Altitude registration analysis of limb scattering observations

1
Altitude registration analysis of limb scattering
observations

• Ghassan Taha and Glen Jaross
• Science Systems and Applications Inc.
• Didier F. Rault
• NASA Langley Research Center
• Maria Tzortziou
• Earth System Science Interdisciplinary Center,
  University of Maryland
• Didier Fussen and Filip Vanhellemont
• Belgian Institute for Space Aeronomy
• Richard D. McPeters
• NASA Goddard Space Flight Center

2
Outline

• Introduction and motivation
• Approach and accuracy
• Simulation results
• SAGE II vs. SAGE III occultation
• Results
• Correlative measurements
• SAGE III LP
• OSIRIS
• SCIAMACHY
• Limb radiance
• GOMOS
• SCIAMACHY
• Summary and conclusion

3
Introduction

• The main objective of this work is to help
  improving OMPS/NPP LP algorithm development in
  order to meet its target specification.
• To investigate the use of correlative
  measurements in evaluating limb scattering
  measurements pointing accuracy. Also investigate
  the accuracy and limitation of various altitude
  registration algorithms.
• In this work we will use SAGE limb, OSIRIS, and
  SCIAMCHY L2 (ozone profiles), as well as GOMOS
  limb, and SCIAMACHY L1C radiances.

4
Approach and simulation analysis
Example of single comparison

• Start with SAGE profile, then smooth it,
  introduce bias and shift it up or down to ? 1 km
• Use maximum correlation over optimal altitude
  range to detect altitude offset
• Figures are histograms of input, output and
  retrieval accuracy
• Retrieval accuracy ? 5 - 9 m

5
Simulation accuracy -2
Altitude offset introduced
Altitude offset retrieved
Summary (414 profile)
6
SAGE II vs. SAGE III altitude accuracy

• Investigate altitude differences of two solar
  occultation experiments (SAGE II III)
• Identify coincidence pairs within given criteria
• instrument altitude uncertainties are 50-100m
• Altitude differences are within 200 m
• Use result to estimate effect of atmospheric
  variability and instrument differences

Atmospheric variability and instrument
differences 200m
7
SAGE III limb altitude registration compared to
LIDAR
Mean Diff. 140 m Std. Dev. 390 m
8
SAGE III limb altitude registration compared to
ozonesondes
Mean Diff. 78 m Std. Dev. 540 m
9
Summary SAGE III limb

• SAGE III LP ozone have been compared to satellite
  (SAGE II, III occultation, POAM III, HALOE,
  GOMOS, and OSIRIS), Lidars, and ozonesondes.
• Altitude registration is accurate to within
  150 m, Standard deviation is 420 m.
• Number and location of measurements is not enough
  to investigate any seasonal or latitudinal
  dependence.

10
OSIRIS altitude registration compared to SAGE II
(June 2004)
Using OSIRIS data version 1.2, and 2.4
Mean Diff. 1.34 km Std. Dev. 310 m
11
OSIRIS altitude registration compared to SAGE III
(June 2004)
Mean Diff. 1.46 km Std. Dev. 290 m
12
OSIRIS altitude offset during (2004)

• Comparison with SAGE II and III show an
  increasing altitude offset in March to mid July.
  Corrections made to ODIN orbit resulted in
  reduction of the altitude offset after July,
  where the offset is random and within 500 600
  m.
• Comparison with SAGE II and III show similar
  results. Difference could be due to number of
  coincidences and location of measurements

13
SCIAMACHY altitude registration compared to SAGE
II (May 2004)

• SAGE profiles were convolved with SCIA (v1.62)
  averaging kernels and a priori profile
• Version 1.62 is not altitude corrected

Mean Diff. 1.72 km Std. Dev. 445 m
14
SCIAMACHY altitude registration compared to SAGE
III (May 2004)
Mean Diff. 1.83 km Std. Dev. 534 m
15
SCIAMACHY altitude registration offset during 2004

• SCIAMACHY limb measurements suffer from
  systematic altitude offset
• Both SAGE II and III comparison show a systematic
  negative altitude offset of 1.7-1.8 km and 0.5 km
  standard deviation
• Latitudinal dependence of the offset is small
  compared to the standard deviation

16
SCIAMACHY new version 1.63 compared to SAGE II
Version 1.63 is altitude corrected using TRUE
17
GOMOS limb measurements

• In order to distinguish stellar light from the
  sky background, the detector is split into
  central (star), upper and, lower bands (limb).
• GOMOS pointing is presumed to be very accurate
• Bright limb measurements (small solar zenith
  angle) are ideal for this study
• We used GOMOS Limb radiances to investigate
  altitude registration techniques

18
GOMOS analysis -Approach

• Use 99 GOMOS events, where solar zenith angle is
  less than 80o, same day and within 250 km of SAGE
  II.
• A forward model is used to calculate Limb
  radiances using a nearby SAGE II ozone, aerosol,
  and NO2 profiles, as well as NCEP temperature and
  pressure.
• Altitude offset is detected using maximum
  correlation for 21 pixels around 312 (ozone knee)
  350 nm (Rayleigh) respectively
• Offset for lower vs. upper band is calculated as
  a quality check

Measurements
Forward model
19
GOMOS altitude accuracy (348-352 nm)

• Compare radiances at 348-352 nm (21) pixels
• Lower-Upper
• offset 0.05 km
• STDEV 0.075 km
• Lower band
• offset 0 to -0.035 km
• STDEV 0.35 km
• Upper band
• offset -0.03 to -0.054 km
• STDEV 0.36 km

20
GOMOS altitude accuracy (312-315 nm)

• Compare radiances at 312-315 nm
• Lower-Upper
• offset 0.01 km
• STDEV 0.018 km
• Lower band
• offset 0.61-0.7 km
• STDEV 0.35
• Upper band
• offset 0.61 -0.67 km
• STDEV 0.36

The difference when using 350 to 312 nm pixels is
600m. STDEV is 350 m
21
SCIAMACHY limb measurements

• Use only coincidences with SAGE II during January
  2004.
• Use calibrated radiances
• Altitude offset is detected using maximum
  correlation for 21 pixels around 300 (ozone knee)
  350 nm (Rayleigh) respectively

22
SCIAMACHY altitude registration accuracy January
2004
L1C 350 nm
L1C 300 nm
L2 vs. SAGE II

• The difference when using 350 to 300 nm pixels is
  450 m.
• Results of 350 nm are more consistent with those
  obtained using L2 vs. SAGE II, agreement 100 m

23
Summary and Conclusion

• If left uncorrected, pointing error could affect
  gaseous retrievals accuracy and precision, up to
  17 for 1 km. Random pointing errors often affect
  precision.
• Correlative measurements provide a very useful
  tool to analyze pointing accuracy of limb
  scattering instruments.
• Accuracy of technique is 200m, mainly caused by
  atmospheric variability and instrument
  differences.
• The maximum correlation technique appears to also
  work well with limb radiance profiles, and
  results are consistent with those obtained from
  correlative measurements.
• Altitude registration information obtained from
  300 nm (ozone knee) are systematically
  overestimating those obtained using 350 nm
  (Rayleigh), by 500-600 m

24
Acknowledgment

• The Authors would like to acknowledge all groups
  who made their data available for this work,
  which includes MCH for Payern ozonesonde, DWD
  for Hohenpeissenberg ozonesonde and lidar, AWI
  for Ny Aalesund ozonesonde, NIWA for Lauder
  ozonesonde, BoM for Macquarie Is. ozonesonde,
  CNRS for OHP lidar, JPL for Table Mountain lidar,
  KNMI for Debilt and Paramaribo ozonesonde, Eurica
  lidar and ozonesonde, Lerwick ozonesonde,
  Lindenberg ozonesonde, Legionowo ozonesonde, NASA
  GSFC and Ann Thomson for the IONS ozonesonde
  campaign which included Beltsville, Houston,
  Huntsville, Narragansett, Pellston, Sable Is. NRL
  for POAM III, NASA LaRC for SAGE II, III, and
  HALOE, ACRI/ESA for GOMOS, University of
  Saskatchewan for OSIRIS, and C. Von Savigny and
  Bremen group for L2 SCIAMACHY and Richard van
  Hees and SRON for L1b data. ENVISAT/NADIR,
  WOUDC, and NDSC, for maintaining the database of
  ozonesonde and lidar measurements.