E83-1005g . .

in the interest of early and wide dis- semination of Earth Resources S u V A Joint Program for . ,

Program icltormtion and without lkMQ ' 7 1 fn8de warn Agriculture and

Resources Inventory Surveys Throug h

Ed:ly k rn ing and Crop J Aerospace

Cundition Assessment Remote Sen sing

t - . CONPARISONS AMONG 4 NEW SOIL INDEX AND OTHER TWO- AND FOUR- D lr1Eff SIONAL VEGETATION INDICES

4

(E83-IC058) CCHPA61SCNS AFGNG A l iEL S C I L ti63- 1 4 3 7 1

I N C E X A N D C I B E R T I O - A K D P C U O - C I H E I I I C N A L V E G E T A Z I O N I I O I C E S / A q r i c u l t u r a l 6 ~ z f a r c h Service) 22 F HC . 4a i /NF A C 1 CSCL 32C Unclas

G3/43 00058

https://ntrs.nasa.gov/search.jsp?R=19830006296 2018-07-30T12:56:19+00:00Z

Compariscns Among a New Soil Index and Other Two- and Four- Dimensional Vegetation Indices

1

3. Rcecocnt's Catalog .b. ! i r

5. R w r t Date

1. R . ~ a t No.

EW-U2-1)4350, JSC-18562

7. Authw(r1

USDA/XRS Wuslacc, TX 78596

2. Gomnnwnt Accas~on No.

8. Performing Orgmizrticm R w No.

C. L. Wiegand and A. J. Richardson

9. Rrfaming Orgmiation Nww nd Addrm

11. Contract or Gmnt No.

4. Titte and Subtitle

10. W a k Unit No.

16. Abrtm Tne 2-D difference vegetation indes *DVI) and perpendicular vegetation index ( P V I ) , and the 4-D green vegetation index (GVI) are compared in Landsat MSS data from grain sorghum (Sorghum bjcolor, L. Xoench) fields for the years 1973 to 1977. PVI and DVI were more closely related to LA1,than was GVI. X new 2-D soil line index (SLI), the vector distance from the soil line origin to the point of intersection of PVT with the soil line, is defined and compared with the 4-D soil brightness index, SBI. SLI (based on MSS5 and YSS7) and SL16 * based on ZISS5 and NSS 6) were smaller in magnitude than SBI but contained similar information about the soil background. These findings indicate chat vegetation and soil indices c-lculated from the single visible and reflective infrared band sensor systems, sucl a the AVHRR of the TIROS-N polar orbiting series of satellites, will be meaningful for synoptic monitoring of renewable vegetation resources.

I

12. Soocacring m y Nmw md Addrcrr USDA/SRS Johnson Space CenterISK Houston, TX 77058

13. T y p of Report and Period k e d

14. Sponoaiq Codc

I 6

20. ~ccur~tv ~ a a i f . (of this pyrl 21. No. of Papn 22. Rice'

Unclass. Unclass . 2 1

15. SuoprwmrV kuta

17. K y Words (Suggutod by 9uthar(s)) 18. Oistribut~on Statement

COMPARiSONS AMONG A NEU SOIL INDEX AND OTHER TWO- AND FOUR-DIMENSIONAL VEGETATION INDICES*

C. L. Wiegand and A. J. Richardson U.S. Department o f Agr icu l ture

Ueslaco, TX 78596

BIOGRAPHICAL SKEXH

Craig Wiegand i s a s o i l s c i e n t i s t i n the Remote Sensing Research Unit, Agr icu l tura l Research Service, USDA, Weslaco, TX. He has 17 years experience i n the agr i - cu l t u ra l appl icat ions o f aer ia l photography, op t ica l mechanical s c m ~ e r , and s a t e l l i t e (LANDSAT, EREP, ant H M ! data. Jerry Richardson i s a phys ic is t w i th the same organization. His research experience includes crop and land use d iscr iminat ion and pat tern recognit ion, spectral vegetation and s o i l indices, atmospheric trans- f e r model 1 i ng, and scene component analysis uz i ng ground, aer ia l , and space observations.

ABSTRACT

The 2-D d i f ference vegetation index (DVI) and perpendicu- l a r vegetation index (PVI), and the 4-D green vegetation index (GVI ) are compared i n LANDSAT MSS data from gra in sorghum (So hum b i co lo r (L.) Moench) f i e l d s f o r the years 1973 *77. t o PVI and D V I were i n r e c lose ly re- l a ted t o LA1 than was G V I . A new 2-D s o i l l i n e index (SLI), the vector distance from the s o i l l i n e o r i g i n t o the point o f in tersect ion o f P V I w i th the s o i l l ine , i s defined and compared w i th the 4-0 s o i l brightness index, SBI. SLI (based or. MSS5 and MSS?) and SL16 (based on MSS5 and MSS6) were smal l e r i n magnitude than SBI but contained s im i l a r information about the s o i l background. These f i n d i nqs indicate tha t vegetation and soi 1 indices calculated from the s ingle v i s i b l e and r e f l e c t i v e i n f r a - red band sensor systems, such as t i le AVHRR o f the TIROS-N po lar o rb i t i ng series o f sate1 li tes, w i l l be meaningful f o r synoptic ran i t o r i ng o f renewable vegetation resources.

*Conzribution from the Remote Sensing Research Unit, Agricul t u ra l Research Service, U. S. Departmer~t of Agri- cul ture, Wes:aco, TX. gesearch supported i n pa r t by National Aeronautics and Space Administration Contracts, S-70251-AG and S-53876-AG.

INTRODUCTION

Vegetation icdices (Rouse e t a1 . , 19?4; Uiegand e t a1 . , 1974; Kauth and Thomas, 1976; Richardson and Wiegand, 1977; Tucker, 1979; Ashburn, 1979; and Jackson e t al., 1980) are used f o r monitoring vegetation development, crop condi t ion and stress, forage production, and gra in y i e l d s (Deering e t al., 1975; Richardson and Uiegand, 1977; Thompson and Wehmanen, 1979; Ashburn, 1979; Badhwar and Henderson, 1981; and Lautenschlager and Perry, 1981 ) . The indices are rat ios, differences, sums/di f f e r - ences, and l i n e a r combinations o f v i s i b l e (0.4 t o 0.7 urn) and r e f l e c t i v e in f ra red (0.75 t o 1.35 um) bands tha t reduce the information about the green vegetation cover and the s o i l background respectively, t o a s ingle numerical index.

Although the indices can be calculated from radiance, ref lectance factor, o r d i g i t a l count as observed from any instrument, the most widely avai lab le and extensive data have come f ran the LANDSAT ser ies o f ear th observation sate1 li tes through the impetus o f the Large Area Crop Inventory Experiment (LACIE) (The LACIE Symposium, 1979). The LANDSAT MSS has two v i s i b l e (0.5 t o 0.6 urn snd 0.6 t o 0.7 urn) and two r e f l e c t i v e i n f ra red bands (0.7 t o 0.8 urn and 0.8 t o 1.1 um) t h a t are referred t o as bands MSS4, MSS5, MSS6, and MSS7, respectively. As the three-year LACIE ended, uncertainty about cont inuat ion o f the MSS and LANDSAT series, i t s rather infrequent coverage, and the desire t o continue operational uses o f such data f o r fore ign crDp production forecast ing (MacDonald and Hal 1 , 198C), prompted consideration o f a1 te rna t ive sate1 1 i t e data sources. One such source i s bands 1 (0.58 t o 0.68 urn) and 2 (0.725 t o 1.10 um) o f the Advanced Very High Resolution Radi m e t e r (AVHRR) aboard the polar-orbi t i n g TIROS-N series o f National Oceanographic and Atmospheric Administration (NOAA) sate l 1 i tes. This ser ies o f fe rs 5-day repeat nadi r coverace and cont inui ty through 1985 (Schwalb, 1978; Gray, McCrary and Armstrong, 1981 ). A pre l iminary t e s t (Gray and McCrary , 1981 ) o f band 2 m i nus band 1 o f the AYHRR versus ( 2 MSS7-MSS5) o f LANDSAT f o r camnon ground s i t e s and overpass dates ;ielderl a l i n e a r c o r r l a t i o n coef f ic ient , r=0.86.

Indices tha t u t i l i z e a1 1 four LANDSAT MSS bands, hence are four-dimensional ( 4 4 ) i n terns o f HSS data space are the " b r i yhtness" o r soi 1 brightness index ( 3 1 ) , "green- ness" o r green vegetation index (GVI), "yellow s t u f f " (YEL) and "non-such' o f Kauth and colleagues (Kauth and Thomas, 1976; Kauth, e t al., 1979). The two-band o r two

dimensional (2-D) ind ices t h a t are a1 so referenced t o a s o i l l i n e include the per,endicular vegetat ion index (PVI) and d i f ference vegetat ion index (DVI ) (Richardson and Wiegand, 1977). Should AVHRR data become widely used, then a x i m a t i c a l l y on ly 2-P ind ices can be ca lcu la ted from i t s s ing le v i s i b l e and s ing le r e f l e c t i v e i n f r a r e d band. The purposes o f t h i s paper are t o (a) der ive a 2-D s o i l index, temed the s o i l l i n e index (SLI) t h a t i s analogous t o the 4-0 SBI, and (b) compare the s i m i l a r i t i e s and d i f fe rences among the named 2-D and 4-D vegetat ion and s o i l background ind ices f o r a common data set.

EQUATIONS AND DEFINITIONS

Kauth and Thomas (1976) def ined the plane of s o i l o r s o i l br ightness index (SBI) i n LANDSAT-1 data transformed by ax i s ro ta t ion . The SBI equation fo r LANDSAT-2 equivalent* d i g i t a l count (DC) data adjusted t o a constant so ia r zen i th angle (SZA) o f 39" (Kauth, e t al., 1979) i s

SBI = .332(MSS4) + .603(MSS5) + .676(MSS6) + .263(MSS7). (1

I n a two-dimensional approach (Figure 1) t h a t uses the red v i s i b l e band (MSS5) and e i t h e r MSS6 o r MSS7, Richardson and Wiegand (1977), the vegetat ion ind ices are referenced t o the l i n e o f so i l . For s a t e l l i t e c a l i b r a t i o n and SZA 39" DC data the s o i l l i n e i s def ined by

and MSS5 = -6.09 + 1.12(MSS6) (2%

The i r d i f fe rence vegetat ion index (DVI) i s

D V I = 0.26 + 2.73(MSS7) - MSS5 (3

which by comparison w i t h (2a) i s the estimated value o f MSS5 on the s o i l l i n e , MS$, minus MSS5 o r

DVI = MSS? - MSSS. (3a)

*Coef f ic ients are ava i lab le f o r convert ing LANDSAT-1 and LANDSAT-3 d i g i t a l counts t o equivalent LANDSAT-2 d i g i t a l counts dur ing the per iod 22 Jan 1975 through 15 Ju l 1975. See the LANDSAT Data Users' Manual, p. AE16 f o r the basic data, o r Richardson (1982).

By reference t o F igure 1, DVI i s the v e r t i c a l d is tance from the s o i l l i n e t o the vegetat ion po in t , Rp 5, and the l eng th o f the hypotenuse, labe led c, o f the r i g h t t r i a n g l e formed by the s o i l l i n e , the perpendicular vegetat ion index (PVI), and DVI.

Kauth e t a l . (1979) def ined the ngreen s t u f f " o r green vegetat ion index ( G V I ! component f o r SZA-corrected equiva- 1 e n t LANDSAT-2 aa t a as

G V I = -0.283(MSS4) - 0.660(MSS5) + 0.577(MSS6j + 0.388(MSS7) ( 4 )

This component i s orthogonal t o the SBI component and i s dominated by l i v e , green vegetation. I n a 2-0 approach, Richardson and Wiegand (1977) defined the perpendicular d is tance from the s o i l l i n e t o the vegetat ion p o i n t Rp 7, Rp 5 by

P V I = [ ( ~ ~ g 5 - ~ p 5 ) ~ + i ~ g g 7 - ~ p 7 ) ~ 1 ' ( 5 )

wherein Rgg7 and Rgg5 a re the coordinates o f the in te rsec- t i o n o f the P V I l i n e w i t h the s o i l l i n e (F igure 1). For the s o i l l i n e o f (2a) Rgg7 and Rgg5 are def ined by

and f o r the s o i l l i n e o f equation (2b)

PVI can a1 so be der ived from the formula f o r the d is tance from a p o i n t t o a l i n e (Jackson et . a1 . , 1980; Lauten- schlager and Perry, 1981 ). For Rp7 on the abscissa and RP5 on the ord inate as i n Figure 1

Pv17 , 5 = (a, Rp7-RpS+ao)/[(l 12+(a1 )'I' (5e)*

*Yhen Rp5 i s p l o t t e d on the abscissa and Rp7 on the ord in- ate, the equation equiva lent t o (5e) i s

For t h i s equation a. = -.095 and a, = .366, so t h a t the equation reduces t o ( 5 f ) .

Since a. = t he s o i l l i n e i n t e r c e ~ t = 0.26 and a1 = the s lope o f the s o i l l i n e = 2.73, the equat ion reduces t o

PVI = .939Rp7 - .344Rp5 + 0.09 = -939MSS7 - .344MSS5 + 0.09 ( s f

S im i la r l y , PVI6 = .746Rp6 - 0.666Rp5 - 4.06. Both PVI and PVI6 equal 0 f o r bare s o i l and increase as green vegetat ion Censi ty increases.

I n F igure 1 we have now def ined sides a and c o f a r i g h t t r i ang le , so t h a t the leng th o f s ide b, the s o i l l i n e l e g (SLL) i s

SLL = cos 20.33"(DVI) = 0.938(DVI). (6a

The length o f the SLL f o r a given data se t depends on the range Sn s o i l color, texture, and moisture content encoun- te red i n t he observations; on the amount o f shadow present; and, on s o i l t i l l a g e , cloddiness, and bleaching by the stin. Data f o r s o i l contaminated by c loud shadows o r c loud tops w i l l extend the lower and dpper ends o f the SLL, respec- t i v e l y , s ince both these features f a l l on the s o i l l i n e . When such data a re included, the l i n e i s p roper l y c a l l e d the shadow-soi 1 -c loud 1 i ne.

The leng th o f the vectar i n i t i a t i n g from the o r i g i n o f the s o i l l i n e and terminat ing a t the po in t (Rgg7, Rgg5) f o r each observat ion p a i r (Rp7, Rp5) i s o f g rea te r i n t e r e s t than SLL. Using s o i l l i n e d e f i n i t i o n s (2a) and (Zb), t h i s distance--termed the s o i l 1 i n e index (SLI) f o r MSS bands 5 and 7, a n J SLI6 f o r bands 5 and 6-- is ca lcu la ted by the respect ive equations

SLI = [(Rgg5-.2612 + (Rgg7)'14.

and

SLI and SLI6 a re the two dimensional analogs o f SBI def inea i n eq. (1).

SLI and SLI6 reduce the in fo rmat ion about the s o i l back- ground t o a s i ng l e index f o r the respect ive bands used. Factors t h a t can af fec t the value of SLI on a given obser- va t i on date o r f o r a p a r t i c u l a r s i t e inc lude s o i l type, so l a r zen i th angle, water content o f the surface s o i l ,

s o i l crust ing, p ropor t ion o f shadows, bleaching o f the s o i l surface, amount o f standing, d r y o r l i t t e r e d p l a n t residue, and c u l t u r a l operations such as c u l t i v a t i o n . Wet s o i l condi t ions, compared w i t h d r y so i 1 background, and shadows lower SLI whereas s o i l b leaching and a more nad i r s o l a r zen i t h angle increase SLI. The s o i l background a lso in f luences p l a n t canopy re f lec tance over most o f the crop season. Reflectance modeling s tud ies (Col we1 1, 1974; Chance and LeMaster, 1977; Kauth, e t a l . , 1979) show t h a t t he s o i l background a f f e c t s v i s i b l e band responses up t o an LA1 o f about 2 i f leaves are un i fo rm ly d i s t r i b u t e d over t he ground bu t up t o an LA1 o f a t l e a s t 6 i n the r e f l e c t i v e i n f r a r e d bands. Consequently ca re fu l study o f the seasonal t r a j e c t o r i e s o f boch the s o i l background index SLI and the vegetat ion ind ices i n response t o agronomic and na tu ra l va r iab les should enhance scene understandi ng.

MATERIALS AND METHODS

We assembled a LANDSAT KSS data se t f o r commercial g r a i n sorghum (Sorghum b i co l o r , (L. ) Moench) f i e l d s t h a t included 9 scenes spanning the years 1973 through 1977 (Table 1). The computer compatible tapes (CCT) were obtained e i t h e r from the Data Handling F a c i l i t y a t Goddard Space F l i g h t Center, Greenbelt, MD o r from the US01 EROS Data Center a t Sioux Fa1 l s , SD. L ine p r i n t e r gray maps o f areas i n t he v i c i n i t y o f study f i e l d s were prepared, the study f i e l d s were i d e n t i f i e d and out l ined, and the d i g i t a l counts (DC) f o r each o f the f ou r MSS bands were ext racted and averaged f o r the i n t e r i o r f i e l d p i x e l s near the ground t r u t h s i tes . The raw data were adjusted t o a so l a r zen i th angle o f 3g0, and the c o e f f i c i e n t s f o r ad jus t ing the d i g i - t a l counts f o r o ther s a t e l l i t e s and per iods t o those f o r LANDSAT-2 f o r the reference per iod January 22, 1975 t o Ju l y 15, 1975, were applied.*

The sample f i e l d s used were selected from two areas i n Hidalso County, TX (26.5 N Lat., 9 8 O W. Long.). One was a dry farmigig (non i r r iga ted) area near McCook i n the northwest p a r t o f the county. The predominant s o i l s are McAl l e n ( A r i d i c Ustochrepts) and Brennan ( A r i d Haplustal f s )

*Coe f f i c ien ts are avai lab1 e f o r conver t ing LANDSAT-1 and LANDSAT-3 d i g i t a l counts t o equivalent LANDSAT-2 d i g i t a l counts dur ing the per iod 22 Jan 1975 through 15 Ju l 1975. See the LANDSAT Data Users' Kanual, p. AE16 f o r the basic data, o r Richardson (1982).

fine sandy loams. Their Munsell colors are lOYR 612 and lOYR 413, respectively, in the dry state and they have a relatively high reflectance. The other group of fields came from the irrigated, eastern part of the county near Edcouch. These soils are finer textured and darker colored than soils a t the nonirrigated site. They are typified. by the Hidalgo (Typic Calciustoll s ) sandy clay loam and Raymondvil l e (Vertic Calciustol 1 s ) clay loam series (Jacobs e t a1 . , 1981 ) . Their Munsell colors in the dry state are lOYR 412 and lOYR 511, and they are less reflective than the dryland soils.

LANDSAT data were available for 108 fields, b u t ground observations of percent veget3tive cover and leaf area index (LAI) were available for only 32 fields within a week of the sate1 l i t e overpass. A t our latitude, grain sorghum typical ly reaches the anthesi s dele! opmental stage by May 10. Grain f i l l ing follows for about 30 days. Near the end of grain f i l l ing leaves and leaf sheaths have chlorophyll degradation because of stresses such as fol ia r di seases and senescence. Consequently, data in Table 1 fcr July 10, 1975, represent standing stubble or t i l led fields and that for October 17 were mainly of value to define the soil line.

The vegetation indices DVI, PVI, and GVI were calculated using equations (3) , ( 5 t o 5d), and ( 4 ) , respectively. The soil indices SBI, SLI, and SLI6 were calculated using equations (1 ) , (7) , and (7a), respectively. Standard s tat is t ical procedures were used fcr 1 inear regression and correlation analyses.

RESULTS AND DISCUSSION

The relationsh;.~ between the variable pairs MSS7 and MSSS, PVI and DVI, PVi and SLI, and GVI and SBI are displayed for the 108 fields in parts A through D of Figure 2. Although we used multidate and multisatellite data sources, the MSS7 vs MSSS scatter diagram has the triangular shape usually observed in single date data (Kauth and Thomas, 1976; Kauth e t a1 . , 1979). The triangle i s bounded on the l e f t side, as in Figure 1 , by the soil line leg (SLL) and across the bottom by the limiting value of PVI. The data points of the third side are Lest f i t by a curved line that drops ini t ia l ly nearly vertically from the soil line, then curves outward as PVI increases. This suggests t h a t the vertical displacement of a vegetation point from the soil line given by DVI (Figure 1) should describe vegeta- tion conditions we1 l a t low vegetative cover. The con- stant relation between PVI and DVI shown in part B holds

f o r p a r t i c u l a r data pa i r s as we1 1 as f o r a whole data s e t so t h a t DVI i s no t more sens i t i ve t o low vegetat ive cover than i s PVI. A st rength o f PVI i s t h a t i t i s i n d i f f e r e n t t o p o s i t i o n o f Rg7, Rg5 along the s o i l l i n e , and depends on ly on the perpendicular distance t o the SLL wherever Rg7, Rg5 fa1 1 i n the data t r i ang le . On the o ther hand, because o f the c u r v i l i n e a r nature o f the s ide o f the data t r i a n g l e opposite the SLL, the i n t e r s e c t i o n o f the ver- t i c a l o r D V I l i n e on the s o i l l i n e can occcr outside the range f o r s o i l - - f o r example, i n the range f o r c loud tops. Consequently, the DVI corresponding t o l a rge PVI may no t be physical l y meaningful.

The r e l a t i o n between G V I and SBI i s s i m i l a r t o t h a t between PVI and SLI. However, G V I and SBI both appear t o be domi- nated by the MSS6 response, because the DC and coefff c i e n t f o r the MSS6 band are l a rge i n both equatiovs ( 1 ) and ( 4 ) . This could happen i n a procedure t h a t op-imizes p l an t and s o i l in format ion because MSS6 i s l e a s t a f fec ted by the atmosphere.

I n F igure 2 the data displayed c l u s t e r by s o i l type. Th is i s expected f o r the s o i l i nd ices since the s o i l o f the dry land and i r r i g a t e d s i t e s a re known t o d i f f e r i n r e f l e c - tance o r brightness. The vegetat ion ind ices c l u s t e r by s o i l type p a r t l y because dense canopies were achieved on ly under i r r i g a t i o n . A1 1 ind ices are sub ject t o scene-to- scene v a r i a b i l i t y i n the mu1 t i temporal and mu1 t i sate1 li t e data set.

The r e l a t i o n between SBI and SLI6 (F igure 3) i s much c l ose r ( r 2 = .975) than between SBI and ! I ( r 2 = .719). This i s explained by the r e l a t i v e dominance o f MSS6 i n eq. (1) due t o i t s l a rge c o e f f i c i e n t and the usual l a rge DC f o r MSS6 r e l a t i v e t o the o ther bands. SL16 i s about 0.9 the magni- tude o f S B I and SLI i s about 0.7 the magnitude o f SBI.

The r a t i o PVIISLI and PVI61SLI6 versus GVIISBI (F igure 4 ) normalize the vegetat ion ind ices t o the corresponding s o i l 'ridex. The sca t te r i s l ess than i n Figure 3 because, i n :his case, ind ices from the same bands are ra t ioed. The PVIISLI versus GVIISBI s c a t t e r diagram i s s l i g h t l y cb r v i - l i n e a r over i t s whole range causing i t t o have a poorer l i n e a r c o r r e l a t i o n ( r 2 = .942) than PVI61SLI6 ( r 2 = .976) versus GVIISBI. Jackson e t a1 , (1980) have noted a s l i g h t c u r v i l i n e a r i t y o f the s o i l l i n e from hand-held radiometer data, and the authors ha342 observed t h a t the b i d i r e c t i o n a l re f lec tance f a c t o r increases as reference panel r e f l e c - tance increases (unpublished data). Consequently, i t may express an a r t i f a c t of MSS sensor c a l i b r a t i o n against a h i gh l y r e f l e c t i v e barium s u l f a t e - c o a t ~ d i n t e g r a t i n g sphere.

I n any event, the reason PVI and G V I become more negat ive i n Figure 2 the greater SLI and SBI o f th2 bare s o i l needs t o be explained.

The p l o t s o f SLI vs PVI, PVI6, and G V I i n F igure 5 d i sp l ay an inverse r e l a t i o n between the vegetat ion ind ices and SLI. We expect t h i s t o be the normal behavior because the amount o f s o i l i n the sensor IFOV w i l l decrease as vegetat ive cover increases. However, Heilman e t a l . (198 ) present data showing t h a t SLI increased as s o i l cover 5 y a l f a l f a increased. This could happen i f the s o i l background bleached w i t h time, o r surface s o i l cons t i tuen ts separated dur ing the s p r i n k l e r i r r i g a t i o n s and a l i gh t - co l o red f rac - t i 3 n was l e f t a t the s o i l surface.

Another way o f comparing the vegetat ion ind ices i s through a d i r e c t measure o f vegetat ion dens i ty such as l e a f area index (LAI), which was ava i l ab l e f o r 14 i r r i g a t e d and 18 non i r r i ga ted f i e l d s dur ing the g ra i n f i l l i n g stage o f matur i ty . The r e l a ion between LA1 and the vegetat ion and s o i l i nd ices P V I , D V I , G V I and SLI (F igure 6) shows t h a t f o r PVI, LA1 was expressed by

LA1 = 0.301 +'0.263 (PVI)

f o r which the l i n e a r c o r r e l a t i o n c o e f f i c i e n t ( r ) was 0.782**. The s t a t i s t i c a l f i t f o r D V I was the same as f o r PVI. For G V I t he corresponding equat ion was

LA1 = 1.43 + 0.160 (GVI)

f o r which r = 0.729**. Thus the l e a s t squares f i t i s s l i g h t l y b e t t e r f o r PVI than for G V I .

The d i s t r i b u t i o n o f data po in ts i n A, B, and C cf Figure 5 i s such t h a t the na tu ra l l ogar i thm of LA1 would express the r e l a t i o n about as we l l as the l i n e a r one presented. We be1 ieve, i n agreement w i t h Brakke and Barker (198-; t h a t the spect ra l data i n Figure 6 i s super ior t o the LA1 data; over 40 spec t ra l samples per band ( i n t e r i o r f i e l d p i x e l s ) were used f o r some f i e l d s and no fewer than 8 whereas LA1 i s based on as few as 8 p l an t s per f i e l d .

The sca t t e r f o r the LA1 vs SLI p l o t (F igure 6D) shows a c l e a r segaration between s o i l types on which the i r r i g a t e d and dryland sorghum was grown. Again, the response o f the s o i l i nd ices t o s o i l type o r c o l o r i s i l l u s t r a t e d ,

I n F igure 6 t h w e a re few observations f o r LA1 < 1.0 because these were ful ly-developed p lan ts i n commercial f i e l d s . The LA1 i n t e r cep t , when PVI and D V I was zero, was estimated t o be 0.3. G V I values are negat ive a t low LA1 values; t he p red ic ted G V I a t LA1 = 0 i s -9.0. The LAi o f 8.1 p red ic ted by the SLI equat ion (F igure 5D) was reasonable f o r the number o f l e a f l ayers t h a t ;an a f f e c t near- i n f ra red r e f 1 ectance o f crop canopies, bu t an SLI o f 0 i s u n r e a l i s t i c , except poss ib ly f o r a r i c e crop canopy standing i n water.

In summary, we have i n t e r r e l a t e d a number o f s o i l and vegetat ion ind ices f o r a common data set, high1 i ghted t h e i r s i m i l a r i t i e s and d i f ferences, i l l u s t r a t e d t h e i r r e l a t i v e magnitudes, and demonstrated t h e i r response t o vegetat ive cover and t o the s o i l background. Note- worthy are: ( a ) the e x p l i c i t d e f i n i t i o n o f the 2-D s o i l l i n e ind ices SLI and SLI6, (b ) the s i m i l a r i t y i n i n f o r - mation content o f 2-D and 4-0 s o i l and vegetat ion inaices, and ( c ) the dominance o f MSS6 on both SBI and G V I . The s i m i l a r i t y i n 2-D and 4-D s o i l and vegetat ion ind ices f o l l ows from the long-recogni zed high c o r r e l a t i o n between t he two v i s i b l e and between the two near- in f rared bands o f the LANDSAT MSS, bu t me r i t s r e i t e r a t i o n w i t h the i n t e r e s t developing i n data from the AVHRR o f the TIROS-N ser ies o f sate1 1 i tes. The AVHRR has or, v i s i b l e and one r e f l e c t i v e in f ra red band so t h a t on ly 2- 1 i nd ices can be ca lcu la ted t h a t a re analogous t o LANDSAT MSS experience.

ACKNOWLEDGEMENTS

We thank Joe Cue l la r and A l v i n Gerbermann f o r f i e l d meas- urements and summarization, and Kathy Wolf and Wayne Swanson f o r s t a t i s t i c a l ana lys is and f i g u r e preparat ion.

REFERENCES

Ashburn, P. 1979, The vegeta t ion index number and crop i d e n t i f i c a t i o n : Proc. LACIE Sympos., I I :843-855, JSC- 16015, NASA, Johnson Space Center, Houston, TX

Badhwar, G. D., and K. E. Henderson, 1981, Est imat ing development stages o f corn from spec t ra l data: Agron. J., 73: 748-755

Brakke, T. W . , and J. L. B a r ~ e r , 1982, LANDSAT MSS spec t r i i l t ransformat ions: A model f o r determining wheat LAI: NASA Goddard TM-Goddard Space F1,lght Center, MD, ( I n press)

Chance, J. E., and E. W. LeMaster, 1977, Theoret ica l and experimental t e s t s o f the S u i t s r e f l e c t a n c e models f o r wheat and cot ton: Appl ied Opt ics, 16:407-412

Colwel l , J. E. , 1974, Vegetat ion canopy re f l ec tance : Remote Sens. o f Envi ron., 3: 175-183

Deering, D. W., J. W. Rouse, R. H. Haas, and J. A. Schel l , 1975, Measuring " forage product ion" o f graz ing u n i t s from LANDSAT MSS data: Proc. 1 0 t h I n t ' i Sympos. Remote Sens. o f Envi ron., II:1169-' 178, Univ. Mich., Ann Arbor

Gray, T. I., Jr., and D. G . McCrary, 1981, Meteoro log ica l sate1 1 i t e data--a t o o l t:, descr ibe the heal t h o f the wor ld ' s a g r i c u l t t i , e: AgRISTARS Rpt. RJ-N1-04042, JSC- 17112, NASA, L. B. Johnson Space Center, Houston, TX

, and T. A. Armstrong, 1981 , c h a r a c t e r i s t i c s o f TlROS, GOES. UMSP dnd LANDSAT systems: AgRISTARS Rpt. EM-NI-04075, JSC-1 71 31, L. B. ~ o h n s o n Space Center, Houston, TX

Heilman, J. L., W. E. Heilman, and D. G. Moore, 138 , Evaluat ing the crop c o e f f i c i e n t us ing spec t ra l measure- ments: Agron. J., (Submitted)

Jackson, R. D., P. J. P in te r , Jr., R. J. Reginato, and S. 8. Idso, 1980, Hand-he1 d radiometry: USDA-SEA, Agr-ic. Rev. and Manuals, ARM-W-i9, 66 p.

Jacobs, J. L., 1981, S o i l survey o f Hidalgo County, Texas: S o i l Conser. Ser., USDA, U.S. Govt. P r in t i : i g O f f i ce , Washington D.C., Pub l ic . No. 1981-328.,.'q7/103

Kauth, R. J., and G. S. Thomas, 1976, The tasse l led cap-- A graphic desc r i p t i on o f the spectral-temporal development o f a g r i c u l t u r a l crops as s2en by LANDSAT: Proc. Sympos. Machine Processing o f Remotely Sensed Data, pp. 41-49, IEEE Cat. 76, CH1103-1-MPRSD, I n s t i . E l e c t r i c a l and E lec t ron ic Engin., New York, NY

Kauth, R. Jay P. F. Lambeck, W. Richardson, G. S. Thomas, and A. P. Pentland, 1979, Feature ex t r ac t i on appl ied t o a g r i c u l t u r a l crops as seen by LANDSAT: Proc. 'LACIE Sympos., I I:705-721, JSC-16015, NASA, Johnson Space Center, Houston, TX

Tne LACIE Symposium, 1979, Proceedings of the technica l sessions: Vols. I and 11. Publ ic. JSC-16015, NASA, L. B. Johnson Space Center, Houston, TX

Lautenschlager, L. F., and C. R. Perry, 1981, Comparison o f vegetat ion ind ices based on sate1 1 i t e acquired spect ra l data: Proc. Ann. Yeeting ( D e t r o i t ) , Survey Res. Methods Section, h e r . S t a t i s t i c a l Assoc., Washington, D.C. ( I n press)

MacDonald, R. B., and F. G. Hal l , 1980, Global crop fore- cast ing: Science, 208:670-679

Richardson, A. J., 1982, Rela t ing LANDSAT d ig f t a l count values t o ground re f lec tance f o r o p t i c a l l y t h i n atrriaspheric condi t ions: App1 i e d Optics, --- : - (Accepted)

Richardson, A. J., and C. L. Wiegand, 1977, D is t ingu ish ing vegetat ion from so i 1 background informat ion: Photogramm. Engin. and Remote Sensing, 43(12) :1541-1552

Rouse, J. W., Jr., R. H. Haas, J. A. Schell , and D. W. Deeri ng, 1974, Moni tor ing vegetat ion systems i n the Great P la ins w i t h ERTS: Proc. 3rd ERTS-1 Sympos., Vol. I, Sect. A, pp. 309-318, NASA-SP 351, S c i e n t i f i c and Tech. Znfor. Of f ice, NASA, Washington D.C.

Schwa1 b, A., 1978, T k TIROS-N/NOAA A-G sate1 li t e ser ies: Na t ' l . Ocean. and Atmos. Admin., N a t ' l Environ. Sat. Ser. Tech. Memo NESS 95, U.S. Dept. Commerce, Washington D.C.

Thompson, D. R., and 0. A. Wehmanen, 1979, Using LANDSAT data t o detect moisture stress: Photagramm. Engin. and Remote Sensing, 45: 201 -201

Tucker, C. J., 1979, Red and photographic in f ra red l i nea r combinatinns f o r m n i to r ing v2getation: Remote Sens. o f Environ., 8:127-150

Uiegand, C. L., H, W. Gausman, J. A. Cuellar, A. H, Gerbermann, and A. J. Richardson, 1974, Vegetation density as deduced from ERTS-1 MSS response: Proc. 3rd ERTS-1 Sympos., Vol. I, Sect. A, pp. 93-116, NASA SP-351, Scien- t i f i c and Tech. Infor . Office, NASA, Washington D.C.

Table 1 . LANOSAT scenes and grain sorghum fields used in this study.

Solar tandsat Zenith No. LA1

Date JD Scene ID No. Angle Cu1 ture Fie1 ds Available

(Degrees)

28

39

33

U

Irrig.

Irrig . Irrig . Dry.

Irrig.

Irrig.

Dry.

Irrig.

Dry.

Irrig.

Dry.

Dry.

Mi xed

Figure 1 Graphical d isp lay o f vegetation indices and de f i n ing ver t i ces and sides c f the t r i a n g l e bounding vegetation i n the v i s i b l e red (MSSS) and r e f 1 zc t i ve i n f r a r e d (MSS7) wave1 ength i n t e r v a l s

ORIGINAL PAGE IS OF POOR QUALm

za 1 -la. c. a ze. na.

Pv;

Figbre 2. Relat ion between ( A ) MSS7 and MSSS, ( B ) DVI and PVI, (C) SLI and P V I , and (0) SBI and GVI

ORIGINAL PAGE lb OF POOR QUALrIY

Figure 3. The soil brightness index (SBI) versus SLI and SL16

OfrlG'MAL PA(:: €3 OF POOR QUALITY

Figure 4. GVIISBI versus PVI/SLI and PVI6/SLI6

ORIGINAL PAY:": 13 OF POOR QUALITY

Figure 5. SLI versus P V I , PVI6, and GVI

,, a s. i m . 1s. am. 2s. 4 P V T JIL

<C> / I I 1 LAX-a. 31S-&OO4:DVt> 0. -a- . S O 0

I

L ~ m . am. am. 40. am. e a 7a

Figure 6. Leaf area index (LAI) versus ( A ) PVI, (B) GVI, (C) DVI, and (0) SLI

C ,

0-

9 : I

1 y 0 I a. . n !

a sm. 4.. sm. em. -e. m a

S L I

LAY-a la-& ~ I ~ < S L : > -2.. I I I

TT I

I

P