Multitemporal AVIRIS-Images of Forested and Agricultural
Units in Southern Germany
J. Verdebout, G. Schmuck, S. L. Ustin* and A. J. Sieber
Commission of the European Communities
Joint Research Centre - Institute for Remote Sensing Applications
21020 - Ispra - Italy
*University of California Davis
Department of Land, Air, and Water Resources
CA 95616
Abstract
An analysis, based on the inversion of a simple non-linear model of the
ground reflectance, was conducted on several AVIRIS scenes. The scenes
were acquired during the MAC EUROPE 91 campaign on the 5th and 22nd of
July, over two test sites (Black Forest and Freiburg). The model consists
in a linear mixing of the soil reflectance and a green vegetation reflectance
described with a Kubelka-Munk formula containing the chlorophyll and water
specific absorption coefficients. Its inversion provides a Green Vegetation
Fraction of the pixel and two parameters related respectively to chlorophyll
and water. The model can then be used to evaluate the magnitude of the
1.7 mm absorption feature which is thought to
be a signature of the vegetation biochemical components. The spatial and
temporal variability of this feature over the scenes is commented.
1. Introduction
Imaging spectrometer data have been viewed as a means for making direct
measurements of chemical components of vegetation like lignin and nitrogen
[1]. A spectral matching technique has been applied to the 1.5 - 1.74 mm
region of AVIRIS spectra to demonstrate that the observed vegetation spectrum
in this wavelength region consists of the spectral component of liquid
water and spectral components which appear clearly only in dry vegetation
material. The main purpose of the present work was to further document
this hypothesis by applying a similar technique to AVIRIS scenes containing
both forested areas and several types of agricultural fields. The temporal
variability of the suspected biochemical signature could also be examined
as scenes were acquired on the same test sites at about two weeks interval.
At the same time, we tested the possibility of inverting a non linear
model of the ground reflectance in which the green vegetation fraction
is described by a Kubelka-Munk formula for an optically thick medium. In
this very simple model chlorophyll and water are taken into account by
using their respective optical absorption coefficients.
2. Test Site Description
In the frame of the MAC EUROPE 91 - campaign AVIRIS over flights have been
performed on the 5th and 22nd of July over two test sites in the southern
part of Germany. An extensive ground truth measurement campaign was set
up to accommodate the airborne measurements. Unfortunately not all of the
collected ground reflectance data have been available for our studies.
The agricultural study area is situated approximately 20 km West of
the City of Freiburg in the Upper Rhine Valley and has an extension of
6 x 4 km. This test site contains both forested areas (19 %) and agricultural
areas (50 %). The agricultural part is intensively cultivated with the
main crops being wheat, corn, barley, potatoes, sugar beet and vine. The
average field size of approximately 1.5 ha is representative for small
scale European farming. The area is topographically flat at an altitude
of 200 m above sea level. The soils are dominated by the quarternary sediments
of the Rhine River and thus showing a great variety of grain size distribution
and high porosity. The latter accompanied by low clay contents results
in high infiltration rates requiring the irrigation of the intensive cultivation
areas of corn.
The Black Forest test site is located near the town of Villingen/Schwennigen
at an altitude ranging from 800 m to 960 m above sea level. Besides some
small areas covered by Scots pine (Pinus silvestris L.) and
silver fir (Abies alba Mill.), the dominant tree species of the
overall region is Norway spruce (Picea abies) with tree ages from
80 to 120 years (tree heights 30 - 40 m). The understory is mainly composed
of blueberries and of young spruce and fir trees for rejuvenation. Soils
are dominated by sandy - loamy acid brown earths; the bedrock consists
of sandstone layers.
3. Modeling Vegetation Spectra
It has been shown that the overall shape of the leaf reflectance spectrum
can be explained by the absorption characteristics of chlorophyll and water,
once they are included in a radiative transfer model. A number of simple
models exist which describes the scattering in various ways (Kubelka-Munk
[2], plate models [3], stochastic model [4]). They are successful in reproducing
the spectrum major features such as the visible reflectance up to the red
edge and the water absorption peaks in the infrared. However, there are
details in the spectrum which are still unaccounted for, such as a small
absorption feature centered around 1.7 mm An
increasing interest is being brought to this feature as it is thought to
be a signature of biochemicals such as lignin, cellulose, starch and proteins.
In the frame of spectral unmixing studies, it is revealed as a systematically
recurring residue [5]; it has also been directly investigated using the
spectral matching technique [1].
We are presently working on radiative transfer models which will explicitly
include the biochemical components. One of these models is based on the
Schuster-Schwarzschild (or "two flow") approximation of the radiative transfer
equation. The biochemicals are introduced by adding their contribution
to the spectral absorption coefficient of the leaf tissue. It is not the
purpose of this paper to present this work which has not yet reached its
conclusions. However, the studies conducted so far on laboratory spectra
have shown:
-
that the 1.7 mm feature cannot be explained
by water alone
-
that it cannot be reproduced by using the specific absorption coefficient
available today for lignin (wood), cellulose, starch or proteins
-
that the model based on chlorophyll and water can reproduce accurately
the spectrum in some spectral regions where the two components dominate
the absorption (0.5 to 0.73 mm and 1.5 to 1.65
mm respectively)
-
that the amplitude of the 1.7 mm residual is
dependent on the type of vegetation (see Fig.
1)
Ultimately, our purpose is to couple a leaf model with a canopy model and
to perform the inversion on imaging spectrometry spectra. In order to document
the feasibility and the interest of such a procedure, we applied it to
the AVIRIS scenes; using though a limiting case of the model for which
the algorithm complexity is drastically reduced.
4. Processing of AVIRIS Data
The surface reflectance was first obtained from the radiance by using the
"Atmosphere Removal Program" developed at the CSES/CIRES-University of
Colorado [6]. This program uses the 5s code to model the aerosols while
the gaseous transmittance calculation allows for a pixel to pixel variable
amount of atmospheric water vapour. The amount of water vapour is obtained
from the intensity of the absorption lines at 0.94 and 1.14 mm.
Figure 2 shows typical reflectance spectra
obtained using this procedure. It can be seen that the near infrared plateau
is still much disturbed by remains of the water vapour features; such an
effect would be typically produced by a slight error in the wavelength
calibration. This is of little importance for this study as it does not
make use of the NIR plateau region.
As the scenes contain both forested areas with a high vegetation cover
and agricultural fields of which some have a low cover, the analysis had
to take into account the soil reflectance. This was done in the simplest
way by assuming a linear mixing of soil and vegetation spectra, we therefore
write:
Rp(l) = as×
Rs(l) + av×
Rv(l)
|
(1)
|
Where
Rp(l ) is the reflectance
of the pixel,
Rs(l )= is the soil reflectance
and as its abundance in the pixel spectrum
Rv(l ) is the vegetation
reflectance and av its abundance in the pixel spectrum
The soil spectrum was taken from the scene as the mean spectrum of a small
area known to be bare soil. The vegetation spectrum was modeled with a
Kubelka-Munk formula for an optically thick homogeneous medium:
where w0 (l)
is the single scattering albedo of the medium
s is the scattering coefficient of the medium (as the scattering
in leaf tissue is mainly due to multiple reflections and refractions, it
can reasonably be assumed to be wavelength independent)
k(l ) is the absorption coefficient
of the medium
We further assume that the absorption in vegetation is due to chlorophyll
and water and write:
where kchl(l ) is the specific
absorption coefficient of chlorophyll; the in vivo absorption coefficient
(expressed in cm2 mg-1)
of ref . 3 was used
kw(l ) is the specific
absorption coefficient of water (expressed in cm-1); the measurement
of Curcio and Petti was used [7]
cchl is the chlorophyll concentration
cw is the water concentration
achl and aw are the above concentrations
divided by the scattering coefficient and are the independent parameters
of the vegetation spectrum model
By combining formula (1), (2), (3) and (4), one obtains a model of the
pixel reflectance as a non linear function of four parameters: as,
av, achl and aw, which were
determined by least mean square fitting on the AVIRIS pixel reflectance
by using a Marquardt algorithm. Two spectral windows were used in the fitting:
0.5 to 0.73 mm and 1.5 to 1.65 mm
where the chlorophyll and water absorption are respectively dominant. Once
the fitting is performed, we can compute a "Green Vegetation Fraction"
of the pixel, defined by:
We also retrieve a measured spectrum of the green vegetation fraction (Rvm
(l )):
If we assume that the 1.7 mm feature is an absorption
due to a component of vegetation, we logically evaluate its magnitude from
the absorptance corresponding to the measured and fitted vegetation spectra
(Avm and Av). The absorptance has been
defined here as k/s, and obtained by inverting equations (2) and (3):
The residual has then been evaluated in the 1.65 to 1.76 mm
spectral interval as:
where the average is taken on the N AVIRIS channels in the spectral window.
This analysis procedure was applied to four AVIRIS scenes (the Black
Forest the Freiburg test sites on two dates). The processing of a scene
took about 8 hours of computing time on a SUN SPARC 10 workstation. It
was found that the above model could reproduce very well the spectra of
the vegetated areas: the mean relative deviation (D
R/R) between the measured and modeled pixel reflectance spectra
was typically 5% (within the fitting windows). Examples of the fit are
illustrated at Fig. 3.
5. Results
Figures 4 and 5
show the results of the analysis on a part of the Freiburg test site containing
both forested regions (2 large areas at the top / right part and a smaller
one in the right corner of the image) and agricultural units. The rectangular
area at the top left corner of the image is a small lake. A small town
is located in the centre of the image. The images in the left column of
the figures represent processed AVIRIS data from the overflights of the
5th of July; the right column of the 22nd of July. By comparing the images
from the two overflights a number of qualitative comments can already be
made, but a detailed interpretation will only be possible by the confrontation
with ground data, which are at the moment not available.
Over the forested regions all the calculated indices and parameters
are very stable from one overflight to the other, this is to be expected
as a forest will not change significantly within a two week period. On
the agricultural units all the calculated parameters reflect the growing
and harvesting cycle of vegetation. Compared to the vegetation indices
NDVI and MSI, the chlorophyll and water parameter (achl
and aw) demonstrate a higher sensitivity to the growing
process. However, the interpretation of these two parameters is difficult
as the hypothesis of the model to have an optically thick canopy probably
does not hold on the fields. The vegetation fraction image seem to contain
essentially the same information as the NDVI image though showing a better
dynamics with respect to the cover type.
Major emphasis has been placed on the residual image in the spectral
region of 1.7 mm because different chemical
components like lignin and cellulose have absorption features near this
wavelength region. From Fig. 5 it becomes
obvious that this residual clearly discriminates between forest and other
type of vegetation. Within the forest the spatial variability of the residual
is clearly correlated with that of the water index MSI (negative correlation)
and the water parameter aw (positive correlation). The
situation is more complex regarding the agricultural units within the test
site. By comparing the two residual images, several fields appear brighter
(higher residual) on the 22nd of July, which could be related to the maturation
of the different crops. A definite interpretation (taking into account
the very low values of the residual) will only be possible by comparing
these images with the agricultural and meteorological data of the local
authorities.
A comparison of the AVIRIS images of the Black Forest test site revealed
no differences between the two overflights regarding the calculated vegetation
indices and parameters. Of major interest are three well documented plots
within this forest, of which two have been fertilized with ammonium sulfate
of different concentrations for the last three years. According to our
analysis, eventual effects of the fertilization on canopy characteristics
like chlorophyll concentration, water content and the biochemical components
in the 1.7 mm region were not detectable.
6. Conclusions
At this point of the study, we can conclude that the 1.7 mm
residual does show a systematic relation with the vegetation cover type.
It is markedly higher on forests than on agricultural crops and significantly
varies within the forest. In this respect, the results obtained on two
different test sites and two dates are reproducible. On the fields, the
variability is very fa and partly obscured by the uncertainties resulting
from the detector noise. Progress in the interpretation of this spectral
feature needs further work both by confronting the remotely sensed data
with ground information and by performing accurate and systematic measurements
in the laboratory.
This work has also shown the possibility of inverting non linear models
of the green vegetation spectrum on imaging spectrometer data. Though the
model used is excessively simple, it contains explicitly the effect of
the two main components which are chlorophyll and water and is able to
describe accurately the spectrum shape in two spectral windows. This result
is encouraging to pursue this approach by using more detailed models which
will make use of the entire spectrum and will provide parameters more easily
interpretable in terms of the canopy characteristics.
7. References
1. Goetz, A. F. H., Gao, B. C., Wessman, C. A. and Bowman, W. D. 'Estimation
of biochemical constituents from fresh, green leaves by spectrum matching
techniques', in Proc. Int. Geosci. and Remote Sens. Symp. (IGARSS '90),
Washington DC, 20-24 May 1990, pp. 971-974, 1990.
2. Allen, W. A. and Richardson, A. J. 'Interaction of light with a plant
canopy', J Opt. Soc. Am., vol. 58, no. 8, pp. 1023-1028, 1968.
3. Jacquemoud, S. and Baret, F. 'PROSPECT: a model of leaf optical properties
spectra', Remote Sens. Environ., vol. 34, pp. 75-91, 1990.
4. Tucker, C. J. and Garratt, M. W. 'Leaf optical properties as a stochastic
process', Appl. Opt., vol. 16, no. 3, pp. 635-642, 1977.
5. Smith, M. O., Ustin, S., Adams, J. B. and Gilliespie, A. R 'Vegetation
in deserts: A regional measure of abundance from multispectral images',
Remote Sens. Environ., vol. 31, no. 1, pp. 1-26, 1990.
6. Gao, B.-C and A. F. H. Goetz, 'Column atmospheric water vapor and
vegetation liquid water retrievals from airborne imaging spectrometers
data', J. Geophys. Res., 95, 3549-3564, 1990.
7. Curcio, J. A. and Petti, C. C., 'The near infrared absorption spectrum
of liquid water', J. Opt. Soc. Am., 41(5), 302-304, 1951.
1998, Center
for Spatial Technologies and Remote Sensing (CSTARS)
University
of California, Davis
