Professional Documents
Culture Documents
35:161-173 (1991)
M o s t vegetation indices (VI) combine informa- tions has been paramount in many agriculture-
tion contained in two spectral bands': red and applied studies in recent years. To minimize the
near-infrared. These indices are established in or- variability due to external t~actors, multispectral
der to minimize the effect of external factors on reflectance data have been transformed and com-
spectral data and to derive canopy characteristics bined into various vegetation indices. The most
such as leaf area index (LAI) and fraction of ab- commonly used vegetation indices utilize the in-
sorbed photosynthetic active radiation (P). The po- formation contained in red and near-infrared
tentials' and limits of different vegetation indices canopy reflectances or radiances. They are com-
are discussed in this paper using the normalized bined in the form of ratios: ratio vegetation index
difference (NDVI), perpendicular vegetation index (RVI) (Pearson and Miller, 1972) or normalized
(PVI), soil adjusted vegetation index (SAW), and difference (NDVI) (Rouse et al., 1974), or in linear
transformed soil adjusted vegetation index (TSA VI). combinations as the perpendicular vegetation in-
The discussion is based on a sensitivity analysis in dex (PVI) (Richardson and Wiegand, 1977). These
which the effect of canopy geometry (LAI and leaf indices have been found to be well correlated with
inclination) and soil background are analyzed. The various vegetation variables including green
calculation is performed on data derived from the leaf area (Wiegand et al., 1974; Holben et al.,
SAIL reflectance model. General semiempirical 1980; Asrar et al., 1984, 1985b; Hatfield et al.,
models, describing the relations between VI and 1985; Clevers, 1989), standing biomass (Tucker,
LAI or P, are elaborated and used to derive the 1979; Elvidge and Lyon, 1985), percent ground
relative equivalent noise (REN) for the determina- cover, amount of photosynthetically active tissue
tion of LAI and P. The performances of VIs are (Wiegand et al.), photosynthetic activity (Baret
discussed on the basis of the REN concept. and Olioso, 1989; Choudhury, 1987; Hatfield et al.,
1984; Sellers, 1985; 1987), and productivity (Asrar
INTRODUCTION et al., 1985a).
The green leaf area index (LAI) is a key
The development of fimctional relations between variable which is functionally linked to spectral
crop characteristics and remote spectral observa- reflectance. Leaf area index is also a variable which
is frequently used by agronomists, crop physiolo-
gists, and crop modelers. A large number of rela-
Address correspondence to F. Baret, INRA Bioclimatologie, BP
tionships have been established between vegeta-
91, 84143 Montfavet Cedex, France.
Received June 1990; revised 12 November 1990. tion indices and LAI. Generally the vegetation
0034-4257/91 / $3.50
©Elsevier Science Publishing Co. Inc., 1991
655 Aw~nue q[ the Americas, New York, NY 10010 161
162 Baret and Guyot
indices approach a saturation level asymptotically Table i. Parameters Used for the Calculation of the Differ-
ent Vegetation Indices. a
for LAI ranging from 2 to 6, depending on the
type of vegetation index used, the crop studied, Sun zenith angle: 45 °
View angle: nadir viewing
and experimental conditions (Wanjura and Canopy geometry:
Hatfield, 1987; Daughtry et al., 1980; Chance, - - l e a f angle distribution function: ellipsoidal
1981; Ahlrichs and Bauer, 1983; Best and Harlan, - - a v e r a g e leaf angle: 30, 35, 40, 45, 50, 55, 60, 65, 70 °
1985). Most vegetation indices are dependent on I L A I : 0.10, 0.20, 0.40, 0.80, 1.60, 3.20, 6.40, 12.80
I,'J.,' Y , /
/ . ,,
i
• O. l ~ / ~ ~ PVI
lit °
0 t~ --SAVI 1 TSAV1
and near-infrared (nir) reflectances or radiances. optical properties of soil background. For a given
RVI = r / n i r amount of vegetation, darker soil substrates result
in higher vegetation index values (Elvidge and
(Pearson and Miller, 1972), (1)
Lyon, 1985; Huete et al., 1985) (Fig. 1). To mini-
NDVI = ( n i r - r) / (nir + r) mize the effect of the soil background, Richardson
(Rouse et al., 1974), (2) and Wiegand (1977) have proposed the perpendic-
ular vegetation index (PVI). It represents the or-
NDVI = (RVI - 1 ) / ( R V I + 1). (3)
thogonal distance from a point corresponding to
These indices enhance the contrast between soil canopy reflectance to the soil line, in red-near-
and vegetation but minimize the effects of illumi- infrared space (Fig. 1). For a given soil, the red
nation conditions. However, they are sensitive to (rn) and near-infrared (nir~) reflectances are re-
164 Baret and Guyot
lated by the equation of the soil line: will not be used in the comparison of the different
indices.
nir,. = a ' r ~ + b. (4) The soil adjusted vegetation index (SAVI) pro-
posed by Huete (1988) is derived from the NDVI:
The parameters a and b vary slightly among soils
(Huete et al., 1984). Findings on soil line parame- SAVI = ( n i r - r ) / ( n i r + r + L)(1 + L). (8)
ters determined in different locations confirms this The constant L is introduced in order to minimize
assertion, and shows that these parameters are not soil-brightness influences and to produce vegeta-
independent and that they can be related by a tion isolines more independent of the soil back-
second-order polynomial (Baret et al., 1989b). ground (Fig. 1). It can vary from zero to infinity as
Moreover, as shown by Baret (1986), the soil line a function of the canopy density. If L = 0, SAVI is
concept is also valid for senescent vegetation, equivalent to NDVI and if L tends towards infin-
which represents the real background of the active ity, it is equivalent to PVI (the vegetation isolines
green vegetation layer during a significant portion are parallel). For vegetation with intermediate
of the growth cycle. density the best adjustment is obtained for L = 0.5.
It is possible to express PVI as a linear combi- SAVI is an exact solution fi)r bare soil only when
nation of nir, and r,.: the soil line parameters are a = 1 and b = 0. This
is not generally the case. As it is important, for a
1
P V I - ~ / e +- - l (nir - a ' r - b). (5) vegetation index, to be error-free for plant canopies
with very low densities, Baret et al. (1989b) pro-
posed the transformed soil adjusted vegetation
Experimental and theoretical investigations show index (TSAVI). This index is a measure of the
that PVI is also affected by the optical properties angle between the soil line and the line which
of soil background: brighter soils result in higher joins the vegetation point and a point (S) behmg-
index values for a given quantity of incomplete ing to the soil line, the abscissa of which is - X
vegetation cover as shown in Figure 1 (Huete (Fig. 1). The following equation is an improved
et al., 1985; Huete, 1988; Major et al., 1990). The version of the initial definition given by Baret
points corresponding to the same canopy do not et al. (1989b):
migrate along lines parallel to the soil line when
the soil brightness is changing. For this reason TSAVI = a ' ( n i r - a ' r - b)
some new indices, which are less influenced by
the soil brightness have been proposed (Huete, /[.'nir+r-.h+ X'(1+.2)], (9)
1988; 1989; Clevers, 1986; 1988; 1989; Baret et al.,
19891); Major et al., 1990). The simplest is pro- where a and b are the parameters of the soil line
posed by Clevers (1986; 1989), who introduced and X corresponds to the negative abscissa of the
the weighted difference vegetation index (WDVI), point S. The value of X has been adjusted to
which is similar to the greenness index of Kauth minimize soil effects (X = 0.08). TSAVI is multi-
and Thomas (1976) fin" the two-dimensional case plied by the parameter a to give a vegetation
(red-near-infrared space): index less dependent on soil parameters (a, b) for
high canopy density (nir >> r). TSAVI equals 0 for
WDVI = n i r - a ' r . bare soil and is close to 0.70 fi)r very dense
canopies. For a = 1 and b = 0, TSAVI is equiva-
This index can also be related to the PVI by lent to N DVI.
combining Eqs. (5) and (6): Tile indices discussed here can be classified
into two categories:
1
PVI - -~ - ( W D V I - b). (7)
--indices characterized by a slope: RV1,
NDVI, SAVI, TSAVI;
The information given by WDVI is not different --indices characterized by a distance: PVI,
from that given by PVI. For this reason this index WDVI, GVI;
Vegetation Indices for LAI and APAR Assessment 165
i
VI~ = vegetation index corresponding to
0 0 ' L
0 0.5 0 0,2 0.4 that of the bare soil,
NDVI PVl Vim = asymptotic value of VI when
Figure 2. Graphical determination of the mutual depen- LAI tends towards infinity (practically
dency of NDVI, PVI, SAVI, and TSAVI. Each data point this limit is always reached for LAI
corresponds to a VI value calculated for the canopy re-
flectance (simulated with SAIL with input variables front greater than 8.0),
Table 1). The Vls are expressed in decimal fractions. Kvl = eoeffleient which controls the slope
of the relationship (equivalent to an
extinction coefficient).
Four of them, that is, NDVI, PVI, SAVI, and The Kvj parameter represents the relative in-
TSAVI, are really not fnnctionally equivalent, al- crease in VI due to an elementary increase in LAI.
though, for certain values of the soil line and of It is called extinction coefficient by analogy to the
the parameters L and X, SAVI and TSAVI can classical Beer's law where the extinction (or ab-
also be considered as distance vegetation indices. sorption) eoeMeient describes the relative varia-
To test the mutual independence of these four VIs, tion of a diffuse mediunfs transmission when its
they were calculated from the same data set (Ta- thickness is submitted to an elementary increase.
ble 1). All possible pairs of VIs (six combinations) In the same way, the product Kvl. LAI is analog
were plotted (Fig. 2). Figures 2a, 2c, and 2e show to an optical thickness.
that PVI, SAVI, and TSAVI individually differ in The parameters VI~ and Kvl depend on irra-
concept and contain different information than diance and view geometries and on leaf inclina-
NDVI, while Figure 2b shows that TSAVI and tion. In this analysis the sun zenith angle and the
SAVI are closely correlated. For this reason we viewing geometries are fixed.
compare SAVI and TSAVI with NDVI in Figures We have used, for each set of leaf inclinations
2c and 2e and with PVI in Figures 2d and 2£ (average leaf inclination ALA), nonlinear fitting
Figures 2c and 2d show that SAVI and PVI give techniques ( N e l d e r - M e a d simplex algorithm) to
similar information. TSAVI is not closely related to obtain Kvx(ALA) and VI=(ALA) for data simulated
NDVI and PVI (Figures 2e and 2f). Nevertheless, with SAIL model. VIg was taken as the average
166 Baret and Guyot
Table 2. Average Value of Vlg Calculated with Data of is not strongly affected by the leaf inclination:
Table 1
PVI~ and SAVI~ decrease slightly with the leaf
VI VI~ StandardDeviation inclination, whereas NDVI~ increases and TSAVI~
NDVI 0.193 0.0646 remains practically constant. The computed values
PVI 0.000 0.0000 of K vi and VI~ are in good agreement with exper-
SAVI 0.122 0.0229
TSAVI 0.000 0,0000 imental results obtained on wheat (Asrar et al.,
1984), maize, and soybean (Baret, 1990).
VIs computed with the simplified model
value (Table 2) calculated from the input data (equation 11) are compared to those computed
displayed in Table 1. The resulting values of Kvi with SAIL model in Figure 4. In the simplified
and VI= are presented on Figure 3. This figure model the adjusted values of KvI(ALA) and
shows that K w decreases when ALA increases. VI~(ALA) are used. For VIg fixed, the indices,
The curves corresponding to the different vegeta- which are significantly affected by the soil back-
tion indices have, practically, the same shape. VI~ ground, present a larger scattering than the in-
Figure 3. Variation of the extinction coefficient (Kvl) and of the infinite value of vegetation
index (VI~) as a function of the average leaf inclination (ALA).
1.6 l
NDVI
NDVI
1.4
t~
0.8 SAVI
1.2
z TSAVI
1 Z 0.6
Z
0.8 ...........................................
0.4 PVI
Z 0.6
>
t~ 0.4 0.2 t i ,i
30 40 50 60 70 30 40 50 60 70
AI~
NDVI PVI
0.4
( a ) ~ 0.3
b~
0.5 s_
0.2
0.1
i
0
0 0,5 0 0.2 0.4
SAIL SAIL
SAVI TSAVI
r~
M Figure 4. Comparison of VI data
0.5 0.5 calculated with SAIL model and
with simplified model (dots). The
solid line is the first bisector. As VI~
has a constant value in the simpli-
i
fled model, the scattering of the
0 0 i
points expresses the effect of the soil
0 0.5 1 0 0.5 1 background on the considered in-
SAIL SAIL dex.
Vegetation Indices for LAI and APAR Assessment 167
Table 3. VI RMS D e t e r m i n e d with Simplified Model as which is mainly determined by species and phe-
Compared to SAIL Simulations.
nology, is often available. For this reason we shall
vl RMS RMS / (VI=- VI~)~ first analyze the sensitivity to soil background for a
NDVI 0.0570 0.0738 given ALA and then evaluate the global sensitivity
PV1 0.0187 0.0548 to both soil background and ALA.
SAVI 0.0204 0.0315
TSAVI 0.0101 0.0141
Equation (11) shows that RENLAI is the largest
when the slope of the relation between LAI and
~The third column gives the normalized RMS.
VI is the smallest. For this reason the uncertainty
on LAI is maximum when the asymptotic level is
dices which introduce corrections of the back- reached.
ground effect. For this reason Figures 4a and 4b,
corresponding to NDVI and PVI, display larger Sensitivity to Soil Background for a Given ALA
scatter than Figures 4c and 4d, corresponding to RENLA~ [Eq. (11)] is computed from data in Table
SAVI and TSAVI. Low NDVI values are mainly 1 but for only four different ALA values (35 °, 45 °,
affected by the soil (Fig. 4a). The effect of the soil 55 °, and 65 °) in order to decouple the effects of
on PVI is the largest for medium values of this soil background and ALA. Figure 5 displays the
index (Fig. 4b). SAVI and TSAVI (Figs. 4c and 4d) results of this sensitivity study. There are some
greatly reduce the scattering and TSAVI is the large differences among the VIs and their variation
better index according to this test. It is clear that with leaf inclination. Since NDVI is strongly
Eq. (10) is a good semiempirical approximation of affected by the soil optical properties, the rela-
the VI(LAI) relation as confirmed by the statistical tive equivalent noise is very large, for this vege-
analyses (Table 3). The root mean square error, tation index, when the vegetation density is low
evaluated for each VI, has been normalized by the (LAI < 0.5), and when LAI is greater than 3, for
amplitude of variation (VI=-VIg), to facilitate the canopies with a low ALA (35 ° and 45 ° on Fig. 5).
comparison. At LAI less than 4, the relative equivalent noise
does not change much with the increasing LAI if
PVI, SAVI, or TSAVI are considered. In this do-
Sensitivity Analysis of the Relations between main the better index is TSAVI (RENLAI = 15%)
VI and LAI followed by SAVI and PVI. These three indices
have the same noise in the LAI range 2-4. For
The sensitivity of VI(LAI, ALA) to soil background
LAI greater than 4 the hierarchy among the in-
or leaf inclination has been characterized by the
dices changes, especially when the leaves have a
standard deviation (~rvi) of the relation between
low inclination. For that case the worst index is
LAI and VI. This noise is translated into relative
TSAVI followed by NDVI, SAVI, and PVI. For
equivalent LAI noise (RENtz I) defined by
erect leaves the best index is NDVI followed by
O'LaI / L A I , using the local slope of the LAI-VI
TSAVI, SAVI, and PVI.
relation (d(VI)/d(LAI))
The large variations observed in the relative
equivalent noise for NDVI and TSAVI, when LAI
RENLAI O'LAI_ ~rvi (d(Vi))-1 is greater than 4, are due to the low values of the
= LA---]- LA~ d(LAI) (11)
slope of the relation VI-LAI, because VI is near
The slope of the relation between VI and LAI its saturation level. A small variation in VI can
can be deduced from Eq. (10): correspond to a large variation in LAI. This result
is mainly due to differences in Kvi (Fig. 3) which
d(VI) determines the magnitude of the slope between VI
d(LAI) - KvI'(VIg-VI=)'exp(- KvI'LAI ).
and LAI [Eq. (12)].
(12)
Sensitivity to Soil Background and Leaf Inclination
In Eq. (11) VIg is fixed at its average value We next discuss the sensitivity of VI to soil back-
given in Table 2. In most cases VIg varies with soil ground and leaf inclination. For a given LAI, O-vi
type, soil roughness, and moisture, and is not is again computed using the data set of Table 1
generally known. However, information on ALA (variations in soil optical properties and ALA). It is
168 Baret and Guyot
ALA = 35 ° ALA = 4 5 °
0 0
0.5 0.5
0 0
10-t 10 o 101 NDVI 10-1 10 o I01
LAI PVI .... LAI
SAVI .......
~ = 55 ° ALA = 65 °
TSAVI -.-- 1
03
O 0
10-1 10 0 101
LAI Conclusion
Figure 6. Effect o f soil b a c k g r o u n d and leaf inclination on
LAI estimation from VI. T h e sensitivity is e x p r e s s e d as LAI
relative e q u i v a l e n t noise as a function o f LAI (logarithmic The VI(LAI) relation can be set in the fi)rm of a
scale). R E N t a I is e x p r e s s e d in decimal fraction. simple semiempirical Beer's law [Eq. (10)]. This
approach was used to facilitate the comparison
between the differing VIs. The sensitivity analysis
then translated into RENI~ I using Eqs. (11) and shows that Vls, devised to minimize soil back-
(12). Figure 6 shows that the variation in relative ground effect (PVI, SAVI, TSAVI), strongly reduce
equivalent noise, due to leaf inclination, signifi- the noise for low leaf area indices (LAI < 2-3).
cantly increases whatever the VI considered• Two For greater leaf area indices this gain can be
domains can be distinguished (Fig. 6), that is, compensated by the magnitude of the slope of the
LAI > 3 (i) and LAI < 3 (ii). VI(LAI) fimetion. For example, TSAVI which was
(i) For NDVI, the noise in the determination the best VI for lower LAI, introduced the largest
of LAI decreases with increasing LAI, up to LAI noise for large LAI because it reached its satura-
= 3. This behavior is mainly attributed to soil tion level before the other VIs (except NDVI).
background effect. For the other indices, the noise The noise due to soil background was amplified
is practically constant and the best index is TSAVI when combined with the noise due to leaf inclina-
followed by SAVI and PVI. The contribution of the tion. In all cases it seems to be very diffieuh to
noise due to leaf angle distribution increases the estimate LAI through VI measurements when VI
RENtAI due to soil background (Fig. 5) by approx- is close to Viol, especially when ALA is not known.
Vegetation Indices for LAI and APAR Assessment I69
ALA = 35 ° ALA = 45 °
10 o 10 o
10-t "~ l O - i
co
,.d
0~
10-t 10 o
t0-t
PVI .... P
SAVI .......
M~ = 55 ° TSAVI --.-- 10 o M.~ = 65 °
10 0 v i i i , , T ~_
!
0
Sensing and IGARSS'90, Vancouver, Canada, 10-14 July Hatfield, J. L., Asrar, G., and Kanemasu, E. T. (1984), Inter-
1989, 4 pp. cepted photosynthetically active radiation estimated by
spectral reflectance, Remote Sens. Environ. 14:65-75.
Best, R. G., and Harlan, J. C. (1985), Spectral estimation of
green leaf area index of oats, Remote Sens. Environ. Hatfield, J. L., Kanemasu, E. T., Asrar, C., Jackson, R. D.,
17:27-36. Pinter, P. J., Jr., Reginato, R. J., and ldso, S. B. (1985),
Leaf-area estimates from spectral measurements over vari-
Brach, E. J., Desjardins, R. L., Lord, D., and Dube, P. A. ous planting dates of wheat, Int. J. Remote Sens.
(1981), Field spectrometer to measure percent ground 6:167-175.
coverage and leaf area index of agriculture crops, in Inter-
Hipps, L. E., Asrar, G., and Kanemasu, E. T. (1983), Assess-
national CoUoquium on Spectral Signatures of Objects in
ing the interception of photosynthetically active radiation
Remote Sensing, Avignon, France, 8-11 Sept., Les Collo-
ill winter wheat, Agric. Meteorol. 28:253-259.
ques de I'INRA N ° 5, Ed. INRA, Versailles, pp. 323-330.
Holben, B. N., Tucker, C. J., and Fan, C. J. (1980), Spectral
Campbell, G. S. (1986), Extinction coefficients for radiation in
assessment of soybean leaf area and leaf biomass, Pho-
plant canopies calculated using an ellipsoidal inclination
togramm. Eng. Remote Sens. 46:651-656.
angle distribution, Agric. Forest Meteorol. 36:317-321.
Holben, B. N., Kimes, D. S., and Fraser, R. S. (1986),
Chance, J. E. (1981), Crop identification and leaf area index Directional reflectance in AVHRR red and near-IR bands
calculation with Landsat multispectral data, Int. J. Remote for three cover types and varying atmospheric conditions,
Sens. 2:1-14. Remote Sens. Environ. 19:213-236.
Choudhury, B. J. (1987), Relationships between vegetation Huete, A. R. (1987a), Soil dependent spectral response in a
indices, radiation absorption and net photosynthesis evahl- developing plant canopy, Agron. J. 79:61-68.
ated by a sensitivity analysis, Remote Sens. Environ.
Huete, A. R. (1987b), Soil and Sun angle interactions on
22:209-233. partial canopy spectra, Int. J. Remote Sens. 8:1307-1317.
Clevers, J. G. P. W. (1986), The application of a vegetation Huete, A. R. (1988), A soil adjusted vegetation index (SAVI),
index in correcting the infrared reflectance for soil back- Remote Sens. Environ. 25:295-309.
ground, in Proc. Syrup on Remote Sensing for Resources
Huete, A. R. (1989), Soil influences in remotely sensed vege-
Development and Environmental Management, Enschede,
tation-canopy spectra, in Theory and Application of Optical
25-29 August 1986 (M. C. J. Damen, G. Sicco Smit, and
Remote Sensing (G. Asrar, Ed.), Wiley Series in Remote
H. TH. Verstappen, Eds.), International Archives of Pho-
Sensing, Wiley, New York, pp. 107-141.
togrammetry and Remote Sensing, Vol. 26 (7/1), Balkema,
Rotterdam, Boston, pp. 221-226. tluete, A. R., and Jackson, R. D. (1987), The suitability of
spectral indices for evaluating vegetation characteristics on
Clevers, J. G. P. W. (1988), The derivation of a simplified arid range-lands, Remote Sens. Environ. 23:213-232.
reflectance model for the estimation of leaf area index,
Remote Sens. Environ. 25:53-69. Iluete, A. R., Post, D. F., and Jackson, R. D. (1984), Soil
spectral effect on 4-space vegetation discrimination, Re-
Clevers, J. G. P. W. (1989), The application of a weighted mote Sens. Environ. 15:155-165.
infiared-red vegetation index for estimating leaf area index
ltuete, A. R., Jackson, R. D., and Post, D. F. (1985), Spectral
hy correcting fur soil moisture, Remote Sens. Environ.
response of a plant canopy with different soil backgrounds,
29:25-37.
Remote Sens. Environ. 17:37-53.
Daughtry, C. S. T., Bauer, M. E., Crecellius, D. W., and Jackson, R. D. (1983), Spectral indices in n-space, Remote
Hixson, M. M., (1980), Effects of management practices on Sens. Environ. 13:409-421.
reflectance of spring wheat canopies, Agron. J.
72:1055-1060. Jackson, R. D. (1986), Spectral response of architecturally
different wheat canopies, Remote Sens. Environ. 20:43-56.
Daught~,, C. S. T., Gallo, K P., and Bauer, M. E. (1983),
Jackson, R. D., Pinter, P. J., Idso, S. B., and Reginato, R. J.
Spectral estimates of solar radiation intercepted by corn
(1979), Wheat spectral reflectance: interaction between
canopies, Agron. J. 75:527-531.
crop configuration, sun elevation and azimuth angle, Appl.
Elvidge, C. D., and Lyon, R. J. P. (1985), Influence of Opt. 18:3730-3732.
rock-soil spectral variation on assessment of green biomass, Jackson, R. D., Slater, P. N., and Pinter, P. J., Jr. (1983),
Remote Sens. Environ. 17:265-279. Discrimination of growth and water stress in wheat by
Gallo, K. P., Daughtry, C. S. T., and Bauer, M. E. (1985), various vegetation indices through clear and turbid atmo-
Spectral estimation of absorbed Photosynthetically active spheres, Remote Sens. Environ. 13:187-208.
radiation in corn canopies, Remote Sens. Environ. Jacquemoud, S., and Baret, F. (1990), PROSPECT: a model
17:221-232. of leal optical properties spectra, Remote Sens. Environ.,
Goel, N. S., and Deering, D. W. (1985), Evaluation of a 34:75-91.
canopy reflectance model for LAI estimation through its Kanemasu, E. T. (1974), Seasonal canopy reflectance patterns
inversion, IEEE Trans. Geosci. Remote Sens. GE- of wheat, sorghum and soybean, Remote Sens. Environ.
23:674-684. 3:43-47.
Vegetation Indices for LAI and A P A R Assessment 173
Kauth, R. J., and Thomas, G. S. (1976), The Tasselled C a p - - a rayonnement solaire absorb6 ou intercept6 par un couvert
graphic description of the spectral-temporal development v6gdtal, Agronomie 9:419-439.
of agricultural crops as seen by Landsat, in Proc. Symp. on Verhoef, W. (1984), Light scattering by leaf layers with
Machine Processing of Remotely Sensed Data, Purdue application to canopy reflectance modeling: the SAIL
University, West Lafayette, IN, pp. 41-51. model, Remote Sens. Environ. 16:125-141.
Kollenkark, J. C., Vanderbilt, V. C., Bauer, M. E., and Hous- Verhoef, W. (1985), Earth observation modeling based on
ley, T. L. (1982), Effect of cultural practices on agronomic layer scattering matrices, Remote Sens. Environ.
and reflectance characteristics of soybean canopies, Agron. 17:165-178.
J. 74:751-758.
Vygodskaya, N. N., Gorshkova, I. I., and Fadeyeva, Ye. V.
Major, D. J., Baret, F., and Guyot, G. (1990), A ratio vegeta- (1989), Theoretical estimates of sensitivity in some vegeta-
tion index adjusted for soil brightness, Int. J. Remote Sens. tion indices to variation in the canopy condition, Int. J.
11(5):727-740. Remote Sens. 10:1857-1872.
Pearson, R. L., and Miller, L. D. (1972), Remote mapping of Wang, Y. P., and Jarvis, P. G. (1988), Mean leaf angles for the
standing crop biomass for estimation of the productivity of ellipsoidal inclination angle distribution, Agric. Forest Me-
the short-grass Prairie, Pawnee National Grasslands, Col- teorol. 43:319-321.
orado, in Proc. 8th Int. Symp. on Remote Sens. of Environ., Wanjura, D. F., and Hatfield, J. L. (1987), Sensitivity of
ERIM, Ann Arbor, MI, pp. 1357-1381. spectral vegetation indices to crop biomass, Trans. ASAE
Richardson, A. J., and Wiegand, C. L. (1977), Distinguishing 30:810-816.
vegetation from soil background information, Photogramm. Weiser, R. L., Asrar, G., Miller, G. P., and Kanemasu, E. T.
Eng. Remote Sens. 43:1541-1552. (1986), Assessing grassland characteristics from spectral
Rouse, J. W., Haas, R. H., Schell, J. A., Deering, D. W., and measurements, Remote Sens. Environ. 20:141-152.
Harlan, J. C. (1974), Monitoring the vernal advancement Wiegand, C. L., and Hatfield, J. L. (1988), The Spectral-Agro-
of retrogradation of natural vegetation, NASA/GSFC, Type nomic Multisite-multicrop Analyses (SAMMA) Project, in
III, Final Report, Greenbelt, MD, 371 pp. Proc. 16th Cong., Int. Soc. Photogramm. and Remote
Sens., Kyoto, Japan, 1-10 July 1988, Int. Arch. Pho-
Sellers, P. J. (1985), Canopy reflectance, photosynthesis and
togramm. Remote Sens. 27(B7):696-706.
transpiration, Int. J. Remote Sens. 6:1335-1372.
Wiegand, C. L., and Richardson, A. J. (1984), Leaf area, light
Sellers, P. J. (1987), Canopy reflectance, photosynthesis and interception and yield estimates from spectral components
transpiration II. The role of the biophysics in the linearity analysis, Agron. J. 76:543-548.
of their dependence, Remote Sens. Environ. 21:143-183.
Wiegand, C. L., Richardson, A. J. (1990a), Use of spectral
Shibayama, M., Wiegand, C. L., and Richardson, A. J. (1986), vegetation indices to infer leaf area, evapotranspiration
Diurnal patterns of bidirectional vegetation indices for and yield: I. Rationale, Agron. J. 82:623-629.
wheat canopies, Int. J. Remote Sens. 7:233-246.
Wiegand, C. L., and Richardson, A. J. (1990b), Use of spectral
Tucker, C. J. (1979), Red and photographic infrared linear vegetation indices to infer leaf area, evapotranspiration
combinations for monitoring vegetation, Remote Sens. En- and yield: II. Results, Agron. J. 82:630-636.
viron. 8:127-150. Wiegand, C. L., Gausman, H. W., Cuellar, J. A., Gerberman,
Vanderbilt, V. C., Kollenkark, J. C., Biehl, L. L., Robinson, A. H., and Richardson, A. J. (1974), Vegetation density as
B. F., Bauer, M. E., and Ranson, K. J. (1981), Diurnal deduced from ERTS-1 MSS response, Third ERTS Symp.,
changes in reflectance factor due to sun-row direction U.S. Government Printing Office, Washington, DC, Vol. I,
interactions, in International Colloquium on Spectral Sig- pp. 93-116.
natures of Objects in Remote Sensing, Avignon, France, Wiegand, C. L., Richardson, A. J., Jackson, R. D., Pinter, P. J.,
8-11 Sept., Les colloques de rlNRA N ° 5, Ed. INRA, Jr., Aase, J. K., Smika, D. E., Lautenschlager, L. F., and
Versailles, pp. 499-508. McMurtrey, J. E., III, (1986), Development of agrometeo-
Varlet-Grancher, C., Gosse, G., Chartier, M., Sinoquet, H., rological crop model inputs from remotely sensed informa-
Bonhomme, R., and Allirand, J. M. (1989), Mise au point: tion, IEEE Trans. Geosci. Remote Sens. GE-24:90-98.