SEARCH

SEARCH BY CITATION

Keywords:

  • ozone;
  • data assimilation;
  • satellite observations;
  • reanalysis

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

This article presents an assessment of the quality of the ERA-Interim ozone analyses produced by the European Centre for Medium-Range Weather Forecasts. Ozone retrievals from a number of satellite instruments and an ad hoc generated mean total column ozone (TCO) reference are used to assess the quality of the ERA-Interim ozone analyses during the period from January 1989 to December 2008. The ERA-Interim TCO is typically within ±5 DU (about ±3%) from the TCO reference, while showing up to 2% lower values than the Ozone Monitoring Instrument TCO between 50°S and 50°N.

Comparisons with SAGE, HALOE and (UARS and Aura) MLS data show consistent results both in the Tropics and Extratropics, with mean residuals typically within ±5% around 5 hPa and within ±10% in the region of the ozone mixing ration maximum at 10 hPa. However, the comparisons with POAM II and III show mean relative residuals ranging from a few percent in summer to about –40% in winter at high latitudes, partly confirming the known problems of accurately modelling the ozone field during the polar night. Mean residuals of about +10% (but up to +20% at times) and within ±20% are found both in the Tropics and Extratropics for all instruments near 30 hPa and in the lower stratosphere around 65 hPa, respectively.

The quality of the ERA-Interim ozone analyses is also compared with that of ERA-40. The study shows that, until December 1995, the ERA-Interim ozone analyses are in better agreement with the independent observations than their ERA-40 equivalent in the upper troposphere and lower stratosphere, but slightly degraded on average in the middle stratosphere. With the start of the assimilation of GOME ozone profiles (January 1996 –December 2002), the agreement between the independent data and the co-located ERA-Interim analyses improves and exceeds that calculated for ERA-40 also in the middle stratosphere. Copyright © 2011 Royal Meteorological Society


1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

In the last two decades, the European Centre for Medium-Range Weather Forecasts (ECMWF) has devoted increasing effort to producing consistent global reanalyses of the state of the atmosphere, land and ocean. Two major reanalysis projects were completed over this period, namely ERA-15 (Gibson et al. 1997) and ERA-40 (Uppala et al. 2005), covering the periods from December 1978 to February 1994, and from mid-1957 to August 2002, respectively. These global fields, generated with stable and invariant versions of the ECMWF data assimilation system, have been used in many studies (e.g. Uppala et al., 2005, and references therein). During the same timeframe, the number of reanalysis initiatives and projects has rapidly increased worldwide, underlining their importance and value.

On the numerical weather prediction (NWP) side, apart from ECMWF, the National Center for Environmental Prediction (NCEP) and the National Center for Atmospheric Research (NCAR) jointly produced a 40-year reanalysis at the end of the 1990s (Kalnay et al. 1996), completed just a few years before the ERA-40 Atlas was released. The Data Assimilation Office at the National Aeronautics and Space Administration (NASA/DAO, now the NASA Global Modeling and Assimilation Office, NASA/GMAO) completed a global reanalysis for the period 1985–1993 (Schubert et al. 1993). More recently, the Japan Meteorological Agency (JMA) conducted a 26-year reanalysis project referred to as the Japanese 25-year reanalysis (JRA-25) (Onogi et al. 2007).

On the observation side, most space agencies have intense reprocessing programmes with the objective of producing long-term series of consistent observations to be used in several fields, e.g. climate studies. In this context, the European Space Agency (ESA) has recently launched two programs, the Long-Term Data Preservation and the Climate Change Initiative (CCI), to preserve and capitalize on the European wealth of measurements from past, present, and future missions and to meet the need of global observations of climate, as recognized by the United Nations Framework Convention on Climate Change (UNFCCC), as well as by the Global Climate Observing System (GCOS). The long-term high-quality records that the CCI will deliver, referred to as essential climate variables, are likely to further enhance and improve future NWP reanalyses.

Alongside its operational activities, ECMWF is currently producing a new global reanalysis, ERA-Interim (Dee and Uppala, 2009), that focuses on the period from January 1989 onwards, extending the temporal coverage beyond the ERA-40 availability (from September 1957 to August 2002). The main aims of this latest effort are to improve the exploitation of the enormous amount of data available, particularly from satellite instruments, and to provide an improved baseline for the future reanalysis production by using an up-to-date stable version of the ECMWF operational system that included several improvements compared with the ERA-40 one. For example, in addition to improvements in the model physics and parametrizations, the ERA-Interim data assimilation system was upgraded to a four-dimensional variational data assimilation scheme (4D-Var), as opposed to the 3D-Var scheme used in ERA-40, and it made use of a variational bias correction scheme (VarBC, Auligné et al., 2007) for satellite radiances, which automatically detects and corrects for observation biases.

Ozone, both in the form of three-dimensional fields and integrated columns, is routinely produced by ERA-Interim. In the present paper, we discuss the quality of the ERA-Interim ozone reanalyses by comparison with independent ozone observations (both in the form of profiles and total columns) retrieved from a number of satellite instruments. The study covers the 20-year period from January 1989 to December 2008. Indeed, it can also be considered as a long-term validation of the independent ozone observations with ozone analyses produced by an up-to-date and invariant NWP system. For completeness, similar comparisons were also produced using the ERA-40 ozone reanalyses during the overlapping period (from January 1989 to August 2002). These further comparisons are used to discuss and highlight the differences between the two latest ECMWF reanalysis projects, and the improvements achieved in ERA-Interim. In addition, deficiencies in the system are identified that need to be addressed by the next reanalysis project in order to provide more accurate ozone analyses. It is believed that the results and findings of this study could be beneficial, particularly in the context of climate change studies, or to address questions regarding the recovery of the ozone hole.

The article is structured as follows. Section 2 briefly describes the main characteristics of the ozone system used in ERA-Interim, mainly focussing on the differences from those used for the ERA-40 reanalysis. Section 3 describes the sources of independent ozone observations used in this study, and summarizes their quality. The diagnostic tools and the matching criteria used in the present study are described in section 4. The results from the validation of the ECMWF ozone reanalyses against the independent ozone data, introduced in section 3, are presented in section 5. Conclusions and remarks follow in section 6. Here, the lessons learnt from analysing ERA-Interim, and from comparing it to ERA-40, are summarized. Finally, a number of recommendations and potential improvements to be considered for future reanalyses are discussed in section 7.

2. The ozone system in ERA-Interim

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

General discussions on the ERA-Interim assimilation system can be found in Simmons et al. (2007a,b) and Uppala et al. (2008). Focusing on the ERA-40 reanalysis project, Dethof and Hólm (2004) have described in detail the main characteristics of the ECMWF ozone system. Most of that discussion still applies to ERA-Interim, although a number of changes and noteworthy improvements were implemented in the latest reanalysis project.

As pointed out by Dethof and Hólm (2004), the ozone first guess used at ECMWF is derived from an updated version of the Cariolle and Déqué (1986) scheme. In this scheme, the ozone continuity equation is expressed as a linear relaxation towards a photochemical equilibrium for the local value of the ozone mixing ratio, the temperature, and the overhead ozone column. An additional ozone destruction term is used to parametrize the heterogeneous chemistry as a function of the equivalent chlorine content for the actual year. Since the Dethof and Hólm (2004) paper, this parametrization has undergone significant upgrades thanks to collaboration with Daniel Cariolle (Météo-France). An account of the implemented scheme improvements can be found in Cariolle and Teyssédre (2007).

Most of the radiance observations assimilated in ERA-Interim were already used in ERA-40, although many improvements were implemented in the latest project. Dee and Uppala (2009) provided a comprehensive discussion on the difficulties in assimilating such a long and inhomogeneous set of data. The ozone data assimilated in ERA-Interim made use of a larger dataset than was used for ERA-40. Whenever possible, the temporal coverage of data from instruments already utilized in ERA-40 was extended. New datasets were also considered, e.g. the Global Ozone Monitoring Experiment (GOME) ozone profiles retrieved at the Rutherford Appleton Laboratory (RAL; Munro et al., 1998; Siddans et al., 2002). Figure 1 schematically shows the ERA-Interim ozone data usage time coverage for the 20-year period from January 1989 to December 2008. The ERA-40 ozone data usage is also marked for the overlapping period (January 1989 –August 2002). For each instrument, the data provider, data version and product type are given in Table I.

thumbnail image

Figure 1. Time coverage of the remotely sounded ozone data actively assimilated in the ECMWF ERA-Interim reanalysis project (black lines). For comparison, the corresponding ERA-40 data usage is also shown (grey lines).

Download figure to PowerPoint

Table I. Data provider, data version and product type for each assimilated ozone dataset. The Ozone Monitoring Instrument product used here is the Differential Optical Absorption Spectroscopy (DOAS) TCO.
SatelliteInstrumentProviderVersionType
  • Total Column Ozone.

  • *

    Version 6 SBUV/2 data were used until 20 Jan 2008, and version 8 SBUV/2 data from 21 Jan 2008.

NIMBUS-7SBUVNOAA6Profile
NIMBUS-7TOMSNOAA7TCO
NOAA-9SBUVNOAA6Profile
NOAA-11SBUVNOAA6Profile
METEOR-3TOMSNOAA7TCO
ERS-2GOMERAL2.1Profile
NOAA-14SBUVNOAA6Profile
ADEOS-1TOMSNOAA7TCO
EPTOMSNOAA7TCO
NOAA-16SBUV/2NOAA6/8*Profile
ENVISATMIPASESA4.61Profile
ENVISATSCIAMACHYKNMI0.43TCO
NOAA-17SBUV/2NOAA6/8*Profile
NOAA-18SBUV/2NOAA6/8*Profile
AuraMLSNASA2.2Profile
AuraOMINASA3TCO

ERA-Interim made use of a 4D-Var data assimilation scheme (DAS). In a 4D-Var multivariate DAS, changes to one variable, e.g. ozone, normally lead to adjustment in all the others (temperature, winds, etc.) thanks to the background-error correlations and the physical equations and parametrizations used in the model. This normally is a welcome property of 4D-Var, as it offers the possibility of constraining a variable in regions where there are no direct observations, as has already been demonstrated in other cases. Peubey and McNally (2009) showed that assimilation of humidity-sensitive geostationary clear-sky radiances can produce wind increments and improve the wind analyses in the troposphere. Another example could be the inferring of stratospheric wind information by assimilating ozone data. Clearly, the underlying requirement is that the assimilated observations are of high quality, as any inaccuracy could equally produce unrealistic increments in other fields where they are least constrained and possibly degrade them.

Preliminary analysis of the quality of the ERA-Interim products showed that the assimilation of ozone profile data, such as those from GOME, could generate large and unrealistic temperature and wind increments where these observations were less accurate. This region corresponded to a deep layer around the stratopause. These unrealistic temperature and wind increments were then generated by 4D-Var in an attempt to accommodate observed large local changes in ozone concentration. Several solutions were contemplated to overcome this problem. Since these unrealistic feedbacks were mainly related to deficiencies in the ozone data, an adequate ozone bias correction scheme should, in principle, help to control these kind of situations. At the time the ERA-Interim reanalysis project started, this option could not be exploited as a bias correction scheme for retrievals in general, and for ozone in particular, was not yet available. Therefore, it was decided to switch off the sensitivity of the mass and wind variables to ozone data until a more adequate solution could be made available. This option was implemented in ERA-Interim and affected the production from 1 February 1996 onwards. This issue need to be revisited before the next reanalysis project starts.

3. The independent ozone data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

Data from several satellite instruments, as well as a global mean total column ozone (TCO) reference created ad hoc from the NASA's merged satellite dataset were used in the present study to assess the quality of the ERA-Interim ozone analyses during the period from January 1989 to December 2008. Figure 2 schematically presents the time coverage of independent satellite observations. Most of the data used were in the form of ozone profiles, with the exception of the Ozone Monitoring Instrument (OMI) which provided TCO. Details on the data version and quality are provided below.

thumbnail image

Figure 2. Time coverage of the satellite ozone data used in this study. With the exception of the OMI instrument, which provided total column ozone, all the independent ozone products used in the validation were ozone profiles.

Download figure to PowerPoint

3.1. OMI-TOMS TCO

The Ozone Monitoring Instrument (Levelt et al., 2006a) was launched on the Aura platform in July 2004. OMI is a nadir sounder meant to continue the TCO measurements provided by instruments such as the Solar Backscatter Ultraviolet (SBUV) and Total Ozone Mapping Spectrometer (TOMS; Levelt et al., 2006b). Compared with its predecessors, OMI has a higher spatial and spectral resolution. Two TCO products are available from OMI: one is retrieved using the version 8 TOMS-like algorithm (Bhartia and Wellemeyer, 2002; Balis et al., 2007) referred to as the OMI-TOMS product; and one is retrieved using the Differential Optical Absorption Spectroscopy (DOAS) algorithm (Veefkind et al., 2006) referred to as OMI-DOAS. In ERA-Interim, the OMI-DOAS product was actively assimilated starting from 2008. In contrast, the OMI-TOMS data were used for the ‘independent’ validation. Although the two retrieval algorithms are different, e.g. in the treatment of clouds (Kroon et al., 2008b), they were derived using the same radiance information. Thereby, the comparisons between ERA-Interim and OMI-TOMS data cannot provide an independent validation in 2008.

Kroon et al. (2008a) presented a validation of the OMI TCO using remote and ground-based observations during the NASA Aura Validation Experiment (AVE) campaigns. Their results showed that the agreement between OMI-TOMS TCO and ground-based observations was generally within 1%, although the standard deviation was about 2–3%, presumably due to the combination of using a tropospheric climatology and taking measurements above clouds. Ziemke et al. (2006) referenced a personal communication with G. Labow regarding results from a validation study of OMI-TOMS TCO measurements against ground-based Dobson data and data from the SBUV/2 instrument on board the National Oceanic and Atmospheric Administration 16 (NOAA-16) satellite. These comparisons showed that OMI-TOMS TCO data are around 0.5% higher than the Dobson measurements and within 1% of SBUV/2 for latitudes between 60°S and 60°N. In Balis et al. (2007), OMI-TOMS TCO retrievals were compared with Dobson and Brewer ground-based measurements for the period between August 2004 and September 2006. They found that on average the mean OMI-TOMS TCO residual from Dobson data was 0.57 ± 3.50%, and that from Brewer ground-based instruments (mainly located at midlatitudes in the Northern Hemisphere (NH) between 30°N and 60°N) was –0.03 ± 3.50%.

3.2. NASA merged satellite TCO

A mean TCO reference was generated from the NASA merged satellite TCO as a monthly mean for five consecutive years. The original NASA merged satellite ozone datasets consist of monthly-mean zonal and gridded average products constructed by merging individual (Nimbus 7 and Earth Probe) TOMS and (Aura) OMI total ozone as well as TCO and ozone profiles retrieved from a number of SBUV instruments (namely those on board Nimbus 7, NOAA-9, -11, -16, and -17). An external calibration adjustment was applied to each satellite dataset to account for and minimize any effect due to instrument changes, thus providing a continuous dataset from the 1970s (depending on the product) to September 2008. Data and more detailed information (including the above mentioned external calibration) are available at http://acdb-ext.gsfc.nasa.gov/Data_services/merged/ and links therein. The mean TCO reference used in this study made use of the gridded monthly mean TCO product available on a 5° lat × 10° long grid, averaged over the latitudinal band between 50°N and 50°S.

3.3. SAGE II

The SAGE II (Stratospheric Aerosol and Gas Experiment II) sensor was launched into a non-sun-synchronous, 57° inclination orbit aboard the Earth Radiation Budget Satellite (ERBS) in October 1984 (Mauldin III et al., 1985; Chu et al., 1989; McCormick, 1987). The instrument provided self-calibrating, near-global measurements of aerosol, ozone, water vapour and nitrogen dioxide until August 2005 covering a period of about 21 years. The instrument used the solar occultation technique to measure attenuated solar radiation through the Earth's limb in seven channels centred at wavelengths from 0.385 to 1.02 μm. The SAGE instrument series has been used extensively to study and validate ozone long-term trends provided by global models (e.g. WMO, 2003). Here, almost 16 years of SAGE II (hereafter referred to as SAGE) version 6.2 ozone data profiles were used. Already in version 6.1 (e.g. Wang et al., 2002), the agreement between SAGE ozone profiles and ozone sondes was found to be approximately 10% down to the tropopause. In particular, it was seen that SAGE tended to overestimate ozone between 15 and 20 km altitude and systematically underestimated ozone in the troposphere between 8 and 2 km by approximately less than 5% and 30%, respectively. Above 18 km, the version 6.2 ozone product shows an agreement within 5% from correlative data (e.g. Nazaryan and McCormick, 2005). The main difference between the version 6.2 and the version 6.1 algorithms was an improvement of the water vapour product (Thomason et al., 2004). For the most part, the ozone density profiles retrieved with the version 6.2 algorithm changed on the order of 0.5% from those retrieved with version 6.1 algorithm (NASA 2003).

3.4. HALOE

The HALogen Occultation Experiment (HALOE) sensor (Russell et al., 1993) was launched in 1991 on board the NASA Upper Atmosphere Research Satellite (UARS) spacecraft (Reber et al., 1993), with the objective of improving understanding of stratospheric ozone, especially ozone depletion due to chlorine chemistry. The latitudinal coverage was from 80°S to 80°N over the course of one year and included extensive observations of the Antarctic region during spring. The altitude range of the measurements extends from about 15 km to 60–130 km, depending on the species.

Several validation studies of the HALOE ozone profiles were published over the years and for the various versions of HALOE retrievals. Here, we have used the Level 3 AT (L3AT) ozone profiles retrieved with the version 19 (v19) algorithm. Normally, these are daily time-ordered data arranged at time intervals of about 65 s, or about 495 km intervals along the tangent track. In the case of HALOE, each L3AT file includes the complete set of retrievals for the sunrise and sunset events which begin during that day. The Level 2 profiles are on a vertical grid. L3AT processing produces profiles on the UARS standard pressure grid (information available at http://disc.sci.gsfc.nasa.gov/UARS/documents/haloe).

HALOE v19 retrievals were compared with the SAGE v6.0 data in Morris et al. (2002) and with the SAGE v6.1 data in Randall et al. (2003). Both studies found small systematic differences between 20 and 30 km, SAGE exhibiting higher values than HALOE. Randall et al. (2003) also found residuals within ±5% from 12 to 50 km, with increasing positive differences above 40 km up to ±11% at 55 km. An agreement typically within ±5% was also found between the level 2 HALOE data and the version 2.2 Aura Microwave Limb Sounder (MLS) ozone profiles, even though larger residuals normally up to 10% were seen at lower stratospheric levels (Froidevaux et al., 2008).

3.5. UARS and Aura MLS

MLS measures naturally occurring microwave thermal emission from the limb of the Earth's atmosphere to remotely sense vertical profiles of selected atmospheric gases, temperature and pressure. The first MLS experiment in space flew on UARS. The primary UARS MLS data products were vertical stratospheric profiles of ozone at 183 and 205 GHz, chlorine monoxide, water vapour, and temperature. In this study we used the MLS L3AT ozone profile data (version 5). Only the retrievals from the 205 GHz measurements were considered as they are more accurate than those at 183 GHz (Froidevaux et al., 1996). Cunnold et al. (1996) discussed the quality of the UARS MLS ozone retrievals compared with SAGE II data. They showed that UARS MLS values were typically larger than SAGE II values by approximately 5% in the region between 1 and 32 hPa. Comparisons against ozone sondes showed that UARS MLS values were approximately 20% too small in the tropical lower stratosphere around 46 hPa, particularly during the first six months (October 1991 to April 1992) of the UARS mission.

In July 2004, the UARS follow-on mission, Aura, was launched (Schoeberl et al., 2006) on a sun-synchronous orbit (with Equator-crossing time roughly at 1345 local time), carrying the second MLS experiment. Here, we have used the version 2.2 (v2.2) Aura MLS ozone profiles (Livesey et al., 2008). Froidevaux et al. (2008) compared these (v2.2) ozone retrievals with matching ozone data from ground-based and satellite observations, and found that the differences are generally a few percent compared with remotely sounded data at stratospheric levels and within 5% when compared with ground-based microwave measurements in the mesosphere.

For simplicity, the plots are produced for both MLS instruments together, but the reader is advised that differences may exist between the two datasets. It is also worth remembering that the Aura v2.2 MLS ozone profiles were actively assimilated in the ERA-Interim reanalyses during 2008, and therefore the comparisons shown below do not represent an independent validation of the quality of the ozone reanalyses during that year.

3.6. POAM II and POAM III

The Polar Ozone and Aerosol Measurement (POAM) II and III (hereafter simply referred to as POAM) sensors (Lucke et al., 1999) were launched on board the third and fourth Satellite Pour l'Observation de la Terre (SPOT-3 and -4), respectively. They were both near-polar, sun-synchronous solar occultation instruments sampling in nine wavelength channels between 0.353 and 1.02 μm. The set of available products included high-vertical-resolution profiles of ozone. POAM made year-round measurements at high latitude in both hemispheres, with nominal latitude ranges between about 55° and 70° in the NH, and between about 65° and 88° in the southern hemisphere (SH). Here, we used the version 6 POAM II and the version 4 POAM III ozone profiles. Deniel et al. (1997) compared NH measurements from the POAM II instrument against electro-chemical cell (ECC) ozone sondes and found that the residuals were about 2–3% near the lowest and highest altitudes of profile overlap (17 and 30 km) and up to 7.6% from 21 to 24 km. Rusch et al. (1997) compared the POAM II ozone data against UARS MLS, SAGE II and HALOE measurements and found residuals typically within 5–7% in the stratosphere between 6 and 50 hPa, and up to 20% in the upper troposphere–lower stratosphere (UTLS) region. Very small biases were found in the comparison of POAM III measurements with both SAGE II and HALOE, being typically of ±5% (or less) between 30 and 60 km (Randall et al., 2003), and between –7 and +10% with targeted aircraft and balloon observations between 14 and 30 km during the SAGE III Ozone Loss and Validation Experiment (SOLVE) campaign (Lumpe et al., 2002). For simplicity, the plots are produced for both instruments together, but the reader is reminded that differences may exist between the two datasets.

4. Matching criteria and diagnostic tools

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

The comparisons with all the independent observations made use of the same matching criteria. The 3D ozone analysis (or 2D TCO analysis) closest in time to the independent measurements was interpolated to the independent observation location. Based on this criterion and given the availability of four analyses per day (at 0000, 0600, 1200 and 1800 UTC), a temporal mismatch of up to 3 h between observation time and analysis valid time should be expected. Then, the ozone analysis profiles and the independent ozone profiles were interpolated on the coarsest vertical grid to be chosen between that provided by ERA-Interim (60 vertical layers spanning from surface to 0.1 hPa) and that of the independent observations. Ideally, one should have compared the observations with the simulated profiles obtained by convolving the ozone analyses with the observation averaging kernel (e.g. Rodgers and Connor, 2003; Migliorini et al., 2004) rather than with the ozone analyses directly. This was not a viable option as the averaging kernel information was not always available for the datasets used in this study. However, as the instruments that provided the independent observations are limb sounders with vertical resolutions that are reasonably comparable to the model grid, averaging kernels are a much less important issue than, for example, comparing to nadir sounders.

The results are presented in terms of mean relative residuals (RRs) computed between the independent ozone observation (O3Obs) and its reanalysis equivalent (O3EI) over the whole period of data availability. The monthly mean (indicated by the overline) RRs were calculated as follows:

  • equation image(1)

In addition, monthly mean root mean square (RMS) differences (RMSD) were computed for given latitudinal bands and as a function of pressure, as follows:

  • equation image(2)

where O3E4 is the ERA-40 ozone analysis profile co-located with O3Obs. The RMSD diagnostics is used to highlight the regions where the ERA-Interim ozone analyses fitted the independent datasets better than the corresponding ERA-40 reanalyses, to investigate the relative differences between the two sets of ozone analyses and their possible origins. This can then inform further improvements to be implemented in the next reanalysis project (Dee, 2010). It should be noted that, in regions where the independent observations are affected by large biases, the RMSD tool could penalize the best analysis. However, the cross-validation studies published in the literature for the independent observations generally show a good level of agreement in the stratosphere, particularly in the Tropics and at midlatitudes.

The RR and RMSD diagnostics were normally calculated for the Tropics (30°S–30°N), and Southern and Northern Extratropics (30°–90°), except in the case of POAM, due to its coverage being limited to the high latitudes.

5. Comparisons with satellite ozone data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

5.1. Validation of the total ozone analyses

Figure 3 compares the time series of the monthly mean ERA-Interim and ERA-40 TCO, averaged over the latitudinal band between 50°N and 50°S, with the mean TCO reference over the same latitudinal band. It is noted that, because some of the data that NASA used to build the merged satellite dataset were also actively assimilated in ERA-Interim, the comparison does not necessary provide an independent validation. However, it is still useful to have a first indication of the quality of the reanalyzed product, but also to identify potential problems and drifts in the ERA-Interim time record due to the assimilation of other data. Figure 3 shows that in general the TCO reanalyses were in good agreement with the TCO reference with differences within ±5 DU (Dobson Units), less than 2% of the mean TCO value. However, it should be noted that the total ozone averaged over the latitudinal band between 50°S and 50°N only varies between 260 DU and 300 DU over the studied period.

thumbnail image

Figure 3. (a) Time series of the monthly mean ERA-Interim TCO (black solid line) and ERA-40 TCO (black dotted line) averaged over the latitudinal band between 50°N and 50°S. The grey solid line (‘Climatology’) shows the monthly mean TCO obtained from the NASA merged satellite dataset averaged over the same latitudinal band. The black dashed line covering the period from January 2003 to December 2008 refers to the mean SCIAMACHY TCO actively assimilated in ERA-Interim averaged between 50°N and 50°S. (b) Time series of the monthly mean relative differences from the TCO climatology of the ERA-40 (black dotted line) and ERA-Interim (black solid line) TCO. The differences were computed as 100×(Climatology –ERA)/Climatology.

Download figure to PowerPoint

Two exceptions, where the mean differences exceeded 5 DU, can be identified. The first case is for ERA-40 during 1989–1990; the second case is for ERA-Interim during the period from April 2004 to 2007. In both cases, the reduced agreement can be related to the particular ozone data usage. In the first case (ERA-40, 1989–1990), the degraded agreement with the TCO reference was due to the lack of ozone observations actively assimilated, so that the ozone analyses were completely unconstrained by observations. In the second case (ERA-Interim, 2004–2007), Figure 3 shows that the mean total ozone analyses have a high correlation with the mean assimilated total ozone columns retrieved from the Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY) instrument (dashed black line, Figure 3(a)) during the years from 2004 to 2007. This suggests that the ozone analyses were more constrained by SCIAMACHY TCO than by NOAA SBUV/2 partial columns, the only other ozone observations assimilated during that period. This is likely a consequence of the larger volume of data provided by the former dataset than by the latter.

Figure 4 shows the comparison between the area-averaged monthly mean ERA-Interim TCO and the area-averaged monthly mean OMI TCO from September 2004 to December 2008. Both the monthly mean ERA-Interim and monthly mean OMI TCO were averaged over the latitudinal band between 50°N and 50°S. Figure 4 shows that the ERA-Interim ozone analyses have, on average, lower values than measured by OMI, with residuals within –1.5 and +5 DU (within –0.5 and +1.8%). As mentioned above, the comparison shown here cannot provide an independent validation during 2008 as the OMI-DOAS TCO was actively assimilated in ERA-Interim.

thumbnail image

Figure 4. (a) Time series of the area-averaged monthly mean total column ozone (DU) from Aura OMI (black) and the ECMWF ERA-Interim (grey) total column ozone reanalyses and (b) their absolute (black) and relative (grey) differences. Only the latitudinal range between 50°N and 50°S was considered in this comparison.

Download figure to PowerPoint

5.2. Validation of the stratospheric ozone analyses

This section discusses the results from the validation of the 3D ERA-Interim ozone analyses against ozone profiles from a number of independent satellite instruments. These instruments were introduced in section 3 together with a review of the data quality. The comparisons with the occultation data were run for both sunrise and sunset measurements, separately. Negligible differences were identified in the level of agreement between the observations and the corresponding ERA-Interim ozone analyses in the two cases (i.e. using either sunrise profiles only or sunset profiles only). Therefore, only the comparisons using the sunset measurements are presented. Figures 5 and 6 show the time series of the RR diagnostics defined in Eq. (1), and computed for the Tropics and Extratropics at four vertical levels in the stratosphere. These four stratospheric pressure levels were selected as follows: one level at 10 hPa, near the typical ozone volume mixing ratio maximum; one level above and one level below the ozone maximum, at 5 and 30 hPa, respectively; and, finally, one level in the lower stratosphere at 65 hPa.

thumbnail image

Figure 5. Time series of the area-averaged monthly mean relative difference between the independent ozone retrievals and their co-located ERA-Interim ozone reanalyses at (a, c, e) 5 hPa and )b, d, f) 10 hPa, for latitudinal bands as indicated in each panel. The relative differences were computed from Eq. (1) for SAGE (black lines), HALOE (red lines), UARS and Aura MLS (1991–1999, and 2004–2008, respectively) (cyan lines), and POAM II (1993–1996) and POAM III (1998–2005) (green lines). The two vertical dashed lines delimit the period (January 1996–December 2002) during which the GOME ozone profiles were actively assimilated in the ERA-Interim reanalyses.

Download figure to PowerPoint

thumbnail image

Figure 6. As Figure 5, but at (a, c, e) 30 hPa and (b, d, f) 65 hPa.

Download figure to PowerPoint

At 5 and 10 hPa (Figure 5), the comparisons show a generally good agreement of the ERA-Interim ozone reanalyses with SAGE, MLS and HALOE measurements, with relative differences mostly within ±5% at all latitudinal bands. The comparisons with POAM data show larger negative departures than those with the other instruments, with values ranging from a few percent in summer to about –40% in winter in both hemispheres. These results might appear inconsistent with the level of agreement between the POAM data and the other independent datasets reported in the literature. The reasons for these different results are not fully understood, but the following points could explain these results. The native units of the ERA-Interim ozone analyses and of the POAM ozone retrievals are different. The former are available as mass mixing ratio, while the latter are retrieved as ozone concentrations. The unit conversion requires the use of a temperature profile co-located with each observation. Any inaccuracy in such a temperature profile is likely to affect the converted POAM ozone data and in turn their level of agreement with ERA-Interim. Here, the temperature profiles provided with the ozone measurements were used. The same practice was also adopted for other instruments, e.g. SAGE. In addition, we note that most of the POAM data were sampled either at latitudes north of 60°N or south of 60°S, so that the POAM comparisons should be regarded as representative of high latitudes rather than an indication of the level of agreement over the whole extratropical regions. The ozone analyses at high latitudes are normally less accurate than in the Tropics and at midlatitudes. The high latitudes are regions where it is normally difficult to model the ozone transport and depletion with high accuracy in winter. Dethof and Hólm (2004) showed that the model transport alone can produce reasonable ozone distributions at most latitudes even when no ozone data are assimilated, except at high latitudes in winter, where it can lead to an accumulation of total ozone. This accumulation is likely related to changes in the atmospheric circulation forced by the assimilation of satellite data (other than ozone). In addition, at these latitudes in winter the ozone vertical distribution, which is mainly determined by the ozone background, might be poorly represented. Dragani and Dee (2008) showed, for example, that the maximum of the ozone analyses in the ECMWF operational model (on which the ERA-Interim system is based) is generally placed at a higher vertical level than that measured by ozone sondes during the polar winter.

At 10 hPa, particularly in the Tropics and in the SH Extratropics, the level of agreement between the ERA-Interim ozone analyses and SAGE, MLS, and HALOE measurements slightly improved during the period from January 1996 to the end of 2002 (marked by the vertical dashed lines in Figures 5 and 6). This is the period during which GOME ozone profiles were assimilated in ERA-Interim. These results clearly show that the assimilation of ozone profiles can provide a very useful constraint on the vertical distribution of the ozone analyses.

At 30 hPa (Figure 6(a, c, e)), the comparisons between the ERA-Interim ozone analyses and all the four independent datasets (including POAM) show the same level of agreement, particularly in the Tropics and in the NH. In the NH Extratropics, the analysis departures from each independent observation is normally within –10 and +20%. All instruments indicate that the ERA-Interim ozone analyses were too low (up to 20% lower than the observations) before January 1996. With the start of the assimilation of GOME ozone profiles in January 1996, the analysis departures were reduced to values within –5 and +10%. A similar evolution can also be seen in the Tropics and in the Southern Extratropics. Here, the analysis departures were around +20% during the pre-GOME period, they decreased to an average value of +10% during the period GOME profiles were assimilated, then after an initial increase at the beginning of 2003, they decreased again to about +5% from mid 2003 until the beginning of 2004, to finally stabilize to a value around +20% in the following years. This evolution is also a consequence of the data usage. Besides the assimilation of GOME profiles, the ERA-Interim ozone analyses also benefitted from the assimilation of about six months of ozone profiles from the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) between 2003 and 2004. Dethof (2003) showed that the assimilation of MIPAS ozone profiles into the ECMWF system can substantially improve the quality of the ozone analyses, particularly in the tropical lower stratosphere, where the ozone peak value is usually underestimated, as well as inside the winter vortex.

In the lower stratosphere near 65 hPa (Figure 6(b, d, f)), the comparisons show that the ERA-Interim ozone analyses generally exhibit lower ozone values than those from the independent observations, particularly in the NH and in the Tropics. The differences are typically within 0 and 20% depending on the instrument. The impact of assimilating the GOME ozone profiles at this vertical level seems quite modest. Some improvements can be found in the comparisons with HALOE in the Tropics (Figure 6(d)) from 1996.

5.3. ERA-Interim versus ERA-40

Figures 7 to 10 show the time series of the RMSD diagnostics defined in Eq. (2), computed over the Tropics and extratropics for SAGE, HALOE, MLS and POAM, respectively. Because the ERA-40 project was run only until August 2002, the plots do not span the whole period of the independent data availability in most cases.

thumbnail image

Figure 7. Time series of the difference (mPa) between the RMS of (SAGE minus ERA-Interim) and that of (SAGE minus ERA-40) as function of pressure, for (a) the NH Extratropics (30–90°N), (b) the Tropics (30°N–30°S), and the SH Extratropics (30–90°S). Negative values (grey shaded areas) mean a better fit of ERA-Interim ozone analyses to SAGE observations than those from ERA-40. Positive values (contours over white areas) mean a better fit of ERA-40 ozone analyses to SAGE observations than those from ERA-Interim. The contour intervals are 0.5 mPa. The vertical dashed line delimits the period (January 1996–December 2002) during which the GOME ozone profiles were actively assimilated in the ERA-Interim reanalysis. The four dashed horizontal lines are located at 5, 10, 30, and 65 hPa and correspond to the four pressure levels analysed in Figures 5 and 6.

Download figure to PowerPoint

thumbnail image

Figure 8. As Figure 7, but for the comparisons with HALOE.

Download figure to PowerPoint

thumbnail image

Figure 9. As Figure 7, but for the comparisons with UARS MLS.

Download figure to PowerPoint

thumbnail image

Figure 10. As Figure 7, but for the comparisons with POAM II and POAM III. Because of the POAM data coverage, comparisons are averaged over (a) NH high latitudes (60–90°N), and (b) SH high latitudes (60–90°S).

Download figure to PowerPoint

In general, the RMSD diagnostics show similar features regardless of the instrument, particularly in the cases of SAGE, MLS and HALOE. At all latitudinal bands, the ERA-Interim analyses compare better with the independent observations then their ERA-40 equivalent in most of the UTLS region, typically below the ozone maximum peak (at pressures greater than 40 hPa), but not at levels above, at least until December 1995. From January 1996, when the assimilation of GOME ozone profiles started, the representation of the ozone vertical distribution was likely improved in the ERA-Interim reanalysis compared with ERA-40, and that led to a general better agreement with the independent observations (SAGE, HALOE, and MLS). The comparisons with POAM (Figure 10) partly confirmed the improved agreement with ERA-Interim over ERA-40 in the UTLS region (below 70 hPa), as well as the problems in the middle stratosphere during the POAM II period (1993–1996) but not in the POAM III period (1998–2002). In the latter case, the comparisons show a periodicity in the agreement with ERA-Interim compared with ERA-40. In particular, in both hemispheres the level of agreement of ERA-Interim with POAM III seems degraded compared with that of ERA-40 in the region of the ozone maximum in winter and in the lower stratosphere during spring, and improved in the rest of the year. This outcome could be explained partly by temperature effects in the unit conversion and with the difference in latitudinal coverage offered by POAM compared with that of SAGE, HALOE, and MLS, as the comparisons with the former are limited to the high latitudes where the quality of the ozone analyses is normally poorer than in the Tropics and at midlatitudes as discussed above. However, the differences shown in the comparisons with POAM II and POAM III might also suggest non-negligible differences between the two POAM retrievals.

6. Summary and conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

We have presented an assessment of the quality of the ECMWF ERA-Interim ozone analyses for the period from January 1989 to December 2008 by comparisons with independent ozone retrievals. Total column ozone from OMI and a mean TCO reference have been used to assess the quality of the ozone column analyses. Ozone profiles from SAGE, HALOE, UARS and Aura MLS, and POAM II and III have been used to validate the ozone vertical distribution.

The comparisons with OMI data for the most recent years of the ERA-Interim production show a good level of agreement. The TCO analyses typically exhibit lower ozone values than measured by the OMI instrument, though the relative residuals are less than 2% over the latitudinal band 50°S–50°N. In addition, comparisons with a mean TCO reference, created ad hoc from the NASA merged satellite dataset, show residuals generally within ±2% during most of the 20-year period under study.

A summary of the results from the validation of the vertical ozone profiles by comparisons with SAGE, HALOE, (UARS and Aura) MLS, and POAM (II and III) is given in Table II for each latitudinal band and pressure level discussed above. The validation generally shows consistent results, particularly in the case of SAGE, HALOE, and MLS. In these cases, the observation minus analysis residuals are typically within ±5% in the region of the ozone mixing ratio maximum at 10 hPa and just above, but larger up to around 20% in the lower stratosphere. The comparisons with POAM data show a bias of a few percent in summer but large negative biases (the ozone analyses exhibit larger values than the observations) up to –40% during winter at 5 and 10 hPa. Two points that could explain these differences in the results have been noted: a possible temperature effect in the observation conversion from their native units, and different latitudinal coverage over which the comparisons were performed. Regarding the latter point, while the SAGE, HALOE, and MLS comparisons were averaged over the whole Extratropics (i.e. latitudes poleward of 30°), those with POAM were only representative of the high latitudes (latitudes generally poleward of 60°), where the quality of the ozone analyses can be poorer than in the Tropics and at midlatitudes. For example, Dragani and Dee (2008) showed that the ozone maximum of the ECMWF ozone analyses at high latitudes in wintertime is normally placed at higher vertical levels than suggested by ground-based measurements. In the lower stratosphere, the comparisons with POAM show similar results to those obtained with the other instruments. Mean residuals within ±20% are typically found at 30 and 65 hPa at all latitudinal bands and for all instruments.

Table II. Summary of the results from the comparisons of the ERA-Interim ozone analyses with SAGE, HALOE, MLS and POAM, as a function of latitude and pressure level (%).
 Pressure (hPa)SAGEHALOEMLSPOAM
 5±5[0,+5][0,+5][−25,+5]
Extratropics NH10[−10,+5][−5,+1][−5,+3][−25,+10]
(30−90°N)30[0,+20][−2,+20][0,+20][−10,+20]
 65[−20,+10][−2,+10][−20,+10][−20,+10]
 5±5±5±5n/a
Tropics10±10±5[−5,+8]n/a
(30°S−30°N)30[0,+20][0,+20][0,+20]n/a
 65[−20,+30][−5,+20][+5,+25]n/a
 5[−8,+5]±2[−8,+5][−40,+1]
Extratropics SH10[−10,+5][−8,+1][−10,+3][−40,−5]
(30−90°S)30[0,+20][0,+20][0,+20][−20,+10]
 65[−20,+20][−2,+20][0,+20][−20,+20]

The analysis carried out in this study has also shown that the ERA-Interim ozone analyses benefited from the assimilation of GOME ozone profiles (January 1996–December 2002), particularly in the tropical middle stratosphere, where the relative residuals between the independent data and co-located ERA-Interim ozone analyses are normaly reduced to about 10%.

The comparisons with the four independent instruments have also been used to discuss the quality of the ERA-Interim ozone analyses with respect to those produced in ERA-40. In particular, the results have suggested a high consistency in the validation against SAGE, HALOE, MLS, and POAM II. Those comparisons have shown that the ERA-Interim ozone analyses are in closer agreement with the independent observations than their ERA-40 equivalent in the UTLS region, but not in the region of the ozone maximum until January 1996. With the start of the assimilation of GOME ozone profiles, the agreement between the independent data and the co-located ERA-Interim analyses has improved and exceeded that calculated for ERA-40 also in the middle stratosphere. This improved agreement of the ERA-Interim ozone analyses led by the assimilation of GOME ozone profiles is not reflected in the comparisons with POAM III during the period 1998–2002, for which the level of agreement with ERA-Interim seems degraded compared with that of ERA-40 in the region of the ozone maximum in winter and in the lower stratosphere during spring, while improved during the rest of the year. This outcome could be explained as a consequence of temperature effects and different latitudinal coverage of POAM compared with that of SAGE, HALOE, and MLS, as discussed above, but it might also suggest differences between the quality of ozone retrievals from POAM II and POAM III.

Our analysis has pointed out the regions of the atmosphere and the areas of the globe where the ERA-Interim ozone analyses have been improved compared with the corresponding ERA-40 ones. Indeed, it was also shown that a number of deficiencies still remain and need to be addressed in the future. Modelling accurately the ozone distribution at high latitudes in the winter polar region is still a challenge. To address this issue, improvements in both the model transport and the characterization of the chemical processes involving ozone are needed. In addition, the assimilation of ozone information other than that provided by UV sensors could be beneficial to provide a constraint on the ozone analyses during the polar winter.

7. Recommendations for future reanalysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

Although the ERA-Interim project is still in full production at the time of writing, ECMWF has already started planning the next reanalysis project expected to cover an even longer period than that spanned by ERA-40 (Dee, 2010). Several scientific and technical improvements could already be identified to fully justify the effort for a new extended reanalysis.

Besides covering a longer period and using a much improved and a more up-to-date model than ERA-Interim, the new reanalysis system will likely benefit from an increased spatial resolution. Much work and effort are already expected to be devoted to improving the data assimilation system and to better exploiting the available observing system, e.g. by including more accurate, reprocessed observations whenever possible. In addition, the ozone retrievals used as independent observations in this study could also be considered for the assimilation, subject to accurate impact studies. The use of a variational bias correction (VarBC) scheme for retrieval products in general and ozone data in particular is already foreseen. Such a scheme was successfully implemented in the ECMWF operational system in September 2009 (operational cycle CY35R3), and represented an extention to retrieval products of the VarBC scheme used in ERA-Interim for radiance data (Dee and Uppala, 2009).

The parametrization of the homogeneous and heterogeneous processes that regulate ozone chemistry is under periodic revision and update. Undoubtedly the ozone analyses will benefit from any further improvement in its accuracy.

Other improvements could also be implemented, but are still subject to further testing to assess their potential impact on the ozone field as well as on the other meteorological variables. As anticipated in section 2, because of unrealistic feedbacks on the temperature and wind products generated by 4D-Var in an attempt to accommodate unrealistic observed local changes in ozone, the sensitivity of the mass and wind variables to ozone observations during the assimilation minimisation was switched off. In this way, the dynamical link between ozone and the other meteorological variables was removed. As these feedbacks are mainly due to deficiences in the assimilated observations, the use of a bias correction scheme could in principle be beneficial to address the problem, at least partly.

A further link that has not yet been exploited, since it is in need of further attention, is the coupling between the ozone field and the short wave radiation package. The current radiation scheme used at ECMWF includes the absorption of short wave radiation by uniformly mixed gases (such as oxygen, CO2, CH4, N2O and O3 itself), aerosol, and cloud particles. Although the ozone absorption is accounted for, the current formulation of the radiation scheme does not make use of the full prognostic ozone field, so that there are no actual feedbacks between ozone and radiation (Morcrette, 2003). If successfully implemented, the contribution of all these changes have the potential to produce a much improved global ozone reanalysis.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References

The OMI-TOMS, HALOE and MLS data used in this study were retrieved from the NASA Goddard Space Flight Center (GSFC) Distributed Active Archive Center. The SAGE II, POAM II and III datasets were obtained from the NASA Langley Research Center Atmospheric Science Data Center. The NASA merged satellite ozone datasets and information on how they were derived can be found at http://acdb-ext.gsfc.nasa.gov/Data_services/merged/. The GOME ozone profiles were provided by Dr B. Kerridge and Dr R. Siddans (Rutherford Appleton Laboratory, UK). The author would like to thank Dr Sakari Uppala for his support during the early stages of this study, and Dr Dick Dee and Dr Peter Bauer for useful discussions and helpful comments on the manuscript. Two anonymous referees are also thanked for their insightful questions and comments that helped to improve the quality of the paper. Robert Hine skilfully improved the figures in this paper. The author was funded through ESA contract number 21519/08/I-OL.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The ozone system in ERA-Interim
  5. 3. The independent ozone data
  6. 4. Matching criteria and diagnostic tools
  7. 5. Comparisons with satellite ozone data
  8. 6. Summary and conclusions
  9. 7. Recommendations for future reanalysis
  10. Acknowledgements
  11. References
  • Auligné T, McNally AP, Dee DP. 2007. Adaptive bias correction for satellite data in a numerical weather prediction system. Q. J. R. Meteorol. Soc. 133: 631642.
  • Balis D, Kroon M, Koukouli ME, Brinksma EJ, Labow G, Veefkind JP, McPeters RD. 2007. Validation of Ozone Monitoring Instrument total ozone column measurements using Brewer and Dobson spectrophotometer ground-based observations. J. Geophys. Res. 112: D24S46, DOI: 10.1029/2007JD008796.
  • Bhartia PK, Wellemeyer C. 2002. Toms-v8 total O3 algorithm. In: OMI Algorithm Theoretical Basis Document, vol. II, OMI Ozone Products, ATBD-OMI-02, Bhartia PK (ed.) NASA: Goddard Space Flight Centre, Greenbelt, Md. 1531. Available at http://eospso.gsfc.nasa.gov/eos_homepage/for_scientists/atbd/ index.php.
  • Cariolle D, Déqué M. 1986. Southern hemisphere medium-scale waves and total ozone disturbances in a spectral general circulation model. J. Geophys. Res. 91: 1082510846.
  • Cariolle D, Teyssédre H. 2007. A revised linear ozone photochemistry parameterization for use in transport and general circulation models: multi-annual simulations. Atmos. Chem. Phys. 7: 21832196.
  • Chu WP, McCormick MP, Lenoble J, Brogniez C, Pruvost P. 1989. SAGE II inversion algorithm. J. Geophys. Res. 94: 83398351.
  • Cunnold D, Froidevaux L, Russell J, Connor B, Roche A. 1996. Overview of UARS ozone validation based primarily on intercomparisons among UARS and Stratospheric Aerosol and Gas Experiment II measurements. J. Geophys. Res. 101: 1033510350.
  • Dee DP. 2010. ‘The funding of ERA-CLIM’. ECMWF Newsletter 123: 6.
  • Dee DP, Uppala S. 2009. Variational bias correction of satellite radiance data in the ERA-Interim reanalysis. Q. J. R. Meteorol. Soc 135: 18301841.
  • Deniel C, Dalaudier F, Chassefíre E, Bevilacqua R, Shettle E, Hoppel K, Hornstein J, Lumpe J, Rusch D, Randall C. 1997. A comparative study of POAM II and electrochemical concentration cell ozonesonde measurements obtained over Northern Europe. J. Geophys. Res. 102: 2362923642.
  • Dethof A. 2003. ‘Assimilation of ozone retrievals from the MIPAS instrument on board ENVISAT’. Tech. Memo. 428. ECMWF: Reading, UK.
  • Dethof A, Hólm E. 2004. Ozone assimilation in the ERA-40 reanalysis project. Q. J. R. Meteorol. Soc. 130: 28512872.
  • Dragani R, Dee D. 2008. ‘Progress in ozone monitoring and assimilation’. ECMWF Newsletter 116: 3542.
  • Froidevaux L, Read WG, Lungu TA, Cofield RE, Fishbein EF, Flower RF, Ridenourne BP, Shippony Z, Waters JW, Margitan JJ, McDermid IS, Stachnik RA, Peckham GE, Braathen G, Deshler T, Fishman J, Hoffman DJ, Oltmans SJ. 1996. Validation of UARS Microwave Limb Sounder ozone measurement. J. Geophys. Res. 101: 1001710060.
  • Froidevaux L, Jiang YB, Lambert A, Livesey NJ, Read WG, Waters JW, Browell EV, Hair JW, Avery MA, McGee TJ, Twigg LW, Sumnicht GK, Jucks KW, Margitan JJ, Sen B, Stachnik RA, Toon GC, Bernath PF, Boone CD, Walker KA, Filipiak MJ, Harwood RS, Fuller RA, Manney GL, Schwartz MJ, Daffer WH, Cofield BJD, Cuddy DT, Jarnot RF, Knosp BW, Perun VS, Snyder WV, Stek PC, Thurstans RP, Wagner PA. 2008. Validation of Aura Microwave Limb Sounder stratospheric ozone measurement. J. Geophys. Res. 113: D15S20, DOI: 10.1029/2007JD008771.
  • Gibson JK, Kållberg P, Uppala S, Nomura A, Hernandez A, Serrano E. 1997. ‘ERA Description’. ERA-15 Project Report Series 1. ECMWF: Reading, UK. Available from http://www.ecmwf.int/publications.
  • Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L, Iredell M, Saha S, White G, Woollen J, Zhu Y, Leetmaa A, Reynolds R, Chelliah M, Ebisuzaki W, Higgins W, Janowiak J, Mo KC, Ropelewski C, Wang J, Jenne R, Joseph D. 1996. The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteorol. Soc. 77: 437471.
  • Kroon M, Petropavlovskikh I, Shetter R, Hall S, Ullmann K, Veefkind JP, McPeters RD, Browell EV, Levelt PF. 2008a. OMI total ozone column validation with Aura-AVE CAFS observations. J. Geophys. Res. 113: D15S13, DOI: 10.1029/2007JD008795.
  • Kroon M, Veefkind JP, Sneep M, McPeters RD, Bhartia PK, Levelt PF. 2008b. Comparing OMI-TOMS and OMI-DOAS total ozone column data. J. Geophys. Res. 113: D16S28, DOI: 10.1029/2007JD008798.
  • Levelt PF, van den Oord GHJ, Dobber MR, Malkki A, Visser H, de Vries J, Stammes P, Lundell J, Saari H. 2006a. The Ozone Monitoring Instrument. IEEE Trans. Geosci. Remote Sensing 44: 10931101.
  • Levelt PF, Hilsenrath E, Leppelmeier GW, van den Oord GHJ, Bhartia PK, Tamminen J, de Haan JF, Veefkind JP. 2006b. Science objectives of the Ozone Monitoring Instrument. IEEE Trans. Geosci. Remote Sensing 44: 11991208.
  • Livesey NJ, Filipiak MJ, Froidevaux L, Read WG, Lambert A, Santee ML, Jiang JH, Pumphrey HC, Waters JW, Cofield RE, Cuddy DT, Daffer WH, Drouin BJ, Fuller RA, Jarnot RF, Jiang YB, Knosp BW, Li QB, Perun VS, Schwartz MJ, Snyder WV, Stek PC, Thurstans RP, Wagner PA, Avery M, Browell EV, Cammas JP, Christensen LE, Diskin GS, Gao RS, Jost HJ, Loewenstein M, Lopez JD, Nedelec P, Osterman GB, Sachse GW, Webster CR. 2008. Validation of Aura Microwave Limb Sounder O3 and CO observations in the upper troposphere and lower stratosphere. J. Geophys. Res. 113: D15S02, DOI: 10.1029/2007JD008805.
  • Lucke RL, Korwan D, Bevilacqua RM, Hornstein JS, Shettle E, Chen DT, Daehler M, Lumpe JD, Fromm MD, Debrestian D, Neff B, Squire M, Knig-Langlo G, Davies J. 1999. The Polar Ozone and Aerosol Measurement (POAM III) instrument and early validation results. J. Geophys. Res. 104: 1878518799.
  • Lumpe JD, Bevilacqua RM, WHoppel K, Randall CE. 2002. POAM III retrieval algorithm and error analysis. J. Geophys. Res. 107: 45755008.
  • Mauldin III LE, Zaun NH, McCormick MP, Guy JH, Vaughn WR. 1985. Stratospheric Aerosol and Gas Experiment II instrument: A functional description. Opt. Eng. 24: 307312.
  • McCormick MP. 1987. SAGE II: An overview. Adv. Space. Res. 7: 219226.
  • Migliorini S, Piccolo C, Rodgers CD. 2004. Intercomparison of direct and indirect measurements: Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) versus sonde ozone profiles. J. Geophys. Res. 109: 1931619324.
  • Morcrette J-J. 2003. ‘Ozone-radiation interactions in the ECMWF forecast system’. Tech. Memo. 375. ECMWF: Reading, UK.
  • Morris GA, Gleason JF , III JMR, Schoeberl MR, McCormick MP. 2002. A comparison of HALOE v19 with SAGE II v6.00 ozone observations using trajectory mapping. J. Geophys. Res. 107: 41774229.
  • Munro R, Siddans R, Reburn WJ, Kerridge BJ. 1998. Direct measurement of tropospheric ozone distributions from space. Nature 392: 168171.
  • NASA. 2003. ‘SAGE II Version 6.2 Data’. From http://www-sage2.larc.nasa.gov/Version6-2Data.html.
  • Nazaryan H, McCormick MP. 2005. Comparisons of Stratospheric Aerosol and Gas Experiment (SAGE II) and Solar Backscatter Ultraviolet Instrument (SBUV/2) ozone profiles and trend estimates. J. Geophys. Res. 110: D17302, DOI: 10.1029/2004JD005483.
  • Onogi K, Tsutsui J, Koide H, Sakamoto M, Kobayashi S, Hatsushika H, Matsumoto T, Yamazaki N, Kamahori H, Takahashi K, Kadokura S, Wada K, Kato K, Oyama R, Ose T, Mannoji N, Taira R. 2007. The JRA-25 reanalysis. J. Meteorol. Soc. Japan 85: 369432.
  • Peubey C, McNally AP. 2009. Characterization of the impact of geostationary clear-sky radiances on wind analyses in a 4D-Var context. Q. J. R. Meteorol. Soc. 135: 18631876.
  • Randall C, Rusch DW, Bevilacqua RM, Hoppel KW, Lumpe JD, Shettle E, Thompson E, Deaver L, Zawodny J, Kyrö E, Johnson B, Kelder H, Dorokhov VM, König-Langlo G, Gil M. 2003. Validation of POAM III ozone: Comparisons with ozonesonde and satellite data. J. Geophys. Res. 108: 43674384.
  • Reber CA, Trevathan CE, McNeal RJ, Luther MR. 1993. The Upper Atmosphere Research Satellite (UARS) mission. J. Geophys. Res. 98: 1064310647.
  • Rodgers CD, Connor BJ. 2003. Intercomparison of remote sounding instruments. J. Geophys. Res. 108: 41164130.
  • Rusch D, Bevilacqua RM, Randall CE, Lumpe JD, Hoppel KW, Fromm MD, Debrestian DJ, Olivero JJ, Hornstein JH, Guo F, Shettle EP. 1997. Validation of POAM ozone measurements with coincident MLS, HALOE, and SAGE II observations. J. Geophys. Res. 102: 2361523627.
  • Russell III JM, Gordley LL, Park JH, Drayson SR, Hesketh DH, Cicerone RJ, Tuck AF, Frederick JE, Harries JE, Crutzen P. 1993. The Halogen Occultation Experiment. J. Geophys. Res. 98 1077710797.
  • Schoeberl MR, Douglass A, Hilsenrath E, Bhartia P, Beer R, Waters J, Gunson M, Froidevaux L, Gille J, Barnett J, Levelt P, DeCola P. 2006. Overview of the EOS Aura mission. IEEE Trans. Geosci. Remote Sens. 44: 10661074.
  • Schubert SD, Pfaendtner J, Rood R. 1993. An assimilated data set for Earth Science applications. Bull. Am. Meteorol. Soc. 74 23312342.
  • Siddans R, Reburn WJ, Kerridge BJ, Munro R. 2002. ‘Height-resolved ozone information in the troposphere and lower stratosphere from GOME’. Technical report, British Atmospheric Data Centre: Harwell, UK. Available from http://badc.nerc.ac.uk/home/index.html.
  • Simmons A, Uppala S, Dee D. 2007a. ‘Update on ERA-Interim’. ECMWF Newsletter 111: 5.
  • Simmons A, Uppala S, Dee D, Kobayashi S. 2007b. ‘ERA-Interim: New ECMWF reanalysis products from 1989 onwards’. ECMWF Newsletter 110: 2535.
  • Thomason LW, Burton SP, Iyer N, Zawodny JM, Anderson J. 2004. A revised water vapor product for the Stratospheric Aerosol and Gas Experiment (SAGE) II version 6.2 data set. J. Geophys. Res. 109: D06312, DOI: 10.1029/2003JD004465.
  • Uppala SM, Kållberg PW, Simmons AJ, Andrae U, da Costa Bechtold V, Fiorino M, Gibson JK, Haseler J, Hernandez A, Kelly GA, Li X, Onogi K, Saarinen S, Sokka N, Allan RP, Andersson E, Arpe K, Balmaseda MA, Beljaars ACM, van de Berg L, Bidlot J, Bormann N, Caires S, Chevallier F, Dethof A, Dragosavac M, Fisher M, Fuentes M, Hagemann S, Holm E, Hoskins BJ, Isaksen L, Janssen PAEM, Jenne R, McNally AP, Mahfouf JF, Morcrette JJ, Rayner NA, Saunders RW, Simon P, Sterl A, Trenberth KE, Untch A, Vasiljevic D, Viterbo P, Woollen J. 2005. The ERA-40 re-analysis. Q. J. R. Meteorol. Soc. 131: 29613012.
  • Uppala S, Dee D, Kobayashi S, Berrisford P, Simmons A. 2008. ‘Towards a climate data assimilation system: Status update of ERA-Interim’. ECMWF Newsletter 115: 1218.
  • Veefkind JP, de Haan JF, Brinksma EJ, Kroon M, Levelt PF. 2006. Total ozone from the Ozone Monitoring Instrument (OMI) using the OMI-DOAS technique. IEEE Trans. Geosci. Remote Sens. 44: 12391244.
  • Wang HJ, Cunnold DM, Thomason LW, Zawodny JM, Bodeker GE. 2002. Assessment of SAGE version 6.1 ozone data quality. J. Geophys. Res. 107: 4691, DOI: 10.1029/2002JD002418.
  • WMO. 2003. Scientific assessment of ozone depletion: 2002. Global Ozone Research Monitoring Project Report 47, World Meteorological Organization: Geneva, Switzerland.
  • Ziemke JR, Chandra S, Duncan BN, Froidevaux L, Bhartia PK, Levelt PF, Waters JW. 2006. Tropospheric ozone determined from Aura OMI and MLS: Evaluation of measurements and comparison with the Global Modeling Initiative's Chemical Transport Model. J. Geophys. Res. 111: D19303, DOI: 10.1029/2006JD007089.