General Disclaimer

One or more of the Following Statements may affect this Document

This document has been reproduced from the best copy furnished by the

organizational source. It is being released in the interest of making available as

much information as possible.

This document may contain data, which exceeds the sheet parameters. It was

furnished in this condition by the organizational source and is the best copy

available.

This document may contain tone-on-tone or color graphs, charts and/or pictures,

which have been reproduced in black and white.

This document is paginated as submitted by the original source.

Portions of this document are not fully legible due to the historical nature of some

of the material. However, it is the best reproduction available from the original

submission.

Produced by the NASA Center for Aerospace Information (CASI)

https://ntrs.nasa.gov/search.jsp?R=19780024582 2020-06-04T01:24:40+00:00Z

N 16iw%SATechnical Memorandum 79620

Red and Photographic Infrared LinearCombinations for Monitoring Veqetation

Compton J. Tucker 1(NASA-TM-79(X) FED AND FH 17T r G g 4PHIC N78-32525

`i INFRARFI) LINEAR CCMFINATTCNS FCF MCNITCRINGVEGETATIO4 (NASA) 37 F HC 4C?/*F A01

CSCL 0eF UnclasG3/41 31096

t

MAY 1978

National Aeronautics andSpace Administration

Goddard Space Flight CenterGreenbelt, Maryland 20771

9

^ J

I M79020

R.'D AND 1'11Of0(;RAI'IIIC INFRARED LINEAR COMB [NATIONS FOR

MONITORING VEGETATION

Compton ! 'PickerFarth Resources Branch

s

May 1975

NASA/Goddard Space Flight CenterGreenbelt, Maryland 20771

t

> y1CONTENTS

Page

AB:. "RACT .......................................................... A si

iN I'RODUCTION ..................................................... 1 js i

PREVIOUS WORK ........... ......................... .............. 1

BASIC PRO"ERTIES OF THE IR AND RED RADIANCES WI"I I I RESPECT TOGREEN VEGETATION ............................... ................. h

DF.SCRIPTIONOF RESEARCH UNDERTAKEN ............................. 9

METHODS AND ANALYSIS ............................................. 9

EXPERIMENTAL RESULTS ............................................. 10

PIIENOLOGICAI. CONSIDERATIONS ..................................... ' !

EVAL UATION OF DIFFERENT I BANDWIDTHS ............... ........ 24

CONCLUSIONS ....................................................... 26

REFERENC!:S .................. ..................................... 26

TABLES

Table

I Statistical Summary of the Biophysical Characteristics of the Sample Plots.A Statistical Description of the Vegetation Canopy Characieristics for (A)The Thirty-Five 1/4 N1 2 Semple Plots of Blue Grama Sampled in June 1972,(B) The Forty 1/4 M 2 Sample Plots of Blue Grama Sampled in September1971, and (C) The Eighteen 1/4 M 2 Sample Plots of Blue Grama Sampledin October, 1972 . ............................................... I1

2 C oe fficients of Uetennination for the Simple Regressions Between the NineGreen and Red Radiance Variables and the Canopy Variables for (A) 35Plots of Blue Grama Grass Sampled in June, 1972; (B) 40 Plots of BrieGrama Grass Sampled in September, 1971. DI = Green - {Zed, SUM =Green + Red, VI = DIF/SUM, TVI = SQRT (VI } 0.5). 13

iii

TABLES ( Continued)

Table

PE3 Coefficients of Determination for the Simple Regressions Between the NIJ._

Red and IR Radiance Variables and the Canopy Variables for (A) 35 Plots,A Blue Gran ► a Grass Sat npleJ in June, 1972; (B) 40 Plots of Btic Gran:aGra. i Sampled in September, 1971 , and (C) 18 Plots of Blue Granta GrassSampled in October, 1972. DIF = IR - RFD, SUM = IR + RED, VI = DIF/SUM, TVI - SQRT (V1 + 0.5) ....................................... 14

_ 4 Correlation Maxrix Between the Sampled Plot Variables for (A) 351 /4 m2Plots of Blue Grama Grass Sampled in June, 1972, (B) 40 1/4 m 2 Plots ofBlue Grama Gass Sampled in September, 1971, and (C) 18 1/4

in Plots of

Blue Grama g rass Sampled in October, 1972 . .......................... 22

Coefficients of Determination for the Simple Regressions Between the NineRadiance Variables and Six Canopy Variables for the 33 Plots of Blue GrartaGrass Sampled with the 0.70 - 1.60 µin grating in June, 1972. (A) is for the0.75 - 0 .80 µm bandwidth, ( B) is for the 0.80 - 0 . 90 pill bandwidth, anti (C)is for the 0.75 - 0. 40 µ ►n bandwidth. ...... ...... I ... I ............. 25

ILLUSTRATIONS

figure Page

Spectral retlectancc s for d ► y soil, wet soil, and asymptotic green reflect-ance. The dry soil and wet soil curves are the average of five. bare soil plotsmeasured when dry and Het, respectively. The asymptotic green reflect-ance curve is from a plot o f blue grama grass having a total dry biomassof 530 g/m 2 . .................... ............................... 5

Radiance plotted against total wet biomass for the (a) 0.63 -- 0.69 and(b) 0.75 - 0 .80 µ ► n intervals for the June data. Similar results were ob-tained for total dry biomass, leaf water content, total wet biomass, andtotal chlorophyll content for this sampling time. The total wet biomasswas predominantly green and contained little dead vegetation (Table 1). .....

3 Comparisons between tl:e green radiance (0.52 - 0.60 µ ► n) and the green/red radiance ratio. Refcr to Tables 2 and 3 for the r2 values associatedwith this portion of the analysis and Figures 4c, 5c and 6c for comparisonsto the it/red ratio for the same data sets . ............................. 15

r^

1LLUS'1 RA"f IONS (Continued )

FigureN;tge

4

The nine radiance variables plotted against the total wet biomass for the 35plots sampled in .tune, 1 1472. (A) red radwii.e, (14) it radiance. ((•) it/redratio, (D) square root of the it/red ratio, (I = ) it-red radiance difference, (F)ir+ red radiance suns, ((,) vegetation index. (11) sum/difference, and (1)transformed vegetation index. Refer to Table 3 for the r2 values betweenthe nine radiance variables and the other five plot variables.

F.r 5 The nine radiance variables plotted against the total dry biomass for the 40Olots sampled in September. 1`)71. (A) red radiance, (B) it radiance ((') ir/red ratio, (D) square root of the it /rnd ratio. (1:) it-red radiance difference,(F) it + red radiance sum, (G) vegetat i on index, ( 11) suns /difference, and (1)transformed vegetation index. Refer to Table 3 for the r 2 values betweenthe nine radiance variables and the other five plot variables . ............... IS

6 The nine radiance variables plotted against the leaf water ^ ontent for the40 plots sampled in September, 197 1. (A) red radiance (13) it radiance,(C) it/red ratio, (D) square root of the ir;'red ratio. (I') it-red difference,(F) it + red radiance suns, (G) vegetation index, (1I) sum/difference. and(1) transformed vegetation index. Refer to Table 3 for the r' values be-tween the nine radiance variables and the other five plot variables. .......... 1`^

7 The nine radiance variables plotted against the total wet biomass for the

IK plots sampled in October. 1072. (A) red radiance, (R) it radiance, (C)it/red ratio. (D) square root of the it/red ratio, (E) tr-red radiance differ-ence, (F) it + red radiance sum. (G) vegetation index, (11) sum/difference,and (1) transformed vegetation index. Refer to Table 3 for the r = valuesbetween the nine radiam:e variables and the ether five plot variables. ........ 20

V

{(

L-

Ri.L),".":1: Hi()TO(;RAPFiI(' INFRARED LINEAR COMBINATIONSFOR MONITORING VEGETATION

Compton J. TuckerEarth Resources Branch

ABSTRACT

In collected spectrometer data were used to evaluate and quantify

_i

the relationships between the various linear combinations of red and photo-

graphic infrared radiances and experimental plot biomass, Ieaf water content,

Mid chlorophyll content. The radiance variables evaluated included the red

end photographic infrared (ir) ► alliance and the linear combinations of the ir/

red ratio, the square root of the it/red ratio, the it-red difference, the vegetation

index, and the transformed vegetation index. In addition, the corresponding

green and red linear combinations were evaluated for comparilive purposes.

Three data sets were used from June, September, and October sampling period,.

Regression analysis showed the increased utility of the it and ► ed linear

combinations vis-a-vis the same green and red linear combinations. The red

and it linear combinations had 7% and 14% greater regression significant than

the green and red linear combinations for the June and September sampling

periods, respectively.

The VI, TVI, and square root of the it/red ratio were the most significant

followed closely by the it/red ratio. Less than 6% difference separated the

highest and lowest of these four it and red linear combinations. The use of

these linear combinations was shown to be sensitive primarily to the green

v ►

ICa) area or green leaf hioniass. As much, t here linear con ► hinationm of tht, red

and 1 1 hotographic it radiances can he employed to monitor tilt , photosynthetically

active hiomamm of plant canopies,

— s

vu

,.

RED AND PIIOTOGRAPHIC INFRARED LINEAR COMBINATIONSFOR MONITORING VEGETATION

INTRODUCTION

The use of photographic infrared (ir) and red linear combinations for monitoring

'

vegetation biomass and physiological status have recently become common in the remotes

sensing community. Accompanying tlus increased usage, however, has been a lack of

detailed analyses concerning limitations of these data and their applicat,.m(s) to vegetation

monitoring. Quantitative information regarding the various it and red linear combinations

and the constraints involved in the use of these methods will enable more advantageous

application of these techniques. It will also prevent over-ambitious use of these techniques

when other methods would be more applicable. I will examine ground-collected grass

canopy spectra in an . ttempt to quantify the relationship between the it and red linear

combinations and properties of plant canopies.

PREVIOUS WORK

The use of near infrared/red ratio method for estimating biomass or leaf area index

was first reported by Jordan (1969) who used a radiance ratio of 0.800/0.675 pill to

derive the leaf area index for forest canopies in a tropical rain forest. This application

of the it/red ratio used the transmitted light at these wavelengths which was sensed on

the forest floor (Jo rdan, 069). Subsequent work was reported by Pearson and Miller

(1972) who developed a hand-held spectral radiometer for estimating grass canopy biomass.

The instrumentation aspect of the hand-held radiometer is described in Pearson et al.

(1976).

h

Colwell 0973 and 1974) presented a detailed study of bidirectional spectral reflect-

Mice of pass canopies. lie concluded that the it/red ratio was effective in somewhat

normalizing the effect of soil background reflectance variation(s) and was useful for

estimating biomass. Colwell also cautioned that the ir!red ratios may worsen angular

effects rather than alleviate them under certain conditions. Smith and Oliver (1974)

^ s, corroborated several of Colwell's ( 1 973 and 1974) conclusions using it stochastic

,anopy model vers e.... Colwell's use of Suits' (1972) deterministic canopy model.

"Ilse it/red ratio method has been applied to Landsat image analysis of range biomass

by Reuse et al. ( 1973 and 1974). C'arneggie et al. ( 1974), Johnson ( 1976), and Maxwell

(1976), among others.

Carneggie et al. (1974) used it ratio of Landsat MSS7;MSS5 and found that the

ratio curves, piotted as a function of time, peaked during the period of greatest forage

production. 'lltereafter, the curves fell off signalling the period of drying following the

maximum green period for their California study site. Once the curves had leveled off,

Carneggie et al. (1974) concluded that all annual vegetation had dried.

Rouse et al. 11973 and 1974) analyzed Landsat MSS data and developed what

they referred to as the vegetation index (V1) and transformed vegetation index (TVI).

They found that although a simple ratio of MSS7/MSS5 could be used as it

of relative greenness, location and cycle deviations would introduce a large error component.

Me difference of the MSS7-MSS5 radiance values, normalized over the suns of MSS7 +

MSS5, was used as an Index value and was christened the V1.

R~ 411110^

VI = MSS7 - MSS5MSS7 + MSS5

(1)

To avoid working with w.gative ratio values and the possibility that the variances

of the ratio would be prolx)rtiomil to the mean values (i.e. a Possion distribution)Fi

the constant of 0.5 was added ant square-root transformation was applied to the VI.

rvi = VI -+0- 5 (2)

r Vie VI and TVi were then applied to [.andsat data. MSS bands 6 and 7 were both

evaluated as the near infrared band. Rouse et al. (1974) reached several conclusions.

Among them were: ( 1) that [Andsat VI and TV[ methods could lx used to monitor

rangelands and wheat crops; (2) the close relationship between g reen biomass and TVI

should allow researchers to follow crop development as ground cover, biomass, and leaf

area indices increase; and (3) phonological inferences :ol.ild possibly be gleaned for certain

crops or range types and used to monitor these types of vegetation (Rouse et al. 1974).

Johnston (1976) and Maxwell (1(,76) also analyzed Landsat imagery using ratio

methods. They concluded that the ratio of MSS6/MSS5 was slightly more statistically

significant than MSS7/MSS5 and that both r atios were useful in monitoring green biomass.

An explanation of the apparent greater utility of MSS6 vs MSS7 for rangeland biomass

estimation in low biomass situations based upon soil-gyeen vegetation spectral contrasts

has been proposed by Tucker and Miller (1977).

Other researchers have also used the VI for Landsat analyses. Blair and Baumgardner

(1977) monitored several hardwood forest sites using Landsat imagery. They used the VI,

which they refer to as the "band ratio parameter," and hound that the gro-enwave effect

could be monitored for these vegetation types using Landsat imagery.

3

Asnlcy and Rea (1975) reported how L mdsat MSS5 and %ISS7 data were used to

depict plhenolokical change. 'They also uwd the VI and found that it increased with

foliage development and decreased with senescence. PIC VI w:rPs found .o reduce in-

fluencing multiplicative effects such as solar elevation differences between overpasses

(Ashley and Rea 1975).

. s

In addition to the above reviewed l.andsat analyses. Kauth and Thomas (1976)

arid Richardson and Wiegand (1 177) have proposed L.andsat vegetational analytical

methods using, at least in part, it and red linear combinations.

Richardson and Wicgand (1977) have proposed a departure from the soil background

line with the perpendicular ,!getation index (PVI):

PVI = v1 -(Red Soil - Red veg)' r OR Soil - I R veg )2(3)

where:

Red S,,il= soil background red reflectance or radiant,.-

Red veg = vegetation back g7ound red retleztance or radiance

IR"Al = soil background it reflectance or radiance

1R Ve2 = vegetation background it retlectance or radiance

Kauth and Thomas (1976) have developed a technique for transforming Landsat M';S

information in foie-dimension data space using the four MSS bands. from this, a soil

brightness index (SBI) and green vegetation index (C.VI) were calculated as follows:

S111 = 0.43*MSS4 + 0.63*MSS5 + 0.5 1)*MSS6 + 0.26*MSS7 (4)

and

(;Vl = -0.29*MSS4 - 0.56*MSS5 + 0.60*MSS6 + 0.49*MSS7 (5)

t

r'

4

F

-- -A

Note that all the Slil independent variable .06ficients are positive while the GVI in-

dependent 7ASS4 and %ISS5 are negative. The SBI eitabiishes the data space

of soils and the G %7 1 departs from it, in a negative or absorptive fashion with \ISS4 and

MSS5, approxim"► tely the same coefficients for MSSO for both models, and a lossitive

departure for MSS7. This aim) follows from Figure I.

r

Deering (1 Q 78), in a recent and comprehensive analysis of Landsat rangeland biomass

monitoring has reported that the VI and TAI approaches hr evaluated were slightly

more significant with respect to green biomass than the PVI of Wiegand and Richardson

(1977) and the GVI of Kautli and Thomas (1970). In addition, Deering ( 1978) reported

9 that the denominator of the V1 and TVI (i.e., MSS5 + MSS6 or MSS5 + MSS7) were

highly correlated (r > 0.05) to Kauth and Thomas' 11976) sail brightness index (Still.

W —.e;

40GREEN VEGETATION PLOT 20074

(TCTAL DRY Sit— MASS = 530 g/m2)

^ 130 DRY SOIL PLOT

SAME SOIL.w ONLY WET

UA

►0

0 35 OAC 0 45 0.60 0.56 0 80 0 66 0 70 0 75 0 80 0 86 0.90 0 96 00

WAVELENGTH (pm)

Figure I . Spectral retlectances for dry soil, wet sail, anLasymptotic green • flectance. The dry soil and wet soilcurves are the average of five bare soil plots measured whendry and wet, respectively. The asymptotic green reflectancecurve is from a plot of blue grama grass having a total drybiomass of 530 g/m2.

k-

ma; -itv of j r and red linear combination work have used Landsat data.

! "Memasu (1974), however, rcpx,its on a ground-bawd reflectance study of crop types

where various ratios were investigated. Wheat, sorghum, and soybean plots were moni-

toted periodically during the growing s.asem using a spectrom.ter. Kanemasu (1974) co n-

eluded the 0.545/0.655 pm wavebands provided useful information regardless of crop

atype. For all crops studied, the green/red ratio closely followed crop growth and develop-

ment and appeared to be inore desirable than the near-infrared reflectance as an index

of growth. The grc:­ /rca ratio will be evaluated in this paper also.

Recent work by Nalepka et al. (1977) and Tucker et al. (1975) have used it and red

Jjta to forecast winter wheat yields and monitor agricultural crop vigor and condiU,)n,

respectively. Nalepka et al. (1977) evaluated various green measures using Landsat data

and cop.-e -eded that most are useful but stress that no new information is created. Tucker

, 19/5 1, monitored several crop types using a hand-held radiometer.

BASIC PRQUIU ILS OF ' HL IR AN12 R RADIANCES %VIFI KESPECT TO GRFENV I-,G ETA I ION

It is perhaps prudent to briefly review the basic properties of the 1r and red radiances

with respect to green vegetation before embarking on a 'Aaded analysis of the various

linear combinations.

Me red radiance exhibits the nonlinear inverse relationship between integrated

spectral radiance an(! green biomass while the near infrared component exhibits a non-

linear direct relationship.

6

The relationship betwe-vii the 0.03 - 0.69 pm radiance and green hioniass results from

strong spectral absorption of incident radiation by the chlorophyll,. It is apparent that a

spectral radiance asymptote is mo :e quickly reached for the 0.0 - 0.09 pin red radiance

than the 0.75 - 0.P0 um near infnncd radmace 01gure 2) (Gausman et al , 1976. Tucker

1977x).

The 0.0 - O.he1 pin radiance is inversely proportion.il to the amount of chlorophyll

present in the plant canopy and thus is sensitive to green or photosynthetically active veget-

ation pr wnt.

17te 0.75 - 0.90 pnf radiance is sensitive to green or photosynthetically active vegeta-

tion and, to a lesser extent, the dead or noriphowsynlhetically active vegetation (Colwell,

1974: Tucker, 1977b and 1978).

The relationship between the 0.75 - OM pin infrared radiance and biomass results

from the lack of appreciabl e spectral absorption in the 0.74 - 1.10 pin region and the higli

Oegree of mtra- and interleaf scattering fit plant canopy. In the absence of spectra!

absorption, p! -.)ortionally more incident spe-ual radiance escapes from the canopy than is

absorbed. Thus the spectral radiance in the 0.74 - 1.10 pin region is said to be en:fanced or

increased over the level t)!' radiance of the background material.

Discrimination of vegetation hioniass is strongly dependent upon the soil surface-veg-

etation spectral reflectance or radiance contrast (C:olwe11. 1974). For this reason, softie

wavelengths are I'm sup .• rior to others I'm discrimination of green vegetation biomass

(Figure I).

7

h

v

-+ o

bO^/ y M

v

^ j

Q

^ ^ C

^ O

U

O V ^V

^ C K

^ P ^\.O

a.

cl

N •.r

~ E

S

O

o^ tn

v3 0 _o

0,2 «•

H .^

3_c`^

^ ro

O

a^ U

°' y F

O

d13

E

^ bUb'a

@ C

^ ^

EE

51

^' N

.4

^ n

^

--

o

_

wx;

O

*LO

N

°E

0

a.Ow

•

N

O (/)

^

w Q

v^ Z

• v

o O

X

cow

0 7

•C

►W

•VO

N0

LU

•

O

O ^

•

•

tl II

^ U

r

O }

c— '4Q

•^

• D

•^

O

O O O O O O O O O O

M

O

Oi

co

h

w L

V

30N

di0

VH

HI

0r

co

m0co

C)kohfi

o 7

co

Ou u

00

00

0I^

(D

LO

1M

N

30N

VIC

IVH

03U

---, oW014OE

m

1^1It0N? U)

O Q

co

09

o co

o

11ZJN w

om ^

r }

Nr 0

0w00

8

h,

I hr Vxecn region, by contrast, has a lower soil-seen vegt • tation rellrctance c nnast

(Figtire 1). 'p his results front fact that the chlorophvlls are slightly absorptive in the

t!reen region (extinction coefficients ~ 10) while much more absorptive in the red region

(extinction coefficients of — 40 - 90) (Salisbury and Iles., 14119). The re•Litionship

1.

F between the green radiance and green biomass is similar to the same relationship between

the red radiance and green biomass (I-igure `a and .t).

DESCRIP" IO N O F RUSFARCH UND RTAKFN

The x^ ii k reported herein will examine )!round -collected in -sit t spectrometer data

and evaluate the green/red ratio method ul haneinasu ( 1074) and contrast that with

the it/red ratio methods) to dcteriune which are superior for the June, SeptenAter.

.uid Oclollet data uts. The utility of the %arious green vegetation measures usin)t the

drffetent it and red linear combinations will also be evaluated.

MI: HODS AND ANALYSIS

The data used in this evaluation haw all heen previously described and w ill not he

redescribed in this report. The .tune anc: Septemher data sets are described in 'Tucker

and Maxweli ( 1970) while t l tc October data is described in Tinker ( 19781.

he narrow handwidiIt radiance curves ((1.005 Nit bandwidth) were ninterically

ii-coated to approximate three bandwidths: 0.5' 0.00 pin for the grecrt, 0.03 0.69

pin fur the red: and 0.75 0.80 pin in the photographic intiared, fhe radiance comes

resulted front the product of the spectral reflectance and spectral irradiance curves (sec

Tucker, 14770.

L_ J

.4

Regression analyses identical to Tucker 1 1977) were performed. The various grass

canopy variables (Table 1) were regressed against the following it and red and green and red

a

radiance variables:

I . red radiance (0.63 - 0.69 pin)

2. it radiance (0.75 - 0.80 µm 1

3. Wired

4. SQRT (ir/red)

5. it - red

6. it + red

7. Or - red)/(ir + red)

8. (ir+ red)/(ir - red)

9, (ir - red)/(ir + red) + 0.5

red radiance (0.63 - 0.69 µm)

green radiance (0.52 - 0.60 µm)

green/red

SQR C (green/red)

green - red

green + red

(green - red)/(green + red)

(green + red)/(green - red)

(green - red)/(green + red) + 0.5

LAPF RIM ENTA L RESULTS

The nine spectral variables involving the it-red data and the green-red data were

regressed against the six canopy variables measured for the June and September data and

the four canopy variables measured for the October data. This resulted in 288 separate

comparisons which defies concise presentation. The results of this analysis will be presented

for the canopy variables total dry biomass and leaf water content.

The linear combinations of the it and red data were regressed against the six canopy

variables as were the -,ame green and red linear combinations. Without exception, the it-red

linear combinations were snore significant in an regression context (Tables 2 and 3, Figure 3).

This supports the majority of Landsat analyses which have used it and red data instead

of green and red data for vegetational analyses. In addition, these results show that the it/red

10

I

c

^,

.googoL

r LA

O N

y^ C

N4 C' tlid

v ^ _

NL

7UQ

r^

4 D

t`

ON

N

.^6..0CQ

6.V i_

G

Lt N

1b

U ^' E

G. f•-' Q

^ y

Q O

O

Ln

LL.

+^ u

.^+

O O

C

O u

c, ;^

u

^4.

^t c

u7 v

N

c ...

ro 4Ctd

^ON

^ ^

er

v1

^O

MCA

M00

^O

rlaO

M000

^r

.rMM^-

O^

in a

cr„

o._

JI.

O^M

^MV1O

OQrJ

V1O

^d

d00

Md

viO

0MN

doo

Ooc

Oo yu

^^'^

M~

cyC

11

^1

11

V Z^

^f

..N

Nrpp+ O

^)

kn

LA

^ca

c^EroE

Eo

ai

^E

^. EE

-EE

3 E

v

InV

"'JJJDODO

H^

b

3D

0D

I•-- s

o ^

11^

.^r!

MM

^ti

6►.^

ei

^P^

UOM

r,n

^r.

C ._

`d

Ot%1

cG

O'--00

—Oa

._opr-

Or-C00

ODO

v1,

^0

..;00

v'^

00000

UO

O0,.,

^rO)

o0

00o0

\

00r^+1

^

M

00~l

NG

^,

o0.M.

^ 0J^

N^'J

N..+

C/I Q

...

C—

o_.

u

VN

Q^

00r!00

O^

r+1

ellM

r!C

t'7O

C !

Or`

OO

`'^N

'Ar—

v i

O00

r~rJ.>

M'^

oen

III

rCA

Mr^

r-O\

Vi

Ji

vf

/^tVE

^C 1

^^

y rC~ .^

.

AC

V^C

' iD

i'J^N

NO

V....

L.fJ

.p ^

A ^

17

G;

G F

►.

Ef O

UJ

^^

^^L

`y

_^+

00

::°

0"

°3

Q^to

p'

.°+w

^A

3F,

^i

0.+0+-'

^....p

J^

^

as3

^D

^--^

UU

3G

.-aJ

1

h

;N,&

M.8

2 Ev G

.. cv-

CA_L

av

vL^

a m

,^b o

0d ,n +c "O ^

a`

NY C r-

f-

6!3 ^

cif

JE

1 N

^ ^

vOG

E C

C L

r r h

^ J

11

C m .,

ca n

'D

O a

Jy M

v

0n

c`"

ts,

L

^ e

_a

0— t

Jc>

00f-x

ao

grM

^O

t+iC

M^+

h—C

OO

CO

C000000

iL

op

a0 00 r^ .Y

`^

C O C O C C

O C C O o 0

'V

^,,^

OM

StM

M^C

•f ^

Dr r^

10

co

o r

- x

so

v r

-

c00000

cam,^c

o rn0oo000

Sin ao

x 0\ ^ a

rJ

M r o

0o

r- oo r- M

o0

M r•! v

i

M0 M

v:000000

0 0 0 0 o 0

r!

r! vl^

r Z

O n

or M

N00

...,of ^

? ^

t '7 N

MM

triN

^O O e 0 0 0

0 0 0 0 o C

F'-

^•! in

c V

l so 00

t70 r r ! 0

t- 30 a

c v

r-cn -- S

D ff) O

try

^',' ^

000000

000000

V'vQ^

N in

'o in

,o

ao

v •

C r

0%.)0

t-- xx

d;r-

M^N

C

I!61

O O o 0 0 00 0 0 0 0 0

VCOs N

N tt 00M

AD` c x

yr- r

- 0

C r

CA

xr!

-7 Q

J000000

o00o0o

00 O

rJ rJ

Vi Q

- ^D

Sr00 w

S 0

0 M

O^1t r

! r et C

M

0 0

0 0

cc

O O

O C

O O

O

r,a

vE >.

I I;

^

a M ..

Eu

o a

0 0

0a

.0.^

ccoo

m^ °

ca c

a ^

^C

C^

C c

3^3

0v

3C

^u

yA

:a .. "

^

:vy

iv r. L

O c

o a

:,0

0 0

13

cE^

O v

^^ a vs

A a

i^. C

G ti

11

4, N

^cgym°.^

p ac

ro O

CL

j m^ fJ

y c

'0 S =

JJ ^

Ccr•^

- GC

y N

^ 1r

w O

Vc mr1

H V

ly O O

N

00 v

40, o

C..1

u>

^F

-

pp v

1 •

! r! r

v,

M

co) C

` MO

Q O

O j

O A

Q O

00D

00v,

O O

O O

E-

000000

00000c

0000

-tr00

f ! MC

^ ON

C^ 00 J

"i r-

885888

0000

r•O C O 0 0 0

0 0 0 0 0 0C O O O

NC

` 30

f! ^^ f

^ rJ

rJ r

8 8

00 3

0 a,

T H

00

Z It

^c ^

00

V 0

0 0 0 0 0 00 0 0 0 0 0

O O O O

00121

)003080

^^

0 0 0 0 0 0O O O O O=

O C C O

;!.C

^.c

0C

,v O

^k

,viv

'^ON

oc r

O1C

00 x C. 00 14 a ,

000000

00 ,q

v,000000

f! fl rJ wt

oo00

WOG

a0^, M

r J O

kn

Z00 00 of C , v) 00

M

knV) M 00

W)

O o

O

MO 8 O O

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0

r-Cm

rn x

x 00 O^ o0

In oo

Wl-

O O O OY

0 0 0 0 0 00 0 0 0 0 0

0 0 0 0

Z '!t N

i W+

O o

c!t r

! 00 ^O

00 r JN

M O

CrJ

a3000-x

30r0

0O

tnA1

^0ro

kn

r-r r

r-"

000000

CO

0000

O000

Q00 O

O f ! r +

^rte

, v-, O ^ r- C

r IC. 00 IG

.r00 o

0 Q

; x M

CfJ

r ^t O

r+iz z .0

z

OG0 0 0 0 0 0

0 0 0 0 0 0O O O O

tr^'

DP

!F

C C

O.

C C

O t

E E

C -'

a u

p

0 0

GF)

0 0 0

•^

G`

O O

O C

-0

mC

nJ..

O C

`^

m^iU

m C

Fy

Om

^o

>^

cv ^

' y y

3 t

^v

^' ^

v 3

,^o

v

^, .^'.^

:,3a

^ ^

u^

3a3

u

0 0

a^ ^-

O-,

O O

v ^

'C

O O

y G

14

+w

r.

• •V •r 1.

•rI. rI.A•

• \

•• •• ••

Tor" •• 1 •wr••t w- 1

e•

••

`.. •

.> L _ L_1_ L^__1 • 1^^

, .° o ^x 1e+

ll•t •AT•• COM I t MI 4 n.

T_T.^- T-r- r T T - ^..-

s'• • f

• ,t

iAT

•

TOT" 011. 911W1. 191.1')

.r T ...^ ^ .. T r- T

T

• ewt. ..

VU..••.1 W

t

• M

e M ^ 1.1 V•

101" 0.160 0040"

..

1• ,

• t•

nr + •

•

, a•• ,ls ^r „r rw

TOTAL ••T •10r••• 4 1

... ...r_T_.1

, ..rw.epr... •a•

•

, . •w r

1EM V•ll•C„N11-11V-

Figure 3. Comparisons between the green radiance (0.52 - 0.60 µm) and the green/redradiance ratio. Refer to Tables 2 and 3 for the r2 values associated with this portion of theanalysis and Figures 4c, 5c and 6c for comparisons to the it/red ratio for the same data sets.

15 ORIGINAL PAGE 15-OZ EOOR QUALITY

ratio, the Aluare rout of chic it/red ratio, the VI, and TVI are sensitive to the photosyntheti-

cally active biomass or the green leaf area present in the grass canopy. This is evident from

the June d.,ta where — 8Wr of the canopy was green or alive and only — 2017r) was standing

dead vegetation (Table 1). For this data set, the dry brown biomass canopy variable had the

lowest r'- when regressed against any of the nine it and red radiance variables evaluated (Table.

&' 3; Figure 4).

The September data, comprised of-52% live and — 48% dead vegetation (Table 1),

ilso showed the lowest r2 values for the dry brown biomass canopy variable when regressed

against any of the nine it and red radiance variables (Table 3; Figures 5 and 6).

This confirms quantitatively that the it/red ratio, the square root of the it/red ratio,

the it-red difference, the V1, and the TVI are primarily sensitive to the green leaf material

or photosynthetically active biomass present in the plant canopy.

The it-red difference, TVI, V1, square root of the it/red ratio, and it/red showed the

greatest regression significance for the June and September data sets (Table 3). The it-red

difference can be excluded from further consideration because it will not compensate for

different irradiational conditions.

A 4% range existed between the it/red ratio, the square root of the it/red ratio, V1, and

TVI for the June data in terms of explaining greater regression variability. A 6% range

existed for the September Ieaf water content variable and the it/red, square root of the it/red,

VI, and TVI regressions, respectively (Table 3).

16

S.

17

LI --

c 1lilC OR Qu —;r

r

tort wr •r•ruN,♦^ rc • Y .. r •rut... N r9 tart •• • •rot•N rN'r

^rr r - • • • • • r •r. s .^ ^ •r err .•

f0•M wr •rOt•••N^'1 rOrr w r •earl•►• I••rr 10rY w• •r+rr••• ^^

• r•

a i'Ob

_IA; Tir. . ^..__m _.l_ r u^_... .•

. • lr tar ro

r. . +r r .r,rr.... TCI•, •'1 r •r•yyr p w.ltart wl r•aY•N br>•'/

Figure 4. The nine radiance variables plotted against the total wet hiomass for the 35 plotssampled in June, 1972. (A) red radiance, (13) it radiance, (C) it/red ratio, (I)) square root ofthe it/red ratio,(E) it-red radiancedifference,(F) it + red radiance sum, (G) vegetation index,(11) sum/difference, and (1) transformed vegetation index. Refer to Table 3 for the r'- valuesbetween tho nine radiance variables and the ether five plot variables.

(:

s

.ptY Wt fM11•N y+VI

••

• • .••

tOtM O•. • ..fWN 1• r'.

rr 1

r -

laid

.••

.rL t • i•

•

.• .__1 _i.-j

• w ^

tulN p•. •.Ir••• ^••f1

•r

trl

• ^ r rt^ u .r a

toy" M. ••fr••• 0."

J►

Ir_ •:1

•r --r .._T _- i . -T• an •

•r

•• f

•r ^

t r-YT T_ - ior

•

r

••

"r•

.r • —.L.L_. _l 1 h ._L.--^totY u•• •uw.•• i•r^

'r r - • ---T^ T^—T•

•

F fff

.01•. p•r bOr••• o ^

Figure S. fhe nine radiance variables plotted against the total dry biomass for the 44 plotssampled in September, 1971. (A) red radiance, (B) it radiance, (C) it/red ratio, (D) squareroot of the it/red ratio, (E) it-red radiance difference, (F) it + red radiance stilt). (G) veget-ation index, (11) sum/difference, and (1) transformed vegetation index. Refer to Table 3 forthe r'- values between the nine radiance variables ant] the other five plot variables.

18pVJGTNAL PAGE I9

OF pooR QU ALITY

L_

it

r.-

46

Y pC 1^

01.

t.

i ^•

' I

•UN NI•• [Ow•l•T 10 . 1

t

•

1 1 J 1rr ^

\1Y ••11 r yr, ► .'1

r• '

r-1 -_T

, Mesr•^^ ^ ••

•

•

•rr • •

i

i

jam•

JLaw ► ..11 1 C.ObT••T » ..I

/.. •• r l. .. 11.r 1µ•Y

010-^. .^

' • •

o

•1•

^ i•

^^i 1 • • I I L

llM •111• COII f/.T 1•.'1

Ir T-- T--T--f'-r---r_'^

r

I

bI

b 1./ A

r•1 bamp CO.T••I WI••I

••

M

re rn

11Y /•tf• CO•TI MI ,•+rl

Figure h. The nine radiance variables plotted against the leaf water content for the 40 plotssampled in 5eptemher, 1071. (A) red radiance, ($) it radiance, (C) it/red ratio, (D) squareroot of the it/red ratio, (E) it-red radiance difference, (F) it + red radiance sum, (G) veget-ation index, (H) sum/difference, and (1) transformed vegetation index. Refer to Tahle 3 for

tine r2 values between the nine radiance variables and the other five plot va:' ahles.

I O

L

I_ 1

r.- A.

f.w•rpwr.••

am

to or.

.. % •f •

i

1 1 = 1_J M^

. .. M. NWr..f I. w 1 IOM ut . ..MM416 W-) .1' Y•a^Y.N N « I

r r— I --r r— T- . »( r_ T _Y _t__r__^ .

a.

/gr...rpW..

° r r.. r ,rpY M/ MOWM M ^'I TOTµ wt •)OWN W ^'I tOtY Nl .wn.M ^t

• ♦

• • •

• I — t t .- —^ o p ..._1_ I.—. 1 I I _L l_ L—_l -_Ja, a r .• x .^ u ,A

.Ot.. Yt, •u.Wf. It « 1o1N, tl.T •IIMMN y.vl ro Ma M 1 NOIM.• W+•'1

Figure 7. The nine radiance variables plotted against the total wet biomass for the 18 plotssampled in Octuhgr, 197 1.. (A) red radiance. (B) it radiance, (C) it/reu ratio, (U) square rootof the it/red ratio, (E) it-red radiance difference, (F) it + red radiance sum, (G) vegetationindex, (H) sum/difference, and (1) transformed vegetation index. Refer to Table 3 for ther2 values between the ninr radiance variables and the other five plot variables.

10

h.

.p.i

The October d. ► ta demonstrated conclusiveh t hat the var..., ► S green vegetation measures

do not have a ► )plicability to dormant vegetation (Table 3; Figure 7).

The l ► so of' the square root transformation for the it/red ratio INalepka et al., 1977) and

TVI (Rouse et al., 1973 and 1974) needs to be examined. Rouse et al. (1973 and 1974)

suggest that (tic distribution of the VI is Possion while Nalepka et al, ( 1 9 77) suggest that the

square root of it/red ratio is more linear. The data analysis for th,, June data shows the same

fun gi tional relationship(s) between the total wet bionic s and it/red ratio and square root of

the it/red ratio with the sanie asymptotic nature for both plots, resp:,ctively. The asymptotic

properties of the it/red ratio, square root of the it/red ratio, VI, and TVI are very sir.:ilar as

are the respective degrees of regression significance (Table 3: Figure 4).

PIIF.NOLO(;I('AI. CONSIDFRATIONS

Thi! spectral manifestations of grass canopy phonology can be inferred from the three

sampling periods used 1'or this study. Phenological development resulted in the gradual

accumulation of more standing vegetation in the grass canopy. By September there were

a pproxintately equal amounts of standing live and dead vegetation. The October data was

composed entirely of standing dead vegetation.

Spectral manifesta , ions of grass canopy phonology can be seen by comparing the various

radiance variables for the three sampling periods. The June analysis results were more signifi-

cant in a regression sense, showed the most nonlinearity, and had the highest degree of inter-

correlation between the six canopy variables ("fables 3 and 4). Canopy composition at this

time was — 80% green vegetation and only " 30%: dead vegetation (Table 1).

,I

r-~

9

aV, f1-r

J ^'

7 C

^ 4

O r

NC .rG

^"

fE E

l M C

wN

o bL'v

O•^

O L

L y

O

3 ^ c

L -`

^ — L

f~d ^b

L ^

J

O

^

y G

00r

G 0

00

0*

00000 W

,0

ca o

•: of as o,O

00 ljo

00 r

00 0

F^

00000-

00000-

sL

b8^8

0^

r°00t8n

3--O

--O --

0000-

5

`n

EO

^ 0 v00

O0000 O

N O

TC

O

000-

coo-

C• m

^j E°o

, o,,

o°,

oOO

o 0m c

0GE

o 00 0

Fc 00

N N

_r

y N

0 o

r'g

°c

OJ

b c.s^ v

^ 3U

^3^ ^

33

^U

a `^

«bca

V7p

Q^-

E-l^ ^

^F

-m

HF

-^D

.^U

22

^. c^ C

O C

a.

G

3 ''c

vvOpj^, N

Q Z

A

.^^ O

C m

v

F—'

GI ^

3 N

«. Oo

?

M W

! O

(71, 0C G C:

C O^ O

0 0 -

QQ0c

^.C

Op

L.O

i°

3^

3 c

V^"

OO

a^

23

The September analysis results were less significant in a regression sense than the J umc

ttsults, were linear. and had a lower degree of canopy variable intercorrelation than tlik•

June results( Table 3 and 4). Canopy composition at this time was-52",' green vegetation

and --48 ' dead vegetation (Table I ).

The October analysis results demonstrated t he need for sufficient chlorophyll absorp-

tion, to occur for the it/red ratio and related transformation% to work. By this sampling

time, canopy composition had simplified again and all the standing crop was standingdead

vegetation. Associated with this phenological condition were direct linear relationships be-

tween tx-)tlt the red and it radiances and each of the four canopy variables sampled at this

time. The regression results were not significant, except I -or three radiance variables, and

there was higher degree of Canopy variable intercorrelation IIimi was the ease for the

Setiternber data (Tables ; and 4).

It should be noted that the "chlorophyll" determination for the Octobe ► ,.milllinp

period does not present in vivo chlorophyll a h b. It is thought to represent chlorophyll

decomposition products for this sampling period.

EVALUATION OF DIf FEREMU IK BANDWll) H IS

Another asl• ect of the study was to evaluate the influence of it bandwidth upon ratio

technique applications for estimating the various canopy variables for the .lime data. In

addition to the orilanal it handwidtIt of 0.75 - 0.80 {ant, the balldwidtllS of 0.80 - 0.90

and 0.75 - 0.90 /gy m were evaluated. No differences were found in regression significance

among the three it bandwidths for the .tune data (Table 5),

,4

r,a E

A ^

'= o

A ^0

CJ.O v^

C t`

A 0

C.^x a

.t

N^'

v ^

v ^

C ^

^V

A r` t7

c i

C C

a W

O^ c

i'J

^ 1

n^ ^

r

y ^ v

v^

pp ^ U

C O

'^

^ t ^

v 3 •3

v ^

a a

r

N ^ ^

.+ b aw Q

-C, V

o ^ o0

A V

c

a,

p o

a ^

O

M

'uw y 3

S

r J C^

r 4 M

ID ID

— Q

` f J M V

1 in—

00 — M

10>

O+ o0 ON ON NC

cc

O, 00 O

, C+ in co

O 00 O

+ Ok

00E"

OO

OO

GO

00

60

00

00

00

00

LLQ-• ••• ••• •—

O N

— oc O

00 DrM

^p t! 00 r J v,O

100n\00cl

w00

oc000000

000000

00

00

700

N Q

\ M M

O r

r J C^ M

M 0

0N

CN

M O

N ^c

...>

a 00 0.

a Z 00

s o0 O^ C` v,

0• oo T C; ^O

0c

000000

00o0000

00

00

00

0 0

0

0 0

oo0

^^

_ f^

,1

0 0

0 0

—0

0^. —

—

r

I!

t/10 0 0 0 0 0

0 0 n 0 0 0

0 00 0 0 0

,y-t M

M M

r-- —Q

M M

rr, r- —r 1 r J rr+

O^

^+

Qc%

a^ C

; C\ r

C`

C, O

N C

^ C

, NO

ON

as C

^ 0

:0

0 O

N

0 0 0 0 0 C

0 0 0 0 0 0

O O p

0 0 0

CG

C"m

0ccr,--t C

CA

CrvtC

- O

rQrt

00 f- 00 N r l a

^00 00 u0 00 r

J C^

O(

00 O

P C

, C,^.;

0000 0

0000000

000000

GL17-

00 Cl M

W) V

1rJ O

s c4 M

r- V,

rdC

A M

In inco O` 00 IC OC

011 30 C, Ch ^D 00O` 00 C , ON 1 10 ac

^O

O O

O O

UO

O O

O C

C' ^

0 0

0 0

0 0

Q` r- ON

00 M C`

r- Y1 r` in M 00

oc I c

00 11

0 M 00

rJ 4

.N

r- n

r•- ^ r

r- r r r`

t ^ h

• r-

lO r-

`0 0 0 0 0 0

0 0 0 0 0 00 0 0 0 0 0

G00 O

O N

rJ ^-00 p

O r J hJ

00 O

C N

CLI --

,_,Lys00 00 C

` co M

a,00 0

0 ^:N

00 M

ON0

0 0

0 ON

00 M rJ

C1CO O O O O C

0 0 0 0 0 0

0 0 0 0 0 0

Cy A

'1 ^-

rZ3

y b

y b

b

0

CL

>E

l C«. y

33G

3poI

j ,. C

O,

«. a

3

-x

uv,

,, y

c C

Gv

^^ .. v 3

^^^

cV

im`"

om

`--M

id id_

cd w

M_

b d

A

1

L_

CONCLUSIONS

1. The it/red ratio and related irand red linear combinations were found to be superior

to the green/red ratio and related green and red linear comhuu► tions for monitoring; vegetation.

2. Thw it/red ratio, syuare root (4 the it /red ratio, it-red difference, VI, and TVI are

i sensitive to the amount of photosynthetically active vegetation present in the plant canopy.

All were found to be very similar for estimating the photosynthetically active biomass.

3. The asymptotic properties of the it/red ratio, square root of the it /red ratio, it-red

difference, V1, and TVI were very similar for high green biomass situations. The square root

transformation did not result in a more linear situation.

4. The accumulation of standing dead vegetation in the canopy had a linearizing effect

upon the various green vegetation measures.

5. The regression significance for the different it bandwidths of 0.75 - 0.80, 0.80 -

0.90 and 0.75 - 0.90 pin were evaluated and I'ound to he extremely similar when used with

the red radiance or used in the various Linear combinations.

REFERENCES

Ashley, M.D. and J. Rea. 1975. Seasonal vegetation dil'ferences from FR I S imagery.

PURS 41(6): 713-719.

Blair. B.O. and M.F. Baumgardner. 1977. Detection of the green and brown wave in hard-

wood canopy covers using multidate n ► ultispectral data from Landsat-1. Agron. J.

69:808-811.

. .M

26