Sei sulla pagina 1di 10

Spectropolarimetric properties of vegetation

Dennis H. Goldstein, Joseph L. Cox


Air Force Research Laboratory, 101 West Eglin Blvd, Eglin AFB, FL 32542

ABSTRACT

Optical properties of vegetation materials are described. Spectral measurements of reflectance and
Mueller matrices in the near infrared are given. The measurement method is discussed, and results
are presented for a variety ofplant material. Samples include both green leaves and bark. Measured
results are compared to published results where possible.

Keywords: Polarimetry, spectropolarimetry, Mueller matrices, reflectance, vegetation.

1. INTRODUCTION

Vegetation is an important and prevalent element in the environment that may be sensed from space or air based
platforms. Sensors that collect polarimetric and/or spectral information are becoming more common. It is the intent of
this study to investigate the spectral and polarimetric properties of various representative vegetation samples using
laboratory instrumentation.

There are a few sources ofnear infrared spectral measurements on plants"234567 but even fewer sources on polarimetric
properties8'9, and none known that provide complete polarimetric information. In fact, it has only become practical to
make high-resolution spectropolarimetric measurements in the last few years.

We describe a study to measure monostatic spectropolarimetric reflectance distribution functions of various vegetative
material samples. The instrument used to make the measurements is briefly described, previous results are summarized,
and current results are presented.

2. SPECTROPOLARIMETRY INSTRUMENTATION

An instrument that measures spectral polarization properties of materials in transmission and reflection has been
previously designed, patented'°, and described".

The spectropolarimetric reflectometer is based around a commercial Fourier transform spectrometer. The
spectropolarimeter generates radiation from the ultraviolet to the far infrared, and has a Visual Basic executive program
that allows control over the spectrometer, control over motorized rotation stages, and spectropolarimetric data
processing. We are using a Bio-Rad FTS-6000 with Win-JR Pro software. For purposes of the measurement program
described here, the spectrometer is used with one source, a tungsten-halogen lamp, and two detectors, a silicon detector
and an indium antimomde detector. The Bio-Rad spectrometer serves as a radiation source for the polarimetric portion
of the instrument and is operated in the conventional absorption spectroscopy mode. The radiation generated by the
spectrometer is brought out through the spectrometer's external port. Figure la shows the basic optical schematic of the
instrument for monostatic reflectance distribution function (MRDF) measurements. For monostatic measurements, the
detector is fixed in the position shown. The sample is mounted on a computer-controlled motorized rotational stage. The
sample can be rotated so that any in-plane incidence angle is obtained. The optical system that collects light for the
detector consists of an off-axis parabolic mirror. This mirror is fixed to look toward the beamsplitter and focus light
onto the detector that is mounted perpendicularly to the light coming from the beamsplitter. The parabolic mirror,
detector, and mounting devices are referred to as the detector assembly. The sample is mounted vertically so that its
surface comprises a vertical plane, and it may be rotated to any position around a vertical axis. Data collection proceeds
by rotating the sample to a set of desired angles, and spectropolarimetric data is then collected at each sample angle. The

Polarization: Measurement, Analysis, and Remote Sensing VI, edited by Dennis H. Goldstein, 53
David B. Chenault, Proceedings of SPIE Vol. 5432 (SPIE, Bellingham, WA, 2004)
0277-786X/04/$15 · doi: 10.1117/12.546808
insirument may be used for transmissive measurements as well by placing the transmissive sample on the sample
rotation stage and lining up the detector assembly with the source beam on the far side of the sample. This is illustrated
in Figure lb. The reflectometer may be used withOut any modification ofthe polarization ofthe source radiation, i.e., no
polarization elements, and in this mode, it is a spectral reflectometer. In order to obtain spectropolarimetric
measurements, a dual rotating retarder Mueller matrix polarimeter, described by Azzam12, is included in the system.
This polarimeter consists of a polarization state generator before the sample and a polarization state analyzer after the
sample. The polarization state generator consists of a linear polarizer followed by a quarter wave retarder. The
polarization state analyzer consists of a quarter wave retarder followed by a linear polarizer and is located on the platter
in front of the detector assembly. Although we use retarders that are nominally quarter wave in the spectral region being
measured, the exact retardance is not critical. When the retarders are rotated in a five to one ratio, all sixteen elements of
the sample Mueller matrix are encoded onto twelve harmonics of the detected signal, which can then be Fourier analyzed
to recover the Mueller matrix elements. Other previous implementations of this Mueller matrix polarimeter have been
described elsewhere'3

The data reduction algorithm for this polarimeter as originally presented by Azzam assumes ideal polarization elements
and no orientation errors. The data reduction algorithms may be generalized to compensate for systematic errors that
result when orientation misalignment and non-ideal retarders are used. If the polarization elements are rotationally
misaligned, or the retarders do not have exactly one-quarter wave of retardance, the changes in Fourier amplitudes and
phases result in errors in the sample Mueller matrix. Even small orientation and retardance errors (<1 O) can lead to large
errors in the measured Mueller matrix (> 1 0% in some matrix elements). These errors become especially important
when the retardance and alignment vary significantly from their nominal values such as in multi-wavelength or spectral
instruments. We have incorporated correction terms for orientation and retardance errors into the dual rotating retarder
data reduction algorithm. Small angle approximation error correction equations are given in Goldstein and Chipman'4,
and this is generalized to larger angles in Chenault et al.'5 These equations are quite lengthy and will not be presented
here.

FOURIER TRANSFORM
SPECTROM ETER SAMPLE
PSG

PSA

I DETECTOR

a. Monostatic reflectance measurement configuration

FOURIER TRANSFORM
SPECTROMETER SAMPLE
PSG PSA

DETECTOR

b. Transmission measurement configuration

Figure 1 . Diagram of spectropolarimeter

54 Proc. of SPIE Vol. 5432


3. VEGETATION
In the range of wavelengths visible to humans, plants are generally green. This is because chlorophyll is better at
absorbing light in two bands centered generally around 450 nm (blue) and 650 am (red), and it is the energy in this
region that plants use. More light is reflected in between these spectral regions, so that yellow and green wavelengths
are apparent to our eyes. Plants also have a requirement to control their temperature. They seem to do this by reflecting
a substantial amount of the near infrared energy out to beyond 2 jtm, a region in which there is still a substantial amount
of solar irradiance, and absorbing most of the radiation beyond 2.5 tm2. Thus vegetation appears to be white in near
infrared imagery, and black in thermal infrared (3-14 tm) imagery. High absorption in the infrared is concommitant
with high emission, and plants use radiant cooling in the infrared to dissipate heat.

In this paper, we examine properties ofplants in the near infrared from 700 to 2300 nm. Measurements have been made
previously in this spectral region"8 and Fig. 2 is an example for some crop leaves (data from Gausman et al.) These
plots show the chorophyll absorption around 650nm and characteristic water absorption around 1400 and 1900 nm.

Data from Gausman et al show that the transmittance follows the same pattern as the reflectance, i.e., transmittance is
high where reflectance is high. Reflectance and transmittance for one crop leaf type, pepper, is shown in Fig. 3 . The
region in between the reflectance and transmittance curves corresponds to the absorbtance.

Reflection from the surface of the plant material may generate polarized light. The outer layer of a plant is called the
cuticle, and the outermost portion of this cuticle consists of a wax9. This waxy portion of the cuticle may be very thin or
fairly thick (5mm), and the character of this wax ranges from amorphous to crystalline8'9. The presence of a crystalline
material introduces the possibility ofthe generation of elliptically polarized light9, at least from individual leaves.

Avocado
Bean
Corn
Cotton
Lettuce
Okra
OnIon
—Orange
Peach
& Pepper
Pumpkin
Sorghum
..
Sugarcane
Sunflower
Tomato
500 1000 1500 2000 2500

Wavelength (nm)

Figure 2. Reflectance of crop leaves in the near infrared (Data from Gausman1).

Proc. of SPIE Vol. 5432 55


100 0

90 10

80 20

70 30
w
C.)
AI
Lw
C.)
C

C.) 50

4Q 60
F-
30 70
20 80

10 90

0 I 00
500 1000 1500 2000 2500
Wavelength (nm)

Figure 3. Reflectance and transmittance ofpepper (Data from Gausman').

4. VEGETATION MEASUREMENTS
4.1 Reflectance Measurements
Reflectance measurements were performed on a variety of plant leaves. Each leaf sample was placed on a sample holder
and held as flat as possible. The spectropolarimeter was operated in monostatic reflectance mode (Fig. la without
polarization elements). A reflectance spectrum was collected at each of 18 sample angles ranging from 60 degrees on
one side of normal incidence to 10 degrees on the other side of normal incidence with a greater density of samples
collected around normal incidence. These angles are given in Table 1 and comprise a set that we normally use for
diffuse samples. They are used for the Mueller matrix measurements as well.

1-60 -45 -301-20 -15 1-10 -8 -6 -4 -2 I-i to Ii 12 4 6 8 10 1

Table 1 Sample angles used in reflectance and Mueller matrix measurements.

A Spectralon sample with 99% reflectance is measured at each of the sample angles prior to a sample data collection for
reference. The MRDF is calculated using the following equation:

MRDF= Sample
Spectralon it
where Sample is the measured sample reflectance value, Spectralon is the measured Spectralon value, and .99 is the
Spectralon hemispherical reflectance furnished by the manufacturer.

56 Proc. of SPIE Vol. 5432


Reflectance spectra from two tree samples are shown in Fig. 4, Bradford pear and southern oak. These spectra show the
characteristic water absorption and chlorophyll absorption as in the crop spectra, and are very similar. Fig. 5 shows plots
for tree bark for the same trees. Water absorption is still apparent.

0.5

0.45

0.4

0.35

0.3
U-

0.2

0.15

0.1

0.05

0
0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3

Wavelength (&m)

Figure 4. Measured reflectance of tree leaves.

0.5

0.45

0.4 Bradford Pear


. . Southern Oak
0.35

0.3
U-

0.2

0.15

0.1

0.05

0
0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3

Wavelength (pm)

Figure 5. Measured reflectance of tree bark

Proc. of SPIE Vol. 5432 57


4.2 Mueller matrix measurements
The Mueller matrices of the same leaf samples were measured using the spectropolarimeter configured as in Fig. la.
Data for the Bradford pear is shown in Fig. 6 for the wavelength range 1. 1 to 2.3 rim. The x-axis is sample angle index,
i.e. angles 1 through 18, so that this is not a linear scale in angle. The y-axis is wavelength index from 1 to 615 where
the index corresponds to a linear scale in wavenumber rather than wavelength. Plots of slices through these 3-
dimensional plots are shown in Figs. 7 and 8.

The Mueller matrix shows that the leaf appears as an absorbing mirror. That is, the leaf will reflect some energy as a
mirror while scattering much of the energy into depolarized states. Note that in the wavelength regions where water
absorption is particularly strong, the mirror-like behavior is emphasized, so that the spectral properties of the leaf are
apparent in the Mueller matrix. This behavior occurs because a greater relative amount of the energy reflected in these
regions is from leaf-air interface, and not from energy that is scattered within the leaf and re-radiated.

Fig. 8 is the Mueller matrix of the pear leaf vs. sample angle for one particular wavelength, 1 .324 m. These plots show
the absorbing mirror behavior peaking around normal incidence. Away from normal incidence at intermediate angles,
the leaf appears to be a good depolarizer. At large angles, a slight increase in the m (linear diattenuation) and m10
(linear polarizance) can be detected, indicating that as the angle goes to a grazing incidence the material will look a little
like a linear polarizer.

Mueller matrices for oak bark that correspond to the plots for pear leaf in Figs. 7 and 8 are shown in Figs. 9 and 10.
These plots indicate that the bark is a better depolarizer than the leaf and its properties are independent of wavelength
and incidence angle.

01... o
.;.. ..

I
I —

01
y:....:.
1J
y3O3D y313D y323D y333D

Figure 6. Mueller matrix ofBradford pear leaf vs. wavelength (1.1 to 2.3 tm) and sample angle.

58 Proc. of SPIE Vol. 5432


I I

O'Z 0

II
I

VI
I

LI E Et T1 E"P.I ELIPtI1 LIPTFT E

______
ti EEELIPI 'C
______
EEELtPII1 'C
TI Pt LI
'C
E ELTPTtt 'C

'C

O1flId L IPMAI X!11I P°JP'HJ° Jd SAJj t$UJAM(11) °J P'1OU oupiorn

co
lIlttItt cc
111111 I.III cc
III 11111
.

cc
III lull

co—_
0 Ztoj.

co—
0
H 0

çQ...
Zw 0
co—
I—
I

I— 1—
1111111111
I—
P IT O9ç9fr6EEZ2T I—
E—OI— $ It U9E9p6UEZ2t c—ot- t it o9cc9t6uEcI c—ct— i ii o9cctucc8I
8
U8

II I I I I I I I I

cc cc
OIA 0 c
cc- Ic I I I I I

E—0l- 1? 8111 EE 6E 9V cc 09

co
tZOCA 0

co-
_______________________
E—OI— I II 81 c E 6E 9t cc 09 c—cl— v11 81 9t6uEc cc 09

II 'eIt II
t8 'e
I II I II II I j I II II II I I I I I I It It
co co co co .

o
_ o

co—
—'----——l H co—
o
H F—
co—_
0
. _________
I I
1—
1111111111
P II 81 c E 6E 9P E 09 IIIIIIIIII1t__
cE E
E—OI— P 11 81 6E9V' 09 E—OI- P II i c E 6E 9P E 09 I—
E—OI-
I1IIIIIIII
P 11 81 c E 6E 9P c 09

mi s ijjnj, xum piojpiujo id SAJ1J jduxs J&I ioj j •iui1

Proc. of SPIE Vol. 5432 59


1.4 17 I 11HTITH3 1H H3 1H4H3HI

___ ___ ___ ___


H3 11H17H3 H3

oi;H PLH
1HH3 7H3
Figure 9. Mueller matrix of oak bark vs. wavelength (tm) for normal incidence.

HH .;FHHHHHHHH
—10—341118253239465360 —1O-34111825323946536( —1O—34111825323946536t
H
—10—341118253239465360

___________

-;HHHH -;EHHHH -;H


—10—3 4 1118253239465360 —10—3 4 11 18253239465361 —10—3
H -;H H
4 11 182532 3946 53 61 —10—3 4 1118253239465360

_____ $ $

—10—3 4 11 18253239465361

4 1118253239465360
8 $ $

Figure 10. Mueller matrix of oak bark vs. sample angle for 1.324 tm.

60 Proc. of SPIE Vol. 5432


5. DISCUSSION

So far as we can determine, there are no previous studies of monostatic spectropolarimetric properties of vegetation.
Spectral measurement results already cited (e.g., Gausman et al) were concerned with spectral properties only, and the
polarimetric studies that exist8' 9, 16, 17 were bistatic studies and did not obtain Mueller matrices. Grant et al dismiss
monostatic measurements by stating, "At near—normal incident angles, specular reflectance from the leaf surface is
presumed minimal. . . and is often negligible..."

The investigation of spectral and polarimetric properties of different types of vegetation was the goal of the current
study. This spectropolarimetric signature has the potential to be readily exploited by remote sensing instruments for a
variety of applications. It is clear that reflectance and the Mueller matrix may be used to discriminate living plant tissue
from nonliving tissue. An examination of Figs. 4 and 5 shows that significant reflectance differences exist between
deciduous bark and leaves. Chlorophyll absorption and water absorption bands are greatly diminished in bark
reflectance as opposed to leaf reflectance data. Considering Mueller matrix information as presented in Figs. 7 through
10, obvious structure along the diagonal elements is observed near the water absorption bands for leaves (Fig. 7) and is
absent for bark (Fig. 9) at normal incidence. The influence of water in the leaves gives rise to a mirror-like polarized
reflection component that adds to the discriminating characteristics. For a single wavelength, 1.324 tm, the mirror-like
component is seen at and near normal incidence for leaves and absent in bark as demonstrated in Figs. 8 and 10. The
ability to discriminate between bark and leaves is much greater when spectropolarimetric information is utilized as
opposed to spectral reflectance information alone.

Proc. of SPIE Vol. 5432 61


REFERENCES

1. H.W. Gausman, W. A. Allen, C. L. Wigand, D.E. Escobar, R. R. Rodriguez, and A. J. Richardson, "The Leaf
Mesophylls of Twenty Crops, Their Light Spectra, and Optical and Geometrical Parameters," Technical Bulletin
No. 1465, Agricultural Research Service, USDA, March 1973.
2. D. M. Gates, H. J. Keegan, J. C. Schieter, and V. R. Weidner, "Spectral Properties ofPlants," Appi. Opt. 4, pp. 11-
20, 1965.
3. W.A. Allen and A. J. Richardson, "Interaction ofLight with a Plant Canopy," J. Opt. Soc. Am. 58, pp. 1023-1028,
1968.
4. W. A. Allen, H. W. Gausman, A. J. Richardson, and J. R. Thomas, "Interaction oflsotropic Light with a Compact
Plant Leaf," J. Opt. Soc. Am. 59, pp. 1376-1379, 1969.
5. W. A. Allen, H. W. Gausman, and A. J. Richardson, "Mean Effective Optical Constants of Cotton Leaves," J. Opt.
Soc. Am. 60, pp. 542-547, 1970.
6. W. A. Allen, H. W. Gausman, A. J. Richardson, and C. L. Wiegand, "Mean Effective Optical Constants of Thirteen
Kinds ofPlant Leaves," Appi. Opt. 9, pp. 2573-2577, 1970.
7. C. E. Olson, Jr., "Seasonal Change in Foliar Reflectance of Five Broadleaved Forest Tree Species," PhD Thesis,
The University ofMichigan, 1969.
8. P. N. Raven, D. L. Jordan, and C. E. Smith, "Polarized directional reflectance from laurel and mullein leaves," Opt.
Eng. 41, pp. 1002-1012, 2002.
9. V. C. Vanderbilt, L. Grant, and C. S. T. Daughtry, "Polarization of Light Scattered by Vegetation," Proc. IEEE, 73,
pp. 1012-1024, 1985.
10. D. H. Goldstein and D. B. Chenault, "Spectropolarimetric reflectometer," Opt. Eng. 41, pp, 1013-1020, 2002.
1 1. D. H. Goldstein, D. B. Chenault, and M. Owens, "Spectropolarimetric Reflectometer," U. S. Patent No. 6.618,145, 9
September 2003.
12. R.M.A. Azzam, "Photopolarimetric measurement of the Mueller matrix by Fourier analysis of a single detected
signal," Opt. Lett. 2, pp. 148-150, 1978.
13. D.H. Goldstein, "Mueller matrix dual-rotating retarder polarimeter," Appi. Opt. 31, pp. 6676-6683, 1992.
14. D. H. Goldstein and R. A. Chipman, "An Error Analysis of a Mueller Matrix Polarimeter," J. Opt. Soc. Am. A 7, pp.
693-700, 1990.
15. D. B. Chenault, J. L. Pezzaniti, and R. A. Chipman, "Mueller matrix algorithms," Polarization Analysis and
Measurement, D. H. Goldstein and R. A. Chipman, eds., Proc. SPIE 1746, pp. 23 1-246, San Diego CA, 19-2 1 July
1992.
16. L. Grant, C. S. T. Daughtry, and V. C. Vanderbilt, "Polarized and specular reflectance variation with leaf surface
features," Physiol. Plant. 88, pp. 1-9, 1993.
17. W. G. Egan, Optical Stokes Parameters for Farm Crop Identification," Remote Sensing of Environment 1, pp. 165-
180, 1970.

62 Proc. of SPIE Vol. 5432

Potrebbero piacerti anche