2. DIRECT TROPOSPHERIC OZONE RETRIEVALS FROM SATELLITE ULTRAVIOLET RADIANCES Alexander D. Frolov, University of Maryland Robert D. Hudson, University of Maryland Anne M. Thompson. NASA/Goddard Space Flight Center

3. OUTLINE INTRODUCTION BASIC CONCEPT NEW ALGORITHM RESULTS

4. INTRODUCTION NADIR LOOKING SATELLITE INSTRUMENTS MEASURE THE TOTAL COLUMN OZONE AMOUNT, I.E. TROPOSPHERE + STRATOSPHERE WE MUST REMOVE THE STRATOSPHERIC CONTRIBUTION IN ORDER TO GET THE TROPOSPHERIC CONTRIBUTION PIONEERING WORK BY FISHMAN AND LARSEN (1978) USED LIMB SCANNING INSTRUMENT (SAGE) TO MEASURE THE STRATOSPHERIC CONTRIBUTION - RESIDUAL TECHNIQUE. THE RESULTING TROPOSPHERIC DATA WERE MONTHLY AVERAGES. HUDSON ET. AL. (1995) AND HUDSON AND THOMPSON (1998) REFINED THIS TECHNIQUE BY INTRODUCING CORRECTIONS FOR THE EFFICIENCY WITH WHICH NADIR INSTRUMENTS SEE TROPOSPHERIC OZONE. DATA BASED ON THIS TECHNIQUE IS ARCHIVED AT http://metosrv2.umd.edu/~tropo. 1979-1992, Nimbus-7 TOMS, -20 to +20 LATITUDE, 15 DAY AVERAGES. 1996-PRESENT, EP TOMS, -30 to +20 LATITUDE, DAILY AND 9 DAY AVERAGES. THE NEW ALGORITHM WHICH DERIVES TROPOSPHERIC COLUMN OZONE DIRECTLY FROM THE NADIR RADIANCES, ON A DAILY BASIS.

5. BASIS FOR THE ALGORITHM (1) $ IF ONE EXAMINES A TOMS IMAGE (MARCH 1990) THEN ONE SEES THREE DISTINCT REGIMES SEPARATED BY THE UPPER TROPOSPHERE SUB-TROPICAL AND POLAR FRONTS. $ THESE FRONTS ARE ASSOCIATED WITH THE JET STREAMS $ THESE FRONTS SHOW UP AS SHARP GRADIENTS IN THE GEOPOTENTIAL HEIGHT FIELDS AND POTENTIAL VORTICITY FIELDS. $ THERE IS A STRONG CORRELATION BETWEEN THE SHARP GRADIENTS IN THE TOTAL OZONE FIELD AND THE METEOROLOGICAL FIELDS. $ WITHIN EACH REGIME THE TOTAL OZONE DISTRIBUTION IS STRONGLY PEAKED WITH A HALF WIDTH OF ABOUT 10 DU. $ THE MEAN VALUE WITHIN EACH REGIME SHOWS LITTLE VARIATION OVER A MONTH BUT DOES HAVE A SEASONAL DEPENDENCE

8. BASIS FOR THE ALGORITHM (2) $ TOMS IMAGE FOR JULY 1999 DOES NOT SHOW GOOD AGREEMENT BETWEEN THE SHARP GRADIENTS IN THE GEOPOTENIAL HEIGHT FIELD AND THE TOTAL OZONE FIELD FOR THE SUBTROPICAL FRONT. $ THIS IS BECAUSE OF ADDITIONAL TROPOSPHERIC OZONE FROM POLLUTION EVENTS. $ AS 90% OF THE TOTAL OZONE IS WITHIN THE STRATOSPHERE, WE ASSUME THAT IT IS THE STRATOSPHERIC DISTRIBUTION THAT HAS THE NARROW DISTRIBUTION. $ THE OZONE PROFILES WITHIN A REGIME HAVE A NARROW DISTRIBUTION AS A FUNCTION OF ALTITUDE. $ HENCE FOR ANY MONTH THE CLIMATOLOGICAL STRATOSPHERIC PROFILE SHAPE CAN BE ASSUMED TO BE CONSTANT. $ WE CAN NOW CALCULATE A SET OF LOOK-UP TABLES THAT ASSUME A CONSTANT STRATOSPHERIC PROFILE, BUT VARYING TROPOSPHERIC PROFILES (THE REVERSE OF THE CURRENT TOMS ALGORITHM)

12. SUMMARY OF THE ALGORITHM ASSUMES A FIXED STRATOSPHERIC PROFILE AND A VARIABLE TROPOSPHERIC PROFILE STRATOSPHERIC PROFILE DERIVED FROM CLIMATOLOGY. TROPOSPHERIC PROFILES CAN ALSO BE BASED ON CLIMATOLOGY. FOR THE RESULTS PRESENTED, WE USED PROFILES CORRESPONDING TO A CONSTANT OZONE MIXING RATIO SET OF LOOKUP TABLES CALCULATED USING THE TOMRAD CODE ULTRAVIOLET ALBEDOS OBTAINED FROM EP-TOMS, LEVEL-2 DATA SET IN THIS TALK WE LIMIT THE ANALYSIS WITHIN THE TROPICAL REGIME, HOWEVER ALGORITHM CAN BE APPLIED TO ANY REGIME.

13. PROCEDURE $ USE ONLY PIXELS WHOSE REFLECTIVITY IS LESS THAN 20%. REFLECTIVITY OBTAINED FROM THE MEASURED ALBEDO AT 360 NM. $ THIS AVOIDS PROBLEMS WITH CLOUDS IN THE FIELD OF VIEW, AND WITH LARGE AEROSOL CONCENTRATIONS. $ SELECT TOTAL OZONE DATA FROM THE LEVEL-2 DATA SET FOR THE TROPICAL REGIME, AND DETERMINE THE MOST LIKELY STRATOSPHERIC MEAN COLUMN OZONE AMOUNT. $ USE TWO WAVELENGTHS (B PAIR) TO DETERMINE THE COLUMN OZONE AT EACH OF THE SELECTED PIXELS. $ GRID DATA ON A 1.0 X 1.25 DEGREE LATITUDE-LONGITUDE GRID.

17. ERROR ANALYSIS We have assumed that the stratospheric column ozone has a fixed and stable monthly value. An error of +1 DU in the stratospheric column ozone will introduce a bias error of -1.15 DU in the retrieved tropospheric column ozone. The standard deviation of the ozone distribution shown for July, 1999 is 10 DU, this translates into an error of 11 DU in the retrieved tropospheric ozone. There is a third error due to our choice of tropospheric profile. To check this we have compared the retrieved tropospheric ozone values with ozonesonde data from Huntsville.

19. FILM NOTES $ WHITE AREA REPRESENTS THOSE PIXELS WHICH ARE NORTH OF THE SUB-TROPICAL FRONT. $ BLACK PIXELS REPRESENT EITHER NO DATA IN THE TOMS RECORD, OR REFLECTIVITIES GREATER THAN 20%

20. CONCLUSION WE HAVE DEVELOPED A PYHSICAL ALGORITHM WHICH RETRIEVES COLUMN TROPOSPHERIC OZONE AMOUNTS FROM BACKSCATTERED ULTRAVIOLET RADIANCES. COMPARISON OF THESE DATA WITH OZONE SONDE MEASUREMENTS ARE IN GOOD AGREEMENT.