Transcript Operational Calibration of NOAA`s Satellites

Vicarious Calibration of
GOES Imager Visible Channels
X. Wu
With contributions from M. Weiner, N. Rao, J. Bremer, C. Dean, F. Yu, D.
Crosby, I.-L. Chang, S.-R. Chung, T. Stone, F. Sun, G. Rancic, D. Han,
K. Mitchell, Z. Li, B. J. Sohn, D. Doelling, X. Xiong, N. Zhang, and …
5th Global Space-based Inter-Calibration System (GSICS)
Joint Meeting of the Data and Research Working Group (GDWG & GRWG)
The Centre National d'Études Spatiales (CNES), Toulouse, France
9-11 February 2010
9-11 Feb 2010 Toulouse, France
GRWG-5
1
The Problem
 GOES Imager visible channel lacks onboard calibration
 Pre-launch calibration gradually becomes obsolete
9-11 Feb 2010 Toulouse, France
GRWG-5
2
Option 1
No Calibration
 Adequate for basic application
 Spatial analysis of cloud and weather system
 Inadequate for advanced applications





Quantitative applications (aerosol, insolation)
Threshold (cloud detection)
Multiple channel (NDVI)
Multiple satellite (Inter-operability)
Long-term monitoring (climate)
9-11 Feb 2010 Toulouse, France
GRWG-5
3
Option 2
Vicarious Calibration
 Choice of reference:
 Evaluation Criteria
 Stars
 Desert
 Empirical Distribution
Function (EDF)
 MODIS
 Moon
 DCC
 Rayleigh scattering
 Cloud modeling
9-11 Feb 2010 Toulouse, France







GRWG-5
Independence
Stability
Traceability
Precision
Availability
Latency
Cost
4
Presentation Outline
Algorithm
 Assumption and basic structure
 Example of product
 Advantages and limitations
Comparison of all methods
 For a common satellite over a common period
 Evaluate with the criteria
Recommendation for way forward
9-11 Feb 2010 Toulouse, France
GRWG-5
5
Star
Stable irradiance available form navigation …
9-11 Feb 2010 Toulouse, France
GRWG-5
6
Star
… but not properly processed for calibration
9-11 Feb 2010 Toulouse, France
GRWG-5
7
Star
… but not properly processed for calibration
9-11 Feb 2010 Toulouse, France
GRWG-5
8
Star
… but not properly processed for calibration
9-11 Feb 2010 Toulouse, France
GRWG-5
9
Star
Reprocessed data look good … too good
9-11 Feb 2010 Toulouse, France
GRWG-5
10
Star
Reprocessed data look good … too good
9-11 Feb 2010 Toulouse, France
GRWG-5
11
A star is observed at different time of
the day on different date of year
star
☼
earth
GOES
Sun
☼
earth
GOES
40-60K
9-11 Feb 2010 Toulouse, France
GRWG-5
12
Star
Degradation Detected by Star Measurements
Issues:
14.00
• Interruption of
observations
12.00
11.00
Signal (Arbitary Unit)
• Intra-annual variation in
addition to inter-annual
degradation
13.00
10.00
9.00
8.00
7.00
6.00
5.00
• Consistency among stars
4.00
1998
1999
2000
2001
2002
Days since 1995/01/01
2003
2004
GOES-08
2005
GOES-10
2006
GOES-12
• Relative (trending) only
9-11 Feb 2010 Toulouse, France
GRWG-5
13
Star
GOES-10 Tracking Star 520
14
12
Signal Intensity (arbitary unit)
10
signal
8
exclusion
Expon.
(signal)
6
4
2
y = 6E+38e-0.0434x
R2 = 0.179
0
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Time (Year)
9-11 Feb 2010 Toulouse, France
GRWG-5
14
Star
GOES-10 Tracking Star 520
14
moved to 60° West
12
Wrong Exlusion Time
Signal Intensity (arbitary unit)
10
signal
8
exclusion
Expon.
(signal)
6
4
2
y = 6E+38e-0.0434x
R2 = 0.179
0
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Time (Year)
9-11 Feb 2010 Toulouse, France
GRWG-5
15
Desert
9-11 Feb 2010 Toulouse, France
GRWG-5
16
Desert
9-11 Feb 2010 Toulouse, France
GRWG-5
17
Desert
Issues:
 General for all – stable TOA reflectance
•
•
•
•
Clouds
Dust/aerosols
Water vapor
BRDF
 Specific for (most) GEO – Unfavorable
viewing geometry
9-11 Feb 2010 Toulouse, France
GRWG-5
18
EDF
Empirical Distribution Function
Statistical distribution of brightness is stable over time
9-11 Feb 2010 Toulouse, France
GRWG-5
19
EDF
4*10^6
79% OF INT ENSITIES ARE LESS THAN 200
2*10^6
6*10^6
SPACE VIEW INTENSITIES
EARTH VIEW INTENSITIES
0
FREQUENCY OF OCCURENCE
8*10^6
HISTOGRAM OF VISIBLE CHANNEL VALUES INTENSITIES
GOES-10 FOR JANUARY 3, 2000
0
50
100
150
200
VISIBLE CHANNEL INTENSITIES (COUNTS)
9-11 Feb 2010 Toulouse, France
GRWG-5
20
EDF
9-11 Feb 2010 Toulouse, France
GRWG-5
21
EDF
0.98
0.92
0.94
0.96
469=INTENSITY VALUE WITH
5 MILLION COUNTS ABOVE
EDF=.978
0.90
NORMALIZED CUMULATIVE FREQUENCY
1.00
UPPER EDF OF VISIBLE CHANNEL INTENSITIES
GOES-10 FOR JANUARY 3, 2000
400
600
800
1000
VISIBLE CHANNEL INTENSITIES
9-11 Feb 2010 Toulouse, France
GRWG-5
22
EDF
400
450
FIT WITH ANNUAL AND SEMI-ANNUAL TERMS
FIT WITH SOLAR CORRECTION
350
Y(t) = INTENSITY WITH 5 MILLION COUNTS ABOVE
PLOT OF INTENSITIES WITH 5 MILLION COUNTS ABOVE
FOR GOES-10
2000
2001
2002
2003
TIME (YEAR)
9-11 Feb 2010 Toulouse, France
GRWG-5
23
EDF
Issues:
 How to account for solar zenith angle
 Assumption is harder to justify, which is
equivalent to stable amount and
reflectance of DCC on daily basis
 Replaced by DCC.
9-11 Feb 2010 Toulouse, France
GRWG-5
24
MODIS
 Method
 SGP4 model for orbit prediction
 GOES/MODIS pixels collocated in space, time and viewing
geometry
• Geological collocation, < 1 pixel (1 km)
• Time difference < 10 minutes
• Nadir and near nadir satellite viewing angle (VZN < 8o)
 Match the empirical cumulative distribution functions (EDF) for the
collocated GOES and MODIS-simulated observations
M is the minimum mismatch between MODIS-simulated
EDF(EDFM) and GOES EDF (EDFG). ρcld is the threshold for
cloud pixel reflectance ρ. C is the correction
 Degradation rate calculation
• C = α*exp(βt)
EDF match method to estimate
the GOES correction to MODIS refl.
9-11 Feb 2010 Toulouse, France
GRWG-5
25
MODIS
GOES Imager
Vis.
MODIS
Level1b
GOES Vis.
SRF
MODIS
Band 1 SRF
SGP4 model
Satellite VZN
<8.1o
Time diff < 10 m
Spectral radiance
relationship
Collocation
ρcld
Collocated GOES image
Collocated MODIS image
Match GOES and
MODIS EDFs
Relationship between simulated
GOES and MODIS spectral radiance
Time-series of GOES corrections
Degradation rate
9-11 Feb 2010 Toulouse, France
Wu, X. and F. Sun, 2005. Post-launch calibration of GOES
Imager visible channel using MODIS. Proc. SPIE, Vol.
5882, doi:10.1117/12.615401
GRWG-5
26
MODIS
MODIS is perfectly calibrated
Match data are available that are:
 Co-located in space
 Concurrent in time
 Identical spatial and
spectral coverage
 Identical view geometry
9-11 Feb 2010 Toulouse, France
GRWG-5
Operational Navigation
Within 10 minutes
MODIS Ch. 1 @ 1KM
Within ~8° from nadir
27
MODIS
 36 Spectral Bands
 20 reflective solar bands (RSB) in VIS, NIR, and SWIR
covering wavelengths from 0.41 to 2.2mm
 16 thermal emissive bands in MWIR and LWIR covering
wavelengths from 3.7 to 14.4mm
 3 Spatial Resolutions (nadir)
 250m (2 bands), 500m (5 bands), and 1km (29 bands)
 Broad Science Applications
 Nearly 40 science data products for studies of the Earth’s
system of land, oceans, and atmosphere
 On-board NASA EOS Terra and Aqua Spacecraft
 Terra launched in December 1999 and Aqua in May 2002
 MODIS follow-on: VIIRS
9-11 Feb 2010 Toulouse, France
GRWG-5
Xiong
28
MODIS Scan Cavity and On-Board
Calibrators (OBC)
SD
SDSM
9-11 Feb 2010 Toulouse, France
BB
GRWG-5
SRCA
SV (moon)
29
MODIS Specifications and Applications
Primary Use
Land/Cloud/Aerosols
Boundaries
Land/Cloud/Aerosols
Properties
Ocean Color/
Phytoplankton/
Biogeochemistry
Band
Bandwidth
(nm)
Spectral
Radiance 1
Required
SNR
1
620 - 670
21.8
128
2
841 - 876
24.7
201
3
459 - 479
35.3
243
4
545 - 565
29
228
5
1230 - 1250
5.4
74
6
1628 - 1652
7.3
275
7
2105 - 2155
1
110
Primary Use
Band Bandwidth (mm)
Surface/Cloud
Temperature
Atmospheric
Temperature
Cirrus Clouds Water
Vapor
Spectral
Radiance 1
Required
NEDT(K)
20
3.660 - 3.840
0.45 (300K)
0.05
21
3.929 - 3.989
2.38 (335K)
0.2
22
3.929 - 3.989
0.67 (300K)
0.07
23
4.020 - 4.080
0.79 (300K)
0.07
24
4.433 - 4.498
0.17 (250K)
0.25
25
4.482 - 4.549
0.59 (275K)
0.25
26
1.360 - 1.390
6
150 (SNR)
27
6.535 - 6.895
1.16 (240K)
0.25
28
7.175 - 7.475
2.18 (250K)
0.25
8
405 - 420
44.9
880
9
438 - 448
41.9
838
10
483 - 493
32.1
802
Cloud Properties
29
8.400 - 8.700
9.58 (300K)
0.05
11
526 - 536
27.9
754
Ozone
30
9.580 - 9.880
3.69 (250K)
0.25
12
546 - 556
21
750
31
10.780 - 11.280 9.55 (300K)
0.05
13
662 - 672
9.5
910
Surface/Cloud
Temperature
32
11.770 - 12.270 8.94 (300K)
0.05
14
673 - 683
8.7
1087
33
13.185 - 13.485 4.52 (260K)
0.25
15
743 - 753
10.2
586
34
13.485 - 13.785 3.76 (250K)
0.25
Cloud Top Altitude
Atmospheric Water
Vapor
–
–
16
862 - 877
6.2
516
35
13.785 - 14.085 3.11 (240K)
0.25
17
890 - 920
10
167
36
14.085 - 14.385 2.08 (220K)
0.35
18
931 - 941
3.6
57
19
915 - 965
15
250
1
Spectral Radiance values are (W/m 2-µm-sr)
20 reflective solar bands (RSB: bands 1-19, and 26) from 0.41 - 2.2mm
16 thermal emissive bands (TEB: bands 20-25, 27-36) from 3.5 - 14.4mm
9-11 Feb 2010 Toulouse, France
GRWG-5
30
MODIS Calibration and Characterization
 On-board Calibrators
 SD and SDSM for reflective solar bands (RSB) calibration
• Weekly (year 1), bi-weekly (year 2-4), tri-weekly (after year 4)
 BB for thermal emissive bands (TEB) calibration
• Normally set at 290K for Terra MODIS and 285K for Aqua MODIS
• Quarterly warm-up and cool-down (WUCD) from 272K to 315K
 SRCA for spectral and spatial characterization
• Monthly for radiometric mode
• Bi-monthly for spatial mode (quarterly after year 4)
• Quarterly for spectral mode (semi-annually after year 3)
 Moon and Ground Targets
 Nearly monthly lunar observations (through space view port) at same phase angles
for VIS/NIR stability monitoring
 Dome C and desert test sites for both TEB and RSB calibration validation and
inter-comparison
9-11 Feb 2010 Toulouse, France
GRWG-5
31
MODIS Reflective Solar Bands Calibration
 Calibration Requirements
 ±2% for reflectance factors and ±5% for radiances (at typical
scene radiances (Ltyp) within a ±45º scan angle range)
 1% additional uncertainty for scenes from 0.3Ltyp to 0.9Lmax
 Calibration Algorithm
 Reflectance based calibration
 A simple linear approach
 Sensor response corrected for instrument temperature,
background, and response versus scan angle (RVS)
 Calibration Parameters
 Combination of pre-launch and on-orbit characterizations
9-11 Feb 2010 Toulouse, France
GRWG-5
32
Calibration Algorithm
EV Reflectance
m1 
2
EV  cos EV   m1  dn*EV  d EarthSun
BRFS D cos S D
2
 dn*S D  dEa
rth S un
 S D   S D
Solar
Diffuser
 SD 
dcSD
dcSun
SRCA
SDSM
SD: SD degradation factor;
SD: SD screen vignetting function
d: Earth-Sun distance
dn*: Corrected digital number
dc: Digital count of SDSM
9-11 Feb 2010 Toulouse, France
Blackbody
Scan
Mirror
GRWG-5
Space
View
33
Page 33
Calibration Parameters
2
EV  cos EV   m1  dn*EV  d EarthSun
Pre-launch
 Solar diffuser BRF
 Instrument temperature effect
 Response versus Scan Angle (RVS)
On-orbit
 Solar diffuser BRF degradation
 Changes in responses at different AOI
• SD observations
• Lunar observations
9-11 Feb 2010 Toulouse, France
GRWG-5
34
Page 34
MODIS RSB Calibration
Traceability
Pre-launch (3 key parameters)
 SD BRF
• Characterized with traceability to NIST
reflectance scale (a scale transfer process)
 Instrument Temperature Effect
• Characterized at 3 instrument temperature
plateaus
 Response versus Scan Angle (RVS)
• Characterized over a number of scan angle
(relative measurements)
On-orbit
 SD BRF degradation
9-11 Feb 2010 Toulouse, France
GRWG-5
35
Page 35
BRF Characterization
NIST Standard
Reference
Lab Secondary
Standard
MODIS SD BRF
Error Sources
1 NIST reference:
2 SBRS scattering goniometer:
3 NIST BRF scale to MODIS SD reference:
4 MODIS SD characterization:
5 SD spatial non-uniformities:
6 Interpolation angular / spectrally:
7 Pre-launch to on-orbit SD BRF change:
8 SD screen (SDS):
9 SDSM and SDS impact:
10 Solar illumination of the SD surrounds
11 Earthshine through the SD door
12 Earthshine through nadir aperture door
RSS
SBRS
0.50
0.70
0.50
0.50
0.70
0.10
0.50
0.20
0.50
0.30
0.30
0.10
1.57
MCST (I)
0.50
0.70
0.50
0.50
0.35
0.10
0.50
0.50
0.00
%1.37
SD BRF characterized at limited wavelengths, angles, and panel locations
GRWG-5
SBRS: Santa Barbara
Remote Sensing
9-11 Feb 2010 Toulouse, France
36
Page 36
Temperature Correction
Coefficients
Spectral Integration Sphere (SIS) used for Rad. calibration
Rad. Calibration
at Cold Plateau
+
Calibration Impact
9-11 Feb 2010 Toulouse, France
Rad. Calibration
at Nominal Plateau
+
Rad. Calibration
at Hot Plateau
dn* = dn·[1+kInst·(TInst-TRef)] / RVS
GRWG-5
37
Page 37
Response Versus Scan Angle
SIS at fixed levels used for RVS characterization (sensor on a rotating table)
Terra MODIS
Aqua MODIS
VIS
VIS
NIR
NIR
1354 frames: scan angle range of
Calibration
9-11 Feb
2010 Toulouse, Impact
France
±55º
dn* =GRWG-5
dn·[1+kInst·(TInst-TRef)] / RVS
38
Page 38
MODIS
9-11 Feb 2010 Toulouse, France
GRWG-5
39
MODIS
9-11 Feb 2010 Toulouse, France
GRWG-5
40
MODIS
9-11 Feb 2010 Toulouse, France
GRWG-5
41
MODIS
9-11 Feb 2010 Toulouse, France
GRWG-5
42
MODIS
9-11 Feb 2010 Toulouse, France
GRWG-5
43
MODIS
GOES is brighter than MODIS for dark scenes
9-11 Feb 2010 Toulouse, France
GRWG-5
44
MODIS
9-11 Feb 2010 Toulouse, France
GRWG-5
45
MODIS
9-11 Feb 2010 Toulouse, France
GRWG-5
46
MODIS
Post-Launch
m post  C t m pre
9-11 Feb 2010 Toulouse, France
GRWG-5
47
MODIS
C t   a exp bt 
R post  C t R pre
9-11 Feb 2010 Toulouse, France
GRWG-5
GOES-10
GOES-12
m (count-toradiance)
0.5582154
0.577103
κ (radiance-toreflectance)
1.98808 x 10-3
1.97658 x 10-3
a
1.2248
1.0875
b
0.04389
0.04890
Starting Date
2000/01/01
2003/04/01
48