Intercomparison of HIRDLS and MLS Ozone Profiles Using OMI

1
Intercomparison of HIRDLS and MLS Ozone Profiles
Using OMI

• Xiong Liu1,2,3, Pawan K. Bhartia3, Kelly Chance2,
  Thomas P. Kurosu2, Robert J.D. Spurr4, Jerry
  Ziemke1,3
• xliu_at_umbc.edu
• 1Goddard Earth Sciences and Technology Center,
  UMBC
• 2Harvard-Smithsonian Center for Astrophysics
• 3NASA Goddard Space Flight Center
• 4RT Solutions Inc.
• HIRDLS Science Team Meeting 2008
• January 30, 2008

2
Outline

• Introduction
• Comparison of OMI ozone profiles with MLS and
  OMI-MLS Data Assimilation
• OMI tropospheric ozone
• Summary
• Proposed work to use HIRDLS Data

3
Introduction

• HIRDLS and MLS
• Vertical resolution 1 km and 3 km
• Typically 1500 km apart
• Difficult to do direct comparison except in the
  zonal mean sense or doing some spatial
  interpolation.
• OMI
• Much coarser vertical resolution
• Complete global coverage, overlap both HIRDLS and
  MLS
• OMI ozone profiles can be used to compare both
  HIRDLS and MLS data.

4
Introduction to OMI Ozone Profile Retrievals

• OMI ozone profiles 6-10 km vertical resolution
  Liu et al., 2005
• Retrieve O3 profiles at 24 2.5-km thick layers
  from surface to 60 km
• 270-330 nm OMI radiances, Optimal estimation
  technique
• Apply x-track and ?-dependent correction to OMI
  derived from 0-20S lat and single day of MLS/OMI
  match-up.

Mid- Latitudes
Tropics
5
Random-Noise and Smoothing Errors

• Precision lt2 above 22 km, increase to 10
  below
• SmoothingPrecision within 3 between 22-40 km,
  increase to 10 above 40 km, and to 15-30 below
  20 km.

Tropics
Mid- Latitudes
6
Comparison of OMI Ozone Profiles with MLS
2006m0711
7
Comparison of OMI Ozone Profiles with MLS
8
Comparison of MLS/OMI 0-215 hPa O3 Columns
15 days of Comparison Negative Bias
0.8-2.5 Std. Dev. (Global) 2.3-3.1 Std.
Dev. (Tropics) 1.9-2.5
9
Comparison with Data Assimilation

• How about other x-track positions?
• OMI/MLS Assimilation Stajner et al., 2007
• Mean bias 0 5 DU
• Std. Dev. 3.2 4.3 DU (1.5 1.8)

10
Identification of Stratospheric Waves
11
Comparison of Surface to Tropopause-2.5 km Mean
O3 Mixing Ratio with Data Assimilation

• Comparison (fc lt 20)
• Global -4.8 4.7 DU, R 0.78
• Tropics -7.0 3.0 DU, R 0.83
• Similar distribution mid-latitude high ozone
  related to synoptic activities low ozone due to
  transport of tropical air tropical wave one
  pattern
• Large biases in the tropics and around dynamic
  structures and southern mid-latitudes

12
Comparison of Ozone Mixing Ratio at 500 hPa
Without applying OMI averaging kernels
13
Column-Averaged O3 Mixing Ratio (Jul 15 Sep 7,
06)
14
Summary

• The 0-200 hPa O3 column and ozone profiles
  directly derived from OMI compares well with MLS
  matchups and assimilated OMI MLS data.
• Stratospheric and tropospheric O3 column can be
  accurately derived from OMI data alone.

15
Proposed Work to Use HIRDLS Data

• Cross-evaluate OMI and HIRDLS stratospheric
  ozone profiles and ozone column
• Use HIRDLS to evaluate the performance of OMI
  retrievals in regions of stratospheric influence
  and evaluate whether we can use HIRDLS to improve
  tropospheric ozone retrievals.
• Use OMI to inter-compare HIRDLS and MLS and
  evaluate their random and systematic differences.
• Combine HIRDLS (high vertical resolution) and
  OMI (global coverage) to study science questions
  (e.g., stratospheric waves, stratospheric
  intrusion)

16

• Acknowledgements
• OMI, MLS, and HIRDLS Science Team
• Support from NASA
• I. Stajner, K. Wargon, S. Pawson for providing
  OMI-MLS Assimilation data

Thank You
17
Validation of OMI UV Radiances

• Simulate OMI radiances using ZM MLS profile,
  trop O3 climat, normalized to ZM total O3 from
  OMTO3, using VLIDORT
• Compare simulations with measured OMI radiances
  (Collection 3)

18
Validation of OMI UV Radiances
UV2

• Smaller x-track dependent biases (2-3)
• Large wavelength dependence at shorter
  wavelengths
• UV1/UV2 jump 4-10 depending on x-track position

19
Ozone Profile Retrieval Algorithm and Validation

• Retrieval algorithm spectral fitting by optimal
  estimation
• Retrieve O3 profiles at 24 2.5-km thick layers
  from surface to 60 km
• Use 270-310 nm (UV1) and 310-330 nm (UV2) OMI
  radiances
• Apply x-track and l-dependent correction to OMI
  derived from 0-20S lat and single day (7/11/06)
  of MLS/OMI match-up.
• Comparison between total column (from OMTO3)
  with profile total column.

Mean Difference 0.8-1.8 DU (0.5) RMSD
3.6-4.7 DU (1.4)
20
OMI Trop O3 Column with and w/o Correction
UV1, No Correction
UV1UV2 with MLS-derived Radiance Correction
Radiance correction is critical for tropospheric
ozone retrievals.
21
Retrieval Errors in Ozone Column
22
Comparison of OMI Ozone Profiles with MLS
2006m0501