Transcript Retrieving Ozone from Limb Scatter Measurements on NPOESS

UV/VIS Limb Scatter Workshop- University of Bremen
April 14-16, 2003
OMPS Limb Profiler
Retrieving Ozone from Limb Scatter Measurements
Jack Larsen, Colin Seftor, Boris Petrenko, Vladimir Kondratovich
Raytheon Information Technology and Scientific Services
Dave Flittner
University of Arizona
Quinn Remund, Juan Rodriguez, Jim Leitch, Brian McComas
Ball Aerospace and Technologies Corp
Glen Jaross
Science Systems and Applications, Inc
Tom Swissler
Swissler Info Tech
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
1
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
Presentation Outline


Ozone limb scattering background
OMPS limb sensor overview
–
–

Limb algorithm overview
–
–
–
–
–

Heritage basis (SOLSE/LORE)
OMPS enhancements to SOLSE/LORE algorithm
Channel selection
Algorithm flow
Optimal estimation
Selected sensitivity studies
–
–
–

Spectral characteristics
Limb viewing geometry
Polarization
Sensor noise
Altitude registration
Conclusions
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
2
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
Ozone EDR profile requirements
Provide profiles of the volumetric concentration of ozone
Performance requirements:
Horizontal cell size : 250 km
Vertical cell size : 3 km
Horizontal coverage : global for SZAs < 80 degrees
Vertical coverage : tropopause height (or 8 km)- 60 km
Measurement range : 0.1-15 ppmv
Measurement accuracy :
tropopause - 15 km : greater of 20% and 0.1 ppmv
15 - 60 km : greater of 10% and 0.1 ppmv
Measurement precision :
tropopause height- 15 km : 10%
15 - 50 km : 3%
50 - 60 km : 10%
Long term stability : 2% over 7-year single sensor lifetime
Maximum local average revisit time : 4 days
Exceptions to EDR performance (precision and accuracy)
Ozone volume mixing ratio < 0.3ppmv
Volcanic aerosol loading - CCD saturation - optical depth
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
3
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
Limb scattering technique has improved
vertical resolution over Nadir profile products
General Description - Basis for SOLSE/LORE and OMPS
Limb Algorithms
Ozone Products

Profiling

By measuring the amount of scatter and absorption of
solar radiation through the atmosphere at different
wavelengths (e.g. UV, visible, near-infrared), profile
scattering instruments can infer the vertical profiles of
a number of trace constituents, including ozone

Limb scatter combines advantages of both BUV and
visible limb occultation methods
–
Limb viewing geometry provides good vertical
resolution
–
Measurements can be made throughout the sunlit
portion of the orbit; not restricted by sun within
FOV
UV, VIS, NIR Limb
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
4
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler


–






Sensor is based on a Prism Spectrometer
Prism spectrometer provides spectral coverage
from 290 nm to 1000 nm
Scene dynamic range accommodated with 4
gain levels:
–
April 14-16, 2003
M325 Model Atmosphere, SZA=40
290 nm
Aperture split provides two images/slit along the
vertical direction of the focal plane
Two integration times for additional discrimination
Wavelength-dependent resolution of prism
spectrometer is consistent with ozone spectral
detail over this range
Three slits provide three cross-track samples
with a single spectrometer and no moving parts
All three slit samples are included on a single
focal plane
Radiances nearly simultaneous in altitude and
wavelength
Limb radiances sampled multiple times within
38 second integration time
Calibration stability maintained on-orbit by
periodic solar observations
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
350 nm
600 nm
1000 nm
5
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
OMPS Limb Sensor Views the Limb Along the
Satellite Track
Photo from GSFC’s SOLSE/LORE Shuttle flight
OMPS limb sampling
065km
2.23
Limb
4.25
250km
Left Slit
Center Slit
Right Slit
• OMPS limb sensor has 3 slits separated by 4.25 degrees
• 38 second reporting period: 250 km along track
• 130 km (2.23 degree) vertical FOV at limb for 0-60 km coverage
plus offsets (pointing, orbital variation, Earth oblateness)
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
6
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
Radiance profiles constructed from 4 gain
level images
Simultaneous imaging of all three slits



Focal plane images as viewed from
behind CCD
Spectral and spatial smiles of ~8 pixels
Inter-image spacing of 50 pixels
(vertical) and 20-35 pixels (spectral)
High Gain
4 gain
levels
Long
Short
Image 1
Image 3
4.55
Low Gain
4.42
Image 2
4.55
Image 4
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
7
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler




April 14-16, 2003
Heritage algorithm provides strong foundation
for OMPS profile ozone retrieval
Successful shuttle flight by GSFC Code 916 demonstrates that SOLSE / LORE
retrieves ozone from space
Adapting the SOLSE / LORE algorithm developed by Ben Herman and Dave
Flittner (U. of Arizona)
Herman code (Applied Optics, v. 34, 1995)
– Multiple scattering solution in a spherical atmosphere
 Molecular and aerosol scattering
 Ozone absorption
– Includes polarization
Combines spherical multiple scattering solution with integration of source
function along line of sight
 (T )
I
Total
Limb

  Je
0
0

d
J - source function
0 - single scatter albedo
 - optical depth
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
8
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
OMPS algorithm enhancements improve
profile retrieval performance

Inverts neutral number density at 350 nm
– Eliminates need for external EDR temperature and pressure above 20 km
– Use external EDR temperature and pressure to derive density from 10 to 20 km
– If external EDR unavailable, use climatology for 10 to 20 km
Inverts aerosol at non-ozone visible wavelengths
– Simple aerosol model interpolates to ozone wavelengths
– Wavelength triplet formulation reduces effects of aerosol on ozone when
aerosol inversion cannot be performed
Solves for visible surface reflectances

Solves for cloud fraction

Multiple scattering tables include clouds at four pressure levels


Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
9
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler


April 14-16, 2003
OMPS channels selected to optimize limb
profile performance
OMPS uses the UV and visible limb scatter spectrum to measure ozone
– Middle and near-ultraviolet channels provide coverage from 28 to 60 km
– Visible channels provide coverage from tropopause to 28 km
Additional channels between 350 and 1000 nm provide characterization of
Rayleigh and aerosol scattering background
Ozone
Normalization Altitudes
Aerosol
Neutral
Number
Density
Surface
Reflectance
Cloud
Fraction
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
10
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
Limb profile algorithm flow
Recent Limb Profile
Nadir Profile
Scene Characterization
Cloud ID Scheme
Cloud Properties
Surface Properties
Cloud Fraction
O3 SDR
Im(z)
Inorm (z)=Im(z)/Im(zNorm)
Initial T, P, Density,
Aerosol, Ozone, R
Surface Reflectances
Density Inversion
Aerosol Inversion
Cloud Fraction
Reflectances
Density Profile
Aerosol Profiles
O3 N.D.
Iterated
Database
No
Yes
Convert O3 N.D. to VMR
Convergence
Criterion
O3 Inversion
(Number Density)
SOLSE/LORE Algorithm
OMPS Enhancements
O3 EDR
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
11
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
Baseline
approach
April 14-16, 2003
Scene Characterization
Cloud
Ground
Radiance multiple
scattering
component depends
on lower boundary
conditions



Terrain
UV
Cloud
Visible
N7 TOMS DB
Iterated
Reflectance
(Herman &
Solution
Celarier)
Pressure/
Altitude
CrIS
UV
Visible
0.8
0.8
VIIRS/
OMPS 1000nm
channel
Spatial variation in cloud and surface reflectivity
Radiances-weighted average (cloud fraction) of clear sky and cloud
Iterated solution for cloud fraction from 347, 353 nm channels
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
12
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler


Profile retrievals employ optimal estimation
Ozone
Kernels define sensitivity of radiances
to atmospheric constituents
Kernel shapes sharply peaked due to
limb geometry - provides high vertical
resolution
–
–

April 14-16, 2003
575 nm
290 nm
Positive kernels: scattering
Negative kernels: absorption
Optimal estimation (Rodgers, 1976)
X n1  X 0  S x K T ( KS x K T  S e ) 1[(Ym  Yn )  K ( X 0  X n )]
Density
Aerosol
347 nm
500 nm
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
675 nm
13
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler
April 14-16, 2003
Accuracy and Precision Error Terms
Precision
Accuracy
Sensitivity studies presented
Sensor
Algorithm
Pointing/Altitude
Albedo calibration
-Independent
Albedo calibration
-Dependent
Pixel-Pixel
Wavelength calibration
Rayleigh Scattering
Coefficients
Ozone Absorption Coefs,
including T dep
Aerosol correction
MS table interpolation and
retrieval error
Boresight alignment
On-orbit wavelength shifts
Polarization
Neutral number density
Non-homogeneous scene
Ephemeris knowledge, radial
Attitude reference knowledge
Straylight
Cloud top/surface pressure
Random noise
Ozone absorption coefs, T
dependence
Neutral number density
Sensor misalignment
Alignment knowledge
Structural/thermal distortion
Altitude registration
Aerosol correction
Ozone inhomogeneity-LOS
Ozone inhomogeneity-cross
track
Cloud fraction/reflectivity
Surface reflectivity
Cloud top/surface pressure
Complete error budget in ATBD-http://npoesslib.ipo.noaa.gov
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
14
UV/VIS Limb Scatter Workshop- University of Bremen
Sensor



April 14-16, 2003
Sensitivity studies find <0.1% ozone error due
to polarization effects
Broad range of
observing conditions
studied
Error for a sensor with
10% polarization
sensitivity reduced to
1.3% by depolarizer
Excess allocation for
polarization stability
reallocated to on-orbit
wavelength
calibration/stability and
pixel-to-pixel
calibration
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
15
UV/VIS Limb Scatter Workshop- University of Bremen
Sensor


April 14-16, 2003
Ozone sensor precision errors meet
allocations for most model atmospheres
TOMS V7 Standard profiles
Background volcanic aerosol (May 9, 1991 30.1N)
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
16
UV/VIS Limb Scatter Workshop- University of Bremen
Pointing Altitude Reg.
April 14-16, 2003
Parallel approaches to altitude registration
RSAS
New baseline
approach : use
S/C attitude only
for first guess
C & Sigma

Limb radiances compared to
predictions based upon
“known” neutral density profiles

Limb radiances compared to
predictions based upon “known”
ozone profiles

Information: 20 km < Z < 45 km

Information: 42 km < Z < 50 km

Registers limb profiles to a
neutral density scale

Registers limb profiles to a
pressure scale

CrIS EDR provides density vs. Z
from temperature and pressure
profiles

CrIS EDR provides pressure vs.
Z; NP provides ozone vs.
pressure
Dual approach
reduces risk
Advantage:

Low variability in Rayleigh
scatter
Disadvantage:

Sensitive to surface and lower
stratosphere

Requires 2 CrIS EDRs
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
Advantage:

Insensitive to surface and lower
stratosphere

Uses multiple NP & LP channels
Disadvantage:

Requires NP SDR radiances

Requires CrIS EDR
17
UV/VIS Limb Scatter Workshop- University of Bremen
Pointing Altitude Reg.
April 14-16, 2003
Limb profile altitude registration
algorithm flow
Z scale
from S/C
attitude
RSAS
Calibrated
LP
Radiances
2 minimization
Write Ref.
Z to SDR
Calculate LP
radiances
CrIS
D, T profiles
NP SDR
radiances
Calculate LP
radiances
C-
Peak fitting
Z scale
from S/C
attitude
Calibrated
LP
Radiances
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
Write Ref.
Z to SDR
18
UV/VIS Limb Scatter Workshop- University of Bremen
Pointing Altitude Reg.
CrIS
uncertainties
dominate RSAS
in the absence of
aerosols
Correctable
errors excluded
from total
Lunar obs. can
reduce accuracy
errors
April 14-16, 2003
Summary of C-sigma and RSAS
accuracy errors
C&
RSAS
LP OOB Stray Light
(no correction)
23 m
13 m
NP long-term drift
100 m
N/A
NP retrieval errors
~ 500 m
N/A
Aerosols (aged
volcanic)
TBD
1000 m
CrIS T = 1K
N/A
26 m
CrIS psurf = 2 mb
55 m
55 m
~ 500 m
~ 1000 m
Total (RSS)
(assuming good
MTF knowledge)
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
19
UV/VIS Limb Scatter Workshop- University of Bremen
Pointing Altitude Reg.
Geophysical
uncertainties
dominate
Ozone volume
match-up
uncertainties
have not been
quantified
TBD terms are
not expected to
be significant
April 14-16, 2003
Summary C-sigma and RSAS of precision
errors
C&
RSAS
LP SNR
0.1 m
3.3 m
NP SNR
2.5 m
N/A
~ 100 m *
N/A
Aerosol variation
TBD
165 m
Surface / ozone
inhomogeneity
TBD
< 1100 m
CrIS T (s = 1K)
N/A
26 m
CrIS p vs. Z
(ssurf = 2 mb)
55 m
55 m
~ 120 m
 1100 m
NP retrieval errors
Total (RSS)
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
20
UV/VIS Limb Scatter Workshop- University of Bremen
Limb Profiler


Polarization errors < 0.1% ozone
Ozone errors due to sensor noise meet requirements
C-Sigma selected as primary approach to altitude registration
 Precision ~ 120 m exceeds error allocation of 55m
 Accuracy ~ 500 m
 Will continue to study RSAS
 May combine both for operational use
OMPS algorithms to be tested on limb scatter observations
–

Sensor SNRs tailored to algorithm/EDR requirements
 Requirements met except for a few model atmospheres-altitude regimes
Sensor-algorithm performance verified with on-going sensitivity studies
–
–
–

OMPS Limb Profiler Summary
Unique sensor design accommodates wide dynamic range of scene
radiances and is spectrally optimized to match ozone absorption features
–

April 14-16, 2003
SAGE III, SOLSE/LORE 2, OSIRIS, SCIAMACHY, GOMOS
Engineering unit being built and tested fall-winter 2002-2003
First NPOESS flight currently planned for 2011
–
Early flight of opportunity on NPP (Launch 2006)
Use or disclosure of this information
may be subject to United States export
control laws. For official use only.
21