MERIS and the red-edge position

JAG

MERIS

l

Volume

and the red-edge position

Jan G P W Cleversl, Steven M de Jongl, Gerrit F Epemal, F-reek van der Meer2, Wim H Bakker2, Andrew K Skidmore Elisabeth A Addink 1 Centre for Geo-Information (CGI), Wageningen-UR, [email protected]) 2 International

P.O. Box 47, NL 6700 AA Wageningen,

KEYWORDS:

red-edge

MERIS, imaging

position,

University of Michigan,

sensitivity

standard

Resolution

Imaging

of Envisat-1.

Spectrometer

The red-edge

acquiring position

foliar chlorophyll

over land with

images with which

role in ecosystem processes. The objectives

a

a 300 m spatial

(REP) is a promising

concentration,

430 East University, Ann Arbor, Ml, 48109-l

115, USA

reflectance. These indices correlate well with plant variables such as biomass, leaf area index (LAI) and the fraction of absorbed photosynthetically active radiation (fAPAR) [Baret & Guyot, 1991; Broge & Leblanc, 2000; Daughtry et al, 20001.

(MERIS) is a payload

MERIS will be operated

15 band setting

resolution. deriving

(e-mail:

information about ecosystem processes and they can be remotely sensed [Wessman et a/, 19881. Many studies have focused on the use of vegetation indices calculated as combinations of near-infrared (NIR) and red

analysis,

spectrometry

ABSTRACT

component

The Netherlands

and

Institute for Aerospace Survey and Earth Sciences, P.O. Box 6, 7500 AA Enschede, The Netherlands

3 School of Natural Resources and Environment,

The Medium

3 - Issue 4 - 2001

variable for

plays an important

of this paper are: (1) to

INTRODUCTION

The wavelength region of the red-NIR transition has been shown to have a high information content for vegetation spectra [Horler et a/, 19831. This region is generally referred to as the “red-edge”. Imaging spectrometers offer possibilities to estimate the chemical composition of plants from the reflected radiation of vegetation [Elvidge, 19901. However, at present, the remote sensing of foliar chemical concentrations, other than chlorophyll and water, has not been very successful. The presence of water in living leaf tissue masks some biochemical absorption features [Curran, 1989; Kokaly & Clark, 19991. Recently, Clevers [I 9991 showed that imaging spectrometry might provide additional information at the red-edge region. This was not covered by the information derived from a combination of a NIR and a visible broad spectral band, and red-edge may yield quantitative information on terrestrial ecosystems.

The Kyoto protocol to the United Nations Framework Convention on Climate Change contains quantified, legally binding commitments to limit or reduce greenhouse gas emissions to 1990 levels and allows carbon emissions to be balanced by carbon sinks represented by vegetation. As such, terrestrial ecosystems are an important link in the global carbon cycle and monitoring of their carbon fixation is an important topic. Remote sensing, used in combination with ecosystem models, offers one approach to estimating ecosystem functioning on a regional scale.

The red-edge is the region of the sharp rise in reflectance of green vegetation between 670 and 780 nm, which can be used for studying the chlorophyll concentration as a measure of plant condition [Horler et al, 19831. Both the position and the slope of the red-edge change under stress conditions, resulting in a shift of the slope towards shorter wavelengths [eg, Horler et a/, 1983; Wessman, 19941. The position of the red-edge is defined as the position of the main inflection point of the red-NIR slope. This is also denoted as the red-edge position (REP).

The amount of biomass and the foliar chemical composition are important characteristics, in that they provide

Within ESA’s Earth Observation programme, MERIS (Medium Resolution Imaging Spectrometer) will be one

study

which

whether ting

factors

effect

the

REP of vegetation,

(2) to study

this REP can be derived from the MERIS standard

and (3) to show what

data. Two different

REP represents

data sets were explored

using MERIS bands: (1) simulated (2) airborne the airborne ear method”,

reflectance

for simulating

data using reflectance

spectra of an agricultural

visible-infrared

imaging

spectrometer

area obtained

mining

the REP and the MERIS bands at 665, 708.75,

778.75

nm could be used for applying

estimation.

Results of the translation

were very promising

for applying

by

(AVIRIS). A “lin-

of the slope, was a robust method the “linear

the REP

models and

assuming a straight slope of the reflectance

around the midpoint

band set-

at the scale of MERIS

spectrum for deter-

753.75

method”

and

for REP

to the scale of MERIS data

MERIS at, for instance, the ecosys-

tem level.

313

MERIS

JAG Volume 3 -lssue4- 2001

and the red-edge position

l

of the main payload components of Envisat-1 to be launched in 2002 [ESA, 19971. MERIS is a 15 band, programmable imaging spectrometer, which allows for

20021 the band definitions, which were updated at the end of 2000, were used [Rast, personal communication 2001; see also www.envisat.com]. In this study we also paid some attention to the interpretation and application

changes in band position and bandwidths throughout its lifetime. It is designed to acquire data at variable bandwidth of 1.25 to 30 nm over the spectral range of 390 1040 nm [Rast & Bezy, 19901. Data will be acquired at 300 m spatial resolution over land, thus vegetation can be monitored at regional to global scales. MERIS will be operated with a standard band setting (as shown in Table 1). However, it has the capability of in-flight selection of bands for specific applications or experiments. Operational constraints, however, and frequency of band changes.

of a REP at the spatial scale of MERIS data.

METHODOLOGY POSITION

Since the REP is defined as the mum slope) of the red infrared mination requires a number of narrow bands in this region.

will limit the number

15 spectral bands of the standard band setting for MERIS[Rast et al, 1999; Rast, personal communication 20011

1

Band centre (nm)

Bandwidth

412.5

10

2

442.5

10

3

490

10

4

510

10 10

5

560

6

620

10

7

665

10

8

681.25

9

708.75

10

753.75

11

760.625

12

778.75

15

(nm)

7.5 7.5 3.75

13

865

20

885

10

15

900

10

inflection point (or maxislope, an accurate deterspectral measurements in Subsequently, the REP is

Shah et al, 19901. However, existing curve fitting techniques are complex. Dawson & Curran [I9981 presented a technique based upon a three-point Lagrangian interpolation technique for locating the REP in spectra that have been sampled coarsely. A problem with this method arises when the reflectance spectrum exhibits more than one maximum in its first derivative. Recently, Dawson [ZOO01 observed an artefact when using the Lagrangian technique on a reduced number of spectral bands. The REP was more sensitive to spectral variation than when calculated from continuous spectra. Clevers et al [ZOO21 also observed a discontinuity in the REP using this technique.

10

14

THE RED-EDGE

defined by the maximum first derivative of the reflectance spectrum in the region of the red-edge. Highorder curve fitting techniques are employed to fit a continuous function to the derivative spectrum [Horler et a/, 1983; Demetriades-Shah & Steven, 1988; Demetriades-

TABLE 1:

Band no.

FOR DETERMINING

For practical reasons, fitting a curve to a few measurements in the red-edge region (requiring an instrument with not so many spectral bands) often is applied for approximating the inflection point. First, a polynomial function may be fitted to the data [Clevers & Bilker, 1991; Broge & Leblanc, 20001. Secondly, a so-called inverted Gaussian fit to the red infrared slope may be applied [Bonham-Carter, 1988; Demarez & GastelluEtchegorry, 2000; Lucas et a/, 2000; Pate1 et a/, 20011. Finally, Guyot & Baret [I9881 applied a simple linear model to the red infrared slope. A comparison of the three methods yielded comparable results [Clevers & Buker, 1991; Bilker & Clevers, 19921. Since the method of Guyot 81 Baret [I9881 uses only four wavelength bands, this method is likely to be applicable to future sensors. Moreover, Clevers et al [ZOO21found for various data sets that the latter method is much more robust than procedures based on the first derivative spectrum because the first derivative spectrum did not have one unique maximum. Therefore it was chosen to use the method of Guyot & Baret for this study.

The MERIS spatial resolution of 300 m should be sufficient to monitor heterogeneous terrestrial surfaces at scales required for global change studies [ESA, 1995; Verstraete et al, 19991. Vegetation indices using red and NIR reflectance can be used for estimating the LAI and the fAPAR [Govaerts et a/, 19991. The red-edge feature may provide useful information on the physiological status of the vegetation under observation. Verstraete et a/ [I9991 noted that the spectral and spatial resolution of MERIS should be particularly appropriate to derive the REP. However, they also stated that much research is needed to design algorithms based on MERIS data. The first objective of this paper was to study which vegetation and external variables affect the REP. The second and main objective was to study whether the REP can be derived from the MERIS standard band setting, as this will be the setting that will be applied under normal operation. The derivation of REP from the MERIS standard setting was initially tackled using simulation Subsequently, data from the airborne models. visible-infrared imaging spectrometer (AVIRIS) were used. In contrast to previously reported results [Clevers et al,

The linear interpolation as described by Guyot & Baret [I9881 assumes that the reflectance curve at the red-

314

MERIS and the red-edge

position

edge can be simplified

JAG

to a straight line centred around

which

can be described in the following

of the reflectance at the inflection

point

(2) Calculation

(1)

of the REP:

REP = 700 + 40

l

((RREP - R7ooMR740

- R700))

- Issue 4 - 2001

Airborne reflectance spectroscopic data used in this study were derived from AVIRIS. The ER-2 aircraft of NASA, carrying AVIRIS, performed a successful flight over the Flevoland test site on 5 July 1991 (being the middle

(RREP): RREP = (R670 + R780)/2

3

The one-layer SAIL radiative transfer model [Verhoef, 19841 simulates canopy reflectance as a function of canopy variables (leaf reflectance and transmittance, LAI and leaf angle distribution), soil reflectance, ratio diffuse/direct irradiation and solar/view geometry (solar zenith angle, zenith view angle and sun-view azimuth angle). The SAIL model has been used in many studies and validated with various data sets [eg, Goel, 19891.

way: (1) Calculation

Volume

put of the PROSPECTmodel can be used as input into the SAIL model.

a midpoint between the reflectance in the NIR at about 780 nm and the reflectance minimum of the chlorophyll absorption feature at about 670 nm. These authors estimated the reflectance value at the inflection point and applied a linear interpolation procedure for the measurements at 700 and 740 nm estimating the wavelength corresponding to the estimated reflectance value at the inflection point. Here we refer to this method as the “linear method”,

l

of the growing season). AVIRIS acquires 224 contiguous spectral bands from 410 to 2450 nm using four spectrometers (with some overlapping bands). Both the spectral resolution and the spectral sampling interval are about 10 nm. The ground resolution is 20 m at an operating altitude of 20 km. The pre-processing which includes correction for dark current, vignetting and radiometric response was done by the Jet Propulsion Laboratory (JPL). Atmospheric correction was also performed by JPL with a version of the LOWTRAN 7 atmospheric model, yielding surface reflectance [van den Bosch & Alley, 19901.

(2)

Rs70, R7eo, R74e and R7ae are the reflectance values at 670, 700, 740 and 780 nm wavelength, respectively, and the constants 700 and 40 result from interpolation in the 700 - 740 nm interval (Figure 1).

The test site was located in Southern Flevoland in the Netherlands, an agricultural area with homogeneous soils reclaimed from the lake IJsselmeer in 1966. The test site comprised ten different agricultural farms, each 45 to 60 ha in size. Main crops were sugar beet, potato and winter wheat. Regular field measurements were collected. FIGURE 1:

Illustration of the “linear method” of Guyot & Baret

(1988). RESULTS AND DISCUSSION SENSITIVITY

DATA SETS Two different data sets were used. (0 Simulated data with a combination of the PROSPECT leaf reflectance model and the SAIL canopy reflectance model; (11)Airborne reflectance spectra of agricultural crops obtained from AVIRIS. Jacquemoud & Baret [1990] developed a leaf model that simulates leaf reflectance and leaf transmittance as a function of leaf properties: the PROSPECT model. PROSPECT allows computing the 400-2500 nm reflectance and transmittance spectra of different leaves using three input variables: the leaf mesophyll structure N, the pigment concentration and the water content. All three variables are independent of wavelength. The out-

ANALYSIS

One of the interesting features of the REP is that it seems to be independent of soil reflectance [Clevers, 19943. Moreover, the atmosphere seems to have only a minor influence on the REP. Both the soil background and the atmospheric influence hamper the use of solely one spectral band in the visible part of the spectrum for estimating leaf chlorophyll concentration. In this section a sensitivity analysis is described using theoretical leaf and canopy reflectance models and atmospheric influence on the REP is also considered. In this study, simulations were performed for a so-called standard crop under standard irradiation and viewing conditions, unless otherwise indicated, using a combined PROSPECT + SAIL model. The inputs for this standard crop are given in Table 2.

315

MEWS and the red-edge

JAG

position

TABLE2: input data for the coupled

Input values

chlorophyll

34.24 pg cm-25

concentration

leaf structure water

N

1.83205

leaf area index (LAI)

leaf angle distribution hot-spot

(LAD)

spherical

19881.

0

soil reflectance

20%

solar irradiation

direct

solar zenith

45O

zenith

cause false blue shifts that might be confused with real shifts because of vegetation stress [Bonham-Carter,

4.0

size

3 - Issue 4 - 2001

one). Systematic errors in the atmospheric calibration of some spectral bands may have effects on the REP. Moreover, bias in the reflectance measurements may

0.01375

content

Volume

This would be a second important characteristic of the REP (its independence of soil reflectance being the first

PROSPECT+SAIL model

Input variables

l

angle

view angle

O0

5 Exampleof a dicotyledonousplant as given by Jacquemoud& Baret [1990]

The effect of changing

one input variable at a time was

studied. All simulations were restricted to vertical viewing. Clevers & Verhoef [I9931 studied the influence of the observation geometry on red and NIR reflectances. They concluded that off-nadir viewing effects become significant at angles larger than about 7O (and most pronounced in the direction of the hot spot). So, in this study we did not investigate the influence of off-nadir viewing on the relationship between REP and chlorophyll concentration.

/ : 80

FIGURE 2: Influence of leaf chlorophyll concentration on simulated REP for several L.41 values of the standard crop.

PROSPECT-SAIL simulation.

Influence of chlorophyll concentration

The simulated REP as a function of the leaf chlorophyll concentration for various LAI values is illustrated in Figure 2. The REP shifted to longer wavelengths with increasing chlorophyll concentrations. Largest effects occurred at the lower chlorophyll concentrations. This confirms the assumption that the REP can be used as a measure for estimating the leaf chlorophyll concentration. Influence of LAI

Figure 2 already showed that the LAI had a clear effect on the position of the REP. The effect of varying LAI on simulated REP for various leaf chlorophyll concentrations is presented in Figure 3. There was a distinct shift of the REP to longer wavelength positions with increasing LAI. This shift was most pronounced at the lower LAI values.

0

1

2

3

4

5

6

7

8

UI

FIGURE 3: Influence of LAI on simulated REP for various leaf chlorophyll concentrations (CHL, pg cm-z) of the standard crop. PROSPECT-SAIL simulation.

Influence of soil background

One of the main advantages of working with the REP is its insensitivity to soil background. Figure 4 illustrates the effect of soil reflectance on simulated values of the REP for varying LAI. Under the condition of our simulations, the REP was not very sensitive to soil reflectance for a given LAI. The same was found as a function of chlorophyll concentration. A very small shift to longer wavelengths occurred with increasing soil reflectance. Atmospheric

influence

FIGURE 4: Influence of the soil reflectance on simulated REPfor various LAI values of the standard crop. PROSPECT-SAIL simulation.

Guyot et a/ [I9881 concluded from simulation studies that the REP was unaffected by atmospheric conditions.

316

JAG l Volume 3 - Issue 4 - 2001

MERIS and the red-edge position

The atmospheric

influence

in airborne

or space-borne

but with the earlier reported MERIS band at 705 nm instead of the band at 708.75 nm. For REP values below

measurements around the red-NIR slope is captured within the transformation from radiance values at the sensor (or digital numbers) to reflectance at ground level. Clevers & Buiten [I9911 described an atmospheric model based on the model of Richards [1986]. However, they extended the model of Richards by including diffuse and

710 nm, considerable deviations between the two methods occurred, which primarily was attributed to using an extrapolation, instead of an interpolation, on the MERIS bands. This extrapolation also caused quite a difference in the relationship using the 708.75 nm spectral band instead of the previously reported 705 nm band [Clevers et al, ZOOZ]. The relationship between the two methods showed a good linear relationship. Using all the data of Figure 6, the regression line yielded: REPMERIS = -807.0 + 2.124 l REP700_740 R* = 0.991.

direct upward fluxes in addition to the direct and diffuse downward fluxes. Moreover, they included multiple reflections between the Earth surface and the atmosphere. Results of the atmospheric correction procedure using this extended model in combination with parameterisation methods of lqbal [I9831 yielded good results for the Flevoland study area in the Netherlands. Similar results were obtained using the model of Verhoef

Using AVIRIS

[1985a, 1985b. 19981.

spectra

Using data set II, MERIS spectral bands were simulated with the AVIRIS bands at the Flevoland test site. It comprised real imaging spectrometer measurements over an agricultural area. In this study we neglected the fact that the radiometric sensitivity of AVIRIS is cruder than MERIS. Clevers et al [ZOO21 reported a dip in the radiance curve of agricultural crops at about 760 nm. This is a well-

The above atmospheric model was used for studying the influence of the atmosphere on the calculation of the REP. The visibility was used as an indicator for the state of the atmosphere. Figure 5 depicts the influence of the atmosphere on the REP for various chlorophyll concentrations. As expected, the influence of the atmosphere was minor.

.-.___--

Other variables The influence of other vegetation variables, like leaf structure, leaf size (hot-spot effect) and leaf angle distribution, and of other external variables, such as solar angle and diffuse/direct irradiation ratio, on the position of the REP appeared to be minor.

-_

-____

----

_

_

_

_

_

/ I

REPSIMULATION WITH MERIS MERIS

simulations

As seen in Table 1, MERIS also has spectral bands in the chlorophyll absorption feature at 665 nm and at the NIR plateau at about 778 nm. These are very close to the wavelengths used in the “linear method” as described by Guyot & Baret [I9881 and can be used for estimating the reflectance value at the inflection point. The next step in the “linear method” is to translate this reflectance value at the inflection point to the corresponding wavelength by a linear interpolation between the reflectances measured at two wavelength positions on the red-edge slope. From the MERIS standard band setting, the spectral bands centred at 708.75 and 753.75 nm are available. In this section the usefulness of these bands for estimating REP was tested. When the MERIS bands are simulated we call the linear method the “MERIS method”. Using data set I, Figure 6 illustrates the relationship between the REPbased on the “linear method” using the interpolation between 700 and 740 nm and the REP based on the “linear method” using the MERIS band setting and interpolating between 708.75 and 753.75 nm. The result was reported before by Clevers et a/ [ZOO21

317

FIGURE 5: Influence of atmospheric conditions (visibility) on simulated REPfor various leaf chlorophyll concentrations (CHL,

pg cm-*) of the standard crop. PROSPECT-SAIL simulation.

of the REPcalculated using the “linear method” interpolating between 700 and 740 nm and the “MERIS method”, thus interpolating between 708.75 and 753.75 nm. PROSPECT-SAIL simulation for a canopy with varying leaf chlorophyll concentration and various IAI values. The regression line REPMER~~ = -807.0 + 2.124 l REPTOO_740 is also plotted. FIGURE6: Comparison

JAG Volume 3 - Issue 4 - 2001

MERIS and the red-edge position

known absorption

l

feature caused by oxygen present in

the atmosphere. Since MERIS has a spectral band at about 753 nm, an atmospheric correction was performed. For the “linear method” the following AVIRIS spectral bands were employed: 667.4 - 697.3 - 736.7 775.3 nm. For simulating MERIS, the bands at 667 and 775 nm were used for estimating the reflectance at the inflection point and the bands at 707 nm, and the average of 746 and 756 nm were used for the linear interpolation. Figure 7 illustrates the relationship between the REP calculated from the interpolation between 697 and 737 nm and that calculated using the simulated MERIS bands for the main agricultural crops. When this relationship is compared with the regression line (for data set I), a good match was also observed.

bands. The resulting REP simulation is also depicted in Plate 1. This image shows that we could not recognise the individual agricultural fields anymore. However, when there tones in the an area with ly degraded

were fields with high REP values (bright image) close to each other, we recognised high REP (bright MERIS pixel) in the spatialimage. The same was true, for instance, for

areas with low REP values (the dark tones). Figure 8 illustrates the relationship

between the REP cal-

culated from the interpolation between 697 and 737 nm and that calculated using the simulated MERIS bands for a number of degraded pixels at the Flevoland test site. When we compared this relationship with the relationship found at 20 m spatial resolution (Figure 7), we saw

FIGURE7: Comparison of the REPcalculated using the “linear method” interpolating between 697 and 737 nm and the “MERIS method” for the main crops and bare soils at the Flevoland site, measured with the AVIRISsensor on 5 July 1991. The regression line from Figure 6 is also plotted. Results for both data sets show that a nearly linear relationship exists for the “linear method” using either bands at about 700 and 740 nm or the MERIS bands. However, since the relationship is not a I:1 relationship, we have to work with instrument-specific REP indices. For MERIS, the above-mentioned linear relationship may be used for translating REP values obtained with one method (using the MERIS bands) into values obtained with the other method (based on bands at 700 and 740 nm). Deviations occurred mainly at high REP values matching vegetation with a high chlorophyll concentration and with very high LAI values. To cope with this, a non-linear function might be fitted. Another possibility is to fit a second linear relationship for the upper part of the curve [Clevers et a/, 20021.

PLATE1 Top: Impression of an REPimage derived from AVIRIS data for the Flevoland test site, 5 July 1991. Dark is low REP; white is high REP. Bottom: Simulated REPimage for MERISwith 300 m spatial res-

Upscaling to the MERIS resolution

In the previous section the REP was calculated for the simulated MERIS bands using AVIRIS data at 20 m spatial resolution (cf. Figure 7). A REP image for the Flevoland test area is depicted in Plate 1. MERIS pixels of 300 m spatial resolution were simulated by taking the average reflectance of 15 x 15 pixels for all AVIRIS spectral

FIGURE8: Comparison of the REPcalculated using the “linear method” interpolating between 697 and 737 nm and the “MERIS method” for AVIRIS pixels degraded to 300 m at the Flevoland site. The regression line from Figure 6 is also plotted.

318

MERIS and the red-edge

JAC Volume 3 - Issue 4 - 2001

position

that the data coincided

l

very well and that the relation-

ACKNOWLEDGEMENTS

ships matched. The range in REPvalues at 300 m was less

We

because there were no points at the lower end in comparison to the range at 20 m. This was caused by the fact that no pure bare soil points were present in the degraded image.

Netherlands 3.1/AP-07).

partial

Remote

financial

Sensing

Board

support

by

the

(BCRS contract

REFERENCES Baret, F. & G. Guyot, 1991. Potentials and limits of vegetation indices for LAI and APAR assessment. Remote Sensing of Environment 35: 161-173.

From this it is concluded that the REP derived for the MERIS spectral bands at the MERIS spatial resolution of 300 m still provided useful information. If the objeicts at

Bonham-Carter, G.F.. 1988. Numerical procedures and computer program for fitting an inverted gaussian model to vegetation reflectance data. Computers and Geosciences 14: 339-356.

the Earth surface are large enough, we may obtain a significant range in REP values. At the Flevoland agricultural test site individual fields could not be recognised, but MERIS will provide informative scale (groups of fields).

acknowledge

Broge, N.H. & E. Leblanc, 2000. Comparing prediction power and stability of broadband and hyperspectral vegetation indices for

REP values at a regional

estimation of green leaf area index and canopy chlorophyll sity. Remote Sensing of Environment 76: 156-I 72.

den-

BSker, C. & J.G.P.W. Clevers, 1992. imaging spectroscopy for agricultural applications. Report LUW-LMK-199206. Department of Land surveying and Remote Sensing, Wageningen Agricultural University, Wageningen.

CONCLUSIONS The “linear method” [Guyot & Baret, 19881 assumes a straight slope of the reflectance spectrum around the midpoint between the reflectance at the NIR plateau and the reflectance minimum at the chlorophyll absorption feature in the red. This midpoint is then defined as the REP. This point may not coincide with the maximum of the first derivative, but it appears to be a very robust definition and it requires only a limited number of spectral bands. Thus, this method is very useful for practical applications, especially for space-borne narrow-band scanners, which do not have a continuous spectral coverage.

Clevers. J.G.P.W., 1994. Imaging spectrometry in agriculture - plant vitality and yield indicators. In: J. Hill & J. Megier (Eds), Imaging Spectrometry - A Tool for Environmental Observations. Kluwer Academic Publishers, Dordrecht, pp. 193-219. Clevers, J.G.P.W., 1999. The use of imaging spectrometry for agricultural applications. ISPRS Journal of Photogrammetry and Remote Sensing 54: 299-304. Clevers, J.G.P.W. & H.J. Buiten, 1991. Multitemporal analysis of optical satellite data. Report 91-02. BCRS, Delft, 64 pp. Clevers, J.G.P.W. & C. BOker, 1991. Feasibility of the red edge index for the detection of nitrogen deficiency. Proceedings, 5th “Physical Measurements and International Colloquium Signatures in Remote Sensing”, Courchevel, 14 - 18 January 1991, Paris: ESA, ESA Publication SP-319, pp. 165-168.

The REP may be considered as a particular “vegetation index” of interest ‘because of its. low sensitivity to disturbing factors such as atmospheric conditions or soil brightness and its high sensitivity to canopy characteristics such as chlorophyll concentration or LAI.

Clevers, J.G.P.W. & W. Verhoef, 1993. LAI estimation by means of the WDVI: a sensitivity analysis with a combined PROSPECT-SAIL model. Remote Sensing Reviews 7: 43-64.

Although the spectral bands of the MERIS standard band setting at the red-edge slope are not optimally located, they can be used for applying the “linear method” for REP estimation. The bands to be used are located at 708.75 and 753.75 nm. However, since the latter band is located very close to the oxygen absorption feature of the atmosphere, an atmospheric correction must be applied previous to calculating the REP using the MERIS bands. Previous studies [eg, Clevers, 19941 have shown that for bands located at about 700 and 740 nm, an atmospheric correction is not necessary since the REPcalculated in terms of reflectances equals the one calculated in terms of radiances.

Curran, P.J., 1989. Remote sensing of foliar Sensing of Environment 30: 271-278.

The

REP derived

MERIS spatial information. enough, these

the

resolution

we may obtain

m still

bands

at the

provided

useful

at the Earth surface specific

This makes

monitoring

MERIS spectral of 300

If the objects

objects.

ecosystem

for

information

MERIS a useful

Clevers, J.G.P.W., S.M. de Jong, G.F. Epema, F. van der Meer, W.H. Bakker, A.K. Skidmore & K.H. Scholte, 2002. Derivation of the red edge index using the MERIS standard band setting. International Journal of Remote Sensing 23 (in press). Remote

Daughtry, C.S.T., C.L. Walthall, M.S. Kim, E. Brown de Colstoun & J.E. McMurtrey Ill, 2000. Estimating corn leaf chlorophyll concentration from leaf and canopy reflectance. Remote Sensing of Environment 74: 229-239. Dawson, T.P., 2000. The potential for estimating chlorophyll content from a vegetation canopy using the Medium Resolution Imaging Spectrometer (MERIS). International Journal of Remote Sensing 21: 2043-2051. Dawson, T.P. & P.J. Curran, 1998. A new technique for interpolating the reflectance red edge position. International Journal of Remote Sensing 19: 2133-2 139. Demarez, V. & J.P. Gastellu-Etchegorry, 2000. A modeling approach for studying forest chlorophyll content. Remote Sensing of Environment 71: 226-238. Demetriades-Shah, T.H. & M.D. Steven, 1988. High spectral resolution indices for monitoring crop growth and chlorosis. Proceedings, 4th International Colloquium “Spectral Signatures of Objects in Remote Sensing”, Aussois, 18 - 22 January 1988, Paris: ESA, ESA Publication SP-287, pp. 299-302.

are large

concerning instrument

chemistry.

Demetriades-Shah, T.H., M.D. Steven & J.A. Clark, 1990. High resolution derivative spectra in remote sensing. Remote Sensing of Environment 33: 55-64.

for

activities.

319

MERIS and the red-edge

JAG

position

optical remote sensing Wageningen Agricultural

ESA, 1997. Envisat-1 mission & system summary. ESTEC-ESA, Noordwijk. Goel, N.S., 1989. Inversion of canopy reflectance models for estimation of biophysical parameters from reflectance data. In: G. Asrar (Ed.), Theory and Applications of Optical Remote Sensing. J. Wiley & Sons, New York, NY, pp. 205-251. Govaerts, Y.M., M.M. Verstraete, B. Pinty & N. Gobron, 1999. Designing optimal spectral indices: a feasibility and proof of concept study. International Journal of Remote Sensing 20: 1853-1873.

Guyot, G., F. Baret & D.J. Major, 1988. High spectral resolution: determination of spectral shifts between the red and near infrared. International Archives of Photogrammetry and Remote Sensing 27(11): 750-760. Horler, D.N.H., M. Dockray, & J. Barber, 1983. The red edge of plant

Iqbal, M., 1983. An Introduction Ontario, 390 pp.

of Remote

Sensing 4:

to Solar Radiation. Academic Press,

Jacquemoud, S. & F. Baret, 1990. PROSPECT: a model of leaf optical properties spectra. Remote Sensing of Environment 34: 75-91. Kokaly, RF. & R.N. Clark, 1999. Spectroscopic determination of leaf biochemistry using band-depth analysis of absorption features and stepwise multiple linear regression. Remote Sensing of Environment 67: 267-287. Lucas, N.S., P.J. Curran, SE. Plummer & F.M. Danson, 2000. Estimating the stem carbon production of a coniferous forest using an ecosystem simulation model driven by the remotely sensed red edge. International Journal of Remote Sensing 21: 619-631. Patel, N.K., C. Patnaik, S. Dutta, A.M. Shekh & A.J. Dave, 2001. Study of crop growth parameters using Airborne Imaging Spectrometer data. International Journal of Remote Sensing 22: 2401-2411.

Wessman, C.A., 1994. Estimating canopy biochemistry through imaging spectrometry. In: J. Hill & J. Megier (Eds), Imaging Spectrometry - A Tool for Environmental Observations. Kluwer Academic Publishers, Dordrecht, pp. 57-69.

RESUME Le spectrometre imageur de resolution moyenne (MERIS) est une composante de la charge utile d’Envisat-I. MERIS sera mis en fonction au-dessus des terres avec un montage standard de 15 canaux spectraux permettant d’acquerir des images de 300 m de resolution spatiale. La position limite du rouge (REP) est une variable prometteuse pour deriver une concentration de chlorophylle foliaire, qui joue un role important dans les procedes des ecosystemes. Les objectifs de cet article sont: (1) d’etudier quels facteurs affectent la REPde vegetation, (2) d’etudier si cette REP peut etre derivee du montage standard des canaux spectraux a partir de MERIS et (3) de montrer ce que REP represente a I’echelle des donnees MERIS. Deux differentes series de don&es ont et@ explorees pour simuler la REP en utilisant les canaux spectraux MERIS: (I) donnees simulees utilisant des modeles de reflectance et (2) des spectres de reflectance d’une zone agricole obtenue a I’aide d’un spectra-imageur aeroporte travaillant dans le visible infrarouge (AVIRIS). En supposant une pente reguliere du spectre de reflectance autour du milieu de la pente, la “methode lineaire”, s’est aver&e une methode robuste pour la determination de REP et les canaux spectraux MERIS a 665, 708.75, 753.75 et 778.75 nm pourraient @tre utilises pour appliquer la “methode lineaire” pour une estimation de REP. Les resultats de la translation a I’echelle des don&es MERIS ont ete tres prometteurs pour appliquer MERIS au niveau de I’ecosysteme, par exemple.

El espectr6metro de imdgenes de resolucibn mediana MERIS es un componente comercial de Envisat-1 En areas continentales, MERIS va a operar con un patr6n regular de 15 bandas, que toman imdgenes de 300 m de resoluci6n espacial. La posici6n de borde de la banda roja (REP) es una variable prometedora para derivar la concentraci6n foliar en clorofila, la cual juega un papel importante en 10s procesos de 10s ecosistemas. Los objetivos de este articulo son: (1) analizar 10s factores que afectan el REPde la vegetaci6n, (2) averiguar si se puede derivar este REPa partir de la combinaci6n estdndar de bandas de MERIS y (3) mostrar lo que el REP representa a la escala de 10s datos de MERIS. Se exploran dos diferentes conjuntos de datos para simular el REP a partir de las bandas de MERIS: (1) datos simulados mediante modelos de reflectancia y (2) espectros de reflectancia aeroportados de un area agricola, obtenidos por el espectr6metro aeroportado AVIRIS que toma imdgenes en el visible y en el infrarrojo. Un “metodo lineal”, el cual asume una pendiente recta del espectro de reflectancia de ambos lados del punto medio de la pendiente, result6 ser un metodo robust0 para determinar el REP. Se pudo usar las bandas de MERIS a 665, 708.75, 753.75 y 778.75 nm para aplicar el “metodo lineal” a la estimaci6n del REP. Los resultados obtenidos de la traducci6n a la escala de 10s datos de MERIS son muy prometedores para aplicar MERIS, por ejemplo, a nivel de ecosistema.

Rast, M., J.L. Bezy & S. Bruzzi, 1999. The ESA Medium Resolution Imaging Spectrometer MERIS - a review of the instrument and its mission. International Journal of Remote Sensing 20: 16811702. Image Analysis:

Verstraete, M.M., B. Pinty, & P.J. Curran, 1999. MERIS potential for land applications. International Journal of Remote Sensing 20: 1747-1756.

RESUMEN

Rast, M. & J.L. B&y, 1990. ESA’s Medium Resolution Imaging Spectrometer (MERIS): mission, system and applications. SPIE Journal 1298: 114-I 26.

Richards, J.A., 1986. Remote Sensing Digital Introduction. Springer-Verlag, Berlin.

of vegetation canopies. PhD thesis. University, Wageningen, 310 pp.

Wessman, C.A., J.D. Aber, D.L. Peterson & J.M. Melillo, 1988. Remote sensing of canopy chemistry and nitrogen cycling in temperate forest ecosystems. Nature 335: 154-I 56.

Guyot, G. & F. Baret, 1988. Utilisation de la haute resolution spectrale pour suivre I’etat des couverts vegetaux. Proceedings, 4th International Colloquium “Spectral Signatures of Objects in Remote Sensing”, Aussois, 18 - 22 January 1988, Paris: ESA, ESA Publication SP-287, pp. 279-286.

Journal

- Issue 4 - 2001

Verhoef, W., 1998. Theory of radiative transfer models applied in

ESA, 1995. MERIS: The medium resolution imaging spectrometer. Report SP-1184. ESA, Paris.

International

Volume 3

Les Arcs, 16 - 20 December 1985, Paris: ESA, ESA Publication SP-247, pp. 143-I 50.

Elvidge, C.D., 1990. Visible and near infrared reflectance characteristics of dry plant materials. International Journal of Remote Sensing 11: 1775-I 795.

leaf reflectance. 273-288.

l

An

van den Bosch, J. & R. Alley, 1990. Application of LOWTRAN 7 as an atmospheric correction to Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) data. In: R.O. Green (Ed.), Proceedings, 2nd Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop. JPL Publication 90-54. JPL, Pasadena, pp. 78-81. Verhoef, W., 1984. Light scattering by leaf layers with application to canopy reflectance modelling: the SAIL model. Remote Sensing of Environment 16: 125-141. Verhoef, W., 1985a. Earth observation modeling based on layer scattering matrices. Remote Sensing of Environment 17: 165178. Verhoef, W., 1985b. A scene radiation model based on four stream radiative transfer theory. Proceedings, 3rd International Colloquium “Spectral Signatures of Objects in Remote Sensing”,

320