Page 1
Validation of different methods of
chlorophyll content estimation by remote
sensing data from AGRISAR campaign
S. Gandía, J. Delegido, J. Moreno, G. Fernández
Department of Earth Physics and Thermodynamics Laboratory for Earth Observation
Faculty of Physics University of Valencia
AGRISAR & EAGLE Final Workshop ESA/ESTEC October, 15-16 2007
Page 2
Definition and validation of methodologies to derive Level 2
products from Sentinel-2 data: leaf chlorophyll content and leaf
area index as focus products
Sensitivity analysis of the defined SENTINEL-2 bands to
different chlorophyll retrieval methods, validated against a
defined reference (full spectrum of CASI)
Page 3
• Regions of interest (ROIs) of 3 pixels area (attending crop
homogeneity) have been defined around GPS coordinates of each
elementary sampling unit (ESU) with chlorophyll data
• Defined ROIs have been used with CASI images and SENTINEL-2
simulated images and mean spectra with standard deviation error have
been obtained
• Some chlorophyll indices have been calculated for both CASI and
SENTINEL-2 sensors and capability of SENTINEL-2 indices to
reproduce CASI indices has been tested
• Correlations between CASI and SENTINEL-2 chlorophyll indices and
ground truth chlorophyll and LAI data have been analyzed
Page 4
++AHSAHS--INTAINTA
CASICASI--15001500
Simulation of Sentinel-2 products to evaluate their spectral bands
capabilities
SentinelSentinel--2 Simulation Image2 Simulation Image
airborne imaging spectrometers
SENTINELSENTINEL--2 2 Bands Bands
ConfigurationConfiguration
available spectral available spectral information frominformation from Exploring the possibilities
offered on retrieving vegetation properties
ChlorophyllChlorophyll
We have focused our analysis on
Page 5
288 spectral bands288 spectral bands450 450 -- 1000 nm1000 nm
CASICASIAHSAHS--INTAINTA
43 spectral bands 43 spectral bands 1000 1000 -- 2500 nm2500 nm
++
The available spectral information from
SentinelSentinel--2 simulation2 simulation
1313 spectral spectral bandsbands443 443 -- 2190 2190 nmnm
Band Unit B1 B2 B3 B4 B5 B6 B7 B8 B8a B9 B10 B11 B12
1610 2190
18090
1375
20
940
20
λλ centrecentre nm 443 490 560 665 705 740 775 842 865
Width band nm 20 65 35 30 15 15 20 115 20
Atmospheric Correction BandsAtmospheric Correction Bands
Page 6
230
140
460102
450
222
UV ESUs July-06
ZALF ESUs July-06
230
222
140
460 102
450
102SB
460SB
140
Rape
230
222Corn
Wheat
Barley450
23 ROI to obtain CASI and S2S
spectra
23 Chlorophyll ESUs at July 06
Page 7
Spectra from CASI + S2S July 5 2006 CASI CASI
S2S S2S
Each region of interest (ROI) has an extension of three pixels centred on the GPS coordinates of the chlorophyll ESU
Page 8
From the image taken on July, 5 2006 (CASI image date), several
chlorophyll indices (taken from the literature) have been calculated
with CASI and S2S bands for the chlorophyll ESUs
Correlations between CASI and S2S indices have been analyzed in
order to check if the defined SENTINEL-2 bands provide us the same
information as CASI in terms of chlorophyll retrievals
The CASI and S2S chlorophyll indices have been correlated with
chlorophyll content and with the product (LAI*Chl) to analyze the
capability of both sensors to estimate crop chlorophyll
Chlorophyll indices from other campaigns has been compared with
AGRISAR campaigns
Page 9
From the numerous chlorophyll indices that appear in the literature, the
following have been chosen :
λ
λλ λ∫
2
1I = R( ) d
⇒ Integral Index:
Area intersected between vegetation
spectrum and soil spectrum is an
indicator of chlorophyll content of the
vegetation
Also, area under vegetation reflectance
curve is related to the amount of
chlorophyll in the leaves (if soil
spectrum can be considered constant)
Page 10
⇒ Band Quotient Index (Gitelson & Merzlyak, 1997)
⇒ Modified Chlorophyll Absorption in Reflectance Index (MCARI)(Daughtry et al, 2000)
750
700
RGMI =R
B6 GMIS =2SB5
749.1
699.8
CAR
GMIS =R
I
( ) ( ) ⎛ ⎞⎡ ⎤ ⎜ ⎟⎣ ⎦
⎝ ⎠700
700 670 700 550670
RMCARI = R -R -0.2* R -R *R
( ) ( ) ⎛ ⎞⎡ ⎤ ⎜ ⎟⎣ ⎦ ⎝ ⎠
B5MCARI = B5-B4 -0.2* B5-B3 *B4
S2S
( ) ( )699.8 669.2 699.8 549.4669.2
⎛ ⎞⎡ ⎤ ⎜ ⎟⎣ ⎦
⎝ ⎠700RMCC ARI = R -R -0.2* R -R *
RASI
Page 11
⇒ Transformed CARI (TCARI) (Haboudane et al, 2002)
( ) ( )⎡ ⎤⎛ ⎞⎢ ⎥⎜ ⎟
⎝ ⎠⎣ ⎦
700700 670 700 550
670
RTCARI =3* R -R -0.2* R -R *R
( ) ( )⎡ ⎤⎛ ⎞⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦
B5TCARI =3* B5-B4 -0.2* B5-B3 *B4
S2S
( ) ( )⎡ ⎤⎛ ⎞⎢ ⎥⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦
699.8699.8 669.2 699.8 549.4
669.2
RTCARI =3* R -R -0.2* R -R *
RCASI
⇒ MERIS Terrestrial Chlorophyll Index (Dash & Curran, 2005)
( )( )
753.75 708.75
708.75 681.25
R - RMTCI =
R -R( )( )
55
B6 - BS2S MTCI =
B
-B4( )( )
753.8 709.2
709.2 681
R - RMTCI =
R C
-RASI
Page 12
⇒ Optimized Soil - Adjusted Vegetation Index (OSAVI) (Rondeaux et al., 1996):
⇒ Ratio TCARI/OSAVI (Haboudane et al, 2002)
( )( )
800 670
800 670
1.16 * R -ROSAVI =
R + R + 0.16
( )( )1.16 * B8-B4
OSAVI =B8 + B4 +
S2S 0.16
( )( )
800.8 669.2
800.8 669.2
1.16 * R -ROSAVI =
RCASI
+R + 0.16
Page 14
SENTINEL-2 (600-705) nm Integral Calculation
Interpolation to obtain a 600 nm band between B3(560 nm) and B4(665 nm)
12 Chl ESU from ZALF team 11 Chl ESU from UV team
0
300
600
900
1200
0 300 600 900 1200
CASI
Inte
gral I
ndex
S2S Integral Index
y = m1 + m2 *xErrorValue
162m1 0.020.88m2
0.99r
0
2
4
6
8
10
0 2 4 6 8 10
CASI
GM
I
S2S GMI
y = m1 + m2 * xErrorValue0.4-1.2m1 0.142.05m2 0.95r
Integral Integral IndexIndex
GMIGMI
Page 15
12 Chl ESU from ZALF team 11 Chl ESU from UV team
0.0
0.1
0.2
0.3
0.0 0.1 0.2 0.3
CASI
MCA
RI
S2S MCARI
y = m1 + m2 * xErrorValue0.0011-0.0003m1 0.0080.316m2
0.99r
0.0
0.1
0.2
0.3
0.0 0.1 0.2 0.3
CASI
TCA
RI
S2S TCARI
y = m1 + m2 * xErrorValue0.003-0.017m1 0.020.69m2 0.99rMCARIMCARI
TCARITCARI
Page 16
12 Chl ESU from ZALF team 11 Chl ESU from UV team
0
2
4
6
8
10
0 2 4 6 8 10
CASI
MTC
I
S2S MTCI
y = m1 + m2 * xErrorValue0.09-1.06m1 0.031.7m2
0.998r
0.5
0.6
0.7
0.8
0.5 0.6 0.7 0.8
CASI
OSAVI
S2S OSAVI
ErrorValue0.012-0.032m1 0.0181.050m2
0.997r
y = m1 + m2*x
OSAVIOSAVI
MTCIMTCI
Page 18
12 Chl ESU from ZALF team 11 Chl ESU from UV team
0
2
4
6
8
10
0
2
4
6
8
10
0 20 40 60 80 100
S2SCASI
S2S G
MI
CASI GM
I
Chlorophyll (μg cm-2)
y = m1 + m2 * xErrorValue0.52.7m1
0.0160.003m2 0.04r
y = m1 + m2 * xErrorValue
1.24.9m1 0.03-0.007m2
0.04r
0
500
1000
1500
0
500
1000
1500
0 20 40 60 80 100
S2SCASI
S2S
Inte
gral I
ndex
CASI Integral Index
Chlorophyll (μg cm-2)
y = m1 + m2 * xErrorValue200900m1
7-6m2 0.16r
y = m1 + m2 * xErrorValue200800m1
6-5m2 0.17r
Integral Integral IndexIndex
GMIGMI
Page 19
12 Chl ESU from ZALF team 11 Chl ESU from UV team
0
0.1
0.2
0.3
0
0.1
0.2
0.3
0 20 40 60 80 100
S2SCASI
S2S
MCA
RI
CASI M
CARI
Chlorophyll (μg cm-2)y = m1 + m2 * x
ErrorValue0.070.30m1
0.002-0.005m2 0.47r
y = m1 + m2 * xErrorValue0.020.09m1
0.0007-0.0016m2 0.45r
0
0.1
0.2
0.3
0
0.1
0.2
0.3
0 20 40 60 80 100
S2SCASI
S2S
TCARI
CASI TCA
RI
Chlorophyll (μg cm-2)y = m1 + m2 * x
ErrorValue0.060.23m1
0.0017-0.0030m2 0.35r
y = m1 + m2 * xErrorValue0.040.16m1
0.0012-0.0024m2 0.40r
TCARITCARI
MCARIMCARI
Page 20
12 Chl ESU from ZALF team 11 Chl ESU from UV team
0
2
4
6
8
10
0
2
4
6
8
10
0 20 40 60 80 100
S2SCASI
S2S
MTCI
CASI M
TCI
Chlorophyll (μg cm-2)y = m1 + m2 * x
ErrorValue1.22.9m1
0.030.02m2 0.14r
y = m1 + m2 * xErrorValue
1.93.5m1 0.060.04m2
0.18r
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0 20 40 60 80 100
S2SCASI
S2S
OSA
VI
CASI O
SAVI
Chlorophyll (μg cm-2)y = m1 + m2 * x
ErrorValue0.060.74m1
0.0017-0.0023m2 0.29r
y = m1 + m2 * xErrorValue0.060.75m1
0.0017-0.0025m2 0.30r
OSAVIOSAVI
MTCIMTCI
Page 21
0
20
40
60
0 1000 2000 3000 4000
SPARC
AGRISAR
Curve fit
Chor
ophy
ll (μ
g cm
-2)
Integral Index
y = m1 + m2 * xErrorValue
1.051.3m1 0.0007-0.0146m2
0.78r
Page 22
0
10
20
30
40
50
60
0 0.1 0.2 0.3 0.4 0.5
Chl - ZALFChl-Haboudane
Chloro
phyll (μg
cm
-2)
CASI (TCARI/OSAVI)
y = m1+m2*ln(x)ErrorValue3 e-7-18.363m1 2 e-7-30.194m2
1r
Page 24
0
50
100
150
200
250
0
50
100
150
200
250
0 400 800 1200
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
CASI Integral
y = m1 + m2 * xErrorValue
60140m1 0.09-0.07m2
0.24r
y = m1 + m2 * xErrorValue
60140m1 0.09-0.08m2 0.25r
0
50
100
150
200
250
0
50
100
150
200
250
0 400 800 1200
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
S2S Integral
y = m1 + m2 * xErrorValue
60150m1 0.08-0.08m2
0.30r
y = m1 + m2 * M0ErrorValue
60150m1 0.08-0.08m2 0.32r
LAI and Chl data: 12 ESUs (ZALF)LAI(1): early in the morning LAI(2): afternoon
Page 25
0
50
100
150
200
250
0
50
100
150
200
250
0 2 4 6 8 10
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
S2S GMI
y = m1 + m2 *xErrorValue11060m1 4010m2 0.09r
y = m1 + m2 * xErrorValue11030m1 4020m2 0.16r
0
50
100
150
200
250
0
50
100
150
200
250
0 2 4 6 8 10
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
CASI GMI
y = m1 + m2 * xErrorValue
80150m1 17-11m2 0.21r
y = m1 + m2 * xErrorValue
90120m1 18-7m2 0.12r
LAI and Chl data: 12 ESUs (ZALF)LAI(1): early in the morning LAI(2): afternoon
Page 26
0
50
100
150
200
250
0
50
100
150
200
250
0.0 0.1 0.2 0.3
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
S2S MCARI
y = m1 + m2 * xErrorValue
30130m1 160-300m2 0.51r
y = m1 + m2 * xErrorValue
30130m1 160-300m2 0.49r
0
50
100
150
200
250
0
50
100
150
200
250
0.0 0.1 0.2 0.3
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
CASI MCARI
y = m1 + m2 * xErrorValue
30130m1 500-900m2
0.46r
y = m1 + m2 * xErrorValue
30130m1 500-900m2 0.45r
LAI and Chl data: 12 ESUs (ZALF)LAI(1): early in the morning LAI(2): afternoon
Page 27
LAI and Chl data: 12 ESUs (ZALF)LAI(1): early in the morning LAI(2): afternoon
0
50
100
150
200
250
0
50
100
150
200
250
0.0 0.1 0.2 0.3
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
S2S TCARIy = m1 + m2 * x
ErrorValue40160m1
200-400m2 0.53r
y = m1 + m2 * xErrorValue
40160m1 200-400m2 0.52r
0
50
100
150
200
250
0
50
100
150
200
250
0.0 0.1 0.2 0.3
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
CASI TCARIy = m1 + m2 * x
ErrorValue30140m1
300-600m2 0.49r
y = m1 + m2 * xErrorValue
30140m1 300-600m2
0.49r
Page 28
0
50
100
150
200
250
0
50
100
150
200
250
0 2 4 6 8 10
Chl*LAI (1)Chl*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
CASI MTCI
y = m1 + m2 * xErrorValue
4010m1 817m2 0.55 r
y = m1 + m2 * xErrorValue
402m1 819m2 0.59r
0
50
100
150
200
250
0
50
100
150
200
250
0 2 4 6 8 10
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
S2S MTCI
y = m1 + m2 * xErrorValue
50-10m1 1430m2 0.56r
y = m1 + m2 * xErrorValue
50-20m1 1433m2 0.60r
LAI and Chl data: 12 ESUs (ZALF)LAI(1): early in the morning LAI(2): afternoon
Page 29
LAI and Chl data: 12 ESUs (ZALF)LAI(1): early in the morning LAI(2): afternoon
0
50
100
150
200
250
0
50
100
150
200
250
0.5 0.6 0.7 0.8 0.9 1.0
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
S2S OSAVIy = m1 + m2 * x
ErrorValue160560m1 200-700m2 0.67r
y = m1 + m2 * xErrorValue190500m1 300-500m2 0.56r
0
50
100
150
200
250
0
50
100
150
200
250
0.5 0.6 0.7 0.8 0.9 1.0
Chlor*LAI (1)Chlor*LAI (2)
Chl*LA
I (1
) Chl*LAI (2)
CASI OSAVIy = m1 + m2 * x
ErrorValue150530m1 200-600m2 0.67r
y = m1 + m2 * xErrorValue170470m1 300-600m2 0.57r
Page 30
• Concerning to SENTINEL-2 simulations, the use of CASI images (VNIR)
together with AHS images gives a more realistic simulation than the use of
AHS data only (particularly for channel 1-9)
• Concerning SENTINEL-2 and CASI chlorophyll indices, all them shows very
good correlations between both sensors (r > 0.95)
• When correlations between chlorophyll indices and ground truth chlorophyll
data are made, for SENTINEL-2 and CASI images, bad correlation
coefficients are found (r < 0.6)
When AGRISAR data are considered with other ESA campaign data (like
SPARC) it can be seen that AGRISAR data inserts well in the set but with a
minor range of variation due to the minor number of different crops
considered in the AGRISAR campaign and a better correlation coefficient can
be found with a large dynamical range of values
Page 31
• Concerning the correlation, for SENTINEL-2 and CASI images, between
chlorophyll indices and LAI*Chl, better correlation coefficients than in the
case of the correlations with chlorophyll data only are found (r < 0.7) what
indicates the necessity of obtaining LAI and chlorophyll measurements
together
The small number of crops and their small range of variability seems to be
responsible for the low correlation coefficients found when considering
chlorophyll indices and ground truth chlorophyll data