AN EIGHT-YEAR RECORD OF OZONE PROFILES AND TROPOSPHERIC COLUMN OZONE FROM GLOBAL OZONE MONITORING EXPERIMENT (GOME) Xiong Liu, Kelly Chance, and Thomas P. Kurosu Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA xliu@cfa.harvard.edu; kchance@cfa.harvard.ed; tkurosu@cfa.harvard.edu ABSTRACT This paper presents an eight-year record (July 1995- June 2003) of ozone profiles and tropospheric column ozone (TCO) retrieved from GOME ultraviolet radiance spectra using the optimal estimation technique, including detailed treatments of wavelength and radiometric calibrations and forward radiative transfer modeling. The agreement with Stratospheric Aerosol and Gas Experiment (SAGE) data is usually within 15% down to 15 km. The total ozone agrees with TOMS and Dobson/Brewer measurements to within 1-2%. The retrievals can capture the spatiotemporal evolution of TCO in response to regional or short time-scale events such as the 1997-1998 El Niño event and a 10-20 DU change within a few days. The mean biases relative to ozonesonde data are usually within 3 DU (15%) and the standard deviations are within 3-8 DU (13-27%). The global distribution of TCO displays nearly zonal bands of enhanced TCO of 36-48 DU at 20ºS-30ºS during the austral spring and at 25ºN-45ºN during the boreal spring and summer. 1. INTRODUCTION Ozone in the troposphere is a key species in air quality, climate change and atmospheric chemistry. In the past, most satellite retrievals determine Tropospheric Column Ozone (TCO) using residual-based approaches [1-8], by subtracting the Stratospheric Column Ozone (SCO) from the total column ozone (TO). Most of the reliable TCO products from these methods are monthly means and are limited to the tropics. The GOME was launched in 1995 on the European Space Agency s (ESA s) second Earth Remote Sensing (ERS-2) satellite to measure backscattered radiance spectra from the Earth s atmosphere in the wavelength range 240-790 nm. Observations with moderate spectral resolution of 0.2-0.4 nm and high signal to noise ratio in the ultraviolet ozone absorption bands make it possible to retrieve the vertical distribution of ozone down through the troposphere. In recent years, several algorithms have been developed to directly retrieve ozone profiles, including tropospheric ozone, from GOME data [9-12]. However, global distributions of TCO from these algorithms have not so far been published. We recently developed our ozone profile retrieval algorithm. The retrieval algorithm and its validation, presented in several papers [13-17], demonstrate that valuable tropospheric ozone information can be derived from GOME. In this presentation, we reported an updated retrieval algorithm that includes degradation correction so that we can derive an eight-year record of ozone profiles from GOME. We highlighted what tropospheric ozone information we can obtain from GOME, summarized the validation results, and presented the 8-year GOME TCO climatology. 2. ALGORITHM DESCRIPTION Our previous retrieval algorithm is summarized as follows. We retrieve profiles of partial column ozone from fitting windows 289-307 and 326-339 nm using the optimal estimation. Wavelengths in between are not due to calibration problems. In addition to standard corrections provided by the GOME data processor software, we determine the variable slit widths and wavelength, perform undersampling correction, and fit wavelength shifts among radiance, irradiance, ozone cross sections. To account for the initial radiance offset in level 1b data [11], we fit a second-order polynomial in 289-307 nm. We use the Linearized Discrete Ordinate Transfer Model (LIDORT) to simulate both radiances and weighting functions [18], with a look-up table to correct for neglecting polarization. We directly model the first-order Ring effect [19] and fit the scaling parameter on line. We use the ozone cross section measured by [21], which can reduce the fitting residual by 30-45% compared to using the other datasets. Clouds are treated as Lambertian surfaces and partial clouds are modeled with the independent pixel approximation. Cloud fraction is derived from nonabsorbing region 368-372 nm. We use monthly mean stratospheric aerosols observed by SAGE-II [22] and tropospheric aerosol fields by the Global Ozone Chemistry Aerosol Radiation and Transport (GOCART) model [23]. Surface albedo is initialized from the GOME database [24]; wavelength-dependent surface albedo is fitted to partly take the residual aerosol and calibration effects into accounts. Since ozone profile retrieval is an ill-posed retrieval, we use the latitudinaland monthly- dependent version-8 ozone profile climatology [25] and GOME random noise errors to initialize and regularize the retrievals. Proc. Envisat Symposium 2007, Montreux, Switzerland 23 27 April 2007 (ESA SP-636, July 2007)
Compared to the previous version, we made several major updates in the second version. First, we use the model fields to constrain the vertical distribution of other trace gases (e.g., SO2, NO2, HCHO, BRO) and fit their weighting functions instead of their cross sections. Second, we correct the solar-i0 effect for all trace gas cross sections, which can reduce the fitting residuals by ~20-30%. Third, we retrieve ozone at 24 ~2.5 km layers instead of 11 layers to better represent the ozone vertical distributions. Most importantly, to account for the large instrumental degradation especially since 2000, we derive the degradation correction by comparing the global average reflectance (60ºS-60ºN) to those in the first six months [16] so that we can derive this 8-year record of dataset. With extensive radiometric and wavelength calibrations and improvements of forward modeling and forward model inputs as described above, we are able to reduce the fitting residuals to < 0.1% in the Huggins bands. The retrieval characterization and error analysis, described in detail in [13], are summarized as follows. The retrieved ozone profiles are best resolved over altitude range of 20-40 km, with vertical resolution of 812 km. Significant tropospheric ozone information can be derived from the measurements. The degree of freedom for signal in the troposphere ranges from ~1.2 in the tropics to ~0.5 at high latitudes. The a priori influence in retrieved TCO ranges from ~15% in the tropics to ~50% at high latitudes. The total errors of the retrievals are estimated to be 20-30% in the troposphere and lower stratosphere and 5-10% in the stratosphere. The globally averaged total errors in the retrieved TO, SCO, and TCO are estimated to be 1.6%, 2.3%, 21%, respectively. The dominant sources of errors are the smoothing, instrument random-noise, systematic temperature errors, and errors in cloud-top pressure. Tropopause pressure is also a significant source of error for TCO and SCO and ozone absorption cross section is an important source of error for TO and SCO. 3. EXAMPLES OF RETRIEVALS Figure 1. An orbit of retrieved profile of partial column ozone. The red line indicates the tropopause. Fig. 1 shows an example of retrieved profiles of partial column ozone for the orbit 71022024 on 22 October 1997. In the stratosphere, we can see ozone hole in the Antarctic with minimal ozone around ~16 km; stratospheric ozone maximizes outside the polar vortex region. In the troposphere, enhanced ozone is found near the surface over the Southern India due to the regular biomass-burning season in South Africa and over Indonesia due to the intense biomass burning activities caused by the 1997-1998 El Niño event. Fig. 2 shows the zonal cross section of tropospheric ozone around 5ºS and the equator on 22-24 October 1997, we can see high ozone values of 90-100 ppbv over the South Atlantic and Indonesia and low values of 10-30 ppbv over the central and eastern Pacific Ocean and India Ocean. Fig. 3 shows a three-day composite global map of TCO for the same period, which discloses similar features in TCO in the southern hemisphere. The large zonal contrast (e.g. ranging from 9 DU over the east Pacific to more than 60 DU over the Atlantic despite of zonally invariant a priori TCO, demonstrates the TCO distribution is determined from GOME measurements. Figure 2. Zonal distribution of tropospheric ozone mixing ratio around 5ºs and 0ºs on 22-24 October 1997. The red line indicates the tropopause. Figure 3. Three-day composite global map of TCO on 22-24 October 1997.
To demonstrate the capability of our retrievals to capture short-term variations in TCO, we show the time series of retrieved GOME TCO in Fig. 4 at two locations during 09/1996-11/1997. Fig. 4 (top) shows the daily GOME TCO and available ozonesonde TCO at Java. GOME TCO well captures the large variation in ozonesonde TCO ranging from ~10 DU to ~60 DU. One exception occurs during October and November 1997, when GOME TCO are usually ~5-10 DU smaller than ozonesonde TCO. This may be due to the reduced sensitivity of GOME retrievals to enhanced ozone near the surface or the large spatial scale of GOME retrievals. The time series of GOME TCO further shows detailed responses to the biomass burning and precipitation, resulting from the 1997-1998 El Niño episode. Fig. 4 (bottom) shows the daily variation in TCO over American Samoa. GOME TCO reproduces well the large variations (10-25 DU) in ozonesonde TCO over a period of 1-2 weeks (e.g., October 1996, middle March 1997, November 1997). The time series of GOME TCO further shows that TCO changes of 10-20 DU or by a factor of two actually occur within one or two days. Figure 4. GOME and ozonesonde TCO from 09/1996 to 11/1997 over Java (top) and American Samoa (bottom). 4. VALIDATION The validation of our previous retrievals during 1996-1999 has been presented in [13-15]. The integrated TO agrees with TOMS and DB TO to within retrieval uncertainty of the various techniques and spatiotemporal variability; the Mean Biases (MBs) are typically within 6 DU (2%) and the 1σ-Standard Deviations (SDs) are within 3-8 DU (<2.5%) vs. TOMS and within 3-16 DU (<5%) vs. DB. The retrieved TCO captures most of temporal variability in ozonesonde TCO at the majority of ozonesonde stations. The MBs and their SDs are within measurement uncertainties and the TCO variability; the MBs are typically <3 DU (15%) and the SDs are in the 3-8 DU (13-27%) range. The comparison with SAGE-II profiles above ~15 km shows a systematic altitude-dependent biases due to inadequate correction for the initial offset in the GOME level 1b data, but the MBs and SDs are usually within 15%. GOME SCO over the altitude range ~15-35 km agrees with SAGE SCO to within 2.5 DU (1.5%) on average usually without significant spatiotemporal variation. There are good agreements between GOME and ozonesonde SCO at most middle- and high-latitude stations (>30ºN/S), but GOME SCO is systematically larger than ozonesonde SCO by 8-20 DU at carbon iodine and most tropical stations within 30ºN-30ºS. The intercomparisons among GOME, SAGE-II and ozonesonde data with additional comparisons with TOMS and Dobson total ozone illustrate that those large SCO biases mainly result from ozonesonde underestimates of stratospheric ozone. The GOME/sonde profile comparison show similar altitudedependent bias patterns at most stations, consistently indicating systematic errors in our GOME retrievals. GOME retrievals are significantly larger than ozonesonde observations over the altitude range ~10-20 km, where ozone concentration is low, for most carbon iodine (30-70%) and tropical (20-55%) stations. However, the GOME/SAGE-II biases are usually 10-20% over ~15-20 km. The uncorrected altitude hysteresis in ozonesonde datasets introduces a 5-15% error over 10-20 km. The remaining biases among various measurements under those low ozone conditions may result from errors from both satellite retrievals and ozonesonde measurements (e.g., background signal removal, measurement technique, sensor solution, normalization procedure). The dependence of GOME/sonde biases on measurement technique, sensor solution and normalization procedure demonstrates the need to homogenize available ozonesonde measurements for reliable satellite validations. For the current retrievals (July 1995-June 2003), the comparison with other correlative measurements is generally similar as that for the previous version except that the comparison of TCO with ozonesonde generally shows higher correlation and smaller SDs. The retrievals during July 1995 and March 1996 contain positive biases relative to the rest of the data period, likely because measurements are integrated over 0.375 s (i.e., the last quarter of 1.5 s normal integration time) but are reported at the geo-location and viewing geometry for the 1.5 s integration time (especially affect the back scan pixel). 5. AN EIGHT-YEAR RECORD OF TCO Fig. 5 shows the seasonal mean GOME TCO averaged during March 1996-June 2003. There is generally lower TCO (except over the Atlantic) in the tropics with a wave one pattern especially in the southern hemisphere (high ozone over the Atlantic and low TOC over the Pacific). There are generally higher ozone in the middle
latitudes around 30ºS/N, especially nearly zonal bands of enhanced TCO of 36-48 DU at 20ºS-30ºS during the austral spring and at 25ºN-45ºN during the boreal spring and summer. In the northern mid-latitude summer, TCO is less zonally invariant than in spring with higher TCO of 48-60 DU over the Middle East, Eastern US, and Eastern China. Figure 7. Time series of zonal and monthly mean GOME TCO during March 1996-June 2003. 6. SUMMARY AND OUTLOOK Figure 5. Seasonal climatological GOME TCO during March 1996-June 2003. Figure 6. Zonal and monthly mean GOME TCO averaged during March 1996-June 2003. The white and black symbols indicate the month of maximum and minimum values. Fig. 6 shows the monthly variation of zonal mean TCO averaged from March 1996 through June 2003. In the northern hemisphere, TCO generally maximizes during late spring and summer and minimizes during fall and winter. With increasing latitude, you can clearly see the migration of ozone maximum from late spring to middle summer. In the southern hemisphere, TCO is generally highest during austral spring and lowest during austral fall. Fig. 7 shows the time series of monthly and zonally mean TCO during March 1996 and June 2003. The seasonality is generally similar from year to year. The most interesting feature is the anomalously low ozone in the tropics during early and middle 1998 and anomalously high ozone in the northern middle latitudes during the middle and later 1998. This anomalous distribution maybe related to the 1997-1998 El Niño event. Ozone profile and tropospheric ozone are retrieved from backscattered radiance spectra in the ultraviolet using the optimal estimation technique, including detailed treatments of wavelength and radiometric calibrations and forward radiative transfer modeling. We performed degradation correction so as to obtain an almost eightyear record of ozone profiles. The retrievals compare well with TOMS, Dobson, SAGE, and ozonesonde measurements, with some biases originating from the initial offset in GOME level 1b data that is not adequately corrected as well as ozonesonde observations. The retrieved TCO is shown to capture the spatiotemporal evolution of TCO on the daily basis. Currently, the initial offset in GOME data is not well corrected, leading to altitude-dependent biases in the retrievals and affecting the quality of retrieved tropospheric ozone. In addition, since the degradation correction is based on the assumption that global reflectance does not change with time (except for seasonal variation), the retrievals maybe inadequate for trend analysis. We are going to improve the retrievals from these aspects in order to derive a unique record of ozone profiles including tropospheric ozone suitable for trend analysis. We also plan to combine with chemical transport models to understand the global distribution of tropospheric ozone. Finally, we are on the way to adopt this algorithm for other measurements like the Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY), Ozone Monitoring Experiment (OMI) and GOME-2 so as to continue the GOME data record. 7. ACKNOWLEDGMENTS This study is supported by the NASA ACMAP and the Smithsonian Institution. We appreciate the ongoing cooperation of the ESA and DLR in the GOME program. We thank the WOUDC, SHADOZ, CMDL,
TOMS and SAGE and their data originators for providing ozone data. 8. REFERENCES 1. Fishman, J., C.E. Watson, J.C. Larsen, and J.A. Logan (1990), Distribution of tropospheric ozone determined from satellite data, J. Geophys. Res., 95(D4), 3599-3617. 2. Fishman, J., A.E. Wozniak, and J.K. Creilson (2003), Global distribution of tropospheric ozone from satellite measurements using the empirically corrected tropospheric ozone residual technique: Identification of the regional aspects of air pollution, Atmos. Chem. Phys., 3, 893-907. 3. Kim, J.H., R.D. Hudson, and A.M. Thompson (1996), A new method of deriving time-averaged tropospheric column ozone over the tropics using TOMS radiances: Intercomparison and analysis using TRACE-A data, J. Geophys. Res., 101(D19), 24,317-24,330. 4. Hudson, R.D., and A.M. Thompson (1998), Tropical tropospheric ozone from TOMS by a modified residual method, J. Geophys. Res., 103(D17), 22,129-22,145. 5. Ziemke, J.R., S. Chandra, and P.K. Bhartia (1998), Two new methods for deriving tropospheric column ozone from TOMS measurements: Assimilated UARS MLS/HALOE and convectivecloud differential techniques, J. Geophys. Res., 103(D17), 22,115-22,127. 6. Newchurch, M.J., D. Sun, J.H. Kim, and X. Liu (2003a), Tropical Tropospheric Ozone Derived using Clear-Cloudy Pairs (CCP) of TOMS Measurements, Atmos. Chem. Phys., 3, 683-695. 7. Chandra, S., J.R. Ziemke, and R.V. Martin (2003), Tropospheric ozone at tropical and middle latitudes derived from TOMS/MLS residual: Comparison with a global model, J. Geophys. Res., 108(9), 4291, doi:10.1029/2002jd002912. 8. Valks, P.J.M., R.B.A. Koelemeijer, M. van Weele, P. van Velthoven, J.P.F. Fortuin, and H. Kelder (2003), Variability in tropical tropospheric ozone: Analysis with GOME observations and a global model, J. Geophys. Res., 108, doi:10.1029/2002. JD002894. 9. Munro, R., R. Siddans, W.J. Reburn, and B. Kerridge (1998), Direct measurement of tropospheric ozone from space, Nature, 392, 168-171. 10. Hoogen, R., V.V. Rozanov, and J.P. Burrows (1999), Ozone profiles from GOME satellite data: Algorithm description and first validation, J. Geophys. Res., 104(D7), 8263-8280. 11. van der A, R.J., R.F. van Oss, A.J.M. Piters, J.P.F. Fortuin, Y.J. Meijer, and H.M. Kelder (2002), Ozone profile retrieval from recalibrated GOME data, J. Geophys. Res., 107(D15), 10.1029/2001JD000696. 12. Hasekamp, O.P., and J. Landgraf (2001), Ozone profile retrieval from backscattered ultraviolet radiances: The inverse problem solved by regularization, J. Geophys. Res., 106(D8), 8077-8088. 13. Liu, X., K. Chance, C.E. Sioris, R.J.D. Spurr, T.P. Kurosu, R.V. Martin, and M.J. Newchurch (2005), Ozone Profile and Tropospheric Ozone Retrievals from GOME: Algorithm Description and Validation, J. Geophys. Res., 110(D20), D20307, doi:10.1029/2005jd006240. 14. Liu, X., et al. (2006), First directly retrieved global distribution of tropospheric column ozone from GOME: Comparison with the GEOS-CHEM model, J. Geophys. Res., 111, D02308, doi:02310.01029/02005jd006564. 15. Liu, X., et al. (2006), Intercomparison of GOME, ozonesonde, and SAGE-II measurements of ozone: Demonstration of the need to homogenize available ozonesonde datasets, J. Geophys. Res., 101, D114305, doi:114310.111029/112005jd 006718. 16. Liu, X., K. Chance, T.P. Kurosu, Improved ozone profile retrievals from GOME data with degradation correction in reflectance, Atmos. Chem. Phys., 7, 1575-1583, 2007. 17. Liu, X., K. Chance, C.E. Sioris, and T.P Kurosu, Impact of using different ozone cross sections on ozone profile retrievals from GOME ultraviolet measurements, Atmos. Chem. Phys. Discuss., 7, 971-993, 2006. 18. Spurr, R.J.D., T.P. Kurosu, and K.V. Chance (2001), A linearized discrete ordinate radiative transfer model for atmospheric remote-sensing retrieval, J. Quan. Spec. Rad. Trans., 68, 689-735. 19. Sioris, C.E., and W.F.J. Evans (2000), Impact of rotational Raman scattering in the O 2 A band, Geophys. Res. Lett., 27(24), 4085-4088. 20. Brion, J., A. Chakir, D. Daumont, and J. Malicet (1993), High-resolution laboratory absorption cross section of O 3. Temperature effect, Chem. Phys. Lett., 213 (5-6), 610-512. 21. Bauman, J.J., P.B. Russell, M.A. Geller, and P. Hamill (2003), Stratospheric aerosol climatology from SAGE II and CLAES measurements: 2. Results and comparisons, 1984-1999, J. Geophys. Res., 108, 4383, doi:10.1029/2002jd002993.
22. Chin, M., et al. (2002), Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sunphotometer measurements, J. Atmos. Sci., 59, 461-483. 23. Koelemeijer, R.B.A., J.F.d. Haan, and P. Stammes (2003), A database of spectral surface reflectivity in the range 335-772 nm derived from 5.5 years of GOME observations, J. Geophys. Res., 10, 4070, doi:10.1029/2002jd002429, 2003. 24. McPeters, R.D., J.A. Logan, and G.J. Labow (2003), Ozone climatological profiles for version 8 TOMS and SBUV retrievals, Eos. Trans. AGU, 84(46), Fall Meet. Suppl., Abstract A21D-0998.