„Atmospheric Correction“

2011/11/30
NOWPAP / PICES / WESTPAC Joint Training Course on
Remote Sensing Data Analysis
Introduction and recent progress in ocean color remote
sensing part II:
Correction of the influence of the atmosphere in
Ocean Colour RS
Roland Doerffer
Helmholtz Zentrum Geesthacht
Institute of Coastal Research
roland.doerffer@hzg.de
Space Shuttle 335 km
20060720 , NASA
„Atmospheric Correction“
•
•
Goal: to determine the water leaving radiance or water leaving radiance
reflectance from top of atmosphere (TOA) radiances or reflectances
The following quantities have to be determined:
– Path radiance of the atmosphere
– Transmittance of solar flux through the atmosphere
– Transmittance of the radiance at the bottom of atmosphere (BOA) to the
satellite
– Solar light which is reflected at the water surface
• Direct: sun glint
• After scattering in the atmosphere: reflected sky light
•
•
Normalisation of the reflectance, i.e. recomputation of the bi-directional
reflectance for a sun in zenith and nadir viewing direction
Further problem is the adjacency effect: sun light, which is reflected by a
bright target in the neighbourhood of the water surface and which is then
scattred by the atmosphere into the sensor
1
2011/11/30
Basic principles of Water Color RS
sensor
sun
air molecules
aerosols
gases
- atmospheric scattering
and absorption
suspended particles
phytoplankton pigment
gelbstoff
- reflection and refraction
- scattering and absorption by
water and its constituents
- bottom reflection
Light paths to the sensor
the satellite observes both the
ocean and the atmosphere
NASA SeaDAS Training Material
2
2011/11/30
Optical Thickness, Transmittance, Angstrom
•
•
•
•
Beam Transmittance T = Iz /I0
Extinction tau  = -log(T), T = exp(- )
T = exp(- /cos())
Aerosol Optical Thickness AOT = ae
TOA

• Angstrom coefficient:
– Alpha  = log(1 /  2) / log(1 / 2)
BOA
Solar Spectrum 1
Composition of earth atmosphere:
• 78.09% nitrogen
• 20.95% oxygen,
• 0.247% water vapor (variable)
• 0.93% argon
• 0.038% carbon dioxide
• traces of hydrogen, helium,
methane, ozone etc.
• Gases cause scattering and
absorption
• Areas with no absorption
are called windows
• Aerosols:
• Salt cristals
• Dust from deserts
• Soot
• Vulcanic ash
3
2011/11/30
Solar Spectrum 2
Atmospheric Transmission
Solar radiation at
the top of the
atmosphere and
the actual radiation at sea level
which has been
reduced due to
absorption by atmospheric gases.
The dashed curve
is a blackbody at
5900K for comparison with the solar
curve outside the
earth's
atmosphere
4
2011/11/30
Solar Flux at TOA (Thullier)
Solar flux by Thullier
2500
irradiance [W m-2 µm]
2000
1500
1000
500
0
100
200
300
400
500
600
700
800
900
wavelength [nm]
Downwelling irradiance
Downwelling irrradiance Poseidon, 05/15/98 Sta 50
1400
1300
irrradiance [W m-2 µ m-1]
1200
1100
1000
900
800
700
600
500
400
400
500
600
700
lambda [nm]
800
900
5
2011/11/30
Aerosol Optical Properties
1.80E+00
1.60E+00
1.40E+00
norm. scattering
1.20E+00
conti
mari99
strato
urba50
sun glint
1.00E+00
8.00E-01
6.00E-01
4.00E-01
2.00E-01
0.00E+00
400
450
500
550
600
650
700
750
800
850
900
wavelength
6
2011/11/30
Annual variation of the Aerosol optical depth 500nm
Frequency of occurrences for Angstrom parameter 440_870
7
2011/11/30
Cosine effect
transmission air-water
Snell‘s Law
sin  a nw

sin  w na
nw = 1.34
8
2011/11/30
Specular reflectance
Fresnel‘s Equation
for unpolarized light
1  sin 2 (   w ) tan 2 ( a   w ) 

r   2 a

2  sin  a   w  tan 2  a   w  
Remote Sensing Reflectance
For comparison with the satellite-sensed signal, it is needed
to consider the above-surface remote-sensing reflectance
which is the ratio of the upwelling radiance to the
downwelling irradiance just above the sea surface
RRS , , j, 0 + ) = Lu , , j, 0 + ) / Ed , 0 + ).
The subsurface upwelling radiance Lu(0 - ) passing through
the sea surface decreases due to reflection and refraction;
the above-surface downwelling irradiance passing through
the sea surface decreases due to reflection but it is
augmented due to internal reflection of the subsurface
upward flux from the sea surface
Lu (0 + ) = (t- /n 2  Lu (0 - );
dW
Ed (0 - ) = t+ Ed (0 + )/(1- R)×
RRS = (t- t+ /n 2  rRS /(1- R); RRS = z rRS /(1- G rRS );
dW/n2
z= t- t+ /n 2 ×; G = Q.
For nadir viewing: z  0.518, G  1.562, (Lee et al.
1998).
9
2011/11/30
Radiances at Top of Atmosphere (TOA)
45
40
35
W m-2 sr-1 µm-1
30
25
Lray
20
15
10
Laer
Lwat
5
0
400
450
500
550
600
wavelength nm
650
700
750
The composition of the Radiance Spectrum at
Top of Atmosphere
Contribution of Lwat, Laer and Lray to Ltoa
1
Relative contribution
to TOA radiance
0.9
Lray
0.8
relative contribution
0.7
water leaving
radiance: Lw
0.6
0.5
0.4
aerosol path
radiance: Laer
Laer
0.3
0.2
Rayleigh path
radiance: Lray
Lwat
0.1
0
400
450
500
550
600
wavelength nm
650
700
750
10
2011/11/30
Ocean color
the atmosphere is 80-90% of the total
top-of-atmosphere signal in bluegreen wavelengths (400-600 nm)
~1% error in instrument calibration or
atmospheric model leads to
~10% error in Lw()
NASA SeaDAS Training Material
Effects of the atmosphere
gaseous absorption (ozone, water vapor, oxygen)
Rayleigh scattering by air molecules
Mie scattering and absorption by aerosols (haze, dust, pollution)
polarization (MODIS response varies with polarization of signal)
Rayleigh (80-85% of total signal)
Aerosols (0-10% of total signal)
• small molecules compared to nm wavelength, scattering efficiency
decreases with wavelength as  -4
• particles comparable in size to the wavelength of
light, scattering is a complex function of particle size
• reason for blue skies and red sunsets
• whitens or yellows the sky
• can be accurately approximated for a given atmospheric pressure
and geometry (using a radiative transfer code)
• significantly varies and cannot be easily
approximated
NASA SeaDAS Training Material
11
2011/11/30
Spectral Attenuation of Water Constituents
3
2.5
k m-1
2
1.5
1
0.5
0
400
450
500
550
600
650
wavelength nm
700
750
800
Diffuse attenuation k
for pure water, chlorophyll (30µg/l), sus. matter (50 mg/l), gelbstoff
Bands for atmospheric correction
Ltoa over water with high SPM and gelbstoff concentration
70
MERIS FR 20030416, x=589, y=194, Elbe/Oste
radiance L [W m^^-^2 sr^-^1 µm^-^1}
60
50
40
30
20
10
0
400
450
500
550
600
650
700
wavelength [nm]
750
800
850
900
12
2011/11/30
Atmospheric correction
TOA
gas
pol
glint
whitecap
air
aerosol
td() Lw() = Lt() / tg() / fp() - TLg() - tLf() - Lr() - La()
brdf
Sun
nLw() = Lw() fb() / td0() 0 f0
But, we need aerosol to get Lw()
Lw(=NIR) ≈ 0 and can be estimated (model extrapolation
from VIS) in waters where Chl is the primary driver of Lw()
NASA SeaDAS Training Material
Magnitudes of Lw(NIR
Lw(NIR) ≠ 0 (turbid or
highly productive water)
Lw(NIR) = 0 (clear water)
NASA SeaDAS Training Material
13
2011/11/30
Aerosol determiniation in visible wavelengths
Given retrieved aerosol reflectance at two ,
and a set of aerosol models fn(,0,).
=
L
F0 · 0
a(748) & a(89)
model
a(NIR)  as(NIR)
 (748,89) =
 (,89) =
as(748)
as(89)
as()
as(89)
 (748,89)
NASA SeaDAS Training Material
Iterative correction for non-zero Lw(NIR)
(1) assume Lw(NIR) = 0
(2) compute La(NIR)
(3) compute La(VIS) from La(NIR)
(4) compute Lw(VIS)
(5) estimate Lw(NIR) from Lw(VIS) + model
(6) repeat until Lw(NIR) stops changing
iterating up to 10 times
NASA SeaDAS Training Material
14
2011/11/30
Level-2 ocean color processing
(1) determine atmospheric and surface contributions to total radiance at TOA and
subtract, iterating as needed.
(2) normalize to the condition of Sun directly overhead at 1 AU and a nonattenuating atmosphere (nLw or Rrs = nLw/F 0).
(3) apply empirical or semi-analytical algorithms to relate the spectral distribution
of nLw or Rrs to geophysical quantities.
(4) assess quality (set flags) at each step
NASA SeaDAS Training Material
Case 2 water reflectance spectra (simulations)
Case 1
High absorption
Low scattering
Clear case 2
High absorption
high scattering
15
2011/11/30
Strange Spectra producing negative reflectances
MERIS FR 20030416
0.06
RL_toa
RL_tosa
RL_path
RLw_bread
0.05
RL [sr-1]
0.04
0.03
0.02
0.01
0
400
450
500
550
600
650
700
wavelength [nm]
750
800
850
900
Main Problems
• Atmospheric
correction often not
sufficient for all 9
bands
• Partly bands 1-2
negativ, or bands
7,8,9 noisy (in case
1 water)
Algal_1
• Result:
0.003
0.002
0.001
0.000
-0.001
-0.002
-0.003
-0.004
-0.005
400
– Noise in data
– Wrong ys or chl
data
Algal_2
450
500
550
600
650
700
750
800
850
900
16
2011/11/30
Design of a Model Atmosphere
direct calc.
Model Atmosphere
Ozone variable
Rayleigh variable
Eotoa
Ltoa
stratosphere
TOA
Eotosa TOSA
Ltosa
MC calculation
Cirrus
troposphere with
fixed continental
aerosol
planetary boundary layer
variable aerosol maritime, urban
Edboa
Lboa
water with scattering
particles
BOA
Attenuation coefficients of boundary layer aerosols
extinction coefficients for urba50 and m ari99 aerosols
2.0
m ari99
urba50
c norm at 550 nm
1.5
1.0
0.5
0.0
400
450
500
550
600
650
700
750
800
850
900
wavelength nm
Attenuations coefficients of maritime and urban aerosols, normalized at 550 nm.
The maritime aerosol has an angstroem coefficient of – 0.065,
the urban aerosol of - 2.15
17
2011/11/30
Variables of the atmosphere
Aerosol
Layer [km]
Optical thickMax mean
ness at 550 nm extinction at
550 nm [km-1]
maritime (99 %
rel. hum.)
0 - 3 km
0 - 0.6
0.2
urban (45% rel.
hum.)
0 - 3 km
0 - 0.6
0.2
continental
2-12
0 – 0.6
0.165
cirrus
8-11
0 - 0.6
0.2
12-50
0 - 0.003
0.0000625
stratosphere
max. optical
thickness
0.6
Max optical thickness at 550 nm: 0.6
Wind speed 0 – 8 m/s
Angström Coefficient
Aerosol Optical Properties used for NN Training data set
Aerosol Optical Thickness (AOT) at 550 nm
18
2011/11/30
Sun glint problem: Hawai 20030705
Cross section Hawai scene
radiance_9 [mW/(m^2*sr*nm)]
200
180
160
140
120
radiance_9 [mW/(m^2*sr*nm)]
100
80
60
40
20
-160
-158
-156
-154
-152
-150
-148
0
-146
longitude (deg)
19
2011/11/30
No glint and high glint TOA reflectance spectra
M ERIS spectra for no sun glint with RL_toa band 865 < 0.004
M ERIS spectra for sun glint with RL_toa band 865 > 0.06
0.08
0.085
0.07
0.080
0.06
0.075
RL_toa [sr-1]
0.04
0.070
0.03
0.065
0.02
0.060
0.01
0.00
400
450
500
550
600
650
700
750
800
850
0.055
400
900
450
500
550
wavelength [nm ]
600
650
700
750
800
850
900
wavelength [nm ]
Simulated Rayleigh path radiance reflectance and sun glint radiance
reflectance
0.04
nadir view
0.035
sun zenith 20 deg
wind 3 m/s
0.03
Radiance refelctance [sr-1]
RL_toa [sr-1]
0.05
Rayleigh path radiance
0.025
0.02
sun glint
0.015
0.01
0.005
0
400
450
500
550
600
650
700
750
800
850
900
wavelength [nm]
20
2011/11/30
NN for atmospheric correction – 3rd version in C2R and Glint
processor
Input
Output
RLtosa
12 bands
Tau_aerosol 412, 550, 778, 865
Sun_glint ratio
a_tot, b_tot
sun zenith
view zenith
azimuth diff
Neural Network
MERIS band 1-9
Trans tosa-surface
[Opt. Wind]
Path radiance reflectance
RLw
errcode
RLw(,) =Lw (,) /Ed
Training of a neural network for atmospheric correction
Atmosphere -optical model
1
Bio -optical model
7
NNforward water
(based on Hydrolight simulations )
RLw
9
12
Transmittance
L_up
5
MC code
2
RLpath_noglint
RLpath_glint
Ed_boa
Tau_aerosole
RLpath
Ed_boa
Tau_aerosole
4
6
Optional
Polarisation
correction
8
Selection
Max sunglint
Max tau_aerosol
Min. Rlw(560)
Etc.
10
RLtosa
RLpath
Ed_boa
RLw
Tau_aerosole
Training &
Test data set
13
11
21
2011/11/30
MERIS 20070429
Standard AC
Algal_2
C2R full spek AC
Algal_2
Radiance reflectance: TOA, path, RLw
MERIS RR 20030422 pix 864 line 253 lat 54.477386 lon 7.5852623
0.06
0.05
RL_toa
RL_tosa
RL_path
RLw_bread
foam 0.00
RL [sr-1]
0.04
0.03
0.02
0.01
0
400
450
500
550
600
650
700
wavelength [nm]
750
800
850
900
22
2011/11/30
TOA
Radiance reflectance
RLw RGB
Shetland
Path radiance at 550 nm
Shetland
23
2011/11/30
C2R Chlorophyll (MERIS FR)
100 km
Shetland
MERIS full resolution: Baltic and North Sea, 20080606
Stockhom
Sun glint
Oslo
Baltic Sea
North Sea
Sun glint
Copenhagen
Spatial resolution: 300 m
Hamburg
Swath: 1200 km, 4800 pixel
24
2011/11/30
Water leaving radiance reflectance
Path radiance reflectance incl. Sun glint band 5 (560 nm)
25
2011/11/30
Water leaving radiance reflectance band 5 (560 nm)
MERIS FR, Area of Gotland, TOA RLw RGB
Stockholm
Estonia
sunglint
Lettland
Ca. 100 km
Gotland
Baltic Sea
26
2011/11/30
Water leaving radiance reflectance RGB
Gotland
Ca. 100 km
Black Sea MERIS RR May 12 2007
Azovskoe
More
Dnipro
Krim
Danube
Sun glint
Full swath
RR 1200 m
RL_toa
Istanbul
27
2011/11/30
Path radiance reflectance band 5 (560 nm)
MERIS RR 20070512 water leaving radiance reflectance
Azovskoe
More
Dnipro
Danube
Krim
Istanbul
28
2011/11/30
Total Suspended Matter
Chlorophyll distribution
29
2011/11/30
MERIS 20070429 LTOA RGB
C2R AGC Path radiance reflectance band 5
30
2011/11/30
Water leaving radiance reflectance band 5
MERIS 20070505: TOA reflectances RGB
New York
31
2011/11/30
reflectance RLpath MERIS
band 5 (560 nm)
reflectance RLw MERIS band 5
(560 nm)
32
2011/11/30
Chlorophyll
MERIS FR USA East Coast 12.6.2008, Signal depth z90
Philadelphia
Chesapeake Bay
Washington
North Atlantic
33