1. Assimilation of EOS-Aura Data in GEOS-5: Evaluation of ozone in the Upper Troposphere - Lower Stratosphere K. Wargan, S. Pawson, M. Olsen, J. Witte, A. Douglass Global Modeling and Assimilation Office (GMAO) Chemistry and Dynamics Branch NASA GSFC

2. A question How much of the ozone that we see in the troposphere is of stratospheric origin? Models disagree on this quite dramatically

3. Plan of Talk Motivation – quantifying sources of tropospheric ozone Ozone data and GEOS-5 Data Assimilation System Results – Evaluation against ozonesonde data – Vertical structure of UTLS ozone fields We look at the lower stratosphere but the goal is to derive some information on tropospheric ozone as well

4. Tropospheric Ozone 2010 boreal summer mean tropospheric ozone column [Dobson Units]

5. Tropospheric Ozone 2010 boreal summer mean tropospheric ozone column [Dobson Units] Maxima along the subtropical jet streams. Summer, Northern Hemisphere higher

6. Tropospheric Ozone 2010 boreal summer mean tropospheric ozone column [Dobson Units] Wave one pattern in the tropics ConvectionLightning

7. Tropospheric Ozone 2010 boreal summer mean tropospheric ozone column [Dobson Units] Tropical Pacific controlled by ENSO and MJO

8. Tropospheric Ozone 2010 boreal summer mean tropospheric ozone column [Dobson Units] Very low over snow-covered Greenland and Antarctica

9. Tropospheric Ozone - Sources NO x / CO / Volatile Organic Compounds chemistry – Biomass burning – Fossil fuel burning – Lightning Transport from the stratosphere Spatial distribution closely tied to meteorology These mechanisms are well understood but quantitative attribution is not precise

10. Constrain ozone in the Upper Troposphere – Lower Stratosphere (abundance, structure, variability) – Stratosphere-Troposphere Exchange Constrain ozone in the Upper Troposphere – Lower Stratosphere (abundance, structure, variability) – Stratosphere-Troposphere Exchange Separate tropospheric and stratospheric ozone columns Accurate tropospheric ozone budget based on global observations Assimilation can help achieve this Model supplies information on dynamics Vertical grid of a DAS can resolve features that data cannot

11. GEOS-5 Data Assimilation System Atmospheric General Circulation Model: Horizontal resolution: flexible - 2.5° to ¼° 72 layers from the surface to 0.01 hPa Parameterized ozone chemistry (stratospheric P&L; dry deposition) No representation of tropospheric ozone sources in the model 3D-Var analysis: Gridpoint Statistical Interpolation Observations: Conventional (surface, sondes, radar, aircraft, MODIS-derived winds,…) Satellite radiance data (TOVS/ATOVS, AIRS, IASI, SSM/I, GOES, GPS-RO) Ozone data (OMI, MLS retrievals)

12. Microwave Limb Sounder (MLS) on EOS Aura Measures temperature and composition of the atmosphere from microwave emissions Limb scanner Vertical range: We assimilate profiles between ~260 hPa – 0.14 hPa Vertical resolution: 2.5 – 6 km 9 years and counting

13. Ozone Monitoring Instrument (OMI) Total ozone information derived from observations of backscattered UV radiation Observes sun-lit atmosphere Total Ozone Monitoring Instrument (TOMS) legacy Operational 2004 - present Sensitivity varies with altitude and local meteorology (no signal from below clouds). This is taken into account by weighting the signal by OMI’s efficiency factors (averaging kernels), ε We have 8 year long, 2°×2.5° assimilation run with this configuration, 2005 - 2008

14. Results in the stratosphere 60N -90N 10S -10N 90S -60S Year Time series of integrated stratospheric ozone column in three latitude bands 60N – 90N: Winter-Spring maximum, interannual variability; “Arctic ozone hole” in 2011 Tropics: Ozone controlled by the QBO 90S – 60S: Austral Spring ozone holes Realistic representation of temporal variability

15. Vertically integrated Observation – Forecast statistics Assimilation significantly reduces the O-Fs for both instruments Negligible bias between analysis total ozone and OMI data Stratospheric column biased low w.r.t. MLS observations but MLS likely overestimates ozone below 200 hPa

16. Total ozone O-Fs and tropospheric response, June - August 2009 2010 O-F Analysis tendency in troposphere Analysis tendencies/increments in the troposphere have similar pattern to total ozone O-Fs: OMI supplies tropospheric ozone information O-Fs and increments positive over land and negative in regions of strong convection

17. Total ozone O-Fs and tropospheric response, June - August 2009 2010 O-F Analysis tendency in troposphere Biomass burning signal over the Amazon varies from year to year. Very low fire counts in 2009 result in lower ozone production Here, OMI makes up for the lack of explicit chemistry in the model Analysis tendencies/increments in the troposphere have similar pattern to total ozone O-Fs: OMI supplies tropospheric ozone information O-Fs and increments positive over land and negative in regions of strong convection

18. Lower Stratosphere and Upper Troposphere; Comparison with ozonesondes Lower Stratosphere, tropopause – 50 hPa Upper Troposphere, 500 hPa - tropopause Excellent agreement of lower stratospheric ozone with sonde data Good agreement in the upper troposphere; assimilation is biased low by 1.4 Dobson Units – missing NO x chemistry in the model? Sonde data are from the WOUDC, NDACC, and SHADOZ databases, 2005 - 2012

19. Ozone in the lower stratosphere – assimilation vs. sondes Sonde data Assimilation Hohenpeissenberg (47.8N, 11E) Assimilation faithfully reproduces the annual cycle as well as day- to-day variability of ozone in the lower stratosphere at this location.

20. Vertical structure – an example Sonde data Assimilation High and low ozone layers in the UTLS often form as a result of transport of ozone poor air from low latitudes. Assimilation reproduces layered structure of this ozone profile as permitted by the vertical resolution of the DAS Hohenpeissenberg (47.8N, 11E) on May 4 th 2005

21. Capturing fine structures in the UTLS Latitude Theta [°K] Satellite data Assimilation Combining observations with assimilated dynamics allows accurate representation of small-scale features unresolved by the data. We use along-track ozone profiles from High Resolution Dynamics Limb Sounder (HIRDLS) Vertical resolution is ~1 km for HIRDLS and ~2.5 – 3 km for MLS

22. Summary Eight year long record of global ozone was obtained by assimilating OMI and MLS observations into GEOS-5 Very good agreement with ozonesondes in terms of vertically integrated ozone in the lower stratosphere Good agreement in the upper troposphere as well. Low bias needs fixing – representation of sources in the model? Good representation of shallow vertical structures in the UTLS The product can be used to quantify stratosphere – troposphere exchange of ozone