Hyperspectral Imagery for Environmental Mapping and Monitoring: Case Study of Grassland in Belgium.
The objective of this study is to show that hyperspectral imagery can be used to characterise grassland as well as its biophysical and biochemical properties.
Call Girls in Munirka Delhi 💯Call Us 🔝9953322196🔝 💯Escort.
Hyperspectral Imagery for Environmental Mapping and Monitoring
1. Hyperspectral Imagery forHyperspectral Imagery for
Environmental Mapping and MonitoringEnvironmental Mapping and Monitoring
Case Study of Grassland in BelgiumCase Study of Grassland in Belgium
Biometry, Data management and Agrometeorology Unit
buffet@cra.wallonie.be – oger@cra.wallonie.be
CENTRE WALLON DE RECHERCHES AGRONOMIQUES
2. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
ContextContext
The objective of this study is to show that hyperspectral imagery
can be used to characterise grassland as well as its biophysical
and biochemical properties.
Inventory of forage production & qualityInventory of forage production & quality
Management practicesManagement practices
Control of application of agri-environmental measuresControl of application of agri-environmental measures
Grassland is an important component of the agricultural landscape.
Monitoring grassland at the regional level is closely linked to the
knowledge of regional:
3. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
PresentationPresentation
Scientific objectivesScientific objectives
MaterialMaterial
Study areaStudy area
Field campaign (before flight)Field campaign (before flight)
Field campaign (during flight)Field campaign (during flight)
MethodMethod
Spectral analysis at Pixel level (intra-parcel)Spectral analysis at Pixel level (intra-parcel)
Spectral analysis at parcel level (inter-parcel)Spectral analysis at parcel level (inter-parcel)
ResultsResults
ConclusionsConclusions
4. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Scientific objectivesScientific objectives
Agri-Environmental Measures (AEM):
• Now compulsory (CAP reform, WR regulations)
• Temporal constraints (Specific cutting or grazing periods)
• Opportunity of using airborne imaging spectroscopy to trace the
recent history of grasslands in terms of managements
Working hypothesis: cutting or grazing actions
• Can be considered as major stresses, which have an impact on the
spectral signatures of grass canopy.
• Changes of spectral signature depend not only on the nature and
the intensity of the action, but also on the time spent between the
action and the remote sensing data acquisition.
5. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Scientific objectivesScientific objectives
Several campaigns with CASI/SASI, CASI-ATM airborneSeveral campaigns with CASI/SASI, CASI-ATM airborne
sensorssensors
A continuum and a possible validation of the first campaign:
• CASI-2 (VIS/NIR) and SASI (SWIR) sensor
• August 2002
• Lorraine test site
• Biophysical (wet matter, biomass, grass height…)
• Biochemical (protein, VEM, DVE…)
Tries to enlighten how imaging spectroscopy are capable to:
• Determine grassland management practices
• Control of application of Agri-Environmental Measures
6. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
PresentationPresentation
Scientific objectivesScientific objectives
MaterialMaterial
Study areaStudy area
Field campaign (before flight)Field campaign (before flight)
Field campaign (during flight)Field campaign (during flight)
MethodMethod
Spectral analysis at Pixel level (intra-parcel)Spectral analysis at Pixel level (intra-parcel)
Spectral analysis at parcel level (inter-parcel)Spectral analysis at parcel level (inter-parcel)
ResultsResults
ConclusionsConclusions
7. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Study areasStudy areas
South-east of Belgium.South-east of Belgium.
Luxembourg ProvinceLuxembourg Province
(Belgian Lorraine).(Belgian Lorraine).
Grassland > 75% of land-Grassland > 75% of land-
use.use.
Study area corners:Study area corners:
Study area =Study area = ±±50 Km²50 Km²
10 flight lines with a rate of10 flight lines with a rate of
covering of 40%.covering of 40%.
Lat. Long
ULC 49°43' 23.16 N 5°27' 34.60 E
LRC 49°39' 00.43 N 5°31' 12.97 E
8. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Study areasStudy areas
South-east of Belgium.South-east of Belgium.
Luxembourg Province (near Arlon).Luxembourg Province (near Arlon).
Located in Natura 2000.Located in Natura 2000.
Grassland > 75% of land-useGrassland > 75% of land-use
Study area =Study area = ±±35 Km²35 Km²
A subset of 33 parcels was followed:A subset of 33 parcels was followed:
Ground cover was scored visually from theGround cover was scored visually from the
middle of May to the airborne flight.middle of May to the airborne flight.
Biophysical parameters (GH, FMY, DMY…)Biophysical parameters (GH, FMY, DMY…)
Field spectroradiometer (ASD)Field spectroradiometer (ASD)
9. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Field campaignField campaign
Biophysical parametersBiophysical parameters
17 parcels17 parcels
4 sampling units / parcel4 sampling units / parcel
During airborne flightDuring airborne flight
2 types of samples2 types of samples
Mower cutting: total canopyMower cutting: total canopy
Scissors cutting: upper canopyScissors cutting: upper canopy
Biophysical parametersBiophysical parameters
Grass heightGrass height
Grass biomass (wet matter)Grass biomass (wet matter)
Field spectroradiometersField spectroradiometers
ASD FieldSpec Pro FrASD FieldSpec Pro Fr
GER 1500GER 1500
10. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Field campaign (before flight)Field campaign (before flight)
PARCEL OBS1 OBS2 OBS3 OBS4 OBS5 OBS6 OBS7 OBS8 OBS9 OBS10 OBS11 MCo LAST CUT
1 * P RP P P P RP RP P RP F P 22/06/2004
2 P P P P P P P P P P F P 22/06/2004
3 * NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF F 18/06/2004
4 * NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF RF F 14/06/2004
5 * NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF F 18/06/2004
6 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF RF F 15/06/2004
7 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF
8 * P P P P P P P P P P P
9 NP NF NP NF NP NF F RF RF RF RF RF RF RF F 28/05/2004
10 P P R R R P P P P P RP P
11 RF RF RF P P P RP P P P P P
12 NP NF NP NF NP NF NP NF NP NF NP NF F RF RF RF RF F 09/06/2004
13 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF F F 22/06/2004
14 NP NF NP NF NP NF F RF RF RF RF RF RF RF F 27/05/2004
15 NP NF NP NF NP NF F RF RF RF RF RF P RP F 28/05/2004
16 * P P P P P P P P P P PP
17 * NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF RF F 14/06/2004
18 * NP NF NP NF NP NF NP NF NP NF F RF RF RF RF F 07/06/2004
19 * P P P PR P P RP RP RP RP PP
20 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF
21 * P P P PR P P P P P RP PP
22 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF
23 P P RP P P P P P P P P PP
24 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF RF F 14/06/2004
25 NP NF NP NF NP NF NP NF NP NF NP NF F RF RF RF RF F 07/06/2004
26 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF RF F 14/06/2004
27 * NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF RF F 14/06/2004
28 * NP NF P P P P P P P P P P
29 * NP NF NP NF NP NF NP NF NP NF NP NF F RF RF RF F 10/06/2004
30 * P P P P P P P P P P PP
31 * P P P P P P P P RP RP PP
32 * NP NF NP NF NP NF NP NF NP NF F RF RF RF RF F 09/06/2004
33 NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF NP NF F RF F 18/06/2004
Table 1. Management practices observed on the 33 monitored parcels and their respective observed
management class (MCo). (R = Restoration, F = Haying, P = grazing, NP NF = No Grazing and No Haying).
13. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Hyperspectral data acquisitionHyperspectral data acquisition
End of JuneEnd of June
CASI resolution:CASI resolution:
- 2.4 x 2.4 m- 2.4 x 2.4 m
-
96 spectral bands96 spectral bands
10 transects by sensors10 transects by sensors
Quality checkQuality check (signal & geometric)(signal & geometric)
Bad weather conditionsBad weather conditions
• Signal problemsSignal problems
• Image correctionImage correction
=> 24 parcels were preserved=> 24 parcels were preserved 0
100
200
300
400
500
600
700
800
0.45
0.47
0.5
0.52
0.54
0.56
0.59
0.61
0.63
0.65
0.68
0.7
0.72
0.75
0.77
0.79
0.82
0.84
0.86
0.88
9_Corr
9_Brute
9_Terrain
0
100
200
300
400
500
600
0.45
0.47
0.5
0.52
0.54
0.56
0.59
0.61
0.63
0.65
0.68
0.7
0.72
0.75
0.77
0.79
0.82
0.84
0.86
0.88
7_brute
7_Terrain
14. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Hyperspectral data acquisitionHyperspectral data acquisition
10 transects by sensors10 transects by sensors
Level II bLevel II b
Quality checkQuality check
(signal & geometric)(signal & geometric)
SASI = some problemsSASI = some problems
3x3 pixel subset3x3 pixel subset
centred over thecentred over the
sampling unit locationsampling unit location
SASI – 1345 nm
15. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Laboratory analysisLaboratory analysis
Biochemical parametersBiochemical parameters
Post field campaignPost field campaign
NIR Spectrometry laboratory-based measurementsNIR Spectrometry laboratory-based measurements
Samples cut with scissors & mowerSamples cut with scissors & mower
Fresh samples & dry samplesFresh samples & dry samples
Biochemical parameters measuredBiochemical parameters measured
Dry matter contentDry matter content
Proteins contentProteins content
Cellulose contentCellulose content
Ashes contentAshes content
Sugar contentSugar content
Energy values which characterised available energy for milkEnergy values which characterised available energy for milk
and meat production: (VEM, VEVI, DVE and OEB)and meat production: (VEM, VEVI, DVE and OEB)
16. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
PresentationPresentation
Scientific objectivesScientific objectives
MaterialMaterial
Study areaStudy area
Field campaign (before flight)Field campaign (before flight)
Field campaign (during flight)Field campaign (during flight)
MethodMethod
Spectral analysis at Pixel level (intra-parcel)Spectral analysis at Pixel level (intra-parcel)
Spectral analysis at parcel level (inter-parcel)Spectral analysis at parcel level (inter-parcel)
ResultsResults
ConclusionsConclusions
17. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Methodology : General overviewMethodology : General overview
CASI & SASI
Images
Pixel
spectral
responses
Moving average
+
First derivative
biophysical or
biochemical
variable
Correlogram
Spectral
indices
Predictive
models
Selected
narrowbands
biophysical or
biochemical
variable
19. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Step 2: Spectral indicesStep 2: Spectral indices (cont.)(cont.)
Photochemical Reflectance Index (PRI):
Red-edge slope, Red-edge step and Red-edge maximum
slope wavelength:
569529
569529
RR
RR
PRI
+
−
=
nm780to677rangein themax
λλ
=
d
dR
RESL
677780 RRREST −=
RESLREMS λ=
This index is related to the green-edge narrowband
20. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Step 2: Spectral indicesStep 2: Spectral indices (cont.)(cont.)
Water Band Index (WBI) :
Normalized Difference Vegetation
Index (NDVI):
This index takes advantage of the water absorption
features and represents canopy water content.
Two reflectance ratios were calculated.
677780
677780
RR
RR
NDVI
+
−
=
( )
971
832902902
1
R
RRR
WBI
−+
=
785902
785902
2
λλ −
−
=
RR
WBI
Reflectance indices
0
100
200
300
400
500
600
700
800
400 500 600 700 800 900 1000
Wavelength (nm)
Reflectance
PRI
NDVI
Red-edge Step
& Slope
WBI1 & WBI2
21. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral analysis at Pixel levelSpectral analysis at Pixel level
Table 2.Table 2. Observed correlation coefficients between the different indices andObserved correlation coefficients between the different indices and
grassland characteristicsgrassland characteristics ..
Leaf water
content
FMY DMY Grass
Height
Biomass
λRESL 0.297 0.078 0.023 0.19 0.087
NDVI 0.814 0.753 0.303 0.417 0.585
PRI 0.716 0.567 -0.016 0.243 0.029
RESL 0.579 0.637 0.215 0.149 0.172
REST 0.55 0.595 0.195 0.127 0.152
WBI1 -0.688 -0.587 -0.163 -0.641 -0.335
WBI2 -0.722 -0.504 -0.074 -0.534 -0.123
Leaf water content is more highly correlated with spectral reflectance thanLeaf water content is more highly correlated with spectral reflectance than
either total wet biomass or total dry biomass.either total wet biomass or total dry biomass.
Results show that ground truth data collection or canopy sampling for remote sensing studiesResults show that ground truth data collection or canopy sampling for remote sensing studies
of grass canopies should measure the total wet biomass and the total dry biomass. This willof grass canopies should measure the total wet biomass and the total dry biomass. This will
allow for the calculation of the leaf water (also Compton conclusions).allow for the calculation of the leaf water (also Compton conclusions).
22. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Step 3: Multiple regression analysisStep 3: Multiple regression analysis
CASI & SASI
Images
Pixel
spectral
responses
Moving average
+
First derivative
biophysical or
biochemical
variable
Correlogram
Spectral
indices
Predictive
models
Selected
narrowbands
biophysical or
biochemical
variable
23. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral analysis at Pixel levelSpectral analysis at Pixel level
This spectral analysis at pixel level is used to investigate and validateThis spectral analysis at pixel level is used to investigate and validate
CASI/SASI-2002 results, with regards to the characterisation of grassCASI/SASI-2002 results, with regards to the characterisation of grass
canopy with quantitative information regarding biophysicalcanopy with quantitative information regarding biophysical
parameters (FMY, DMY, GH).parameters (FMY, DMY, GH).
CASI-ATM pixel responses were average within 3x3 pixel subsetCASI-ATM pixel responses were average within 3x3 pixel subset
centered around the sampling unit.centered around the sampling unit.
In addition to the standard channels a number of channel ratios andIn addition to the standard channels a number of channel ratios and
normalized channel difference indices were developed.normalized channel difference indices were developed.
Photochemical Reflectance Index (Photochemical Reflectance Index (PRIPRI).).
Red-edge slope (Red-edge slope (RESLRESL), Red-edge step (), Red-edge step (RESTREST) and Red-edge maximum) and Red-edge maximum
slope wavelength (slope wavelength (REMSREMS).).
Water Band Index (Water Band Index (WBIWBI))
Normalized Difference Vegetation Index (Normalized Difference Vegetation Index (NDVINDVI).).
……
24. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral analysis at Pixel levelSpectral analysis at Pixel level
Table 2.Table 2. Observed correlation coefficients between the different indices andObserved correlation coefficients between the different indices and
grassland characteristicsgrassland characteristics ..
Leaf water
content
FMY DMY Grass
Height
Biomass
λRESL 0.297 0.078 0.023 0.19 0.087
NDVI 0.814 0.753 0.303 0.417 0.585
PRI 0.716 0.567 -0.016 0.243 0.029
RESL 0.579 0.637 0.215 0.149 0.172
REST 0.55 0.595 0.195 0.127 0.152
WBI1 -0.688 -0.587 -0.163 -0.641 -0.335
WBI2 -0.722 -0.504 -0.074 -0.534 -0.123
Red-Edge indicators (Slope, Step, REMS) have bad correlation coefficientsRed-Edge indicators (Slope, Step, REMS) have bad correlation coefficients
compared to 2002 campaign. REST had 0.63 for GH and 0.68 for Biomass.compared to 2002 campaign. REST had 0.63 for GH and 0.68 for Biomass.
This difference is probably the consequence of the poor images quality resulting of badThis difference is probably the consequence of the poor images quality resulting of bad
meteorological conditions during the flight.meteorological conditions during the flight.
25. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral IndicesSpectral Indices
Relationship of the grass height and the step value of the
reflectance spectra in the red edge band
y = 0.0469x - 9.2337
R2
= 0.63
0
5
10
15
20
25
200 250 300 350 400 450 500 550 600 650 700
Red edge step
Grassheight(cm)
Relationship of the green weight and the step value of the
reflectance spectra in the red edge band
y = 37.858x - 10179
R2
= 0.6821
0
4000
8000
12000
16000
20000
200 250 300 350 400 450 500 550 600 650
Red edge step
Greenweight(kg/ha)
Red-Edge Step:Red-Edge Step:
Height (R²=0.63)Height (R²=0.63)
Green weight (R²=0.68)Green weight (R²=0.68)
Water Balance Index:Water Balance Index:
Dry matter content (R²=0.61)Dry matter content (R²=0.61)
Relationship of the water balance indice and the dry matter
content of grassland canopy
y = 61.984x - 11.559
R2
= 0.615
10
20
30
40
50
0.4 0.5 0.6 0.7 0.8 0.9
WBI2
Drymattercontent(MS%)
26. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral analysis at Pixel levelSpectral analysis at Pixel level
Red-edge indicators seems to be too sensible to the meteorologicalRed-edge indicators seems to be too sensible to the meteorological
conditions and can not be considered for an operational system.conditions and can not be considered for an operational system.
Inspection of the regression results obtained in the 2002 and 2003Inspection of the regression results obtained in the 2002 and 2003
campaigns indicates that NDVI, WBI1 and PRI are the best indicatorscampaigns indicates that NDVI, WBI1 and PRI are the best indicators
to estimate biophysics characteristicsto estimate biophysics characteristics..
Confirm CASI-SASI 2002 results on biophysical parameters (FMY, DMY, GH)Confirm CASI-SASI 2002 results on biophysical parameters (FMY, DMY, GH)
directly linked to the age and the management practices supported bydirectly linked to the age and the management practices supported by
grasslands.grasslands.
27. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral analysis at Parcel levelSpectral analysis at Parcel level
In a second step, the project has analysed the possibility to classifyIn a second step, the project has analysed the possibility to classify
grassland in 3 management classes:grassland in 3 management classes:
• Haying grasslands (P)Haying grasslands (P)
• Grazing grasslands (F)Grazing grasslands (F)
• Neither haying nor grazing grasslands (NP NF)Neither haying nor grazing grasslands (NP NF)
The grassland classification is based on the hypothesis that managementThe grassland classification is based on the hypothesis that management
practices can be identified by the combination ofpractices can be identified by the combination of
• Vegetation indices (Vegetation indices (quantitative parametersquantitative parameters) selected at the pixel) selected at the pixel
levellevel
• Textural indices (Textural indices (qualitative parametersqualitative parameters))
Different textural indices were calculated:Different textural indices were calculated:
• Global approach (global variance)Global approach (global variance)
• Local approach (moving windows of 3x3 pixels)Local approach (moving windows of 3x3 pixels)
28. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral analysis at Parcel levelSpectral analysis at Parcel level
Step 1:Step 1: DDiscriminant analysis to identify regions of interest in the reflectanceiscriminant analysis to identify regions of interest in the reflectance
spectra and to choose the relevant textural indicesspectra and to choose the relevant textural indices
Probability level of the differences between grassland classes
Global texture parameterGlobal texture parameter
=> 2 regions of interest=> 2 regions of interest
- Below 500 nmBelow 500 nm
- Above 725 nm.Above 725 nm.
Local textureLocal texture
=> Quite different
- Around 500 nm
- Above 750 nm
- BUT observed level are
not significant (P<0.05).
29. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral analysis at Parcel levelSpectral analysis at Parcel level
Step 1:Step 1: Based on these results, 2 wave bands from the global texture
analysis have been selected (450 nm and 725 nm) together with
previously defined vegetation indices.
Step 2:Step 2: Discriminant analysis with all the selected dependent variables
(vegetation indicators & textural indicators)
• The stepwise discriminant analysis identified 3 vegetation indices,
(PRI, NDVI and WBI1) and one textural global index (450nm) as
significant.
• Cross validation procedure show that about 83% of correct
classifications may be obtained.
30. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Results: Cross classificationResults: Cross classification
All the parcels with AEM are well identified MCo = MCi = NP NFAll the parcels with AEM are well identified MCo = MCi = NP NF
Only 4 parcels have been classified in a bad management classOnly 4 parcels have been classified in a bad management class
(MCo(MCo ≠≠ MCi).MCi).
Classification
results
Observed management classes
F NP NF P
F 13 0 1
NP NF 2 3 1
P 0 0 4
N total 15 3 6
N correct 13 3 4
Proportion 86,70% 100% 66,70%
31. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Results: Cross classificationResults: Cross classification
Only 4 parcels have been classified in a bad management classOnly 4 parcels have been classified in a bad management class
2 parcel with2 parcel with MCo = FMCo = F classified inclassified in MCi = NP NFMCi = NP NF
1 parcel with1 parcel with MCo = PMCo = P classified inclassified in MCi = NP NFMCi = NP NF
1 parcel with1 parcel with MCo = PMCo = P withwith MCi = FMCi = F
These misclassifications can easily be explained by a regrowth of grass after
a long period without pasture or after haying.
These results also show that if the 24 parcels had been declared inThese results also show that if the 24 parcels had been declared in
AEM, and 18 parcelsAEM, and 18 parcels of them would be irregular (MCoof them would be irregular (MCo ≠ NP NF),≠ NP NF),
only 3 parcels would not have been identified as irregular by remoteonly 3 parcels would not have been identified as irregular by remote
sensing (MCi = NP NF).sensing (MCi = NP NF).
32. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Spectral IndicesSpectral Indices (cont.)(cont.)
Red-Edge Step & Water Balance Index can be used toRed-Edge Step & Water Balance Index can be used to
discriminate between grassland management typesdiscriminate between grassland management types
Chemical characteristic of grass was not clearly linked toChemical characteristic of grass was not clearly linked to
vegetation indices.vegetation indices.
Except for VEM and VEVI energy values which seem wellExcept for VEM and VEVI energy values which seem well
correlated with NDVI (R²=0.60 )correlated with NDVI (R²=0.60 )
In most of cases results from scissors cutting samples whichIn most of cases results from scissors cutting samples which
represent the upper part of the canopy were best correlated withrepresent the upper part of the canopy were best correlated with
vegetation indices.vegetation indices.
33. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Multiple regression analysisMultiple regression analysis
CASI sensor = best resultsCASI sensor = best results
CASI + SASI = not better predictionCASI + SASI = not better prediction
Good predictive quality in generalGood predictive quality in general
Predictive quality of the models
Comparison of the different sensor data
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
D
ry
m
atter
P
rotein
C
ellu
lose
S
ugar
V
EM
V
EV
I
D
V
E
O
E
B
H
eight
B
iom
ass
W
etm
atter
Grass characteristic
R2
34. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
Multiple regression analysisMultiple regression analysis
Predictive quality of the models
Comparison of the two cutting systems
0
10
20
30
40
50
60
70
80
Dry
matter
Protein Cellulose Sugar VEM VEVI DVE OEB
Grass characteristic
Relativerootmeansquareerror
Scissors
Mower
RMSE scissors lower than mowerRMSE scissors lower than mower
RMSERMSE ≤≤ 10% => good predictions10% => good predictions
Except for OEB & SugarExcept for OEB & Sugar
35. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
ConclusionsConclusions
Higher potential
Lower potential
< 1000 kg/ha
> 15 000 kg/ha
Estimation of the wet matter
production at regional level.
Question 1:Question 1:
Regional monitoring?Regional monitoring?
36. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
ConclusionsConclusions
Question 1:Question 1:
Regional monitoring.Regional monitoring.
Question 2:Question 2:
grassland discrimination?grassland discrimination?
Pure ray-grass not yet harvested
Just mowed
Just pastured
Regrown
38. Airborne Imaging
Spectroscopy Workshop –
Bruges, October, 8 2004
ConclusionsConclusions
These study assess the ability of Imaging Spectroscopy to be reliableThese study assess the ability of Imaging Spectroscopy to be reliable
method for estimating grassland management practices and to control ifmethod for estimating grassland management practices and to control if
AEM are correctly applied.AEM are correctly applied.
3 vegetation indices (PRI, NDVI and WBI13 vegetation indices (PRI, NDVI and WBI1 ) are confirmed as goodare confirmed as good
quantitative parametersquantitative parameters
Textural indices (qualitative parameters) and vegetation indicesTextural indices (qualitative parameters) and vegetation indices
(quantitative parameters ) are linked to the age and the management(quantitative parameters ) are linked to the age and the management
practices supported by grasslandspractices supported by grasslands
Changes of spectral signature depend not only on the nature and theChanges of spectral signature depend not only on the nature and the
intensity of the action, but also on the time spent between the action andintensity of the action, but also on the time spent between the action and
the remote sensing data acquisition.the remote sensing data acquisition.
Classification results are promising and must be validated with betterClassification results are promising and must be validated with better
images quality due to the bad meteorological conditions.images quality due to the bad meteorological conditions.
Editor's Notes
Currently available techniques for non destructive herbage measurements have a limited accuracy and in the absence of fast and automatic means to monitor grassland, the quality of grassland management strongly depends on visual judgments (systematic ground observation campaigns ) and laboratory analyses (to determine grass quality e.g. proteins or cellulose content).
Unfortunately, this approach needs a long time to be realised and can be expensive considering the important intra- or inter-parcel variability typical for this type of agricultural production.
With the recently airborne imaging spectroscopy development, a new method for non-destructive grassland characterization can be explored.
With the CAP reform, the European Agriculture should become more competitive and at the same time environmental friendly.
AEM are now compulsory and consolidates in many rural development plans in EU. In the Walloon Region, AEM are the subject to specific regulations, and differents constraints (Temporal, Spatial, Technical) have been defined.
This study is focused on the temporal constraints and evaluates the opportunity of using airborne imaging spectroscopy to calculate different indices to trace the recent history of grasslands and headlands in terms of managements.
The 2003 campaign is:
On the one hand a continuum and a possible validation of the first flight campaign using CASI-2 (VIS/NIR) and SASI (SWIR) sensors performed in August 2002 on an other Belgian site (Lorraine test site).
This previous study had determined that the good spectral resolution of these sensors allows to estimate the quantity (e.g. wet matter, biomass, grass height) and the quality (e.g. protein, VEM, DVE) of grass canopy and so to establish regional inventories on grass production/quality potentials.
These first results were very promising and this new campaign were an opportunity to validate these results in different conditions (different places, different moments)
On the other hand, this study is innovative and tries to enlighten how imaging spectroscopy are capable to facilitate monitoring and assessment of policies and AEM.
The study area of this project is the Municipality of Attert, located in the southest part of Belgium (with center coordinates).
Attert is a representative grassland region where environmental aspects are important as it is located in a Natrura 2000 area.
In the year 2002, three quaters of the declared parcels (about 600) are grasslands and the total acreage of grasslands represent two thirds of the total agricultural area.
33 parcels were followed :
Ground cover was scored visually from the middle of May to the airborne flight
In addition to these observations, ground measurements were realized simultaneously with the airborne flight.
A subset of 17 parcels were selected, and 4 sampling units were defined per parcel and georeferenced with a dGPS.
Ground measurements were performed simultaneously with the hyperspectral airborne flight.
They included a measure of:
- grass height,
- grass biomass,
- floristic composition and
- management or cropping techniques used (pasture, meadows, date of cutting, etc.).
Two types of samples were considered, depending on the cutting technique used:
- a harvest of the whole canopy with a mower and
- a harvest of the upper part of the canopy with scissors.
Additionnal ground measurements were performed with field spectroradiometers.
For each dates management practices were observed that allow to identify an Observed Management Class (MCo) for each parcel.
In total 3 parcels with AEM (MCo=NP NF) were present on the study area with no cut and no grazing authorized before 30/ 06.
12 parcels (MCo=P) were grazing land.
The flight has been scheduled during the last two weeks of June and the beginning of July. This period coincides with the cutting and grazing periods and the control of the agri-environmental constraints.
Priority was given to the use of spectral bands from the CASI sensor. However, the project required high spatial resolution therefore data were acquired with a minimum of 2.4m by 2.4m spatial resolution along the nadir line. This resolution can indeed be considered as adequate to spectrally identify homogeneity and/or heterogeneity per pixel and per parcel.
A total of 96 spectral bands with approximately 6 nm spectral sampling interval in the region from 400 to 900 nm and the ATM thermal radiation emitted by the earth surface in one spectral TIR band were recorded.
Level IIb = radiometrically and geometrically corrected.
The project has checked the signal and geometric quality of the images
We compare field spectrometer measures with spectral response coming from images and field measurements and GCP were used to ensure the sampling unit were located as accurately as possible on the imagery.
To obtain representative spectral response for each sampling unit, the CASI-SASI pixel responses were averaged within a 3x3 pixel subset, centrered over the sampling unit location.
Level IIb = radiometrically and geometrically corrected.
The project has checked the radiometic and geometric quality of the hyperspectral images
We compare field spectrometer measures with spectral response coming from images and
Field maps and GCP were used to ensure the sampling unit were located as accurately as possible on the imagery.
To obtain representative spectral response for each sampling unit, the CASI-SASI pixel responses were averaged within a 3x3 pixel subset, centrered over the sampling unit location.
Samples from each elementary sampling unit were also analysed to determine chemical characteristics of grass. They concern, dry matter content, proteins content, cellulose content, ashes content, sugar content and energy values (VEM, VEVI, DVE and OEB). NIR Spectrometry laboratory-based measurements (on fresh samples havested with scissors) were realised in the same wavelengths as airborne sensors.
To determine best hyperspectral wavebands for the study of grasslands over the spectral range of 400-1300 nm, and
To assess grasslands classification accuracies achievable using various combinations of hyperspectral narrow wavebands
The methodology is illustrated by this figure. We have considered two approaches
- With spectra indices (typical multispectral analysis)
- With narraw-waveband du to the spectral resolution
This two approaches were realised on pretraited data with a moving average of reflectance and the first derivative calcul
CASI:
Most of the biochimical characteristic are closely linked with narrowbands located in the green-edge domain
On the other hands, physical characteristics as grass height & wet matter are more related with red-ege and NIR domain
The structure of the chlorophyll red-edge was best represented by calculating the first derivative of the reflectance spectra with respect to wavelength
SASI:
- Only reflectance are significative except for the derivative computed in the range of 1200-1300 nm
In addition to the standard CASI-SASI channels a number of channel ratios and normalised channel difference indices were developed.
It was decided to limit the investigation to published vegetation indices
In addition a « red-edge » transformation was calculated from the bands defining the begining and the end of the red-edge of the grass vegetation response.
Combination of reflectance, spectral indices and first derivative results
Scatterplots of measured grassland characteristics versus calculated vegetation indices
- Best results were found using the red-edge step (REST) to estimate height and green weight of grassland canopy
- the water balance index (WBI2) for dry matter content.
Different textural indices were calculated depending on the type of approach considered.
In a global approach, all the pixels of a parcel are analysed simultaneously while in a local one the analysis is made with a moving windows of 3x3 pixels. In the first case, the texture parameter is simply the variance of all the pixels reflectances for each narrow band and in the second case, this parameter is calculated from moving windows determining the variance of the 3x3 pixels windows which are then aggregated over all the parcel to produce a local variance (intra window variability) which is used in complement of the total or global variance.
As the number of bands available for analysis is important with hyperspectral data, many of them are highly correlated and provide redundant information. Band selection refers here to the use of a subset of the most important wavelengths for identification of grassland management practices. In a first step, we used a discriminant analysis to identify regions of interest in the reflectance spectra and to choose the relevant vegetation indices.
For the global texture parameter, two regions of interest in the reflectance spectra are clearly identified. The first one is located below 500 nm and the second one above 725 nm. The response obtained for the local texture is quite different and presents a minimum value of the probability around 500 nm and above 750 nm but the observed level is not significant (P&lt;0.05). Based on these results, two wave bands from the global texture analysis have been selected (450 nm and 725 nm) together with previously defined indices to proceed in a second step to a discriminant analysis with all the selected dependent variables.
Even with hyperspectral images characterised by a poor quality, these results are very good and promising
It is satisfactory for a first screening and identification of the parcels which have to be controlled in the field, minimising by this way the cost of the field campaign.
Red-edge step and WBI2 where also successfully used to discriminate between grassland management types and different canopy structures.
Chemical characteristic of grass was not clearly linked to vegetation indices, except for energy values VEM and VEVI which seem well correlated with NDVI (R²=0.60 )
In most of cases results from scissors cutting samples which represent the upper part of the canopy were best correlated with vegetation indices.
In general the CASI sensor gives better results than the SASI sensor.
Except for 3 cases:
Total dry matter production (called biomass)
The Dry matter content
The Wet matter content
The combination of the 2 sensors doesn’t give a better prediction
The prediction level est good in general except for cellulose and total dry matter content
The relative RMSE is lower for the scissors samples so the prediction is better with the upper part of the canopy
For the most parameters, the Relative RMSE is lower than 10 % and show a godd quality of prediction
Except for sugar and one energy value (OEB)
The results of this study showed the potentialities of hyperspectral imagery for the characterization of grasslands.
The quality of relations between physico-chemical parameters and various spectral components (eg. reflectance curve, first derivative, spectral indices) allows to estimate the quality of grass canopy and so to establish regional inventories on grass production potentialities, which constitute a potential decision tool for local agencies.
They allow also to establish a discrimination between the various types of meadows (pasture, mowed meadows, etc.).
And to ….
If we consider the parcel limited by the blue lines, we can clearly see thatthe farmer has devided it in 2 parts
The lower part with actual pasture
The upper one for a next pasture which would occur later in the season
Finally, this study also showed the interest of data collected with the CASI sensor compared with the SASI sensor. Qualitative differences observed in results can be explained by the geometrical quality of images received by the project and the SASI sensor sensibility regard to background noises generated by environment (eg. soil, atmosphere).