PLEASE SCROLL DOWN FOR ARTICLE This article was downloaded by: [Delegido, Jesús] On: 9 November 2008 Access details: Access Details: [subscription number 905243299] Publisher Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK International Journal of Remote Sensing Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713722504 Retrieval of chlorophyll content and LAI of crops using hyperspectral techniques: application to PROBA/CHRIS data J. Delegido a ; G. Fernández a ; S. Gandía a ; J. Moreno a a Department of Earth Physics and Thermodynamics, Faculty of Physics, University of Valencia, Valencia, Spain Online Publication Date: 01 December 2008 To cite this Article Delegido, J., Fernández, G., Gandía, S. and Moreno, J.(2008)'Retrieval of chlorophyll content and LAI of crops using hyperspectral techniques: application to PROBA/CHRIS data',International Journal of Remote Sensing,29:24,7107 — 7127 To link to this Article: DOI: 10.1080/01431160802238401 URL: http://dx.doi.org/10.1080/01431160802238401 Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.
22
Embed
Retrieval of chlorophyll content and LAI of crops using hyperspectral techniques: application to PROBA/CHRIS data
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
PLEASE SCROLL DOWN FOR ARTICLE
This article was downloaded by: [Delegido, Jesús]On: 9 November 2008Access details: Access Details: [subscription number 905243299]Publisher Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK
International Journal of Remote SensingPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713722504
Retrieval of chlorophyll content and LAI of crops using hyperspectraltechniques: application to PROBA/CHRIS dataJ. Delegido a; G. Fernández a; S. Gandía a; J. Moreno a
a Department of Earth Physics and Thermodynamics, Faculty of Physics, University of Valencia, Valencia,Spain
Online Publication Date: 01 December 2008
To cite this Article Delegido, J., Fernández, G., Gandía, S. and Moreno, J.(2008)'Retrieval of chlorophyll content and LAI of cropsusing hyperspectral techniques: application to PROBA/CHRIS data',International Journal of Remote Sensing,29:24,7107 — 7127
To link to this Article: DOI: 10.1080/01431160802238401
URL: http://dx.doi.org/10.1080/01431160802238401
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.
The study of different indices (Richardson et al. 2002, Haboudane et al. 2004,
Jimenez-Munoz et al. 2005) provides information on the product of chlorophyll
content and Fractional Vegetation Cover (FVC) and on the chlorophyll product for
LAI for FVC (Gandia et al. 2004) but these indices are not able to decouple factors
for the isolated study of chlorophyll or LAI. In fact, the effect of chlorophyll content
variations on the vegetation index that calculates LAI remains an unsolved problem
(Haboudane et al. 2004). A method for studying these variables separately is
proposed in the present work, based on third-degree polynomial functions between
500 nm and 750 nm, in the case of LAI, and calculating the areas under the
reflectance curves in the interval from 600 nm to 700 nm to retrieve chlorophyll.
2. Methods
The main objective of this paper is to analyse a methodology to separate retrievals of
LAI and leaf chlorophyll by exploring the multi-angular and enhanced spectral
capabilities of PROBA/CHRIS data. In order to retrieve leaf chlorophyll we need to
look at specific chlorophyll absorption features. However, we also need to have
spectral signature not contaminated by chlorophyll absorption to determine the
amount of absorber by means of some kind of ‘differential absorption’ algorithm.
The problem in this case is that such reference reflectance, which is free from leaf
chlorophyll absorption effects, is in fact contaminated by LAI effects. That is, it is
not a true absolute reference but a variable reference that is also a function of LAI.
Thus, alternative techniques are needed.
2.1. LAI
In order to derive LAI from spectral reflectance curves, we need to consider a
spectral range with variable spectral absorption by some pigment in the leaf or by
the water contained in the leaf. LAI comes from the ‘differential absorption’ as a
function of the also varying scattering. In fact, we need at least one absorption
feature to retrieve LAI, but we also need a strong contrast in the absorption levels
across the considered spectral range to increase the accuracy in LAI ditermination.
For the retrieval of LAI, we have analysed the reflectance curves of different
crops, particularly in the spectral range between 500 nm and 750 nm—this being the
region where plant physiological characteristics have a more direct impact on the
shape of the spectral reflectance curve. Although different functions have been
tested (Vuolo et al. 2006), including Gaussian (Maire et al. 2004) potentials,
logarithmic functions and others, in this particular spectral region pixel reflectance
as a wavelength function can be better adjusted to third-degree polynomials. This
function has been selected as the simplest option that offers the best adjustments,
because it is the function with the fewest number of coefficients requiring adjustment
and yields higher correlation coefficients. The four polynomial coefficients provide
information on LAI. We propose a potential relationship between these coefficients
and LAI. The results obtained provide information on the spatial variability of LAI
by remote sensing techniques when applied to full images.
2.2. Chlorophyll content
In order to retrieve leaf chlorophyll content we need as a minimum two spectral
bands, one fully free from chlorophyll absorption and one quite affected by such
Chlorophyll content retrieval and LAI of crops 7109
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
absorption. However, because the absorption varies with the amount of absorber in
a nonlinear way, several absorption bands are preferable for a better compensation
of such nonlinear effects. On the other hand, measuring only the depth of the
absorption is not enough in most cases, while other indicators such as spectral slopes
or the area under the reference continuum tend to provide most stable estimates.
In the chlorophyll absorption spectrum (figure 1) we observe two absorption
peaks in the blue and red regions. Since in the blue region the contribution of
atmospheric aerosols is important (Niang et al. 2006) and introduces distortion due
to the coupling of several pigment absorptions, we have focused the derivation of
chlorophyll content information on the study of the area under the spectral
reflectance curve between 600 nm and 700 nm (Miller et al. 1990, Curran et al. 1991,
1997, Richardson et al. 2002). The analysis was centred on this wavelength interval,
which is large enough to provide feasible results, though without including spectral
regions where other factors and absorption features can exert an influence.
The greater the chlorophyll content, the greater the absorption observed in that
spectral region, and therefore reflectance will be smaller. Figure 2 represents the
typical vegetation reflectance curve without chlorophyll absorption and the different
curves as a function of leaf chlorophyll content. These curves have been represented
starting from the spectra obtained by PROBA/CHRIS, selecting crops with different
chlorophyll contents from our experimental data. A strong negative correlation is
evident, as expected, between chlorophyll content and the area under the curve. This
relationship, which can be calculated experimentally (based on a highly
representative number of data with broad variability), provides a method to
evaluate plant chlorophyll content using remote sensing techniques. A similar
procedure originating from the Chlorophyll Absorption Integral (CAI) concept has
been used by Oppelt and Mauser (2004) to monitor wheat chlorophyll during the
vegetation period.
Figure 1. Leaf pigment absorption spectrum, showing the optimum spectral range forchlorophyll retrievals where chlorophyll is the sole dominant absorber (600–700 nm).
7110 J. Delegido et al.
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
We propose the calculation of the area under the spectral reflectance curve,
defined as
S~
ð700
600
r dl ð1Þ
where r is the spectral reflectance and l the wavelength (nm). Based on statistical
analysis, it can be demonstrated that the relationship between chlorophyll and this
area is linear. Equalling this equation to zero, we would obtain the value of the areaunder the curve for the case of a crop with chlorophyll content zero. Therefore, it
can be introduced as a constant, area0, yielding the following expression:
Ch~k S0{Sð Þ ð2Þ
where Ch is chlorophyll content and k is an experimental constant. Therefore, we
can use this approach for monitoring crop chlorophyll content by remote sensingtechniques.
3. Materials
3.1. In situ measurements
The data used in this study were obtained during the SPARC-2003 (Spectra Barrax
Campaign) and SPARC-2004 campaigns in Barrax, La Mancha, Spain (coordinates
30u 39 N, 2u 69 W, 700 m altitude). The selection of the Barrax site was justified by
extensive experience in using this area as a test site for many previous satellite and
airborne experiments. The test area has a rectangular form and an extent of 5 km 610 km, and is characterized by a flat morphology and large, uniform land-use units.
Differences in elevation range from 5 m to 10 m only. The region consists of
Figure 2. Typical crop reflectivity (r) curves with different chlorophyll contents in the 600–700 nm interval. l is wavelength.
Chlorophyll content retrieval and LAI of crops 7111
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
approximately 65% dry land and 35% irrigated land. The climatic conditions show
Mediterranean features: important precipitation in spring and autumn, and
minimum rain in summer. The annual rainfall average is about 400 mm.
Furthermore, the region has typical continental climate, with thermal oscillations
during all seasons. La Mancha represents one of the driest regions in Europe
(Moreno et al. 2004).
The 2003 campaign was carried out from 12 to 14 July, and included
measurements of different physical and physiological variables of a large number
of crops. LAI was measured with a digital analyser (LI-COR LAI-2000, Lincoln,
Nebraska, USA), which works by comparing the intensity of diffuse incident
illumination measured at the bottom of the canopy with that arriving at the top (LI-
COR technical report) (Welles and Norman 1991). Each LAI value used in the
present study was obtained as a statistical mean of several measures with variable
standard errors between 5% and 10% (Fernandez et al. 2005).
The chlorophyll a + b measurements were made with a CCM-200 Chlorophyll
Content Meter specifically calibrated during the campaign by means of laboratory
analyses of about 100 samples based on spectroscopic and High Performance Liquid
Chromatography (HPLC) methods (Gandia et al. 2004). The three determinations
of leaf chlorophyll concentration (by CCM/Minolta devices, by direct chemical
spectroscopic analysis and by HPLC methods) were extensively analysed and
compared, yielding a consistent dataset of well established chlorophyll values with
identified accuracy and established measurements errors and intrinsic variability for
each elementary sampling unit (ESU).
For calibration of the CCM-200 instrument, during the SPARC-2003 campaign
50 samples were obtained of leaves of alfalfa, corn, beet, wheat, garlic, onion and
potato. Chlorophyll content was analysed following the methodology described by
Inskeep and Bloom (1984), based on the determination of the extinction coefficients
of chlorophyll a (Cha), chlorophyll b (Chb) and total chlorophyll (Chab) dissolved in
N,N-dimethylforamide (DMF).
The absorptivity (A) for 647 nm and 664.5 nm (A647 and A664.5, respectively)
wavelengths was measured for each sample by means of a CARY-UV-Visible
spectrophotometer, allowing calculation of the chlorophyll contents (mg l21) by
means of the following equations:
Cha~12:70 A664:5{2:79 A647 ð3Þ
Chb~20:70 A647{4:62 A664:5 ð4Þ
Chab~17:90 A647z8:08 A664:5 ð5Þ
Finally, and after testing different calibration functions, the best data fit for
chlorophyll versus the measurements with the CCM-200 instrument (in DC, digital
counts) was found to be a logarithmic function of the following kind (in mg m22):
Chab~m1zm2|log DCð Þ ð6Þ
where m1 and m2 are calibration constants presenting values (with the statistical
errors of the fit) shown to be (in mg m22): m15(120¡30) and m25(345¡19).
7112 J. Delegido et al.
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
The error associated with the total chlorophyll content obtained on applying this
calibration function varied between 5% for the highest values with the CCM-200
and 10% for the lowest values.
A total of 113 ESUs were measured for different crops. In the 2004 campaign,
carried out on 15 and 16 July, a total of 140 ESUs of LAI and 18 chlorophyll
measurements were collected among the different crops. The methodology applied
to obtain the in situ chlorophyll data consisted of measuring about 50 samples with
the CCM-200 meter, to characterize each ESU previously selected for the
characterization of LAI. The analysis of the chlorophyll values obtained in the
SPARC 2003 campaign shows good agreement with the values obtained inthe SPARC 2004 campaign (Gandia et al. 2005). Table 1 shows the plot number of
each of the crops in the two campaigns. Globally for the two years we have a total of
nine crops, with field-measured values of LAI that vary between 0.4 and 6.3, and
chlorophyll contents of between 20 mg and 550 mg m22. Further details on the
measurements and their statistical study can be found in Fernandez et al. (2005).
Figure 3 shows the LAI and chlorophyll values measured in the two campaigns.
The most important information reported in figure 3 is that we have a good and
even distribution of chlorophyll and LAI measured data for the whole interval.
Also, it is observed that for some crops (for example wheat and vineyard), there is
little dispersion of measured data.
Figure 4 shows an image of the study area with the 2004 campaign crops. The
2003 campaign study area was the same, though some crops changed, because the
plot number of each crop was smaller, and there were also some changes in land use.
For the present work we selected around six ESUs for each crop (less than six for
wheat and vineyard, and more than six for the rest of the species) with the aim of
better encompassing the whole interval of measurements, for both LAI and
chlorophyll. A total of 54 ESUs were selected, taken randomly from the available
database, and recruiting half in each year in the case of those crops that have beenmeasured both years. The ESUs have been chosen trying to recruit all the different
plots randomly, far from the borders and with available measurements on all
relevant biophysical variables. Likewise, they have been selected to cover the entire
possible interval of LAI and chlorophyll values.
3.2. Reflectance measurements
The ESA’s Project for On Board Autonomy (PROBA) is intended to demonstrate a
range of innovations in the design, construction and operation of small satellites.
Table 1. Plot (field) number of each crop in the 2003 and 2004 Spectra Barrax Campaigns(SPARC).
Chlorophyll content retrieval and LAI of crops 7113
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
Figure 3. Leaf area index (LAI) and chlorophyll measured values in the 2003 and 2004Spectra Barrax Campaigns (SPARC).
Figure 4. Land use map for selected crops in the 2004 Spectra Barrax Campaign (SPARC).The points marked with ( + ) indicate points where LAI was measured. Chlorophyll was alsomeasured at some of those points where LAI was characterized. The grid reflects the UTM-projection coordinates, where north is towards the top of the image.
7114 J. Delegido et al.
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
The Compact High Resolution Imaging Spectrometer (CHRIS) sensor on the
PROBA platform constitutes a low-cost new generation of satellite remote sensing
systems. CHRIS provides high spatial resolution hyperspectral/multiangular data,
acquiring five consecutive images from five different view zenith angles in one single
satellite overpass (0u, + 36u, 236u, + 55u and 255u). Also, CHRIS measures over the
visible/near-infrared spectra from 400 nm to 1050 nm. It can operate in different
modes, thus compromising the number of spectral bands and the spatial resolution
because of storage reasons. CHRIS Mode 1 (62 bands, full spectral information) has
a spatial resolution of 34 m at nadir. The spectral resolution provides a bandwidth of
1.25 nm at 415 nm, increasing to 11.25 nm at 1050 nm. This multiple-view-angle
imaging capability, in conjunction with high spectral and spatial resolution,
provides new opportunities for scientific investigation of the Earth’s surface and
atmosphere (Barnsley et al. 2004).
For this study we have used PROBA/CHRIS data in Mode 1 for the four
mentioned campaign days, where in situ measurements of surface properties were
measured simultaneously to satellite overpass. The images were geometrically
corrected (Alonso and Moreno 2005), followed by atmospheric correction according
to the method proposed by Guanter et al. (2005a). The method simultaneously
derives a set of calibration coefficients and an estimation of water vapour content
and aerosol optical thickness from the data themselves. The atmospheric correction
of the data was validated by direct comparison of CHRIS/PROBA-derived
reflectance retrievals with simultaneous ground-based measurements acquired
during the campaign, as described in Guanter et al. (2005a, b).
Based on these images, the 5465 pixel spectra corresponding to 54 selected ESUs
and five observation angles have been obtained. Since we have two images per year,
we have randomly selected images corresponding to one or the other day for each
crop, so that the total number of spectra is distributed evenly among the four days.
Three bare soil pixels have also been included in the study, assigning zero values of
LAI and chlorophyll to those points.
4. Results and discussion
4.1. Leaf area index
Figure 5 illustrates typical reflectance values of some crops and bare soil for
wavelengths in the range between 511.53 nm (CHRIS band 9) and 751.36 nm
(CHRIS band 37). All the reflectance spectra, in the given wavelength range, have
been adjusted to third degree polynomial functions in the following way:
Rr~AzB lzC l2zD l3 ð7Þ
where Rr is reflectance r 610 000, l is wavelength (in nm), and A, B, C and D are
the coefficients obtained by the minimum squares method. Use has been made of r6 10 000, since this is the presentation offered by the PROBA/CHRIS images,
with the purpose of using integers rather than decimals. In all cases, and for all
studied crops, correlation coefficients greater than 0.97 have been obtained.
Figure 6 shows the curves obtained for typical values of some crops. Although as
can be seen the curves do not exhibit a detailed fit, for the present study we are
interested in establishing functions that are representative for the entire interval,
with good correlation coefficients, even if exact point-by-point reproduction is
lacking.
Chlorophyll content retrieval and LAI of crops 7115
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
Once coefficients A, B, C and D were all empirically calculated, they were
processed with a statistical software package for additional mathematical study—a
very strong correlation being observed among the four coefficient values. As an
example, figure 7 shows the values of coefficient B as a function of the values of
coefficient A. The relationship of A to C and A to D is similar.
On establishing a linear fit (by means of a least squares method) of B, C and D as
a function of A, we obtain:
B~{15:86{0:0050804|A R~0:99996ð Þ ð8Þ
C~0:05548z8:5512|10{6|A R~0:99985ð Þ ð9Þ
D~{3:9622|10{5{4:7547|10{9|A R~0:99969ð Þ ð10Þ
where R is the correlation coefficient. Equations (8), (9) and (10) are practically
simple proportional relationships (see figure 7), since the independent term in
practice is negligible in all three equations, because A varies from 0 to 266105.
From figure 7 and from the previous correlation coefficients, we can conclude that
the relationship among the four coefficients is linear, and that the spectrum of a
certain pixel image is perfectly defined by a single coefficient. This is explained
because as can be seen in figure 6, most of the curves have the same maximum,
minimum and inflection point values, as can also be calculated based on the
previous relationships: taking first derivatives and making equation (7) equal to
zero, we see that on starting from relationships (8), (9) and (10), the maximum is
543 nm and the minimum 656 nm. The maximum and minimum values of the
polynomial function do not correspond exactly to the maximum and minimum
reflectance curves, which can vary with chlorophyll content. In fact, it is known that
the red edge in the spectral reflectance curve moves towards infrared when the
Figure 5. Typical spectrum of bare soil and of the most representative crops in the studyarea. r is reflectivity and l is wavelength.
7116 J. Delegido et al.
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
chlorophyll content increases (Curran et al. 1995, Gitelson et al. 2005). This implies
an overall displacement of the spectral reflectance curve towards infrared, as a
consequence of the chlorophyll increase when keeping the LAI value as constant.
Differentiating (7) twice and making the result equal to zero, the inflection point
of the spectral reflectance curve is given by 2C/(3D)5599.5 nm.
Figure 6. Fitting of the more representative spectra to a third degree polynomial curve. r isreflectivity and l is wavelength.
Figure 7. Relationship among polynomial coefficients B and A.
Chlorophyll content retrieval and LAI of crops 7117
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
From all the above, we can conclude that all equation (7) coefficients are equally
representative of crop spectra in the considered interval. Therefore, we will now try
to relate any of these coefficients to LAI values. Figure 8 shows LAI values as a
function of B coefficient. This coefficient has been chosen because B has direct
dependence upon LAI (a similar situation being observed for D, and inversely for A
and B). Figure 8 shows that the mathematical relationship can be approached to a
potential function as:
LAI~9:9584|10{6|B1:6806 R~0:84595ð Þ ð11Þ
Considering that the data number is high (N5209), and that a correlation
coefficient of 0.85 is obtained, we can conclude that the relationship equation (11)
provides a method allowing us to obtain LAI based on hyperspectral images (with
sufficient spectral resolution), provided we have sufficient band numbers in the
interval considered. The method provides LAI estimates with a mean error one of
,1 (e(LAI),1). Taking into account that in the adjustment procedure chlorophyll
values have been introduced distributed in the whole interval (from 0 for bare soil to
555 mg m22), it can be affirmed that the method allows the retrieval of LAI
independently of chlorophyll content. In fitting equation (11), we have eliminated
the three points corresponding to wheat, since they deviated considerably from the
point cluster—probably because of the condition of wheat at that time of year
Figure 8. Leaf area index (LAI) as a function of coefficient B, classified according to croptype.
7118 J. Delegido et al.
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
(harvesting), when chlorophyll content is practically zero (senescence), while LAI is
high (see figure 3).
The retrievals are also almost independent of the observation angle, since
significant differences have not been observed among the five angular images of
each pixel. To demonstrate this, figure 9 shows the graphic representation of
figure 8, though with classification of the points by observation angle.
Accordingly, figure 9 shows that the different observation angles are distributed
uniformly over the entire graph area, thus indicating that LAI is not dependent
upon this variable. To justify this affirmation, equation (11) was subjected to
minimum squares fitting, though previously separating the points according to the
angle of observation. Table 2 reports the results obtained.
As it can be seen in table 2, no correlation exists between the observation angle
and the coefficients of the equations obtained. Furthermore, the correlation
coefficients are similar to those of equation (11), with slight improvement for the
case of 0u. This suggests that we could improve correlation for equation (11) by
considering only the image obtained at 0u, though the method would be limited to
this case. However, the adopted approach allows the calculation of LAI under any
angle of observation—thus improving the method, since in some cases not all vision
angles are available.
To summarize, the proposed method would allow implementation of a PROBA/
CHRIS product to automatically obtain a LAI image of a given crop zone. This
would require imaging in CHRIS Mode 1, under any angle of observation—though
preferably close to nadir—with software transformation into an image of coefficient
Figure 9. Leaf area index (LAI) as a function of coefficient B, classified according to viewangle.
Chlorophyll content retrieval and LAI of crops 7119
Downloaded By: [Delegido, Jesús] At: 10:13 9 November 2008
B, obtained on fitting the bands from 9 to 37 to a third-degree polynomial function.
Based on this image, and applying equation (11), a LAI map would be obtained.
4.2. Chlorophyll content
As we have pointed out above, the increase in chlorophyll content in crop leaves
must cause a decrease in the area under the spectral reflectance curve, in the interval
between 600 nm and 700 nm. With the objective of studying this phenomenon, the
areas under the curves have been calculated by numerical integration in this interval,
which corresponds to bands 18 to 28, both inclusive, of CHRIS sensor Mode 1,
corresponding to 604.97 and 694.19 nm, respectively:
S~
ð700
600
Rr dl ð12Þ
where S is the area between l 5 600 nm and l 5 700 nm and Rr is the reflectance (r6 10000) after atmospheric correction.
Figure 10 shows the chlorophyll content as a function of this area S, for all crops
in the zone, including wheat.
These values can be fitted to a linear equation:
Ch~531:29{0:00154|S R~0:81134ð Þ ð13Þ
where Ch is the leaf chlorophyll a + b content (mg m22). This fit, which has been
carried out with N5231 points, provides an acceptable correlation coefficient (0.81),
allowing us to conclude that chlorophyll content can effectively be estimated by
means of this approach, with a precision of ¡ 100 mg m22. Leaf chlorophyll content
is difficult to measure using remote sensing techniques, though the proposed method
seems to be simple and to offer good results. The method allows calculation of
chlorophyll independently of LAI.
In order to understand the physical significance of such a relationship, it is
convenient to write equation (13) in the form:
Ch~0:00154 S0{Sð Þ ð14Þ
where S0 (5344993.5) represents the area in the case of zero chlorophyll content.
Such a baseline area can be estimated for a given series of geographical/climatic
conditions by means of surface reflectance modelling or prior knowledge of the
study area conditions, while the slope 0.000154 is derived from the actual
relationship between area and chlorophyll content.
Table 2. Adjustment of LAI as a function of B, based on the exponential function accordingto the different observation angles