suspended within the acetone). The process was repeated three times in order to
completely extract as much chlorophyll pigment as possible from the leaf material.
The absorbance at 447, 645, and 662 nm of each finished sample was measured using
a spectrophotometer (Thermo Scientific, Detroit, MI). The chlorophyll concentration
of each sample was then calculated using the fresh sample weight, sample extraction
volume, and absorption coefficients as reported by Lichtenthaler (1987). The
Chl a + b concentration is given in micrograms per milligram. The chlorophyll
concentration data for the same species were then averaged and converted to the
chlorophyll content of the species (in grams per square meter).
7.3.2 QuickBird Image Acquisition and Preprocessing
In this study, a multispectral QuickBird image with a spatial resolution of 2.4 m was
acquired on July 5, 2010, from the Digital Globe archival collection. The spectral
bandwidth of QuickBird includes the blue (430–545 nm), green (466–620 nm), red
(590–710 nm), and near-infrared (NIR) (715–918 nm) regions of the spectrum. The
image was geometrically and radiometrically corrected by the distributor. Atmospheric correction was conducted using the ATCOR-2 algorithm in PCI Geomatica
10 using weather conditions obtained from Environment Canada’s National
Climate Data.
7.3.3 Spectral Indices
The reflectance values in the wavelength regions of 700 and 750 nm were combined
to calculate the red edge index (i.e., R750/R700) since it was found to be well
correlated with total chlorophyll content of different types of leaves (Datt, 1999;
Gamon and Surfus, 1999; Sims and Gamon, 2002; Wong, 2012). These calculations
were performed using the hyperspectral reflectance gathered in the lab and field. For
the QuickBird image acquired at the spaceborne level, the broadbands red (band 3)
and NIR (band 4) were used to calculate SR [R, NIR].
7.3.4 Scaling Up from Leaf to Canopy and Landscape Levels
Although hyperspectral data at the canopy level have been used to detect changes in
canopy chlorophyll content in dense vegetation (e.g., crops and forests) (Daughtry
et al., 2000; Yoder and Pettigrew-Crosby, 1995; Zhang et al., 2008; Wu et al., 2008),
it has proven to be difficult to achieve the same outcome on grassland ecosystems
characterized by heterogeneous vegetation (He and Mui, 2010). Therefore, the leaflevel chlorophyll measurements were scaled to the canopy level to produce a canopyintegrated chlorophyll content. Specifically, the chlorophyll content in the middle plot
at each arm in each site (a total of four plots in each site) was calculated by multiplying
leaf Chl a + b content for each dominant species with its percentage cover and
summing the resulting values for all species considered. Only one plot in each arm
was selected to remove the autocorrelation between plots within each arm. Furthermore, the canopy Chl a + b content (Chl canopy , in grams per square meter) is averaged
130
ESTIMATING GRASSLAND CHLOROPHYLL CONTENT
completely extract as much chlorophyll pigment as possible from the leaf material.
The absorbance at 447, 645, and 662 nm of each finished sample was measured using
a spectrophotometer (Thermo Scientific, Detroit, MI). The chlorophyll concentration
of each sample was then calculated using the fresh sample weight, sample extraction
volume, and absorption coefficients as reported by Lichtenthaler (1987). The
Chl a + b concentration is given in micrograms per milligram. The chlorophyll
concentration data for the same species were then averaged and converted to the
chlorophyll content of the species (in grams per square meter).
7.3.2 QuickBird Image Acquisition and Preprocessing
In this study, a multispectral QuickBird image with a spatial resolution of 2.4 m was
acquired on July 5, 2010, from the Digital Globe archival collection. The spectral
bandwidth of QuickBird includes the blue (430–545 nm), green (466–620 nm), red
(590–710 nm), and near-infrared (NIR) (715–918 nm) regions of the spectrum. The
image was geometrically and radiometrically corrected by the distributor. Atmospheric correction was conducted using the ATCOR-2 algorithm in PCI Geomatica
10 using weather conditions obtained from Environment Canada’s National
Climate Data.
7.3.3 Spectral Indices
The reflectance values in the wavelength regions of 700 and 750 nm were combined
to calculate the red edge index (i.e., R750/R700) since it was found to be well
correlated with total chlorophyll content of different types of leaves (Datt, 1999;
Gamon and Surfus, 1999; Sims and Gamon, 2002; Wong, 2012). These calculations
were performed using the hyperspectral reflectance gathered in the lab and field. For
the QuickBird image acquired at the spaceborne level, the broadbands red (band 3)
and NIR (band 4) were used to calculate SR [R, NIR].
7.3.4 Scaling Up from Leaf to Canopy and Landscape Levels
Although hyperspectral data at the canopy level have been used to detect changes in
canopy chlorophyll content in dense vegetation (e.g., crops and forests) (Daughtry
et al., 2000; Yoder and Pettigrew-Crosby, 1995; Zhang et al., 2008; Wu et al., 2008),
it has proven to be difficult to achieve the same outcome on grassland ecosystems
characterized by heterogeneous vegetation (He and Mui, 2010). Therefore, the leaflevel chlorophyll measurements were scaled to the canopy level to produce a canopyintegrated chlorophyll content. Specifically, the chlorophyll content in the middle plot
at each arm in each site (a total of four plots in each site) was calculated by multiplying
leaf Chl a + b content for each dominant species with its percentage cover and
summing the resulting values for all species considered. Only one plot in each arm
was selected to remove the autocorrelation between plots within each arm. Furthermore, the canopy Chl a + b content (Chl canopy , in grams per square meter) is averaged
130
ESTIMATING GRASSLAND CHLOROPHYLL CONTENT