INT. J.

REMOTE SENSING, 10 DECEMBER, 2004,

VOL. 25, NO. 23, 5415–5427

Effect of senescent leaves on NDVI-based estimates of fAPAR:

experimental and modelling evidences

C. M. DI BELLA {{*, J. M. PARUELO§, J. E. BECERRA{§,

C. BACOUR} and F. BARET}
{Instituto de Clima y Agua, INTA, Los Reseros y Las Cabañas S/N (1712),
Castelar, Buenos Aires, Argentina
{Departamento de Producción Animal, Cátedra de Forrajes, Facultad de
Agronomı́a, Universidad de Buenos Aires, Av. San Martı́n 4453, C1417DSE,
Ciudad Autónoma de Buenos Aires, Argentina
§Laboratorio de Análisis Regional y Teledetección (LART), IFEVA, Cátedra
de Ecologı́a-Facultad de Agronomı́a, UBA Av. San Martı́n 4453, C1417DSE,
Ciudad Autónoma de Buenos Aires, Argentina
}INRA-CSE, Site Agroparc, Domaine Saint-Paul, 84914 Avignon Cedex 9,
France

(Received 9 October 2003; in final form 10 May 2004 )

Abstract. Spectral indices from remotely sensed data, such as the Normalized
Difference Vegetation Index (NDVI), are often used to estimate biophysical
characteristics of vegetation. The objective of this study is to evaluate the effect
of senescent leaves on the estimation of the fraction of photosynthetically active
radiation absorbed by the green elements of the canopy (fAPARg) from NDVI
measurements. An experiment was conducted under controlled conditions over
grass canopies. Both NDVI and fAPARg were measured when the cover fraction
by senescent leaves was changed. The results demonstrated that the effect of
senescent leaves was significant on NDVI values. A similar effect was observed
on the fAPARg values. In these conditions, simple models were developed to
relate the fAPARg to NDVI values when the cover fractions of the senescent
leaves vary. When combining these models, a linear relationship was found
between NDVI and fAPARg, which was in good agreement with our
experimental observations. Complementary radiative transfer model simulations
were then run and confirmed these results. In addition, the vertical and
horizontal distribution of the senescent leaves was investigated. They show that
the relationship between NDVI and fAPARg was only marginally affected by the
distribution of the senescent leaves.

1. Introduction
In the last 20 years, remotely sensed data have become an important tool to
study the temporal and spatial variations of structural and functional attributes of
ecosystems (Davis and Roberts 2000, Running et al. 2000). The use of remotely
sensed data in such studies is often based on the contrast observed over
photosynthetic tissues in the red and infrared reflectance values. Because of

*Corresponding author; e-mail: [email protected]

International Journal of Remote Sensing

ISSN 0143-1161 print/ISSN 1366-5901 online # 2004 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/01431160412331269724
5416 C. M. Di Bella et al.

chlorophyll absorption, green leaves reflect a small proportion of the incoming

radiation in the red and a high proportion in the near infrared bands (Guyot 1990).
Several indices have been generated combining information from these two bands.
One of the most widely used indices is the Normalized Difference Vegetation Index
(NDVI) (Rouse et al. 1974, Bannari et al. 1995) (equation (1)):
NIR R
NDVI~ ð1Þ
NIRzR
where R is the red band and NIR is the near infrared band. NDVI values range
between -1 and 1. Thus high values could be associated with photosynthetic
activity, due to a reduction in leaf reflectance in the red band because of chlorophyll
absorption and an increase in infrared reflection due to the leaf structure (Baret
1990). Strong relationships have been reported between NDVI and some structural
and functional characteristics of vegetation such as biomass (Tucker 1977a,b,
Gerberman et al. 1984, Ripple 1985, Sellers 1985), Leaf Area Index (LAI) (Curran
1983, Asrar et al. 1984, Baret et al. 1989), Aerial Net Primary Productivity (ANPP)
(Prince 1991, Paruelo et al. 1997) or Absorbed Photosynthetic Active Radiation
(APAR) (Goward et al. 1994, Gamon et al. 1995). Such relationships lead to many
applications of NDVI in regional studies, including land use and land cover
classifications (Guerchman et al. 2003), estimates of livestock stocking density
(Oesterheld et al. 1998) or actual evapotranspiration (Di Bella et al. 2000).
Nevertheless, the NDVI may not be sensitive in some cases to changes in vegetation
structure, i.e. LAI or biomass. Variations in soil background, atmosphere
variations, leaf orientation, canopy structure, sun or viewing angle and leaf optical
properties may affect the relationship between NDVI and such structural variables
(Baret et al. 1989, Baret and Guyot 1991).
Based on theoretical considerations, Sellers (1989) demonstrated that the nature
of the exponential relationship between the absorbed fraction of Photosynthetic
Active Radiation (fAPAR) and LAI is described by an extinction coefficient
(kfAPAR), which explains the radiation attenuation into the canopy. Similarly,
NDVI is also related to LAI by an exponential relationship characterized by an
extinction coefficient kNDVI (Asrar et al. 1984). Because kfAPAR and kNDVI values
are similar, it follows that fAPAR and NDVI are almost linearly correlated (Sellers
et al. 1992). Despite the empirical and theoretical evidence supporting the use of
spectral indices to estimate structural or functional variables of the canopy, some
critical questions still remain unsolved. A particularly important question for grass-
dominated systems is the influence of dead material on the radiative transfer in the
canopy and the possible consequences on the relationship between NDVI and
canopy characteristics.
Few studies have analysed the influence of standing dead biomass on canopy
reflectance under experimental conditions. Asner (1998) highlighted the importance
of increasing radiation transfer measurements at the canopy level. On the other
hand Gamon et al. (1995) showed that additional estimates of dead tissue biomass,
which can be dominant in certain types of vegetation, are needed to estimate
canopy green biomass from NDVI values. Serrano et al. (2000) indicated the
importance of removing non-photosynthetic materials, like dead or senescent
biomass, when spectral measurements are carried out in the canopy to estimate
characteristics associated with the green parts. Most of the investigations on canopy
structure effects on the estimates of biomass or fAPAR from vegetation indices are
based on models and generally only green elements are taken into account.
Senescent leaf effect on fAPAR 5417

In this article we investigated the effect of standing dead biomass on NDVI and
fAPAR using two approaches, manipulative experiment and simulation analyses using
a Radiative Transfer Model (RTM). We analysed to what extent the effect of senescent
leaves on NDVI is due to changes in the spectral characteristics of the whole canopy or
to changes in the amount of canopy absorbed radiation by the canopy (fAPAR).

2. Materials and methods

Two sets of experiments were conducted to study the effect of the presence of
senescent leaves on the NDVI-fAPAR relationship. The first one corresponded to
actual measurements under controlled conditions. The second one consisted of
simulating a range of conditions with a RTM.

2.1. Experiments under controlled conditions

Plants of ryegrass (Lolium perenne L. var. Manhattan) were cultivated in pots of
15 cm diameter and 15 cm height. Plants grew at 20–25‡C during the day and
15–20‡C during the night and were fertilized at tillering with urea (44 kg N ha21) to
eliminate possible nitrogen deficiencies. Pots were watered to field capacity every
two days, eliminating major water stress.
A basic matrix of 16 pots (464 pots) was designed to simulate the canopy
(figure 1(a)). Above the canopy, a structure (figure 1(c)) bordered by a neutral filter
was placed to eliminate external sources of diffuse radiation. A special wire mesh
mounted on this structure (figure 1(c), position A) allowed strips of senescent
material of ryegrass to be placed on top of the canopy. Each strip was 60 cm
long66 cm wide and 0.3 cm deep and covered 10% of the total surface of the
canopy of 60 cm660 cm (figure 1(a)). The wire mesh on which we placed the strips
intercepted a small fraction of radiation (less than 3.6%).
Radiometric and photosynthetic active radiation (PAR) measurements were
performed using this experimental design. A SKYE SKR100 (Skye Instruments
Ltd, UK) radiometer was used to measure the R/NIR ratio (660/730 nm), from
which the NDVI value was derived. A reference panel surface (Spectralon) was used
every three measures to normalize for changes in illumination conditions and to
obtain calibrated NDVI values. A one-metre linear sensor LICOR (Li Quantum
Sensor) was used to measure PAR. Taking into account that the canopy was 60 cm
wide, we covered 40 cm of the sensor with black polyethylene. All measurements
were performed at noon during 1 hour under completely clear sky conditions.
Measurements of NDVI were performed with nine replicates, each one covering
1/9 of the total surface (figure 1(b)). Five replicates of PAR were measured at
different levels (figure 1(c)). The fraction f of PAR absorbed by a horizontal layer
above a background ( f) could be expressed as:
f ~1{t{r{trs ð2Þ
where r is the canopy hemispherical reflectance, rs is the background hemispherical
reflectance, and t is the transmittance, r, t and rs being computed within the PAR
domain. Because in the PAR domain, the leaves are absorbing most of the light, the
terms (r2trs) could be neglected. In these conditions, the fraction of PAR
radiation absorbed by the canopy can be simply expressed as:
f ~1{t ð3Þ
Applying equation (3) to our experimental set-up leads to the following
5418 C. M. Di Bella et al.

Figure 1. Controlled experiment design. (a) vertical view of the 464 pots canopy and a
strip of dead biomass covering 10% of total surface; (b) vertical view of the nine
radiometric measurements made with the radiometer (the circles represent the actual
field-of-view), and the five replicates of PAR measurements (the grey rectangles); and
(c) side view of the experimental design. The dashed line represents the level where
the wire mesh tray was placed. I, D, G are measurement positions. H and h are the
heights of sensor placement.

definitions of transmittances:
PAR 9
tt ~ PARgi >
>
=
PARd
td ~ PARi ð4Þ
>
>
PAR ;
tg ~ PARdg
where tt is the transmittance of the whole canopy, td that of the senescent layer, and
tg that of the green layer. PARi is the incoming radiation measured above the
canopy (figure 1(c), position I), PARd is the PAR registered by the sensor between
the strips of dead biomass and the green leaves (figure 1(c), position D), and PARg
is the PAR measured by the sensor below the canopy (figure 1(a), position G).
The fraction of PAR absorbed by the green and dead components of the canopy
(fAPAR) was approximated by:
f APAR~1{tt ~1{tg td ð5Þ
Senescent leaf effect on fAPAR 5419

The fraction of incoming PAR absorbed only by the green biomass (fAPARg)
could similarly be expressed as:

f APARg ~ 1{tg td ð6Þ

All the plants were harvested at the end of the experiment. The samples were
dried on a stove for 48 hours (65‡C) and weighed to determine dry weight.
Using this experimental design two experiments were carried out to evaluate the
effect of senescent biomass:
First experiment: we generate 11 levels of senescent leaf cover, ranging from 0 up to
100%, by adding strips of senescent tissue on top of the wire mesh. Each strip
placed on the structure (figure 1(c), position A) covered 10% (figure 1(b)). The
position of the strips was randomly assigned.
Second experiment: we considered two factors: the cover of senescent biomass and
the amount of green biomass present in the canopy. Four levels of senescent
biomass were generated by placing strips on top of the wire mesh, as in the first
experiment. In addition, we included 2, 4, 8 or 10 pots where the biomass was
uniformly reduced by cutting the leaves to one cm height two days before the
measurements. This reduction allowed us also to generate a green biomass
gradient.

2.2. Model simulation

Simulations of NDVI and fAPAR have been performed with the SAIL radiative
transfer model (Verhoef 1984, 1985) to confront the experimental and theoretical
results. In addition, the effect of the dead biomass distribution on canopy
reflectance and fAPAR will be analysed. The SAIL canopy radiative transfer model
(Verhoef 1984, 1985) calculates the reflectance, transmittance and absorptance of
the canopy from a limited number of variables describing its architecture and the
optical properties of the soil and leaves. Thus, the following input variables are
required: the LAI, the mean leaf inclination angle (ALA) assuming an ellipsoidal
distribution of foliage elements (Campbell 1990), the hot spot parameter (sl), soil
reflectance and leaf optical properties. The latter are computed thanks to the
PROSPECT model (Jacquemoud and Baret 1990), which simulates leaf reflectance
and transmittance spectra in the optical domain, given the leaf structure parameter
(N), the chlorophyll a and b contents Cab (mg cm22), the equivalent water thickness
Cw (cm), the dry matter content Cm (g cm22) and the brown pigment concentration
(Cbp) (Jacquemoud and Baret 1990, Baret and Fourty 1997). Finally, the multi-layer
version of SAIL used in this investigation (Weiss et al. 2001) allows canopies
composed of foliar elements exhibiting different optical properties to be considered
(green and senescent leaves in particular).
The values of the input variables were chosen to mimic the actual values
observed under the natural conditions for the grass canopies considered in the three
experiments. The soil reflectance corresponded to a typical dark soil with a high
organic matter content. When the variables were not precisely known, typical
values were selected. Table 1 shows the results. The simulations were performed
from 400 to 750 nm with a 5 nm step. The fAPAR values were computed by
integrating the absorptance vertically through the canopy depth, and spectrally
between 400–700 nm. Only the green elements were used in this fAPAR
computation in good agreement with the fAPARg values measured in the
5420 C. M. Di Bella et al.
Table 1. Values of the PROSPECTzSAIL input variables describing the green canopy
(green layer) and the mixed canopy (green layerzsenescent layer). N~leaf structure
parameter; Cab~chlorophyll a and b content; Cm~dry matter content; Cw~equi-
valent water thickness; Cbp~brown pigment concentration; LAI~Leaf Area Index;
ALA~mean leaf inclination angle; HOT~hot spot parameter.

Inputs Variables Green layer Senescent layer

N 1.3 4
Cab (mg cm22) 60 0
Cm (g cm22) 0.002 0.002
Cw (cm) 0.004 0.004
Leaf optical properties Cbp 0 5
LAI 4 3
ALA 57‡ 0‡
Canopy structure HOT 0.25 0.25

experiments. It also obviously corresponds to the required quantity when using

fAPAR values within photosynthetic models. Conversely, the NDVI values were
obviously computed by simulating the reflectance at 660 nm and 730 nm,
accounting for the contribution of all the elements (green and senescent).
Three types of canopies were simulated (figure 2):
Canopy 1: The canopy is composed of a green leaf homogeneous layer under
one incomplete layer of senescent leaves. This corresponds to the actual conditions
in the controlled experimental setup. Let C be the fraction of senescent leaves, with
C varying between 0 and 1.0 by 0.1 steps. The resulting canopy reflectance, rT, was
approximated as the weighed sum of the reflectance of the pure green layer (rG) and
that of the mixed canopy composed of the green layer under the complete senescent
layer (rGS):
rT ~rG ð1{C ÞzrGS C ð7Þ

Similarly, absorption by the green foliar elements aG

T of the total canopy is given
by:
aT ~aG ð1{C ÞzaGS C ð8Þ
where aG is the absorption by the green elements in the complete green layer and
aGS is that of the complete green layer under the complete senescent layer.
Canopy 2: The canopy is composed of a complete and homogeneous layer of

Figure 2. Scheme of the three types of canopies simulated by the radiative transfer model
(RTM). (a) Canopy 1, (b) canopy 2, (c) canopy 3.
Senescent leaf effect on fAPAR 5421

green leaves under a complete and homogeneous layer of senescent leaves. This
situation corresponds better to actual natural conditions, although it is more
difficult to simulate experimentally. The LAI of the senescent layer varied from 0 to
3 with 11 equidistributed levels.
Canopy 3: The canopy consisted of 20 layers alternatively made up of green and
senescent leaves. This corresponds to the situation where the green and senescent
leaves are homogeneously distributed within the canopy depth. It should be
considered as a possible actual situation, and also as an intermediate between the
two first situations. The total LAI for the senescent layer was similarly varied from
0 to 3 with 11 equidistributed levels. The absorption by the green elements was
computed as the sum of the absorption of each 10 elementary green layers.
Except for the senescent leaf cover fraction (C) in canopy 1 and the LAI of the
senescent layers for canopies 2 and 3, the other input variables remain the same
(table 1).

3. Results
The NDVI measured under the senescent leaf layer (NDVIg) in the first
experiment remained constant (around 0.9) for all the senescent cover fractions
(figure 3). In contrast, the NDVI values measured over the senescent layer decreased
when the cover fraction of the senescent leaves increased (figure 3). The high
variability observed for the average NDVI values was due to the location of the
senescent leaf strips that were integrated in each one of the measures. It is obviously

Figure 3. Relationship between the senescent cover fraction (C) and the NDVI measured at
different positions: integrating the effect of senescent and green biomass over the
senescent layer (NDVI — ¤) and below the senescent biomass layer (NDVIg —% ). Error
bars represent¡1 SD of nine averaged measurements.
5422 C. M. Di Bella et al.

maximal for the cover fractions of the senescent leaves close to 0.5, and minimal for
extreme senescent leaf cover fractions. This linear decrease of NDVI with the
senescent leaf cover fraction is similar to the increase of NDVI with the vegetation
fraction. This similarity is explained by the optical properties of the dead
vegetation: very little transmission and a low NDVI value associated with the dead
materials (in this case about NDVId~0.2). Conversely, the full green vegetation has
a NDVI value close to NDVIg~0.85. It is therefore possible to relate the NDVI
value to the senescent leaf cover fraction C:
NDVIðC Þ~NDVIð1ÞzC ðNDVIð0Þ{NDVIð1ÞÞ ð9Þ
where NDVI(0) and NDVI(1) correspond respectively to the NDVI values for a full
green cover layer (C~0), and to that of a complete senescent leaf layer (C~1).
Incoming PAR (PARi) in the first experiment was approximately constant
during the experiment, around 850 mmol s21 m22 (figure 4) confirming the stable
illumination conditions. The amount of PAR measured below the senescent leaf
layer (PARd) decreased linearly as the senescent cover fraction increased (r2~0.99;
n~11; pv0.001). Note that when the senescent leaf cover fraction is complete
(C~1), there is still a significant level of radiation (PARd(1)~200 mmol s21 m22 )
because of possible small gaps in the senescent leaf cover, and also because the layer
is still transmitting some radiation, particularly in the longer wavelengths, where
absorption by the brown pigments is weaker (figure 4). This behaviour could be
simply formalized by the following equation, expressed in transmittance td by

Figure 4. Relationship between senescent cover fraction (C) and PAR measured at different
positions. PARi~above the canopy or intercepted PAR (+); PARd~between the
senescent and green biomass (&) and PARg~measurement made down the canopy
(e).
Senescent leaf effect on fAPAR 5423

normalizing the PARd fluxes by the incident PAR (PARi):

td ðC Þ~1{C ð1{td ð1ÞÞ ð10Þ

Similarly, the amount of radiation measured below the green layer linearly
decreases when the senescent leaf cover fraction increases, because of the reduction
in the amount of incoming PAR (PARd) above the green layer (r2~0.94; n~11;
pv0.001).
Using equations (7), (8) and (10) leads to the following expressions for the
fractions of PAR flux absorbed by the whole canopy (fAPAR) and only the green
layer (fAPARg):
f APARðC Þ~1{tg ½1{C ð1{td ð1ÞÞ
~f APARð0Þ{C ðf APARð0Þ{f APARð1ÞÞ ð11Þ

f APARg ðC Þ~ 1{tg ½1{C ð1{td ð1ÞÞ

~f APARg ð0Þ{C f APARg ð0Þ{f APARg ð1Þ ð12Þ

We observed that the relationship between fAPAR and fAPARg and the
senescent leaf cover fraction is linear. By replacing the senescent leaf cover fraction
C in equations (11) and (12) with its expression as a function of NDVI (equation
(9)), we obtain the following expressions:
NDVIðC Þ{NDVIð1Þ
f APARðC Þ~f APARð0Þ{ ðf APARð0Þ{f APARð1ÞÞ ð13Þ
NDVIð0Þ{NDVIð1Þ
NDVIðC Þ{NDVIð1Þ
f APARg ðC Þ~f APARg ð0Þ{ f APARg ð0Þ{f APARg ð1Þ ð14Þ
NDVIð0Þ{NDVIð1Þ

It shows that both fractions of absorbed PAR are linearly related to NDVI
values. The total fraction of absorbed photosynthetically active radiation (fAPAR)
is negatively correlated to NDVI since fAPAR(0)vfAPAR(1). Conversely, the
fraction absorbed by the green layer is positively correlated with NDVI values since
fAPARg(0)wfAPARg(1). These theoretical results are confirmed by our experi-
mental observations (figure 5).
In the second experiment we controlled both senescent and green material
amounts by introducing strips of senescent biomass and pots with reduced green
biomass into the structure. Such experimental design allowed us to evaluate the
relationship between NDVI and green biomass together with changes in senescent
and green biomass. We found no relationship between green biomass and NDVI
(pv0.3). As in the first experiment, the NDVI was highly correlated to fAPARg
(r2~0.78; n~16 and pv0.001).
The robustness of the relationship between fAPARg and NDVI was checked by
comparison with our simulations for the three canopies considered. Results show a
very good agreement between these simulations and the experimental values
(figure 6). However, we noticed slight departures for the lower NDVI and fAPAR
values, presumably because of divergence in the optical properties of the senescent
leaves when simulating both fAPAR and NDVI.
5424 C. M. Di Bella et al.

Figure 5. Fraction of PAR absorbed by the whole canopy (fAPAR,z) and by the green
elements (fAPARg, o) as a function of NDVI. The lines correspond to the theoretical
relationship derived from equations (13) and (14).

Figure 6. Relationship between the fraction of the absorbed PAR by green biomass
(fAPARg) and NDVI for different conditions. (6) Measurements of first experiment
(MD) (fAPARg~0.03z0.96(NDVI)); (#) simulations of the first experiment
(senescent leaves incomplete layer – green leaves layer (SLGL)—canopy 1)
(fAPARg~0.2z0.7(NDVI)); (&) multilayer simulation (ML—canopy 3)
(fAPARg~0.07z0.9(NDVI)); and (+) senescent leaves – green leaves complete
layer (SLVGL) simulation (canopy 2) (fAPARg~0.18z0.67(NDVI)). n~11, pv0.001.
Senescent leaf effect on fAPAR 5425

4. Conclusions
Our experimental results demonstrated that the effect of senescent leaves on
NDVI values is significant. These results agree with Sellers (1985), van Leeuwen
and Huete (1996) and Asner (1998) indicating that the presence of any significant
fraction of dead material reduces vegetation indices such as NDVI. A significant
effect is also observed on the fraction of photosynthetically active radiation (PAR)
absorbed by the canopy. In this study, the focus was put on the absorption by the
green elements which are those involved in the photosynthesis process and
associated to the fraction of PAR absorbed by the green elements only (fAPARg).
In these conditions, simple models were developed to relate fAPARg to NDVI
values. The experiment used mainly consisted of changing the cover fraction by
senescent leaves over a complete green canopy. It shows a strong linear correlation
between fAPARg and NDVI as expected from the simple models extracted from the
relationship between the cover fraction of senescent leaves and both fAPARg and
NDVI. Radiative transfer model simulations were then run to confirm these results.
In addition, the vertical and horizontal distributions of the senescent leaves were
manipulated and affect only marginally the relationship between NDVI and
fAPARg.
In this study, the green and senescent leaf area index was varied trying to
simulate realistic conditions. However, further work is needed to investigate a
concurrent change in vertical and horizontal distributions that could be
encountered over the Earth’s surface.
NDVI does not provide a reliable estimate of biomass or LAI. On the contrary
its linear relationship with fAPARg strongly supports the use of this index to
estimate primary production. Monteith’s (1981) model provides the conceptual
framework to integrate the spectral index device from the remote sensor into
carbon gains estimations.

Acknowledgments
Paola Moglia, Gervasio Piñeiro, Rodolfo Golluscio, Gonzalo Irisarri and
Gabriel Kadarián helped during data collection.This work was supported by
CONICET, UBA and INTA, FONCYT (PICT99-06761) and project IICA-BID
FONTAGRO/RF-01-03-RG.

References
ASNER, G. P., 1998, Biophysical and biochemical sources of variability in canopy reflectance.
Remote Sensing of Environment, 64, 234–253.
ASRAR, G., FUCHS, M., KANEMASU, E.T., and HATFIELD, J. L., 1984, Estimation absorbed
photosynthetic radiation and leaf area index from spectral reflectance in wheat.
Agronomy Journal, 76, 300–306.
BANNARI, A., MORIN, D., and BONN, F., 1995, A review of vegetation indices. Remote
Sensing of Environment, 13, 95–120.
BARET, F., 1990, Factors and mechanisms governing canopy spectral reflectance: application
for agriculture. Technical Note (Avignon: INRA).
BARET, F., and FOURTY, T., 1997, Radiometric estimates of nitrogen status in leaves and
canopies. In Diagnosis of the Nitrogen Status in Crops, edited by G. Lemaire (Berlin:
Springer), pp. 201–227.
BARET, F., and GUYOT, G., 1991, Potentials and limits of vegetation indices for LAI and
APAR assessment. Remote Sensing of Environment, 35, 161–173.
BARET, F., GUYOT, G., and MAJOR, D. J., 1989, Crop biomass evaluation using radiometric
measurements. Photogrammetria, 43, 241–256.
5426 C. M. Di Bella et al.
CAMPBELL, G. S., 1990, Derivation of an angle density function for canopies with ellipsoidal
leaf angle distribution. Agricultural and Forest Meteorology, 49, 173–176.
CURRAN, P. J., 1983, Multispectral remote sensing for the estimation of green leaf area
index. Philosophical Transactions of the Royal Society of London A, 309, 257–270.
DAVIS, F. W., and ROBERTS, D., 2000, Stand structure in terrestrial ecosystems. In Methods
in Ecosystem Science, edited by O. E. Sala, R. B. Jackson, H. A. Mooney and R. W.
Howard (Berlin: Springer), part 1 (4), pp. 7–30.
DI BELLA, C. M., REBELLA, C. M., and PARUELO, J. M., 2000, Evapotranspiration estimates
using NOAA AVHRR imagery in the Pampa region of Argentina. International
Journal of Remote Sensing, 21, 791–797.
GAMON, J. A., FIELD, C. B., GOULDEN, M. L., GRIFFIN, K. L., HARTLEY, A. E., JOEL, G.,
PANUELAS, J., and VALENTINI, R., 1995, Relationships between NDVI, canopy
structure, and photosynthesis, in three Californian vegetation types. Ecological
Applications, 5, 28–41.
GERBERMAN, A. J., CUELLAR, J. A., and GAUSMAN, H. W., 1984, Relationship of sorghum
canopy variables to reflected infrared radiation for 2 wavelengths and 2 wavebands.
Photogrammetric Engineering and Remote Sensing, 50, 209–214.
GOWARD, S. N., WARING, R. H., DYE, D. G., and YANG, J., 1994, Ecological remote
sensing at OTTER: satellite macroscale observations. Ecological Applications, 4,
322–343.
GUERSCHMAN, J. P., PARUELO, J. M., DI BELLA, C. M., GIALLORENZI, M.C., and PACIN, F.,
2003, Land cover classification in Argentine Pampas using multitemporal Landsat
TM data. International Journal of Remote Sensing, 24, 3381–3402.
GUYOT, G., 1990, Optical properties of vegetation canopies. In Applications of Remote
Sensing in Agriculture, edited by M. Steven and J. Clark (London: Butterworths),
pp. 19–43.
JACQUEMOUD, S., and BARET, F., 1990, PROSPECT: a model of leaf optical properties
spectra. Remote Sensing of Environment, 34, 75–91.
MONTEITH, J. L., 1981, Climatic variation and the growth of crops. Quarterly Journal of the
Royal Meteorological Society, 107, 749–774.
OESTERHELD, M., DI BELLA, C. M., and KERDILES, H., 1998, Relation between NOAA-
AVHRR satellite data and stocking rate in grasslands. Ecological Applications, 8,
206–212.
PARUELO, J. M., EPSTEIN, H. E., LAUENROTH, W. K., and BURKE, I. C., 1997, ANPP
estimates from NDVI for the central grassland region of the US. Ecology, 78,
953–958.
PRINCE, S. D., 1991, A model of regional primary production for use with coarse resolution
satellite data. International Journal of Remote Sensing, 12, 1313–1330.
RIPPLE, W. J., 1985, Asymptotic reflectance characteristics of grass vegetation. Photogram-
metric Engineering and Remote Sensing, 43, 1915–1921.
ROUSE, J. W., HAAS, R. H., SCHELL, J. A., and DEERING, D. W., 1974, Monitoring
vegetation systems in the Great Plains with ERTS. Proceedings of the Third Earth
Resources Technology Satellite-1 Symposium (Greenbelt: NASA), SP-351,
pp. 301–317.
RUNNING, S. W., THORNTON, P. E., NEMANI, R., and GLASSY, J. M., 2000, Global
terrestrial gross and net primary productivity from the Earth observing system. In
Methods in Ecosystem Science, edited by O. E. Sala, R. B. Jackson, H. A. Mooney
and R. W. Howard (Berlin: Springer), part 1 (4), pp. 44–57.
SELLERS, P. J., 1985, Canopy reflectance, photosynthesis, and transpiration. International
Journal of Remote Sensing, 8, 1335–1372.
SELLERS, P. J., 1989, Theory and applications of optical remote sensing. In Vegetation—
Canopy spectral reflectance and biophysical processes, edited by G. Asrar (New York:
John Wiley and Sons), pp. 297–333.
SELLERS, P. J., BERRY, J. A., COLLATZ, G. J., FIELD, C. B., and HALL, F. G., 1992, Canopy
reflectance, photosynthesis, and transpiration. III. A reanalysis using improved leaf
models and a new canopy integration scheme. Remote Sensing of Environment, 42,
187–216.
SERRANO, L., GAMON, J. A., and PAÑUELAS, J., 2000, Estimation of canopy photosynthetic
and non-photosynthetic components from spectral transmittance. Ecology, 81,
3149–3162.
Senescent leaf effect on fAPAR 5427
TUCKER, C. J., 1977a, Resolution of grass canopy biomass classes. Photogrammetric
Engineering and Remote Sensing, 43, 1050–1067.
TUCKER, C. J., 1977b, Asymptotic nature of grass canopy spectral reflectance. Applied
Optics, 16, 1151–1157.
VAN LEEUWEN, W. J. D., and HUETE, A. R., 1996, Effects of standing litter on the
biophysical interpretation of plant canopies with spectral indices. Remote Sensing of
Environment, 55, 123–138.
VERHOEF, W., 1984, Light scattering by leaf layers with application to canopy reflectance
modeling: the SAIL model. Remote Sensing of Environment, 16, 125–141.
VERHOEF, W., 1985, Earth observation modelling based on layer scattering matrices. Remote
Sensing of Environment, 17, 165–178.
WEISS, M., TROUFLEAU, D., BARET, F., CHAUKI, H., PREVOT, L., OLIOSO, A., BRUGUIER,
N., and BRISSON, N., 2001, Coupling canopy functioning and radiative transfer
models for remote sensing data assimilation. Agricultural and Forest Meteorology,
108, 113–128.