actual FAPAR utilized in initiating photosynthesis in the photosystems (i.e. both photosystem I (PSI) and photosystem II (PSII); hereafter referred to as FAPARps). Asner et al. (1998) reported that in forests with LAI < 3, the nonphotosynthetic components of the canopy (e.g. stem) increased the actual canopy FAPAR by 10–40%. This overrepresentation of FAPAR absorbed by photosynthetic components of the canopy would lead to errors in terrestrial vegetation primary productivity estimates predicted by PEM models that use these datasets. Therefore, there is a need to formulate new approaches to derive only FAPAR absorbed by the photosynthetic components of the canopy (i.e. FAPARps).

Although FAPARps is a crucial input for quantifying plant productivity, it is often ignored or generalized by many researchers when accounting for the amount of PAR used for photosynthesis in PEM at regional to global scales. This is mainly because of a lack of adequate techniques/data to derive this variable at large spatial scales. As measuring FAPARps in the field is quite challenging, few studies have attempted to derive FAPARps. For example, Zhang et al. (2005) used simulated data from a coupled leaf and canopy radiative transfer model (i.e. PROSAIL-2; Jacquemoud & Baret, 1990; Braswell et al., 1996) and MODIS reflectance data to derive different components of FAPAR (i.e. FAPARcanopy, FAPARleaf and FAPARchlorophyll). Their study showed that the estimates of FAPARchlorophyll were indeed lower than those of FAPARcanopy. Hanan et al. (2002) used an inversion approach to derive FAPAR absorbed by canopy chlorophyll from net ecosystem exchange (NEE; CO2) data from eddy covariance flux tower measurements. However, these studies did not suggest how FAPARps could be generated at large spatial scales.

This paper presents an attempt to derive FAPARps using NEE data measured using the eddy covariance method. The derived FAPARps data were then related to contrasting spectral vegetation indices (i.e. two multi-band based vegetation indices – the NDVI (Rouse et al., 1973) and the Enhanced Vegetation Index (EVI; Huete et al., 2002); and a red-edge based vegetation index, the MERIS Terrestrial Chlorophyll Index (Dash & Curran, 2004)) derived from satellite sensor data. The aim of this exercise was to determine the potential of using the vegetation indices as surrogates of FAPARps at large spatial scales. Finally, the paper compares the estimated FAPARps with two operational FAPAR products (i.e. MODIS FAPAR and CYCLOPES FAPAR) with the aim of determining the disparity between these products and the FAPARps.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study sites

The inversion algorithm to derive FAPARps, ecosystem light-use efficiency (LUE) and respiration was applied to eddy covariance data acquired at two contrasting Ameriflux tower sites (i.e. at the Mead irrigated site and Harvard Forest). The Harvard Forest flux tower site (42.5378ºN, 72.1715ºW, elevation 340 m) is located in western Massachusetts, in the USA. It primarily consists of mature deciduous broadleaf forest. The area around the tower is dominated by red oak (Quercus rubra L.), and red maple (Acer rubrum L.), with scattered stands of Eastern hemlock (Tsuga canadensis (L.) Carr.), black birch (Betula lenta L.), and white pine (Pinus strobus L.; Urbanski et al., 2007). The Mead irrigated site (41.1651ºN, 96.4766ºW, elevation 361 m) is located in Mead, NE, USA. The site is under maize crop and is irrigated with a centre pivot system and is large enough to provide sufficient upwind to measure mass and energy fluxes using eddy covariance techniques (Verma et al., 2005).

Eddy covariance flux tower data

The eddy covariance method has become the main approach for sampling ecosystem carbon, water, and energy fluxes from hourly to interannual timescales (Baldocchi et al., 2001). At the flux tower sites, a number of variables, including fluxes of CO2, sensible heat, latent heat and momentum, mean air temperature, humidity, horizontal wind speed, incident and reflected solar radiation, PAR, net radiation, soil heat flux, and soil temperature, are measured using various methods and instruments (Baldocchi et al., 2001; Baldocchi, 2003). Five years of data (2001–2005) on NEE (daytime), PAR, mean air temperature, vapour pressure deficit (VPD), and the Q10-modelled daytime ecosystem respiration (Reco) for the two sites (i.e. Mead irrigated cropland and Harvard Forest) used in this study were acquired from the Ameriflux website (http://public.ornl.gov/ameriflux).

Estimation of the fraction of PAR absorbed by photosynthetic elements of the canopy (FAPARps) through inversion of eddy covariance NEE data

Description of the inversion approach

The typical relationship between the rate of photosynthesis (and, in turn, CO2 uptake) by C3 and C4 species and the incident PAR is often described as a nonrectangular hyperbola (Lambers et al., 2006). As light intensity increases, photosynthesis levels off towards the asymptote, especially in C3 plants. The inversion methodology used in this paper is based on the concept that at a light intensity below the saturation point, there is a near linear relationship between incidence PAR and CO2 uptake. Therefore, the implementation of the inversion approach requires a maximum light limit to be imposed on photosynthesis. This limit can be adjusted to suit the climatology of the area under investigation and the photosynthetic capacity of plants in that ecosystem. The NEE (μmol m−2 s−1) estimated using the eddy covariance techniques can be represented as follows (Hanan et al., 2002):

  • display math(Eqn 1)

where αe (mol mol−1) represents the ecosystem LUE (the number of moles of CO2 fixed per mole of PAR incident on the canopy) and Reco (μmol m−2 s−1) is the whole-ecosystem respiration. For PAR below the light saturation point, (Eqn 1) can be used to estimate Reco and αe by regression of measured CO2 flux against incident PAR. In this regression, the slope of the relationship has been shown to represent the ecosystem LUE and the intercept (PAR = 0) represents the whole-ecosystem respiration (Suyker & Verma, 2001; Hanan et al., 2002). The ecosystem LUE term (αe) can further be divided into a physiological component (α, the intrinsic quantum yield) that can be estimated and an unknown structural component which represents the efficiency of absorption by photosynthetic components of canopy (i.e. the FAPARps; Hanan et al., 2002). The relationship in (Eqn 1) can be rewritten as:

  • display math(Eqn 2)

where αa (mol mol−1) is the ‘actual quantum yield’ (the number of moles of CO2 fixed per mole of PAR absorbed by photosynthetic elements in the canopy). The first part of the equation (i.e. PAR × FAPARps × αa) represents gross primary productivity (GPP). In C4 plants where Rubisco oxidation (i.e. addition of oxygen to the Rubisco enzyme through photorespiration, hence reducing photosynthesis efficiency) is minimal, the actual quantum yield is similar to the intrinsic yield of photosynthesis. In the C3 plants where Rubisco oxidation occurs, the actual quantum yield depends on temperature and leaf internal CO2 concentrations (Ehleringer et al., 1997; Hanan et al., 2002). Therefore, in mixed canopies, the NEE can be written as:

  • display math(Eqn 3)

where PC3 is the proportion of C3 species in the canopy, α3 is the intrinsic quantum yield of C3 species, Ψe (unitless) is a function of temperature and leaf internal CO2 concentration in C3 species, and α4 is the intrinsic quantum yield of C4 species. In both the C3 and C4 plants, the quantum yield is often influenced by VPD, with the effect being more pronounced during the hours around midday (Farquhar & von Caemmerer, 1982; Tezara et al., 1999). The mechanism for this influence is not agreed on, but it could be because evaporation at high VPD causes water stress (Shirke & Pathre, 2004). The water stress could affect several components of photosynthetic metabolism (e.g. electron transport, ATP synthesis, light dissipation, Rubisco, and carbohydrate metabolism), hence reducing photosynthesis efficiency (Lawlor & Cornic, 2002). The influence of VPD on quantum yield can be represented as follows:

  • display math(Eqn 4)

where VPD (KPa) is the instantaneous (e.g. hourly) value. This formulation is based on measurements of stomatal conductance in tropical, temperate and boreal ecosystems on crops, grasses, shrubs deciduous forests and evergreen forests (Aber & Federer, 1992; Hollinger et al., 1994; Leuning, 1995). It has been shown that high VPD often only occurs for a short period around the midday hours and hence its stress effects may not persist during the rest of the day (Monteith, 1995; Shirke & Pathre, 2004). As such, the influence of VPD on the quantum yield was restricted to the hours around midday (10:00–14:00 h) and the mean VPD data at these times of the day was used in parameterizing the ƒD term. To include effects of VPD, (Eqn 3) is rewritten as follows:

  • display math(Eqn 5)

Regression between NEE and PAR can be used to estimate the slope term (αe) and by using the terms in (Eqn 5), FAPARps can be estimated as:

  • display math(Eqn 6)

The maximum intrinsic quantum yield terms for α3 (i.e. 0.08 mol mol−1) and α4 (i.e. 0.06 mol mol−1) were used to parameterize α in (Eqn 6) (Collatz et al., 1991, 1992; Hanan et al., 2002). Laboratory measurements and kinetic analyses using chloroplast suspensions have shown that these values accurately describe the maximum intrinsic quantum yield of C3 and C4 plants, respectively (Collatz et al., 1991, 1992). The Ψe term in Eqns 3, 5 and 6 describes the influence of leaf temperature (T1) and leaf chloroplast CO2 partial pressure (Ci) on actual quantum yield in C3 plants. The Ψe term is derived as a function of the response of C3 photosynthesis to CO2 partial pressure and compensation point, and the leaf temperature as follows (Ehleringer & Bjorkman, 1977; Collatz et al., 1991; Hanan et al., 1998):

  • display math(Eqn 7)

where Pi is the CO2 partial pressure, (Pi = leaf CO2 concentration (Ci) × atmospheric pressure (in Pa)), and Γ* is the CO2 compensation point, which depends on leaf temperature (T1) and oxygen concentration [O2]. The Γ* term can be derived as follows:

  • display math(Eqn 8)

where O2 is the oxygen partial pressure in the chloroplast (c. 20 900 Pa) and τ is the relative specificity of ribulose 1,5 bisphosphate to CO2 vs O2. Refer to Hanan et al. (2002) for information on how to calculate the τ term. The responses of Ψe to temperature and CO2 across a range of likely leaf temperatures and CO2 concentrations(Ci), ranging from near ambient (350 mol mol−1) to below ambient (100 mol mol−1), were calculated by Hanan et al. (2002).

The polynomial equation derived for 275 mol mol−1 leaf CO2 concentration (Hanan et al., 2002) was used to parameterize Ψe. It was shown that at this leaf CO2 concentration, there was less likelihood of saturation in the leaves as a result of excessive drawing down of CO2 from the atmosphere (Hanan et al., 2002). Therefore, during the implementation of (Eqn 6) to calculate FAPARps, the term Ψe was expressed as:

  • display math(Eqn 9)

Mean daily air temperature from the flux tower sites was used to parameterize the values of T in the equation. It was assumed that the daily mean air temperature at the flux tower sites was not very divergent from the leaf temperature, and it has also been shown that the difference between them is often minimal (e.g. < 2°C; Field & Mooney, 1986; Dillaway & Kruger, 2011). However, the influence of varying the mean temperature on the calculation of FAPARps was tested during sensitivity analysis.

Implementation of the inversion approach

The inversion process was performed by regression of the measured CO2 flux (NEE) against incident PAR to estimate ecosystem LUE (αe) as the slope and whole ecosystem respiration (Reco) as the intercept (see (Eqn 1)). During the inversion of NEE data, it is necessary to set an upper limit for the incident PAR so as to ensure that the photosynthetic capacity of the plants under consideration is not inhibited by light saturation. A maximum incident PAR limit of 550 μmol m−2 s−1 was imposed at the Harvard Forest study site and 1000 μmol m−2 s−1 was imposed at the Mead irrigated study site. These values were determined by plotting the 5 yr of NEE against PAR for the two sites and then taking the upper limit where the relationship ceases to be linear (Fig. 1). The coefficient of determination (R2) of the relationship between NEE and PAR was calculated for the entire growing season for the study period. The inversion process was performed for each day of the 5 yr study period using a 7 d moving-average window of measurements centred on the day in question and a standard least-squares linear regression procedure (Sokal & Rohlf, 1981; Hanan et al., 2002). As no in situ FAPARps data was available, an indication of the performance of the inversion process was gauged by comparing the respiration data from the inversion procedure with Q10-modelled respiration data at the two flux tower sites. The sensitivity of FAPARps to changes in the input variables (i.e. maximum PAR, mean temperature, VPD and maximum intrinsic quantum yield) was evaluated.

image

Figure 1. Relationship between net ecosystem exchange (NEE) and photosynthetically active radiation (PAR) for the Harvard Forest (a) and the Mead site (b); and the coefficients of determination (R2) from the 7 d moving window regression for the Harvard Forest (c) and the Mead site (d). Dotted vertical lines in (a) and (b) represent the upper PAR limit used in this study. DOY, day of the year.

Download figure to PowerPoint

Satellite-derived vegetation indices

The relationship between three vegetation indices (i.e. NDVI (Rouse et al., 1973); EVI (Huete et al., 2002); and the MTCI (Dash & Curran, 2004)) and the estimated FAPARps was evaluated at the two flux tower sites. The NDVI is a numerical indicator of whether a target has live green vegetation (Rouse et al., 1973). Previous studies have shown NDVI to be nearly linearly related to FAPAR (Hall et al., 1992). NDVI is calculated as the ratio between reflectance values in the red and near-infrared bands:

  • display math(Eqn 10)

where NIR and Red are reflectances in the near-infrared and red bands of the electromagnetic spectrum. respectively. The MODIS surface reflectance data acquired from the MODIS products website (http://modis.gsfc.nasa.gov/data/dataprod/) were used to calculate NDVI at 8 d temporal resolution.

The EVI was developed to optimize the vegetation signal with improved sensitivity in high biomass regions and improved vegetation monitoring by decoupling the effects of background signal and a reduction in atmosphere influences (Huete et al., 2002). The EVI is calculated as:

  • display math(Eqn 11)

where NIR, Red and Blue are atmospherically corrected surface reflectances in the near-infrared, red and blue bands, respectively (Huete et al., 2002). MODIS surface reflectance data acquired from the MODIS products website (http://modis.gsfc.nasa.gov/data/dataprod/) were used to calculate EVI at 8 d temporal resolution.

The MTCI is calculated as the surrogate index for the red edge position (REP) of the vegetation spectral reflectance and provides information on canopy chlorophyll content (i.e. a product of LAI and leaf chlorophyll concentration; Dash & Curran, 2004, 2007). The MTCI is calculated using reflectance data from the red edge position of the MERIS sensor data as follows (Dash & Curran, 2004):

  • display math(Eqn 12)

where R753.75, R708.75, and R681.25 are reflectances in the centre wavelengths of the MERIS standard band setting. The operational MTCI data is produced at 1 km2 spatial resolution and 8 d temporal resolution. The MTCI data were provided by the European Space Agency (ESA) and processed by Astrium Geo information services and acquired from the NERC Earth Observation Data Centre (http://www.neodc.rl.ac.uk).

FAPAR products

Two FAPAR products (i.e. the MODIS and CYCLOPES FAPAR products) were evaluated against the FAPARps derived through the inversion of NEE. Biome maps and atmospherically corrected MODIS spectral reflectances are used in the inversion of a radiative transfer model to retrieve the FAPAR product at 1 km2 spatial resolution (Knyazikhin et al., 1998; Myneni et al., 2002). Collection 5 MODIS FAPAR data at 8 d temporal resolution and 1 km spatial resolution were acquired from the Earth Resources Observation System (EROS) Data Centre Distributed Active Archive Centre (https://lpdaac.usgs.gov/lpdaac/products/). The CYCLOPES FAPAR product was generated using top of canopy reflectance values of the SPOT/VEGETATION sensor based on a neural network approach and is provided at 1/112° (c. 1 km2 at equator) ground sampling distance and a 10 d temporal sampling, in a Plate Carrée projection (Baret et al., 2007). Version 3.1 CYCLOPES FAPAR data were acquired from Medias France (http://postel.mediasfrance.org).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Relationship between PAR and NEE from in situ eddy covariance measurements

As expected, the relationship between NEE and incoming PAR at the two study sites (Fig. 1) shows an initial linear relationship and then becomes asymptotic after light saturation points. The Harvard Forest site had a lower light saturation point (c. 550 μmol m−2 s−1) than that at the Mead site (c. 1000 μmol m−2 s−1). These values were used as the maximum PAR values for each of the sites during the inversion process. Fig. 1(c,d) represents the coefficient of determination (R2) between the PAR and NEE across all the evaluated years at the two sites. The R2 values were relatively high (R2 > 0.6) at the two sites, indicating good correlation between the two parameters. Higher R2 is envisaged if NEE is compared with PAR for only the portion where the relationship is linear.

Ecosystem respiration (Reco)

Fig. 2 shows the relationship between the 7 d mean respiration data for the study period (2001–2005) predicted by the inversion algorithm and the mean of Q10-modelled respiration from night-time variation in NEE at the two study sites. The strong positive correlation (i.e. R2 = 0.81 at Harvard Forest (Fig. 2a), and R2 = 0.95 at the Mead irrigated cropland site (Fig. 2b)) and low root mean square error (RMSE) values indicate that the inversion approach can be used to predict daytime ecosystem respiration at the two sites. The inverted respiration values were slightly lower than the Q10-modelled respiration, suggesting a slight underprediction of respiration for the two sites by the inversion process.

image

Figure 2. Relationship between predicted respiration and in situ measured respiration for the Harvard deciduous broadleaf forest site (a) and the Mead irrigated crop site (2001–2005) (b). RMSE, root mean square error.

Download figure to PowerPoint

Evaluations of the fitted intercept (i.e. the ‘whole ecosystem respiration’, Reco) from (Eqn 1) for the two study sites (i.e. Harvard Forest and Mead irrigated cropland) are shown in Fig. 3(a), and Fig. 3(b) respectively. The fitted whole-ecosystem respiration tracked the seasonal growth patterns at the two sites relatively well. The lowest daytime respiration at both sites was experienced during winter months, while the highest respiration was during the summer months.

image

Figure 3. Fitted intercept (‘whole ecosystem respiration’, Reco) for the Harvard deciduous broadleaf forest site (a) and the Mead irrigated crop site (b) (black dots represent the 8 d mean respiration, and grey shading represents the standard error).

Download figure to PowerPoint

The rates of respiration for biomes located in temperate climatic regions are generally dependent on temperature (Atkin et al., 2000; Hanan et al., 2002; Atkin & Tjoelker, 2003). At both sites, respiration rates were shown to increase with increasing average temperature (Fig. 4a,b). However, there were differences in the pattern of evolution of rates of respiration with temperature between the two sites. At the Harvard Forest site (Fig. 4a), the increment of respiration with temperature was steady from c. −5°C to 20°C, whereas at the Mead cropland site (Fig. 4b), the increment was minimal between −5 and 10°C; thereafter, the respiration rates steeply increased with rising average air temperature. The steady rise in respiration rates at the Harvard Forest site represents a steady rise in green leaf area and root activity (Atkin et al., 2000; Hanan et al., 2002; Atkin & Tjoelker, 2003), whereas the steeper rise at the Mead irrigated site represents the fact that planting does not take place until the temperatures are > 10°C and then, once the maize crop has germinated (around the end of May), the leaves undergo a rapid increase in leaf area – hence the steeper respiration curve.

image

Figure 4. Relationship between predicted respiration and average temperature for the measurements used in the regression for the Harvard deciduous broadleaf forest site (a) and the Mead-irrigated crop site (2001–2005) (b).

Download figure to PowerPoint

Ecosystem LUE (αe)

Fig. 5 displays the fitted slope (representing realized ‘whole ecosystem LUE’, αe), derived from the inversion of the regression between incoming PAR and NEE for the Harvard deciduous broadleaf forest (Fig. 5a) and the Mead irrigated crop site (Fig. 5b) for 5 yr (2001–2005). As expected, the ecosystem LUE increased in spring, peaked in summer, fell in the autumn, and was at a minimum during the winter season.

image

Figure 5. Fitted slope (‘whole ecosystem light-use efficiency’, αe) of the regression between net ecosystem exchange (NEE) and photosynthetically active radiation (PAR) for the Harvard deciduous broadleaf forest site (a) and the Mead irrigated crop site (b) (black dots, slope; grey shading, standard error).

Download figure to PowerPoint

Even though the two biomes are characterized by different plant functional types, the predicted ecosystem LUEs were similar. Theoretically, C4 plants should have higher LUE or quantum yield than C3 plants. However, because of additional energy requirements to regenerate phosphoenolpyruvate from pyruvate in the operation of the C4 cycle (Farquhar & von Caemmere, 1982), they tend to have lower LUE than the C3 plants. The LUE of C3 plants has been shown to be reduced from maximum values by photorespiration, and as photorespiration is temperature-dependent, the LUE of C3 plants declines with increasing temperature (Osborne & Garrett, 1983; Long et al., 1993; Ehleringer et al., 1997). Therefore, the high photon requirements of C4 plants and the photorespiration impact on quantum yield in C3 plants offset each other, resulting in similar ranges of LUE as observed in Fig. 5.

The fraction of photosynthetically active radiation (PAR) absorbed by photosynthetic elements in the canopy (FAPARps)

Fig. 6 shows the estimated FAPARps derived through the inversion of NEE data for the Harvard Forest (Fig. 6a) and the Mead irrigated site (Fig. 6b) using (Eqn 6).

image

Figure 6. Estimated fraction of photosynthetically active radiation absorbed by photosynthetic elements in the canopy (FAPARps) estimated by inversion of net ecosystem exchange data (2001–2005) for the Harvard Forest (a) and the Mead irrigated cropland site (b).

Download figure to PowerPoint

The FAPARps at the two sites increased in spring, peaked in summer, fell in autumn and was at a minimum during the winter season. The changes in FAPARps tracked the seasonal growth patterns of the two study sites with minimal interannual variations across the 5 yr. Between the sites, the Harvard Forest site had slightly lower FAPARps values than those at the Mead irrigated crop site. However, the FAPARps curves were narrower in the Mead irrigated site, indicating that the inversion approach captured the relatively shorter growing season for cultivated crops compared with the longer growing season of natural vegetation at the Harvard site.

Sensitivity of FAPARps to input variables

Fig. 7 shows the variation of FAPARps with changes in the input variables (maximum PAR, mean temperature, mean VPD and maximum intrinsic quantum yield) in the inversion process at the two study sites. For this task, each of the input variables was changed incrementally, keeping the rest constant. Reduction of maximum PAR at the two sites resulted in higher FAPARps (Fig. 7a,e) and had greater influence on the FAPARps at the Harvard site than at the Mead site. The Mead site is dominated by maize, whose photosynthetic rate has been shown not to saturate with increasing PAR (Crafts-Brandner & Salvucci, 2002). The Harvard site, on the other hand, is dominated by C3 plants which saturate with increasing PAR values (Fig. 1). The influence of changes in the mean temperature and mean VPD on FAPARps was minimal at both sites. Increasing the maximum intrinsic quantum yields at the two sites resulted in low FAPARps values, with the effect more pronounced at the Harvard site. Overall, the calculation of FAPARps using the inversion process seems to be more sensitive to two variables (i.e. the intrinsic quantum yield (α) and solar irradiance).

image

Figure 7. Sensitivity of fraction of photosynthetically active radiation absorbed by photosynthetic elements in the canopy (FAPARps) to changes in maximum photosynthetically active radiation (PAR), mean temperature, mean vapour pressure deficit (VPD) and maximum intrinsic quantum yield at the Harvard Forest site (a–d) and the Mead irrigated site (e–h). DOY, day of the year.

Download figure to PowerPoint

Relationship between the FAPARps and satellite-derived vegetation indices

Fig. 8 displays the relationship between FAPARps and the three vegetation indices (i.e. NDVI, MTCI and EVI) at the two flux tower sites. All the vegetation indices were positively correlated (R2 = 0.67–0.88) with the FAPARps. The MTCI had the strongest correlation with FAPARps at both sites, while the NDVI had the weakest correlation. Overall, the positive correlation between the three vegetation indices and the FAPARps implies they could be used as surrogates to FAPARps at large spatial scales, with the MTCI, a red-edge band based index, offering the best prospect. All the vegetation indices showed high values when FAPARps was zero (Fig. 8), indicating effects of background noise in these indices, with the largest noise seen in the NDVI. Finally, it should be noted that the footprint of FAPARps in this study is c. 500 m × 500 m to 1 km ×1 km, whereas the three indices were produced at a spatial resolution of 1 km2, which would reduce confidence in these results.

image

Figure 8. Relationship between fraction of photosynthetically active radiation absorbed by photosynthetic elements in the canopy (FAPARps) and three operational vegetation indices (i.e. MERIS Terrestrial Chlorophyll Index (MTCI), Enhanced Vegetation Index (EVI) and Normalized Difference Vegetation Index (NDVI)) for the Harvard deciduous broadleaf forest site and the Mead irrigated cropland site (2001–2005).

Download figure to PowerPoint

Comparison of FAPARps with two operational FAPAR products

Fig. 9 shows the relationship between FAPARps and two operational FAPAR products (i.e. MODIS and CYCLOPES FAPAR products). At the Harvard Forest site, the CYCLOPES FAPAR had a stronger correlation to the FAPARps (R2 = 0.73; RMSE = 0.32, Fig. 9a) than the MODIS FAPAR (R2 = 0.56; RMSE = 0.40, Fig. 9b). The relationship between both the FAPAR products and the FAPARps was nonlinear, rising initially and then becoming asymptotic at higher FAPARps ranges. Furthermore, both products overestimated the FAPARps at this site (values above the 1 : 1 line; Fig. 9a,b). At the Mead irrigated crop site (Fig. 9c,d), the relationship between the two FAPAR products and the FAPARps was more linear than was observed at the Harvard Forest site (Fig. 9a,b). The CYCLOPES FAPAR showed better agreement (R2 = 0.88; RMSE = 0.14, Fig. 9c) with FAPARps at this site than did the MODIS FAPAR (R2 = 0.81; RMSE = 0.23, Fig. 9d). Even though the two FAPAR products were closer to FAPARps in the Mead site, they still overestimated FAPARps, especially in the lower ranges.

image

Figure 9. Comparison between fraction of photosynthetically active radiation absorbed by photosynthetic elements in the canopy (FAPARps) and two operational FAPAR products (i.e. CYCLOPES and MODIS FAPAR) for the Harvard deciduous broadleaf forest site (a, b) and the Mead irrigated cropland site (c, d) (2001–2005). The relationship between the FAPARps and the two FAPAR products was polynomial at the Harvard site and linear at the Mead site.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The absorbed PAR represents the available light energy for plant productivity and is therefore the key variable influencing plant physiological processes represented in photosynthesis, transpiration and energy balance in most productivity efficiency models (PEM; Prince & Goward, 1995; Ruimy et al., 1999; Myneni et al., 2002; Morisette et al., 2006; Gobron et al., 2006). Accurate estimation of the fraction of PAR absorbed by photosynthetic elements of the canopy is critical, as only this PAR is utilized for photosynthesis. The inversion methodology employed to derive FAPARps depends directly on the photosynthetic uptake of carbon dioxide. It relies on the relationship between NEE data and incoming PAR. The maximum PAR range used in the inversion process was set at the point where the relationship between NEE and PAR ceases to be nonlinear (Fig. 1). For the whole PAR range, a high positive correlation (R2 > 0.6) was observed between NEE and PAR at the two study sites (Fig. 1). Apart from deriving the FAPARps, the inversion approach was also used to estimate daytime ecosystem respiration and ecosystem LUE.

The estimated daytime ecosystem respiration from the inversion approach was comparable to the daytime respiration inferred from tower NEE measurements by modelling night-time variation in NEE as a function of night-time temperature (i.e. the Q10 model) at the two sites (R2 > 0.8; Fig. 2). Similar findings were reported by Suyker & Verma (2001) at a tall-grass prairie ecosystem. The daytime respiration data derived using the inversion approach also tracked recognized seasonal respiration patterns (i.e. increase in spring, maximum in summer and reduced in autumn) and also increased with increasing temperature (Figs 3, 4). Therefore, the inversion approach could offer a means of deriving daytime respiration from NEE data. However, the daytime respiration values from the inversion approach were slightly lower than the Q10-modelled daytime respiration data at the two study sites (Fig. 2). One explanation for the lower values could be the inhibition of respiration by light that is experienced during the day. It has been shown that light could inhibit respiration by 30–40% (Villar et al., 1995; Atkin et al., 1998; Lambers et al., 2006). Another explanation could be the fact that derivation of daytime respiration from modelling night-time NEE using the Q10 model approach has been shown to overestimate daytime respiration by up to 25% (Reichstein et al., 2005). This overestimation has been attributed to the use of long-term annual temperature datasets whose sensitivity often exceeds short-term temperature sensitivity in the Q10 modelling approach (Reichstein et al., 2005). Finally, the Q10 modelling approach does not account for respiration because of daytime growth and maintenance, which may in turn lead to lower daytime respiration values (Davidson et al., 2006). Owing to the uncertainties in the Q10-modelled respiration data, it is recommended that future studies should compare the respiration output from the inversion approach with actual in situ respiration measurements.

The fitted slope, representing the whole ecosystem LUE, varied from 0 to 0.055 mol C mol−1 PAR (Fig. 3). This range is slightly higher than those reported at a grassland and a wheat field site (i.e. 0–0.04 mol C mol−1 PAR) by Hanan et al. (2002). The slightly higher values were expected, as the deciduous broadleaf forest site (Harvard Forest, Fig. 5a) and the maize crop site (Mead Irrigated, Fig. 5b) evaluated in the present study are more productive than the grassland and wheat sites. Furthermore, a higher maximum PAR threshold (i.e. 550 and 1000 μmol m−2 s−1) was used, compared with the 300 μmol m−2 s−1 used by Hanan et al. (2002). The mean of the growing season (day of the year 152–289) ecosystem LUE for the Harvard Forest site was comparable but slightly higher than those reported by Zhang et al. (2009) (Table 1). The main reason for this disparity is that the LUEs in the present study are for the whole ecosystem, whereas Zhang et al. (2009) calculated LUE for the chlorophyll components of the canopy only. Overall, the inversion approach resulted in an LUE that was within the ranges of values reported by other studies that implemented different approaches (Hanan et al., 2002; Zhang et al., 2009).

Table 1. Comparison between growing season mean (day of the year 152–259) light-use efficiency (LUE) calculated in the present study with those from Zhang et al. (2009)
YearLUE_chl – from Zhang et al. (2009)Ecosystem LUE – from present study
20010.0260.034 432
20020.0260.032 087
20030.0240.037 69
20040.0310.035 438
20050.0260.034 577

The FAPARps derived through the inversion process showed a clear seasonal variation (Fig. 6). The Harvard Forest site had slightly lower FAPARps than the Mead crop site. This disparity is explained by the differences in the physiology of the plants found at the two sites. Maize cultivated at the Mead site is a C4 plant and has leaves with larger surface area, which would result in more photon capture and productivity compared with C3 deciduous broadleaf trees at the Harvard site. Overall, the maximum FAPARps values for the two sites were slightly higher than those reported by Hanan et al. (2002) but were within the standard error bounds of their study. A study by Zhang et al. (2005) on the fraction of PAR absorbed by chlorophyll at the Harvard site using data from a coupled leaf and canopy radiative transfer model (PROSAIL-2) and MODIS surface reflectance data reported maximum FAPARps values ranging from 0.6 to 0.8, which are within the range of the results from the present study. Therefore, the inversion approach was relatively reliable for deriving FAPARps. Evaluation of the variation of FAPARps with changes in input parameters (i.e. maximum PAR, mean temperature, VPD, and maximum intrinsic quantum yield; Fig. 7) showed varying results. Increasing maximum PAR value led to a reduction of FAPARps, especially at the Harvard Forest. Photosynthesis rate of C3 plants, such as those found at the Harvard site, often saturate at high PAR values as a result of photorespiration (Osborne & Garrett, 1983; Long et al., 1993; Ehleringer et al., 1997; Lambers et al., 2006). Therefore, increasing PAR would result in light saturation and reduction in photosynthetic efficiency that would lead to a reduction of FAPARps observed at the Harvard site. Increasing the maximum PAR value had a minimal effect on FAPARps at the Mead irrigated site. The Mead site is dominated by maize, whose photosynthetic rate has been shown not to saturate with increasing PAR (Crafts-Brandner & Salvucci, 2002). Variation of mean temperature and VPD resulted in minimal variation of FAPARps at both sites. Increasing the maximum intrinsic quantum yields at the two sites resulted in low FAPARps values, particularly pronounced at the Harvard site. The reduction of FAPARps is explained by the fact that when changing the maximum quantum yield, the incoming PAR was kept constant and hence the inversion process predicts that the concomitant low NEE values must have been a result of low absorption of PAR (i.e. low FAPARps). It is envisaged that increasing the intrinsic quantum yield and the maximum PAR concurrently would result in higher values of FAPARps until the light saturation point is reached. Overall, the FAPARps was more sensitive to two of the input variables (i.e. the maximum quantum yield (α) and PAR irradiance) than the rest of the variables.

To provide the groundwork for future derivation of FAPARps at large spatial scales, FAPARps at the two study sites was compared with three vegetation indices (NDVI, MTCI and EVI). All the vegetation indices were positively correlated (R2 > 0.6) with the FAPARps (Fig. 8), with the MTCI having slightly higher R2 than both EVI and NDVI. Overall, the three vegetation indices had a positive correlation with FAPARps and hence they could be used as surrogates for FAPARps at large spatial scales. However, the MTCI offers a better prospect, as it had a stronger positive correlation and a more linear relationship with the FAPARps than both the EVI and NDVI. One reason for the better relationship between the MTCI and FAPARps could be the fact that it has been shown to track canopy chlorophyll (which captures most of the PAR utilized in photosynthesis) more closely than the EVI and NDVI (Dash & Curran, 2007).

Comparison of the FAPARps with two operational FAPAR products (i.e. the MODIS and CYCLOPES FAPAR products; Fig. 9) showed notable differences at the two study sites. At the Harvard Forest site, the relationship between the two FAPAR products and the FAPARps was asymptotic, whereas at the Mead site, the relationship was linear. Furthermore, the overestimation of FAPARps by the two products was greater at the Harvard site than at the Mead site. This difference can be explained by the physiology of the plants found at the two sites. At the Mead site, only one species is cultivated (i.e. maize) and hence there is less variation in the reflectance signature used to derive the two FAPAR products. At the Harvard site, there are different species of deciduous broadleaf trees with varying canopy structures which would lead to greater variation on the reflectance signature used to derive the two FAPAR products. Furthermore, maize cultivated at the Mead site has lower nonphotosynthetic elements in its canopy (e.g. woody stem and branches) compared with mature trees found at the Harvard site. This implies that the FAPAR estimated by the two products is likely to be influenced mainly by the photosynthetic elements in the canopy, resulting in FAPAR values closer to the FAPARps at the Mead site than at the Harvard site. Overall, the two FAPAR products overestimated FAPARps at the two sites. The main reason for this overestimation can be attributed to the manner in which the two products are derived. The two FAPAR products utilize the reflectance signature from the whole canopy, amounting to inclusion of the influence of nonphotosynthetic components of the leaf (e.g. cell wall, veins, and brown pigments) in the calculation of FAPAR (Zhang et al., 2005; Weiss et al., 2007). This would lead to overestimation of the actual FAPAR utilized in photosynthesis (i.e. the FAPARps). Nevertheless, the findings of this study indicate that the FAPAR products are probably more accurate at predicting the actual FAPAR used in photosynthesis in plants with low nonphotosynthetic elements in their canopies than those with high nonphotosynthetic elements.

Conclusion

The estimation of FAPAR absorbed by photosynthetic elements of the canopy (i.e. FAPARps) is fundamental, as only this FAPAR is utilized in photosynthesis. This paper presented an approach to generate only the fraction of PAR absorbed by photosynthetic elements of the canopy (FAPARps). This approach involved inversion of NEE data measured through eddy covariance at two flux tower sites. The inversion approach provides a means of determining the actual PAR utilized by vegetation for photosynthesis. In addition, it was shown that the inversion process could predict daytime respiration and can be used as a method to estimate daytime respiration at flux tower sites. The values of the FAPARps derived using the inversion approach were lower than those from operational FAPAR products. This was expected, as these products do not account for the fraction of PAR absorbed by nonphotosynthetic elements in the canopy. Comparison of the FAPARps with three satellite-derived vegetation indices (i.e. NDVI, EVI and MTCI) resulted in a positive correlation, implying that they could be used as surrogates for FAPARps at large spatial scales. However, to fully operationalize the derivation of FAPARps from these indices, more sites with various vegetation types should be evaluated in future studies. Even though the inversion technique was successful in determining FAPARps, it should be noted that its application was limited to the PAR range where photosynthesis is not light-saturated.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors thank the University of Southampton's Geography and Environment Academic Unit and the Overseas Research Students Awards Scheme (ORSAS) for providing funding to B.O.O. for this study. We also thank NASA and ESA for providing the MODIS and MERIS data, POSTEL for providing the CYCLOPES datasets and the many researchers within the AmeriFlux consortium for making the flux tower data available for use in the advancement of science. Finally, we thank the anonymous reviewers whose input greatly improved this manuscript.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Aber JD, Federer CA. 1992. A generalized, lumped-parameter model of photosynthesis, evapotranspiration and net primary production in temperate and boreal forest ecosystems. Oecologia' 92: 463474.
  • Asner GP, Wessman CA, Archer S. 1998. Scale dependence of absorption of photosynthetically active radiation in terrestrial ecosystems. Ecological Applications 8: 10031021.
  • Atkin OK, Edwards EJ, Loveys BR. 2000. Response of root respiration to changes in temperature and its relevance to global warming. New Phytologist 147: 141154.
  • Atkin OK, Evans JR, Siebke K. 1998. Relationship between the inhibition of leaf respiration by light and enhancement of leaf dark respiration following light treatment. Functional Plant Biology 25: 437443.
  • Atkin OK, Tjoelker MG. 2003. Thermal acclimation and the dynamic response of plant respiration to temperature. Trends in Plant Science 8: 343351.
  • Baldocchi DD. 2003. Assessing the eddy covariance technique for evaluating carbon dioxide exchange rates of ecosystems: past, present and future. Global Change Biology 9: 479492.
  • Baldocchi DD, Falge E, Gu L, Olson R, Hollinger D, Running S, Anthoni P, Bernhofer C, Davis K, Evans R et al. 2001. FLUXNET: a new tool to study the temporal and spatial variability of ecosystem–scale carbon dioxide, water vapour, and energy flux densities. Bulletin of the American Meteorological Society 82: 24152434.
  • Baret F, Hagolle O, Geiger B, Bicheron P, Miras B, Huc M, Berthelot B, Niño F, Weiss M, Samain O et al. 2007. LAI, ƒAPAR and ƒCover CYCLOPES global products derived from vegetation: part 1: principles of the algorithm. Remote Sensing of Environment 110: 275286.
  • Braswell BH, Schimel DS, Privette JL, Moore B III, Emery WJ, Sulzman EW, Hudak AT. 1996. Extracting ecological and biophysical information from AVHRR optical data: an integrated algorithm based on inverse modeling. Journal of Geophysical Research 101: 2333523348.
  • Collatz GJ, Ball JT, Grivet C, Berry JA. 1991. Physiological and environmental regulation of stomatal conductance, photosynthesis and transpiration: a model that includes a laminar boundary layer. Agricultural and Forest Meteorology 54: 107136.
  • Collatz GJ, Ribas-Carbo M, Berry JA. 1992. Coupled photosynthesis-stomatal conductance model for leaves of C4 plants. Functional Plant Biology 19: 519538.
  • Crafts-Brandner SJ, Salvucci ME. 2002. Sensitivity of photosynthesis in a C4 plant, maize, to heat stress. Plant Physiology 129: 17731780.
  • Dash J, Curran PJ. 2004. The MERIS terrestrial chlorophyll index. International Journal of Remote Sensing 25: 54035413.
  • Dash J, Curran PJ. 2007. Evaluation of the MERIS terrestrial chlorophyll index (MTCI). Advances in Space Research 39: 100104.
  • Davidson EA, Janssens IA, Luo Y. 2006. On the variability of respiration in terrestrial ecosystems: moving beyond Q10. Global Change Biology 12: 154164.
  • Deng F, Chen JM, Plummer S, Mingzhen C, Pisek J. 2006. Algorithm for global leaf area index retrieval using satellite imagery. IEEE Transactions on Geoscience and Remote Sensing 44: 22192229.
  • Dillaway DN, Kruger EL. 2011. Leaf respiratory acclimation to climate: comparisons among boreal and temperate tree species along a latitudinal transect. Tree Physiology 31: 11141127.
  • Ehleringer J, Björkman O. 1977. Quantum yields for CO2 uptake in C3 and C4 plants: dependence on temperature, CO2, and O2 concentration. Plant Physiology 59: 8690.
  • Ehleringer JR, Cerling TE, Helliker BR. 1997. C4 photosynthesis, atmospheric CO2, and climate. Oecologia 112: 285299.
  • Farquhar GD, von Caemmere S. 1982. Modelling of photosynthetic response to environmental conditions. In: Lange OL, Nobel PS, Osmond CB, Ziegler H, eds. Physiological plant ecology II. Berlin, Germany: Springer, 549587.
  • Fensholt R, Sandholt I, Rasmussen MS. 2004. Evaluation of MODIS LAI, fAPAR and the relation between fAPAR and NDVI in a semi-arid environment using in situ measurements. Remote Sensing of Environment 91: 490507.
  • Field CB, Mooney HA. 1986. The photosynthesis-nitrogen relationship in wild plants. In: Givnish TJ, ed. On the economy of plant form and function. Cambridge, UK: Cambridge University Press, 2255.
  • Gobron N, Pinty B, Aussedat O, Chen JM, Cohen WB, Fensholt R, Gond V, Huemmrich KF, Lavergne T, Mélin F et al. 2006. Evaluation of fraction of absorbed photosynthetically active radiation products for different canopy radiation transfer regimes: methodology and results using Joint Research Center products derived from SeaWiFS against ground-based estimations. Journal of Geophysical Research 111: D13110; 13111–13115.
  • Gobron N, Pinty B, Melin F, Taberner M, Verstraete M. 2002. Sea wide field-of-view sensor – an optimized FAPAR algorithm-theoretical basis document. EUR Report 20148 EN, European Commission – DG Joint Research Centre, Institute for Environment and Sustainability.
  • Gobron N, Verstraete MM. 2009. Assessment of the status of the development of the standards for the terrestrial essential climate variables: Leaf Area Index (LAI). Global Terrestrial Observing System Report. Rome, Italy: Food and Agriculture Organization.
  • Gower ST, Kucharik CJ, Norman JM. 1999. Direct and indirect estimation of Leaf Area Index, fAPAR, and Net primary production of terrestrial ecosystems. Remote Sensing of Environment 70: 2951.
  • Hall FG, Huemmrich KF, Goetz SJ, Sellers PJ, Nickeson JE. 1992. Satellite remote sensing of surface energy balance: success, failures, and unresolved issues in FIFE. Journal of Geophysical Research 97: 1906119089.
  • Hanan NP, Burba G, Verma SB, Berry JA, Suyker A, Walter-Shea EA. 2002. Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR absorption. Global Change Biology 8: 563574.
  • Hanan NP, Kabat P, Dolman AJ, Elbers JA. 1998. Photosynthesis and carbon balance of a Sahelian fallow savanna. Global Change Biology 4: 523538.
  • Hollinger DY, Kelliher FM, Byers JN, Hunt JE, McSeveny TM, Weir PL. 1994. Carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere. Ecology 75: 134150.
  • Huemmrich KF, Privette JL, Mukelabai M, Myneni RB, Knyazikhin Y. 2005. Time-series validation of MODIS land biophysical products in a Kalahari woodland, Africa. International Journal of Remote Sensing 26: 43814398.
  • Huete A, Didan K, Miura T, Rodriguez EP, Gao X, Ferreira LG. 2002. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment 83: 195213.
  • Jacquemoud S, Baret F. 1990. Prospect: a model of leaf optical properties spectra. Remote Sensing of Environment 34: 7591.
  • Knyazikhin Y, Martonchik JV, Myneni RB, Diner DJ, Running SW. 1998. Synergistic algorithm for estimating vegetation canopy leaf area index and fraction of absorbed photosynthetically active radiation from MODIS and MISR data. Journal of Geophysical Research 103: 257276.
  • Lambers H, Chapin FS III, Pons TL. 2006. Plant physiological ecology. New York, NY, USA: Springer Science.
  • Lawlor DW, Cornic G. 2002. Photosynthetic carbon assimilation and associated metabolism in relation to water deficits in higher plants. Plant, Cell & Environment 25: 275294.
  • Leuning R. 1995. A critical appraisal of a combined stomatal-photosynthesis model for C3 plants. Plant, Cell & Environment 18: 339355.
  • Long SP, Postl WF, Bolhár-Nordenkampf HR. 1993. Quantum yields for uptake of carbon dioxide in C3 vascular plants of contrasting habitats and taxonomic groupings. Planta 189: 226234.
  • Monteith JL. 1995. Accommodation between transpiring vegetation and the convective boundary layer. Journal of Hydrology 166: 251263.
  • Morisette JT, Baret F, Privette JL, Myneni RB, Nickeson JE, Garrigues S, Shabanov NV, Weiss M, Fernandes RA, Leblanc SG et al. 2006. Validation of global moderate-resolution LAI products: a framework proposed within the CEOS land product validation subgroup. IEEE Transactions on Geoscience and Remote Sensing 44: 214231.
  • Myneni RB, Hoffman S, Knyazikhin Y, Privette JL, Glassy J, Tian Y, Wang Y, Song X, Zhang Y, Smith GR et al. 2002. Global products of vegetation leaf area and fraction absorbed PAR from year one of MODIS data. Remote Sensing of Environment 83: 214231.
  • Osborne BA, Garrett MK. 1983. Quantum yields for CO2 uptake in some diploid and tetraploid plant species. Plant, Cell & Environment 6: 135144.
  • Prince SD, Goward SN. 1995. Global primary production: a remote sensing approach. Journal of Biogeography 22: 815835.
  • Reichstein M, Falge E, Baldocchi D, Papale D, Aubinet M, Berbigier P, Bernhofer C, Buchmann N, Gilmanov T, Granier A et al. 2005. On the separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved algorithm. Global Change Biology 11: 14241439.
  • Rouse JW, Hase RH, Schell JA, Deering DW. 1973. Monitoring vegetation systems in the great plain with ERTS. Proceedings of the third ERTS Symposium. Washington, DC, USA: US Government Printing Office, NASA.
  • Ruimy A, Kergoat L, Bondeau A, The Participants Of The Potsdam NPP Model Intercomparison. 1999. Comparing global models of terrestrial net primary productivity (NPP): analysis of differences in light absorption and light-use efficiency. Global Change Biology 5(Suppl 1): 5664.
  • Shirke PA, Pathre UV. 2004. Influence of leaf-to-air vapour pressure deficit (VPD) on the biochemistry and physiology of photosynthesis in Prosopis juliflora. Journal of Experimental Botany 55: 21112120.
  • Sokal PR, Rohlf EJ. 1981. Biometry. New York, NY, USA: Freeman.
  • Suyker AE, Verma SB. 2001. Year-round observations of the net ecosystem exchange of carbon dioxide in a native tallgrass prairie. Global Change Biology 7: 279289.
  • Tezara W, Mitchell VJ, Driscoll SD, Lawlor DW. 1999. Water stress inhibits plant photosynthesis by decreasing coupling factor and ATP. Nature 401: 914917.
  • Urbanski S, Barford C, Wofsy S, Kucharik C, Pyle E, Budney J, McKain K, Fitzjarrald D, Czikowsky M, Munger JW. 2007. Factors controlling CO2 exchange on timescales from hourly to decadal at Harvard Forest. Journal of Geophysical Research 112: G02020.
  • Verma SB, Dobermann A, Cassman KG, Walters DT, Knops JM, Arkebauer TJ, Suyker AE, Burba GG, Amos B, Yang H et al. 2005. Annual carbon dioxide exchange in irrigated and rainfed maize-based agroecosystems. Agricultural and Forest Meteorology 131: 7796.
  • Villar R, Held AA, Merino J. 1995. Dark leaf respiration in light and darkness of an evergreen and a deciduous plant species. Plant Physiology 107: 421427.
  • Weiss M, Baret F, Garrigues S, Lacaze R. 2007. LAI and ƒAPAR CYCLOPES global products derived from vegetation. Part 2: validation and comparison with MODIS collection 4 products. Remote Sensing of Environment 110: 317331.
  • Zhang Q, Middleton EM, Margolis HA, Drolet GG, Barr AA, Black TA. 2009. Can a satellite-derived estimate of the fraction of PAR absorbed by chlorophyll (FAPARchl) improve predictions of light-use efficiency and ecosystem photosynthesis for a boreal aspen forest? Remote Sensing of Environment 113: 880888.
  • Zhang Q, Xiao X, Braswell B, Linder E, Baret F, Moore B III. 2005. Estimating light absorption by chlorophyll, leaf and canopy in a deciduous broadleaf forest using MODIS data and a radiative transfer model. Remote Sensing of Environment 99: 357371.