Lessons Learned from EO-1 Calibration · PDF fileLessons Learned from EO-1 Calibration...
Transcript of Lessons Learned from EO-1 Calibration · PDF fileLessons Learned from EO-1 Calibration...
National Aeronautics and Space Administration
Lessons Learned from EO-1 Calibration Activities
Advanced Land Imager (ALI) band misregistration impact on spectral indices
Radiometric stability of lunar test sites based on Hyperion observations
Ungar, Stephen a,b, Ong, Lawrence c, and Thome, Kurtis a
a NASA/Goddard Space Flight Center (GSFC), Greenbelt, MD 20771, USA b Universities Space Research Assoc. (USRA), Columbia, MD 21044, USA c Science Systems and Applications, Inc. (SSAI), Lanham, MD 20706, USA
IGARSS 2014 Quebec, Canada July 16,2014
CONCLUSIONS! • Push broom observing systems of the ALI variety can not
produce inherently (band-to band) co-aligned measurements. This is not a design flaw, but rather a “feature” designed to produce higher SNR.
• This “”feature” does not necessarily hamper uses of the data in a variety of applications. However, it does introduce significant uncertainty in determining vegetation indices.
• The moon serves as a solar diffuser monitor for several orbiting missions. The EO-1 CalVal Team, in collaboration with EO-1 Operations, is refining strategies to use specific “vicarious calibration sites” on the lunar surface, enhancing use of orbiting imaging spectrometer missions to serve as transfer radiometers for other passive optical missions.
Lets cut to the chase and get directly to the bottom line
Wavelength
Cross Track Sample (Pixel)
Depiction -Grids represent the detectors
-Spots represent the IFOV centers
-Colors represent the wavelengths
Pushbroom Observing System Sensor Chip Array (SCA) “cartoon”
Pushbroom Observing System
Pushbroom systems come in two flavors
Pushbroom Observing System
Areas viewed simultaneously by each band
Pushbroom Observing System
Areas viewed simultaneously by each band
Pushbroom Observing System
Areas viewed simultaneously by each band
Pushbroom Observing System
Areas viewed sequentially by each band
Pushbroom Observing System
Areas viewed sequentially by each band
Pushbroom Observing System
Areas viewed sequentially by each band
Pushbroom Observing System
Areas viewed sequentially by each band
Pushbroom Observing System
Areas viewed sequentially by each band
E
S
E
Adjusting Attitude and Frame Rate to Ensure Band-to-Band Co-registration
• The ground sampling distance (GSD) is a function of the sub-satellite “ground” speed and detector sampling rate.
• Pixel “size” is a function of the detector angular field of view (IFOV), integration time, range (distance) to target, and ground “velocity”.
• EO-1’s strategy to ensure inherent band-to-band registration for the Advanced Land Imager (ALI) is to: – Align the ALI sensor chip array (SCA) with the ground velocity
vector direction by yawing the spacecraft; – Adjust sampling rate, based on ground speed, such that an
integral number (N) of GSD’s exactly equal the projected ground distance between simultaneously collected bands.
How ALI Achieves Inherent Band-to-Band Coregistration
Adjusted (ALI) Sampling Rate
Fixed (L8/OLI) Sampling Rate ~236 frames/sec
Adjusting Attitude and Frame Rate to Ensure Band-to-Band Co-registration
• The EO-1 approach uses a fixed value of N which is based on maintaining a GSD equal to the nominal pixel size. However, any integer value of N ensures band-to-band co-registration.
• Lowering the value on N leads to under-sampling and decrease in data volume, while increasing N results in oversampling, increased data volume, and possible reduction in SNR.
Further Considerations
Adjusting Frame Rate to Ensure Band-to-Band Co-registration
( )
hDhf
fND
Dhff
DhRf
NDDf
hRND
hRhR
h
h
E
E
E
EE
∆∆ 3
3'0
0
0
'
0
0
'0
'
222'
10000,30
227
10,227
23,30
)(
cos
−
−=
=
≈⇒+=
≈≈
≈=
+=
+=
=
+=
++
+≈
ω
ω
ω
ω
θωωω
ecoincidenc band tillframesNseperation band frame withinD
ationsize/seper pixelor GSDDrate frame 1-EOf
velocityrotational Earthω velocityorbitalangular ω
latitude target groundθaltitude terrainΔRh
radius EarthmeanRradius orbital 1EOR
0
E
E
E
o
≡≡≡≡≡≡≡
+≡≡
−≡Oblate Earth Approximation
𝑎𝑎𝑎 ∆ω’ < 3 𝝬10-3 ω
All you need to know about spectral/spatial alignment impact on derived-parameter uncertainty in 10 minutes
Steve Ungar – NASA/GSFC Scientist Emeritus HyspIRI Science Symposium – NASA GSFC – May 4, 2010
This initial characterization of a synthetic scene, composed of two landscape components, represents a landscape- based radiometric parameterization which is independent of any specific remotely–sensed (pixel-oriented) observation strategy. This is followed by characterizations of the same scene which are based on (pixel-oriented) observation strategies.
28
Synthetic Scene Composition
0102030405060
0,4 0,6 0,8 1,0
Ref
lect
ance
(%)
Wavelength (μm)
Vegetation Bright "Soil"
.3 .3 .5 .5
.3 .3 .5 .5
.5 .3 .3 .5
.5 .3 .3 .5
NIR Reflectance
.3 .3 .1 .1
.3 .3 .1 .1
.1 .3 .3 .1
.1 .3 .3 .1
VIS Reflectance
Landscape Reflectance Values Synthetic Scene Scenario
29
30
Landscape Reflectance Ratios Synthetic Scene Scenario
NIR Reflectance (RNIR)
.3 .3 .5 .5
.3 .3 .5 .5
.5 .3 .3 .5
.5 .3 .3 .5
.3 .3 .1 .1
.3 .3 .1 .1
.1 .3 .3 .1
.1 .3 .3 .1
=
1 1 5 5
1 1 5 5
5 1 1 5
5 1 1 5
= VI RNIR
RVIS
VIS Reflectance (RVIS)
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
.1 .1 .1
.1 .1 .1
.1 .1 .1
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.1 .1 .1
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.1 .1 .1
.1 .1 .1
.1 .1 .1
31
Nominal Position VIS Band Nominal Position NIR Band
Pixel Reflectance Values Aligned Bands Scenario
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
.5 .5 ..5
.5 .5 .5
.5 .5 .5
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.3 .3 .3
.5 .5 .5
.5 .5 .5
.5 .5 .5
Nominal Position NIR Band
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.1 .2 .1
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
.3 .2 .3
“Half-pixel” Shift VIS Band
Pixel Reflectance Values Misaligned Bands Scenario
32
Results of half pixel misalignment and correction through linear re-sampling
Scenario Category 1 Ratio Value
Category 1 Discrepancy
Category 2 Ratio Value
Category 2 Discrepancy
VIS and NIR co-aligned 1.00 0% 5.00 0%
VIS and NIR misaligned 1.17 +17% 4.17 -17%
VIS realigned by resampling 1.13 +13% 3.89 -22%
33
What’s missing from this picture? How we can provide consistent CalVal across Decadal Missions!
The Moon can serve as a virtual solar diffuser monitor to validate HyspIRI solar calibration!
Well characterized test sites facilitate further validation and transfer of calibration to other Decadal Survey and International Missions !
As explained in the following slides
How EO-1 uses Lunar Images • Lunar Calibration
– Calculate Lunar spectral irradiance (EM(λ)) – Compare to the USGS Robotic Lunar
Observatory (ROLO) lunar irradiance model
• Lunar Calibration Team – Jim Butler – Brian Markham – Lawrence Ong – Kurt Thome – Steve Ungar
-- Jack Xiong
Typical Lunation (aka Lunar Cycle)
1 total lunation takes ~29.5 days
ROLO Model USGS Robotic Lunar Observatory
EO-1 Lunar Cal/Val
kM
kk
N
1iki,pk
EΩIπA
LΩIp
⋅=
= ∑=
-6.0
-1.0
4.0
9.0
14.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0
Per
cent
Diff
eren
ce fr
om R
olo
Mission Year
426.82 508.22 599.80 803.30 1205.07 1497.63 2002.06 2254.22
Quasi-annual variations not understood but appears somewhat correlated with the Sun-Moon and Spacecraft-Moon selenographic coordinates.
Hyperion Lunar Trends Comparison of Hyperion integrated lunar responses with the
USGS Robotic Lunar Observatory (ROLO) model for selected bands.
Hyperion Lunar Observations shows Spectral Variation with Phase Angles
0.8
0.9
1
1.1
1.2
1.3
1.4
400 900 1400 1900 2400
Nor
mal
ized
Res
pons
e
Wavelength [nm]
Normalized with 7.434 Deg Phase Angle-80.991 -53.679 -40.057 7.434 23.051 35.106 46.105Phase Angles:
Comparisons with RoLO at various Phase Angle
-6
-3
0
3
6
9
12
15
18
-85 -65 -45 -25 -5 15 35
Diff
eren
ce fr
om R
oLO
[%]
Lunar Phase Angle [Degrees]
447.17487.87569.27660.85793.13864.351245.361648.912213.93
Some bands show signs of phase angle dependencies, eg 569, 660, 793, 864 and 1648 nm
Summary of ROLO Comparisons
• The ROLO model provides a convenient avenue to conduct overall trending of instrument performance.
• Unable to characterize individual detectors • Quasi-periodic trends observed – under investigation • Absolute calibration?
Hyperion Lunar Observations
• Hyperion Lunar observations are potentially highly valuable for the lunar calibration of instruments on both polar-orbiting and geostationary satellites.
• Lunar spectra can be convolved with the spectral response function of any given instrument.
• The relationship between moon phase angle and spectral changes for selected channels require further investigation.
Objective: develop a satellite-based lunar calibration strategy which will serve as the basis for cross calibrating space borne passive optical observing systems.
The EO-1 Hyperion imaging spectrometer will be used to develop an exo-atmospheric spectral radiometric database for a range of lunar phase angles surrounding the fully illuminated moon.
Initial studies will include a comprehensive analysis of the existing 12 year collection of monthly (plus some additional) lunar acquisitions.
Further studies will select specific lunar surface areas, such as lunar maria, and characterize their stability in the presence of lunar nutation and libration using a newly developed observing strategy to expand the EO-1 lunar dataset to include more phase angles during the next 2 years.
Future Use of EO-1 Lunar Images Hyperion Lunar Calibration Activities
Hyperion is now being used to slowly scan the lunar surface at a rate which results in a 32X oversampling to effectively increase the SNR. Several strategies, including comparison against the USGS RObotic Lunar Observatory (ROLO) model, will be employed to estimate the absolute and relative accuracy of the measurement set.
There is an existing need to resolve discrepancies as high as 10% between ROLO and solar based calibration of current NASA EOS assets. Analysis of this dataset will lead to the development of strategies to ensure more accurate cross calibrations when employing the more capable, future imaging spectrometers.
Hyperion Lunar Calibration Activities
Full Moon
(EO-1 ALI Pan band) (Selected sites)
Hyperion views the moon monthly
USGS Lunar Map
Mare Imbrium Mare
Serenitatis Mare Crisium Mare
Tranquilitatis
Hyperion Lunar Views Oversampled
by 8X
1/01/10 4/28/10 6/27/10 12/21/10
Hyperion Lunar Views
Averaged (Aggregated) Oversampled Images
Rotated Averaged (Aggregated) Oversampled Images
1/01/10 4/28/10 6/27/10 12/21/10
1/01/10 4/28/10 6/27/10 12/21/10
Mare Tranquilitatis Cal Site
Preliminary
Date Mean Stdev2013027 35.72 2.182013086 42.32 8.352013145 59.09 11.082013204 36.24 3.302013292 39.29 2.472013322 39.94 2.10
These values need adjustment for solar and selenographic ranges, nutation, libration, etc
CONCLUSIONS! • Push broom observing systems of the ALI variety can not
produce inherently (band-to band) co-aligned measurements. This is not a design flaw, but rather a “feature” designed to produce higher SNR.
• This “”feature” does not necessarily hamper uses of the data in a variety of applications. However, it does introduce significant uncertainty in determining vegetation indices.
• The moon serves as a solar diffuser monitor for several orbiting missions. The EO-1 CalVal Team, in collaboration with EO-1 Operations, is refining strategies to use specific “vicarious calibration sites” on the lunar surface, enhancing use of orbiting imaging spectrometer missions to serve as transfer radiometers for other passive optical missions.