1. H. Worden Jet Propulsion Laboratory, California Institute of Technology J. LoganHarvard University TES ozone profiles compared to ozonesondes ABSTRACT: The Tropospheric Emission Spectrometer (TES) on the EOS-Aura spacecraft, launched July 15, 2004, is making direct, global observations of tropospheric ozone. Routine observations began in late August 2004. Ozone profiles are retrieved from TES infrared radiances with vertical resolution of roughly 6 km for nadir observations. The principal source of validation for TES O3 measurements is ozonesondes. For this analysis, we have focused on data from fall, 2004. In some cases, we have ozonesonde data from dedicated launches timed to match the Aura overpass, while other comparisons are performed with standard data available from the SHADOZ and WOUDC data archives. In order to make TES-ozonesonde comparisons, we must account for TES measurement sensitivity and the disparities in vertical resolution. This is done by applying the TES averaging kernel and constraint to the ozonesonde profile. Differences between TES profiles and the resulting (smoothed) sonde profiles are then compared to expected differences due to estimated retrieval errors, climatological variability and spatial/temporal discrepancies. TES ozone profiles are systematically higher than sondes in the upper troposphere, but compare well in the lower troposphere. TES measurements are able to differentiate tropospheric ozone profiles over a range of (coarse) vertical structures. II. Comparison Examples http://tes.jpl.nasa.gov A41A-0010 PROCEDURE FOR COMPARISONS 1.Map O3 sonde profile to TES 87 pressure levels grid: 2.Apply averaging kernel, A TES, and a priori constraint: 3.Compare to TES profile using retrieval error terms: Measurement Error + Cross-State Error O 3 Sonde from Ascension Island 7.95°S, 14.37°W Launched 10/11/2004, 13:50 UTC Closest TES nadir measurement (Step/Stare run 2199) 7.756°S, 12.934°W (160 km away) DOFS = 3.7 in troposphere < 200 mb Vertical res: 6 to 9 km (avg = 7.2 km) TES RETRIEVAL ERROR ( Total Error Covariance ) (Smoothing Error) (Cross-Term Error includes T, H2O for joint retrieval with O3) (Measurement Error) (Systematic Errors) Where A is the averaging kernel, M is a linear mapping matrix on pressure levels, G is the gain matrix and K is Jacobian matrix. I. Method and Coincidence Criteria Sonde/Nadir Measurement Coincidence Criteria applied for fall 2004 (9/20 - 11/17) TES data both global survey & step/stare runs) Initial: 600 km, 48 hours (800 km for AVE), Chi 2 55 coincidences More selective: Sonde-TES T_diff 44 coincidences Trajectory analysis for comparisons with Boulder, Natal and Houston (AVE) sonde sites. Although some cases are inconclusive, differences > 5K in the temperature profiles generally indicate disparate trajectories and an unusable TES-sonde O3 comparison (not included in the statistics). Backward trajectories to sonde site (Houston ) Backward trajectories to TES nadir measurement Sonde vs. TES: Temperature and Ozone Dist |Sonde-TES|: 580 km Time difference: 1hr Dist |Sonde-TES|: 269 km Time difference: 1 hr Backward trajectories to sonde site (Natal, Brazil ) Backward trajectories to TES nadir measurement Sonde vs. TES: Temperature and Ozone Dist |Sonde-TES|: 193 km Time difference: 15 hrs Backward trajectories to sonde site (Boulder ) Backward trajectories to TES nadir measurement Sonde vs. TES: Temperature and Ozone Comparison of 3 sonde profiles (with TES AK) and 3 TES retrieved profiles in the sub-tropics/mid-latitude. sonde TES retrieval initial guess sonde TES retrieval initial guesss Comparison of 3 sonde profiles (with TES AK) and 3 TES retrieved profiles in the tropics.

2. TES Global Survey with 2 nadir, 3 limb scans/sequence. Nadir scans are averaged (limb scans not shown). 14-15 Nov. 2004 TES Global Survey with 3 nadir scans/ sequence. Nadir scans are not averaged. No limb scans taken. 21-22 May 2005 III. Ensemble Comparisons IV. Outlook & Conclusions http://tes.jpl.nasa.gov A41A-0010 Acknowledgements The work described here was performed at the Jet Propulsion Laboratory, California Institute of Technology and Harvard University, under contracts from the National Aeronautics and Space Administration. Thanks to Frank Schmidlan, Mike Newchurch and Gary Morris for dedicated sonde launches. Gary Morris of Valparaiso Univ. had sponsorship for AVE from the Shell Center for Sustainability and NASA's IONS program. The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model (http://www.arl.noaa.gov/ready/hysplit4.html) used in this presentation. Upper Trop (UT) 500mb>p>p TP Lower Trop (LT) p > 500mb 28 coincidences 16 coincidences __ 1 to 1 O3 Profile Differences [ppbV] O3 Profile Relative Differences O3 Profile Error-weighted Differences Temperature Differences [K] Improved L1B Calibration (see poster: A41A-0007) Current L1B has differences with AIRS > 1K Improved L1B will be < 0.5K comparison with current L1B comparison with L1B calfit prototype Note improvement is mainly in the upper trop. as expected since lower radiances had the largest correction. Sonde comparison with current L1B L1B calfit prototype comparison Payerne (47° N, 7° E) Natal, Brazil (6° S, 35° W) Conclusions  Comparisons to O3 sondes for current TES (v1, beta quality) data available at DAAC show:  TES is able to detect expected variability in the lower troposphere.  Bias in the upper troposphere is: - latitude dependent - peaked at 200 - 150 hPa  Comparisons for data with improved L1B calibration (only 3 cases available) show less bias in the upper troposphere.  Coincidence criteria are tricky for tropospheric comparisons but:  Using the TES-sonde tropospheric temperature differences as a filter seems to be a good proxy for trajectory analysis.  Will have more matches with denser nadir coverage (started in May 2005) IV. Expected Variability Hohen., Payerne: 330 km apart Legion., Linden: 470 km Hoh/Pay. to Lind./Leg: 540-1190 km Payerne and Hohen. very similar up to 400 hPa. Spatial variability on the same day Payerne, Oct. 2004 Natal Sep/Oct 2004 Ascension Sep/Oct 2004 Tropical Atlantic sonde sites It is not unusual to have profiles outside mean ± one s.d., because variability is very high in UT/LS. Variability in ozone in the lower strat. is higher in spring than in fall. TES Global Surveys with 2 vs. 3 nadir scans per sequence