Remote Estimation of Chlorophyll Content and Gross Primary

In honor of Alexander F.H. Goetz
Remote Estimation of Chlorophyll
Content and Gross
Primary Production in Crops.
Anatoly A. Gitelson, Shashi Verma, Donald C.
Rundquist, and Andres Vina
University of Nebraska-Lincoln
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Objective
To develop a quantitative technique
for remote estimation of Chlorophyll
Content and Gross Primary
Production in agro-ecosystems
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Methods and Techniques
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Data: UNL Carbon Sequestration Project
CO2 Fluxes: eddy covariance flux system
and soil C stocks
Radiation Fluxes: upwelling and
downwelling above and under canopy
Biomass: total and green components of
leaves, stems and reproductive organs
LAI: total and green components
Plant height, phenological development
Temperature, precipitation, soil moisture
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Data: UNL Carbon Sequestration Project
Leaf Level
• CO2 Fluxes
• Reflectance 400-900 nm
- Ocean Optics radiometer
attached to leaf clip
• Pigment content and composition
- analytical technique
- non-destructive technique (Gitelson & Merzlyak, 1994)
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Data: UNL Carbon Sequestration Project
Community Level
• Reflectance 400-900 nm
Dual-fiber Ocean Optics radiometers
• Vegetation fraction
Video camera imagery:
‘Excess Green’ technique to retrieve VF
• Temperature
Rundquist et al., 2004
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Data: UNL Carbon Sequestration Project
Field Level
Radiance & Reflectance Imagery
440- 850 nm
AISA Hyperspectral
Imaging System
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Sampling Areas for Close-Range Hyperspectral
Measurements
3a
3
Irrigated continuous
maize
Rain fed
maize/soybean
rotation
Irrigated
maize/soybean
rotation
Sampling
~ 20 m
3a – Bt Maize
3 – Non Bt Maize
2001 18 campaigns 2002 31 campaigns 2003 36 campaigns
2004 35 campaigns 2005 32 campaigns
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Gross Primary Production
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
GPP ∝ fAPAR×PAR×LUE
GPP is a function of
• the efficiency of the leaf light-harvesting apparatus
(fAPAR)
• amount of PAR captured by plant
APAR = PAR×fAPAR
• the capacity of plant to utilize absorbed radiation
(LUE)
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
GPP ∝ fAPAR×PAR×LUE
GPP is a function of
• the efficiency of the leaf light-harvesting apparatus
(fAPAR)
• amount of PAR captured by plant
APAR = PAR×fAPAR
• the capacity of plant to utilize absorbed radiation
(LUE)
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
GPP ∝ fAPAR×PAR×LUE
GPP is a function of
• the efficiency of the leaf’s light-harvesting
apparatus (fAPAR)
• amount of PAR captured by plant
APAR = PAR×fAPAR
• the capacity of plant to utilize absorbed radiation
(LUE)
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
3.5
Vegetative Stage
Maize
Reproductive Stage
3.0
-2 -1
GPP (mg.m s )
Senescence Stage
Is LUE constant?
2.5
2.0
1.5
1.0
0.5
2.0
0.0
0
500
15001.8
1000
-2 -1
APAR (mmol m s )
Reproductive Stage
1.6
GPP (mg m-2 s-1)
GPP ∝ APAR*LUE
Vegetative
Stage
2000
Senescence Stage
1.4
1.2
1.0
0.8
0.6
0.4
0.2
Soybean
0.0
0
500
1000
1500
2000
-2 -1
APAR (mmol m s )
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
The Carnegie-Ames-Stanford Approach (CASA)
GPP = NDVI×PAR×ε×g(T)×h(W)
fAPAR
LUE
g(T) and h(W) are functions that account for
effects of temperature and water stress
Light use efficiency was assumed to be constant
for individual biomes
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
GPP ∝ NDVI×PAR×LUE
Light use efficiency is
not constant per biome.
3.5
Soybean
2
R = 0.6514
Maize
2.5
sPRI ∝ LUE ?
GPP vs. NDVI×PAR×sPRI
2
GPP (mg/m s)
3.0
2.0
1.5
1.0
R2 = 0.681
0.5
3.5
0.0
0
500
1000
3.0
1500
2000
R2 = 0.69
2.5
2
NDVI x PAR (mmol/m s)
GPP (mg/m s)
2
Soybean
Maize
PRI is not a proxy
of LUE
for crops studied
2.0
1.5
1.0
0.5
R2 = 0.72
0.0
0
200
400
600
800
1000
2
NDVI x sPRI x PAR (mmol/m s)
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
LUE = GPP/APAR
0.003
LUE depends on
ecosystem type,
temperature, nutrients,
water stresses,
and leaf physiology
0.002
0.002
0.001
0.001
0.000
-0.001
-0.001
-0.002
Irrigated and rainfed maize
2001-2003
-0.002
120
140
160
180
200
220
240
260
DOY
Uncertainty in LUE assessment
is a primary source of error
in GPP estimates for crops
280
Light Use Efficiency (mg/mmol)
Light Use Efficiency (mg/mmol)
Temporal behavior of LUE
0.002
0.002
300
0.001
0.001
0.000
-0.001
-0.001
Irrigated and rainfed soybean 2002
-0.002
120
140
160
180
200
220
240
260
280
DOY
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
GPP ∝ fAPAR×LUE×PAR
∝ Chl
fAPAR depends on the amount of photosynthetically active
biomass, the primary source of variability in chlorophyll
Chlorophyll is an indicator of the capacity of the plant to
utilize absorbed radiation
Our hypothesis:
in crops, fAPAR×LUE is closely related to total
chlorophyll content
GPP ∝ Chl×PAR
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
GPP vs. Chl×PAR
Irrigated and rainfed maize and soybean
3.5
Maize
GPP (mg/m2s)
3.0
Soybean
2.5
2.0
1.5
2
1.0
2
RMSE = 0.242 mg/m s, R =0.98
0.5
0.0
0
1000
2000
3000
4000
5000
6000
7000
8000
Total Crop Chl (g/m2) x PAR (mmol/m2s)
Total Chl = LAI×Chlleaf
IGARSS’06, August 2, 2006, Denver
Gitelson et al, JGR, Atm., 2006
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
To estimate remotely GPP in crops
one should find a way to accurately
retrieve chlorophyll content from
remotely sensed data
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Algorithm Development
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Pigment Content Estimation
Leaf level
R-1(λ) ∝ [αp (λ) + α0(λ)]/bb
Cp ∝ α pigm (λ1 ) ∝ [ R (λ1 ) − R (λ2 )]R (λ3 )
−1
−1
Chlorophyll: Gitelson and Merzlyak, 1994, 1996; Gitelson, et al., 2003
Anthocyanin: Gitelson et al., 2001
Carotenoids: Gitelson et al., 2002
Flavonoids: Merzlyak et al., 2004
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Chlorophyll Content Estimation
Leaf level
600
Chlorophyll Estimate
500
R2 = 0.9622
400
300
200
100
[(R700)-1 - (RNIR)-1]*RNIR
0
0
100
200
300
400
500
600
700
800
900
Total chlorophyll content, μmol/m2
Gitelson and Merzlyak, 1994
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Canopy, community, field levels
[(ρgreen)-1-(ρNIR)-1]ρNIR ∝ Chl
[(ρred edge)-1-(ρNIR)-1]ρNIR ∝ Chl
Gitelson, Viña, Ciganda, Rundquist, Arkebauer, Geophys. Res. Lett., 2005
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Maize: canopy community and field levels
Chl ∝ [(ρgreen)-1-(ρNIR)-1]ρNIR Chl ∝ [(ρred edge)-1-(ρNIR)-1]ρNIR
(RNIR/Rred edge)-1 & (RNIR/Rgreen)-1
14
Irrigated and rainfed maize, 2001-2003
12
Green and NIR bands
10
R2 = 0.9185
8
6
R2 = 0.9173
4
Red Edge and NIR
2
0
0.0
1.0
2.0
3.0
4.0
5.0
Total Chlorophyll, g/m2
Gitelson, Viña, Ciganda, Rundquist, Arkebauer, Geophys. Res. Lett., 2005
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Soybean: canopy community and field levels
Chl ∝ [(ρgreen)-1-(ρNIR)-1]ρNIR Chl ∝ [(ρred edge)-1-(ρNIR)-1]ρNIR
(RNIR/Rred edge)-1 & (RNIR/Rgreen)-1
14
Irrigated and rainfed soybean, 2002
12
Green and NIR bands
R2 = 0.9132
10
8
6
Red Edge and NIR
4
R2 = 0.9405
2
0
0.0
0.5
1.0
1.5
2.0
2.5
Total Chlorophyll, g/m2
Gitelson, Viña, Ciganda, Rundquist, Arkebauer, Geophys. Res. Lett., 2005
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Having an accurate proxy of
chlorophyll content in a crop
canopy, we can estimate
Gross Primary Production
GPP ∝ Chl×PAR
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Model Calibration
RMSE = 0.30 mg/m2s
2
R = 0.9053
20000
GPP ∝ Chl×PAR
15000
10000
MODIS Bands
5000
Soybean
0
0.0
0.5
1.0
1.5
2.0
GPP (mg/m2s)
Maize
2.5
3.0
[(RNIR/Rgreen )-1] x PAR (mmol/m2s)
[(RNIR/Rgreen )-1] x PAR (mmol/m2s)
25000
12000
RMSE = 0.20 mg/m2s
10000
2
R = 0.8943
8000
6000
4000
2000
0
0.0
Gitelson et al., JGR, 2006
IGARSS’06, August 2, 2006, Denver
3.5
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
GPP (mg/m2s)
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Model Validation
MODIS Green and NIR Bands
3.0
Frequency
2.5
2
0.20
0.15
0.10
0.05
9
>1 0
00
<10
0
-9
0
-7
0
-5
0
-3
0
-1
0
50
70
0.00
2.0
10
30
Predicted GPP (mg/m s)
0.25
Residuals (%)
1.5
1.0
2
RMSE=0.284 mg/m s
0.5
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Observed GPP (mg/m2s)
Irrigated and rainfed maize
IGARSS’06, August 2, 2006, Denver
Gitelson et al., JGR, 2005
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Model Validation
MERIS Red Edge and NIR Bands
9
>1 0
00
<10
0
-9
0
-7
0
-5
0
-3
0
-1
0
10
30
50
2.0
0.30
0.25
0.20
0.15
0.10
0.05
0.00
70
Frequency
2.5
2
Predicted GPP (mg/m s)
3.0
Residuals (%)
1.5
RMSE=0.267 mg/m2s
1.0
0.5
D
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Observed GPP (mg/m2s)
Irrigated and rainfed maize
IGARSS’06, August 2, 2006, Denver
Gitelson et al., JGR, 2005
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Field Level
July 2, 2003
GPP retrieved from AISA imagery
Maize irrigated
IGARSS’06, August 2, 2006, Denver
Maize rainfed
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
GPP retrieved from AISA imagery in
MODIS and MERIS bands
Maize
Soybean
June 21
July 12
July 15
(RNIR/Rgreen-1) & (RNIR/Rred edge-1)
May 3
3.5
MODIS
MERIS
3.0
2.5
y = 0.4458x + 0.6661
R2 = 0.964
2.0
1.5
1.0
y = 0.5168x + 0.1611
2
R = 0.9796
0.5
0.0
September 7
IGARSS’06, August 2, 2006, Denver
0
1
2
3
2
4
5
6
GPP (mg/m s)
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Conclusions
• Chlorophyll content in crops is closely related to
and may be used as an surrogate for GPP
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Conclusions
• Chlorophyll content in crops is closely related to
and may be used as an surrogate for GPP
• A conceptual model, developed originally for
pigment content retrieval in plant leaves, has
been used to accurately estimate GPP in crops
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Conclusions
• Chlorophyll content in crops is closely related to
and can be use as an surrogate for GPP
• Conceptual model, developed originally for
pigment content retrieval in plant leaves, has
been used to accurately estimate GPP in crops
• Our results provide evidence that this model may
be considered as a general solution for assessing
pigment content in optically deep media,
independent of the type of medium
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Chlorophyll in leaves
Anthocyanins in leaves
Gitelson & Merzlyak, 1994, 1997
Gitelson et al., 1996; 2003
Chl, Car, Anth & Flavonoids
in fruits
Merzlyak et al., 2003
IGARSS’06, August 2, 2006, Denver
Carotenoids in leaves
Gitelson et al., 2002
Gitelson et al., 2001
Chla in turbid productive
waters
Dall’Olmo & Gitelson, 2005 ; 2006
Total Chl and GPP
in crops
Gitelson et al., 2005
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Acknowledgements
• Center for Advanced Land Management
Information Technologies (CALMIT) – UNL
• Carbon Sequestration Program – UNL
• NASA Land Cover Land Use Change Program
• NASA EPSCoR (Airborne Remote Sensing)
• NASA Nebraska Space Grant
• NSF EPSCoR (Airborne Remote Sensing)
• U. S. Department of Energy
‰ EPSCoR Program
‰ Office of Science (BER)
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz
Thank you
http://www.calmit.unl.edu/calmit.html
[email protected]
IGARSS’06, August 2, 2006, Denver
Environmental Applications of Imaging Spectroscopy
In honor of Alexander F.H. Goetz