1. Assimilation of TES O 3 data in GEOS-Chem Mark Parrington, Dylan Jones, Dave MacKenzie University of Toronto Kevin Bowman Jet Propulsion Laboratory California Institute of Technology

2. TES Global Survey Observations, July 1-31, 2006 TES O 3 464.16 hPa ppb Enhanced O 3 abundances from central Asia, across the Middle East, and over the subtropical Atlantic High O 3 over the southeastern USA

3. TES Global Survey Observations, July 4-31, 2005 TES O 3 464.16 hPa ppb Enhanced O 3 abundances from central Asia, across the Middle East, and over the subtropical Atlantic High O 3 over the southeastern USA

4. ppb  TES ozone profiles assimilated from 4 July 2005 through to 31 August 2006.  GEOS-Chem captures the dominant features seen in the TES data, although it tends to underestimate the summer-time abundance.  Assimilation of TES data improves the simulated ozone abundance, which enables us to better quantify the processes controlling the distribution. % GEOS-Chem, no assimilationGEOS-Chem, with TES assimilation Percent difference GEOS-Chem ozone, 500hPa, July 26 2005

5. Modelled O 3 Over the Southeastern USA GEOS-Chem ozone (75°W, 12 GMT, July 26) GEOS-Chem ozone analysis  Assimilation increases the upper tropospheric maximum over the southeastern USA.  Assimilation produces a lower “ozone tropopause”, which reflects the broad averaging kernels of TES and the poor representation of stratospheric O 3 in GEOS-Chem. ppb Altitude (km) Latitude

6. Comparison of GEOS-Chem with Ozonesonde Data Assimilation improves the O 3 distribution in the UTLS region O 3 plume is redistributed throughout column in assimilation Sonde Assimilation GEOS-Chem Ozone (ppb) Eureka (85°W, 80°N) 20 July 2005 Churchill (95°W, 58°N) 20 July 2005 Wallops (76°W, 38°N) 26 July 2005 Wallops (76°W, 38°N) 19 July 2005 Pressure (hPa)

7.  The standard synthetic ozone, Synoz, parameterization of stratospheric ozone in GEOS-Chem can lead to uncertainties in the ozone analysis in the upper troposphere.  Over long model integrations, with assimilation, ozone can accumulate in the lower stratosphere with no sinks to remove it.  Due to the coarse vertical resolution of the TES data, this can lead to unrealistic ozone values being introduced into the analysis.  We have implemented the linearized ozone, Linoz, parameterization of stratospheric ozone into GEOS-Chem which gives a much improved distribution of ozone above the tropopause compared to Synoz. Synoz Sonde data Linoz Stratospheric ozone distribution Wallops (76°W, 38°N) 26 July 2005 Ozone (ppb) Pressure (hPa)

8. Timeseries of O 3 over southeast USA  With the Linoz in the stratosphere we can run the assimilation for extended periods.  O 3 over the southeast is at a maximum in July-Sept and decreases into winter. The assimilation enhances the summer ozone maximum indicating possible issues with lightning NO x emissions in the model. Dash line: no assimilation, Solid line: O 3 assimilation, Level: 500 hPa Difference: assimilation - no assimilation Ozone difference (ppb) Ozone (ppb)

9. Boulder (40.3 N, 105.2 W), 21 July 2006 Trinidad Head (40.8 N, 124.2 W), 7 July 2006 Linoz Sonde Synoz Comparison to Ozonesonde Profiles Pressure (hPa) Ozone (ppb)  Over North America the impact of Linoz on O 3 in the troposphere is small (compared to other regions where transport from the stratosphere is a more important in the ozone budget)  The improvement representation of the stratosphere by implementing Linoz provides a more realistic ozone distribution in the UTLS over North America (especially at high latitudes) Wallops (76°W, 38°N) 26 July 2005 Ozone (ppb) Churchill (95°W, 58°N) 20 July 2005 Pressure (hPa)

10.  Residuals (TES - model, black line, and TES - assimilation, red line) time-series at 350 hPa from July 2005 to August 2006.  Area average over northern mid-latitudes: -180° to 180° W and 30° to 60° N.  The bias between the data and model is much reduced in the summer following the assimilation.  Negative bias in the winter due to stratospheric sudden warming in mid-January. Assimilation performance at mid-latitudes O 3 Difference (ppb)

11.  Area average over -180° to 180° W and 60° to 80° N.  The bias between the data and model is much reduced in the summer following the assimilation, as also shown for North America.  In this case the stratospheric sudden warming is much more pronounced and the assimilation does not have much impact on the bias.  This time-series in the upper troposphere implies possible issues with tropopause height in GEOS-Chem and/or TES data?  The bias in the mid-troposphere is also negative but is not as pronounced as in the upper troposphere. High-latitude residuals time-series Upper troposphere: 350 hPa Mid troposphere: 500 hPa O 3 Difference (ppb)

12. Conclusions  GEOS-Chem captures the dominant features in the distribution of tropospheric O 3 as observed by TES, but the abundance is generally under-estimated with respect to these values.  The TES data has sufficient information to improve the ozone abundance in GEOS-Chem relative to ozonesonde profiles.  A linearized ozone chemistry scheme has been incorporated into GEOS-Chem which significantly improves the stratospheric ozone distribution.  This gives us more confidence in the assimilation of ozone in the troposphere where the width of the TES averaging kernels can smooth information from the stratosphere into the analysis.  In the Arctic, GEOS-Chem tends to overestimate the ozone abundance in the upper troposphere, which is reduced by the assimilation.  This could reflect errors in either the GEOS-Chem troposphere or the ozone tropopause in the TES retrievals in the Arctic.  A major sudden warming in the stratosphere in January 2006 enhanced ozone in the upper troposphere, beyond levels observed by TES. Simulation of O 3 by the AM2 model (constrained by NCEP winds) does not produce a pronounced intrusion.