Cross Evaluation of OMI, TES, and GEOS-Chem Tropospheric Ozone

Xiong Liu¹, Lin Zhang², Kelly Chance¹, John R. Worden³, Kevin W. Bowman³, Thomas P. Kurosu¹, Daniel J. Jacob²

​

¹ Harvard-Smithsonian Center for Astrophysics

² Harvard University

³ Jet Propulsion Laboratory

​

3^rd GEOS-Chem Users’ Meeting 2007

Harvard University

April 11, 2007

1

Outline

• Motivation
• TES retrievals and GEOS-Chem simulation
• Preliminary OMI ozone profile retrievals
• Comparison methodology
• OMI/TES/GEOS-Chem comparison
• Summary

2

Motivation

• Tropospheric ozone retrievals can be greatly improved with joint UV/IR retrievals [Worden et al., 2007].
• Are OMI/TES data consistent?
• How well does GEOS-Chem simulation compare with OMI/TES ozone?

3

• OMI/TES: both on EOS-AURA and measure tropospheric ozone
• OMI: ~10-12 km FWHM in the troposphere, 13×24 km², global coverage
• TES: ~6 km FWHM in the troposphere, 5×8 km²

Worden et al., 2007

TES Retrievals and GEOS-Chem Simulation

• TES: V 2.0
• Compares well with ozonesonde observations: generally biased higher by ~10% [Ray et al., 2007].
• Compares well with DIAL obtained during the INTEX-B: Positive bias of 5-15% and a negative biases of up to 20% in the upper troposphere [Richard et al., 2007].

​

• GEOS-Chem simulation: V7-04-09 with GEOS-4
• Lightning NOx: 6 Tg/yr, rescaled with OTD/LIS climatology
• Increase Chinese NOx emission by ~70% (2006)

​

4

Preliminary OMI Ozone Profile Retrievals

• OMI retrievals: O₃ at 24 ~2.5 km layers with optimal estimation
• Fitting window: 270-310 nm (UV-1), 310-330 nm, 368-372 nm (UV-2)
• O₃ climatology (month, lat, Z) [McPeters et al., 2007] as a priori

5

May 8 2006

Overpass US

​

Partial

Column

Ozone (DU)

​

(a) Retrieval

(b) A priori

Preliminary OMI Ozone Profile Retrievals

• OMI calibration: wavelength and cross-track dependent errors
• Derive soft correction (λ, θ, multiplicative) by simulating OMI radiances: McPeters (strat.) and Logan (1999) trop. O₃ clima.

Assumption: climatology represents ozone fields on global average

• Remove remaining systematic stripes

​

6

May 8, 2006

​

1. Original
2. Soft calib.
3. Soft calib. +destriping
4. A Priori
5. 600 mb

f_c < 0.3

Comparison Methodology

• OMI/TES retrievals: different retrieval grid and a priori
• Relatively coarser vertical resolution (vs. ozonesonde)
• TES: more tropospheric ozone information (two defined peaks)
• OMI: more stratospheric ozone information

​

7

Comparison Methodology

• Use GEOS-Chem as an intermediate, also evaluate GEOS-Chem:
• Interpolate GEOS-Chem/TES to OMI grid (coarsest)
• Append GEOS-Chem with TES stratospheric ozone
• Compare GEOS-CEHM with TES (TES AKs + OMI a priori)
• Compare GEOS-CEHM with OMI (OMI AKs + OMI a priori)
• Compare OMI/TES directly, similar to Luo et al. [2007]:
• Interpolate TES to OMI grid and adjust TES with OMI a priori
• Apply OMI AKs to TES data
• Compare OMI/TES with ozonesonde observation later (not here)
• Present the comparison on May 08, 2006 (similar on other days)
• Remove poor retrievals (i.e., TES master flag, emission layer flag, OMI fitting residuals) and cloudy pixels (OMI f_c > 0.3)
• 550 coincidences

8

OMI/TES/GEOS-Chem TCO on May 8, 2006

9

Generally consistent spatial distribution despite systematic biases

OMI/TES/GEOS-Chem TCO on May 8, 2006

10

OMI/TES/GEOS-Chem Comparison

11

1. Difference due to OMI/TES AKs can be up to 10-15% especially in UT
2. Large negative (10°N-20°N, high sun) and positive (40°S-25°S, low sun) biases may be caused by non-linearity of the OMI calibration.

Mainly systematic OMI/TES differences

OMI/TES Comparison

12

(b) Those biases are not caused by a priori

(a, c, d) Mostly systematic differences

Summary

• The spatial distribution of OMI, TES, and GEOS-Chem tropospheric ozone is similar on the global scale.
• TES shows a systematic positive bias of ~10% relative to GEOS-Chem.
• OMI shows a negative bias of ~15% relative to TES except for 10°N-20°N (~ -30%) and 45°S-25°S (~20%). The large negative bias at high sun and positive bias at low sun may be related to the non-linearity calibration of OMI.
• OMI/TES differences cannot be explained by a priori and different averaging kernels, and are mainly systematic.

Acknowledgements

• OMI and TES science team, GEOS-Chem community
• NASA and Smithsonian Institution

13

OMI/TES/GEOS-Chem Ozone (600 mb)�(North Pacific during INTEX-B, May 05-09, 2006)

14

05/05

05/06

05/07

​

05/08

​

05/09

OMI TCO (North Pacific on May 05-10, 2006)

15

OMI tropospheric column ozone

​

f_c < 0.3

Gridded to 2.5°×2°

16

Persistent high O₃ over Northern India from OMI, not clear from TES, not shown in GEOS-Chem. Maybe be due to OMI retrieval artifacts: (a) absorbing aerosols (b) incorrect terrain height

OMI/TES/GEOS-Chem Ozone (600 mb) on May 04-12, 06

05/04

05/06

05/08

05/10

05/12

17

Append GEOS-Chem with TES stratospheric ozone

How does the Appending of Different Stratospheric Ozone to GEOS-Chem Affect the Comparison?

Append GEOS-Chem with OMI a Priori stratospheric ozone