172
In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
2. MATERIALS
2.1 Study area and sampling
The study site is located in a National Park in Italy (latitude
41°52' to 42°14 , N, longitude 13°50' to 13°14'E). The park
covers an area of 74.095 ha and extends into the southern part
of Abruzzo, at a distance of 40 km from the Adriatic Sea. The
region is situated in the massifs of the Apennines. The flora of
the park includes more than 1800 plant species, which
approximately constitute one third of the entire flora in Italy. A
total of 45 plots (30 m by 30 m) were selected. For each plot,
the relevant biophysical and biochemical parameters were
measured within few randomly selected subplots. In each plot
the species varied in terms of leaf shape, size and the amount of
leaves.
2.2 Vegetation parameter measurements
A SPAD-502 Leaf Chlorophyll Meter was used to assess leaf
chlorophyll content. A total of 150 leaves were randomly
selected in each plot representing the dominant species and their
SPAD readings were recorded. From the 150 individual SPAD
measurements, the average was calculated. These averaged
SPAD readings were converted into leaf chlorophyll contents
[pg cm' 2 ] by means of an empirical calibration function
provided by (Markwell et al. 1995). The total canopy
chlorophyll content (CCC) [g m' 2 ] for each plot then have been
obtained by multiplying the leaf chlorophyll content with the
corresponding leaf area index
In each plot, leaf area index was measured using the Plant
Canopy Analyzer LAI-2000 (LICOR Inc., Lincoln, NE, USA).
To prevent direct sunlight on the sensor, samples of below- and
above-canopy radiation were made with the sun behind the
operator and using a view restrictor of 45°. Table 1 reports
summery statistics for some of the measured variables of the
plots.
Measured variables
STDV
Min
Mean
Max
SPAD (unit-less)
3.7
24.2
32.7
41.0
Leaf chlorophyll (pg cm' 2 )
4.7
18.9
28.7
40.9
Canopy chlorophyll (g m' 2 )
0.56
0.21
0.86
2.3
LAI (m 2 m' 2 )
1.59
0.72
2.87
7.54
Table 1. Summary statistics for some measured variables of
sample plots.
2.3 Hyperspectral images
HyMap images of the study area were acquired by DLR,
Germany’s Aerospace Research Centre and Space Agency. The
sensor contained 126 wavelengths, operating over the spectral
range of 436 nm to 2485 nm. The spatial resolution of the data
was 4 m. The data were collected in four image strips, each
covering an area of about 40 km by 2.3 km. The image
acquisition was close to solar noon. The image strips were
atmospherically and geometrically corrected by DLR. A 7 by 7
pixel window centred around the central position of a plot was
used for collection of grass spectra from each sample plot and
its average spectrum was calculated.
3. METHODS
3.1 PROSAIL & Inversion
The commonly used PROSAIL radiative transfer model which
is a combination of the SAILH canopy reflectance model
(Verhoef, 1984; Verhoef, 1985) and the PROSPECT leaf
optical properties model (Fourty et al., 1996; Jacquemoud and
Baret, 1990; Jacquemoud et al., 1996) was selected for canopy
parameter retrieval. The PROSPECT model calculates the leaf
hemispherical transmittance and reflectance as a function of
four input parameters: the leaf structural parameter N (unitless);
the leaf chlorophyll a + b concentration LCC (pg cm' 2 ); the dry
matter content Cm (g cm' 2 ); and the equivalent water thickness
Cw (g cm" 2 ). The SAILH model, apart from the leaf reflectance
and transmittance, requires eight input parameters to simulate
the top-of-canopy bidirectional reflectance. These are sun zenith
angle, ts (deg); sensor viewing angle, to (deg); relative azimuth
angle between sensor and sun, phi (deg); fraction of diffuse
incoming solar radiation, skyl; background reflectance (soil
reflectance) for each wavelength, rsl; LAI (m 2 m' 2 ); average leaf
inclination angle, ALA (deg); and the hot spot size parameter,
hot (m m" 1 ). To account for the changes induced by moisture
and roughness in soil brightness, we used a soil brightness
parameter, scale (Atzberger et al., 2003). Sensor viewing angle,
azimuth angle, sun zenith angle and fraction of diffuse
incoming solar radiation were fixed.
The inversion of PROSAIL radiative transfer model was
considered by using a look-up table (LUT). To build the LUT,
100,000 parameter combinations were randomly generated and
used in the forward calculation of the PROSAIL model. The
ranges (minimum and maximum) for each of the eight “free”
model parameters are reported in Table 2. The maximum and
minimum values of LAI, LCC and ALA were fixed based on
prior knowledge from the field data collection (Combal et al.,
2003; Darvishzadeh et al., 2008). To find the solution to the
inverse problem for a given canopy spectra, for each modelled
reflectance spectra of the LUT the root mean square error
between measured and modelled spectra (RMSEr) was
calculated.
Parameter
Min
Max
Leaf area index
0
8
Mean leaf inclination angle
40
70
Leaf chlorophyll content
15
45
Leaf structural parameter
1.5
1.9
Dry matter content
0.005
0.010
Equivalent water thickness
0.01
0.02
Hot spot size
0.05
0.10
Soil brightness
0.5
1.5
Table 2. Specific ranges for eight input parameters used for
generating the LUT.
4. RESULTS
To find the solution to the inverse problem, the LUT is sorted
according to the cost function and the set of variables providing
the minimum RMSE is considered as the solution. Figure 1
illustrates measured and simulated canopy reflectance spectra
found in this way for two plots with contrasting LAI values.