1 23/07/2013 1 Optical/Thermal: Principles &Applications Jose F. Moreno University of Valencia, Spain [email protected]Lecture D1T2 – 1 July 2013 OPTICAL PRINCIPLES AND APPLICATIONS Information content of optical data: retrievable information Forward modelling of surface reflectance: soil, leaf and canopy models Pre-processing aspects Information retrieval techniques, validation and scaling issues Data usage as inputs to models and applications Perspectives
48
Embed
Optical/Thermal: Principles& Applications- Angular signatures - Spatial signatures - Temporal signatures - Other signatures (i.e., fluore scence, polarization, etc.) What we measure
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.
Information content of optical data: retrievable information
Forward modelling of surface reflectance: soil, leaf and canopy models
Pre-processing aspects
Information retrieval techniques, validation and scaling issues
Data usage as inputs to models and applications
Perspectives
2
Understanding theactual information content
of the optical data:forward modelling
of the signal
1
3
OPTICAL SYSTEMS:- Panchromatic:
- very high spatial resolution (broadband)
- Multispectral: - “colour” imaging
- Hyperspectral:- chemical composition
- Multi-angular:- structure
Many systems availableas a function of spatial, temporal and spectral resolutions
4
0
10
20
30
40
50
60
70
0 2000 4000 6000 8000 10000 12000 14000
Wavelength (nm)
Sola
r R
adia
nce
(µW
/cm
2 /nm
/sr)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Atmosphere
Solar 1.0 Reflectance
Earth 300 K, 1.0 Emisivity
Earth R
adiance (µW
/cm2/nm
/sr)Available Signal
Signatures of natural targets:
- Spectral signatures
- Angular signatures
- Spatial signatures
- Temporal signatures
- Other signatures (i.e., fluorescence, polarization, etc.)
What we measure is always radiance, either reflected and / or emitted by the land surface, which variations depend on the optical properties of land targets(and illumination conditions)
5
1 0-2
1 0-1
1 00
1 01
1 02
1 03
1 04
1 05
1 07
1 06
(m)
1 00
1 01
1 0- 1
Re
al (
n)
75342
1 TM 6TM
PLC
WATER SCATTERING
1 0-2
1 0-1
1 00
1 01
1 02
1 03
1 04
1 05
1 07
1 06
1 01
1 00
1 0-1
1 0-2
1 0- 3
1 0-4
1 0-5
1 0-6
1 0-7
1 0-8
Ima
g (
n)
(m)
TM 6
75
1
WATER ABSORPTION
PL
C
2
3
4
TM
Optical properties of elementary constituents determinethe spectral reflectance of land elements
6
SCATTERING BY VEGETATION MATERIAL
High modelling variability !
Are all pigmentsseparable in the signal ?
Key issues:- existing model parameterisations do not account for the observed variability
- high variability set limits to the possible decomposition of effects due to different pigments
Signal composed by multiple contributions(soil+vegetation)
Multiple scattering effectsplay a major role
SURFACE MODEL PARAMETERISATION:(a) Leaf inputs:
- Leaf effective thickness - Leaf water content- Total leaf chlorophyll (a+b) - Specific leaf weight- Ratio Ca/Cb - Leaf cellulose content- Fraction of Ca in LHCP - Leaf lignin content- Leaf carotenes content
MODEL INVERSION STRATEGIESThe problem of model inversion can be considered from different perspectives:
(a) Root finding of a given function
(b) Solving non-linear set of equations
(c) Function minimisation
(d) Non-linear least-squares modeling of data
Root finding and solving non-linear set of equations would require that the function is “exact”, and for this reason function minimisation is normally preferred.
p
t
pmodmes
t
modmes VVCVVVRRWVRR 112 )()(
a priori Covariance Matrix
Residuals
a priori Covariance Matrix
Observations Model variables
Merit function:Incorporation of the uncertainties in the inversion process
Residuals ResidualsResiduals
Use of constrained minimization procedures that guarantee the minimal variation of model variables to produce the same output, and a robust initialization procedure of such variables (consistency even if model has global bias).
31
THE SHAPE OF THE MERIT FUNCTION
Absolute minimum at first guess (most probable value)
(location of minimum is variable)
X0
Xmin X max
maximum range of possible values
Ymin
Yref
Ymax
maximum range of probable values
Xminref Xmax
ref
f (X)
non valid solutions
valid solution
absolute minimum
relative minimum
Neural networkmethods
Training becomes the critical issue
32
Spectral fitting methods are especially useful because we can use the well-known shape of spectral features.
Usage of the derived information:- Tendency: from proxies to quantitative information- Multi-resolution spatial inputs and time series
- First approach: Land cover mapping, classification and tables of biophysical variables assigned to each class
- Second approach: Retrievals of biophysical variables as direct inputs to models
- Third approach: direct assimilation ofradiances/reflectances into models
- Mapping Applications - cartography - thematic mapping - Monitoring Applications - ecosystems dynamics - natural hazards (fires, floods, desertification) - Research about Land Surface Processes - heat and mass exchange at Land/Atmosphere interface - photosynthesis and net primary production - hydrologic processes - Land/Atmosphere exchange of biochemicals
REMOTE SENSING OF LAND SURFACE PROCESSES
}
37
SPOT-5 / 2005
MOSAICS
38
INTEGRATION IN A GIS ENVIRONMENT
39
2003 growingSeasonBarrax
Validation 15/07/2003 LANDSAT LAI
0.0
1.0
2.0
3.0
4.0
5.0
6.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
LANDSAT retrieved
Gro
un
d m
eas
ure
men
ts
ALFALFA
CORN
PAPAVER
POTATO
SUGARBEET
GARLIC
ONION
CORNALFALFA
MULTITEMPORAL SERIES
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
0
0.001
0.002
0.003
0.004
400 450 500 550 600 650 700 750 800 850 900
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
CASI-2
MERIS
TM
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
0
0.001
0.002
0.003
0.004
400 450 500 550 600 650 700 750 800 850 900
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
CASI-2
MERIS
TM
9
6
3
Re
fle
cta
nc e
(%
)
Wavelength (nm)
12
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
0
0.001
0.002
0.003
0.004
400 450 500 550 600 650 700 750 800 850 900
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
CASI-2
MERIS
TM
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
0
0.001
0.002
0.003
0.004
400 450 500 550 600 650 700 750 800 850 900
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
CASI-2
MERIS
TM
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
0
0.001
0.002
0.003
0.004
400 450 500 550 600 650 700 750 800 850 900
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
CASI-2
MERIS
TM
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
0.005
400 500 600 700 800 900
Longitud de onda (nm)
Ref
lect
anci
a (R
rs, s
r-1)
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
0
0.001
0.002
0.003
0.004
400 450 500 550 600 650 700 750 800 850 900
0
0.001
0.002
0.003
0.004
400 500 600 700 800 900
CASI-2
MERIS
TM
9
6
3
Re
fle
cta
nc e
(%
)
Wavelength (nm)
12
Optical signature of water targets
40
Visible bands
CHRIS/PROBA IMAGERY OVER ALBUFERA DE VALENCIA LAKE
New capabilities for fires monitoring with upcoming sensors
– Bi-temporal and multitemporal detectors
Change detection applications
43
Las Vegas urban development
Athens, PROBA image
44
THERMAL INFRAREDAPPLICATIONS
wat
er c
ycle
carb
on
cyc
le
45
RESOLVED SPATIAL SCALES
global<1 - 300meters Local site Global sampling
severalyears
fewdays
Requires very large time series
RESOLVED TIME SCALES RESOLVED
46
from local measurements to global models
NEW GENERATION OF SENSORS
- Well calibrated (more suitable for multitemporal studies)
- Increased spatial resolution (0.5 m PAN now available)
- Increased spatial coverage (global mapping in highspatial resolution (as ESA GMES/Sentinel-2)
- New type of information (i.e., vegetation fluorescence)
- Time series: gap filling using multi-sensor data, bettertemporal resolution with high spatial resolution
- Integration of multi-resolution data with diverse spectral information in common temporal databases
47
Sen
tin
el-2
Sen
tin
el-3
SE
NT
INE
L-3
SE
NT
INE
L-2
PERSPECTIVES IN DATA EXPLOITATION
- Adequate exploitation of the different data sources: multi-source (multi- resolution data integration).
- Focus on systematic data assimilation approaches exploiting the time-series concept and synergy among simultaneously available satellite systems
- Consistent incorporation in the modelling approaches of processes covering time scales from weeks to decades and exploiting spatially distributed inputs
- Accounting for spatial variability and temporal dynamics as the main contributions