SEARCH

SEARCH BY CITATION

Keywords:

  • Aqua-AIRS;
  • hyperspectral sounding;
  • atmospheric ozone

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[1] An evaluation of the Atmospheric Infrared Sounder (AIRS) version 4 (V4) and version 5 (V5) retrieved ozone profiles and the total column ozone is performed using collocated ozonesonde (O3SND) profiles and total ozone measurements from the World Ozone and Ultraviolet Radiation Data Center (WOUDC) archives. Using O3SNDs and Brewer/Dobson Network (BD) measurements as the truth, bias and root-mean-squared (RMS) difference statistics are computed for the AIRS ozone profile retrievals and the derived total ozone. In addition, global monthly maps of total ozone generated for the AIRS retrievals are compared with the Ozone Monitoring Instrument (OMI) and the Solar Backscatter Ultra Violet (SBUV/2) instrument derived maps to evaluate the characteristic trends and seasonal cycle depicted by the AIRS retrieval. The results of the validation exercise reveal that the AIRS V5 algorithm significantly improves the ozone profile retrieval biases and RMS differences for the lower troposphere and, especially, over the tropical region where the V4 algorithm shows larger discrepancies with the O3SND measurements. The V5 retrieval biases with global O3SNDs are less than 5% for both the stratosphere and the troposphere. The RMS differences are less than 20% for the upper stratosphere and are close to 20% for the lower stratosphere and the troposphere. Total ozone amounts from both the V4 and V5 versions agree well with the global BD station measurements with a bias of less than 4% and an RMS difference of approximately 8%. Analysis of V5 total ozone monthly maps reveals that the V5 ozone retrievals depict seasonal trends and patterns in concurrence with OMI and SBUV/2 observations.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[2] The Atmospheric Infrared Sounder (AIRS) is the first in a new generation of high spectral resolution infrared sounder instruments flown aboard the National Aeronautics and Space Administration (NASA) Earth Observing System (EOS) Aqua satellite. The instrument measures outgoing radiances from 650 cm−1 to 2675 cm−1 at a nominal spectral resolving power (λλ) of 1200. The AIRS is accompanied by the Advanced Microwave Sounding Unit–A (AMSU-A). Aumann et al. [2003] provide details of the instrument characteristics and the AIRS product retrieval software (APS) to derive many AIRS/AMSU-A products. The APS suite has been put into operation at the NASA Jet Propulsion Laboratory (JPL), the National Oceanic and Atmospheric Administration National Environmental Satellite Data and Information Service (NOAA/NESDIS) center for Satellite Applications and Research (STAR), and at the NASA Goddard Space Flight Center (GSFC) Distributed Active Archive Center (DAAC). These facilities provide wide dissemination of AIRS data products to the user community. NOAA/NESDIS operates a near-real-time AIRS processing system to process and disseminate AIRS products to many weather centers [Goldberg et al., 2004]. The products include Level 1B radiance products and Level 2 retrieval core products. The Level 2 core products include profiles of temperature, water vapor and ozone, total ozone and many other surface and cloud parameters. Apart from these core products, the retrieval algorithm generates additional trace gas research products. All the core products generated and distributed so far have used the AIRS V4 algorithm (http://disc.sci.gsfc.nasa.gov/AIRS/). Numerous publications on the details of the AIRS V4 algorithm are available in the special issue published by IEEE Transactions (Institute of Electrical and Electronics Engineers, Special issue on the EOS Aqua Mission, IEEE Transactions on Geoscience and Remote Sensing, 41(2), 171–493, 2003). Also, a compendium of publications on the validation of AIRS V4 products is available as a special section published by the American Geophysical Union (Validation of Atmospheric Infrared Sounder Observations, Journal of Geophysical Research, 111, 2006).

[3] The AIRS science team members have updated the retrieval algorithm to version 5.0.14 (hereafter referred as V5). Details of the V5 algorithm are documented at http://disc.sci.gsfc.nasa.gov/AIRS/documentation/. The NASA/DAAC and the JPL have plans to reprocess the AIRS data with the V5 algorithm for distribution to user communities. The AIRS data starting from September 2002 to date have been processed at NOAA/NESDIS with both the V4 and V5 algorithms and a website to display various geophysical products is located at http://www.orbit.nesdis.noaa.gov/smcd/spb/iosspdt/iosspdt.php.

[4] In an earlier paper [Divakarla et al., 2006], the authors have validated the AIRS V4 retrieved temperature and moisture profiles with a collocated data set consisting of 82,000 matches of AIRS V4 retrievals, radiosonde (RAOB) measurements, and the European Centre for Medium-Range Weather Forecasts (ECMWF) analysis fields. The paper has a detailed discussion on the procedures used in the generation of match-up data sets, and the AIRS validation system operating at NOAA/NESDIS.

[5] In this paper, the authors present the validation of AIRS ozone profile retrievals and derived total ozone from both the V4 and V5 algorithms. The procedures adapted in generating the ozone validation data sets are similar to those discussed in the earlier publication [Divakarla et al., 2006]. In a recent paper, Monahan et al. [2007] have examined the quality of the AIRS V4 ozone profile retrievals in capturing the variability of ozone in the upper troposphere lower stratosphere (UTLS) region. The results of their study indicate that although the AIRS V4 algorithm has large uncertainty in retrieving the ozone profiles, the retrieval captures the variability of ozone in the UTLS region. The main objective in this paper is to evaluate both the V4 and V5 ozone product with many correlative data sets and characterize the accuracy for different regions of the globe. Another aim of this paper is to analyze the implications of the retrieval methodology adapted in the AIRS V4 and V5 algorithms, and suggest possible solutions and enhancements to the upcoming version of the AIRS retrieval algorithm.

[6] Section 2 describes the validation data used in the ozone validation. Details on the data set preparation and the statistical metrics implemented are presented in section 3. Results on the validation of the AIRS V4 and V5 ozone retrievals with ground measurements and intercomparison with other correlative data sets are discussed in section 4.

2. Validation Data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[7] The AIRS ozone validation is performed using two sets of data.

2.1. Collocated Data Sets

[8] The collocated data set consists of point observations from many correlative data sets matched within a radius of 100 km and ±3 h of time coincidence. Two sets of collocated data are generated; one for the validation of AIRS retrieved ozone profiles, and the other for the validation of AIRS total ozone. The AIRS V4 and V5 retrieved ozone profiles are validated with the global ozonesonde launches (here after referred as O3SNDs) archived by the World Ozone and Ultraviolet data Center (WOUDC, http://www.woudc.org) and matched ECMWF ozone fields [Dethof and Holm, 2004]. The data set is generated from 3 years of AIRS data (2002–2005). A relaxed time match criteria of ±12 h is also used to evaluate the differences in the validation of AIRS ozone profiles with ±3 h matches. The ozone profile validation data set is referred to as O3PRF_DSET hereafter.

[9] There are uncertainties in the estimation of ozone amounts for the levels beyond O3SND bursting altitude (usually near 10 hPa), and the computed total ozone uses an estimate of the column above the balloon burst altitude. Hence, a direct measurement of total ozone from the WOUDC Brewer/Dobson spectrometer measurements (hereafter referred to as BD measurements) is used to evaluate AIRS total ozone. The data set contains AIRS V4 and V5 total ozone, BD measurements, and the ECMWF total ozone matches obtained for the period 2002–2005. In addition, simultaneous observations of daytime Aqua-AIRS and the Aura Ozone Mapping Instrument (OMI) total ozone retrievals collocated with BD measurements available for the period 2004–2005 are analyzed to make a relative performance assessment and test the assertion that the OMI data could be used as a transfer standard for the truth. The total ozone validation data set is referred as TO3_DSET hereafter.

2.2. Gridded Total Ozone Monthly Data Sets

[10] Gridded monthly averages of total ozone derived from the AIRS V4 and V5 retrievals are analyzed with the monthly maps from the Aura-OMI and the NOAA 16 Solar Backscatter Ultra Violet (SBUV/2) instrument retrievals. The SBUV/2 and OMI instruments produce retrievals only for the sunlit portion of the globe and consequently the global monthly average data sets are limited to the sunlit portion of the globe. The OMI and the SBUV/2 monthly maps are used as a reference to evaluate trends and seasonal cycles depicted by the AIRS retrievals.

3. Data Set Preparation

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[11] The following sections discuss the satellite retrievals, data from the WOUDC O3SNDs and BD measurements, the ECMWF matches, and the statistical metrics implemented in the validation.

3.1. Satellite Retrievals (EOS Aqua-AIRS, EOS Aura-OMI, and NOAA 16 SBUV/2)

3.1.1. AIRS V4 and V5 Retrievals

[12] The AIRS retrieval matches corresponding to the O3PRF_DSET and TO3_DSET data sets are generated at NOAA/NESDIS. Daily Level 1B AIRS/AMSU radiance files obtained from the AIRS near-real-time processing system [Goldberg et al., 2004] are used with the AIRS off-line retrieval version (http://www.star.nesdis.noaa.gov/smcd/spb/iosspdt/iosspdt.php) to produce AIRS V4 and V5 retrievals. The components of the V4 retrieval system are discussed in detail by Susskind et al. [2003]. In short, the V4 retrieval system includes microwave retrieval [Rosenkranz, 2003], cloud clearing, initial IR fast regression first guess (hereafter referred as REG_FG), [Goldberg et al., 2003], and a final physical retrieval (hereafter referred as PR) [Susskind et al., 2003]. The cloud-clearing procedure employed in the AIRS retrieval uses observed radiance of a set of channels from adjacent 3 × 3 fields of view (FOV) to specify a cloud-cleared radiance for all channels. The retrieval algorithms utilize cloud-cleared radiances to produce a single retrieval for each 3 × 3 array of the AIRS FOV. Thus, the noise in the cloud-cleared radiance due to spatial distribution and quantity of clouds within each AIRS FOV affects the retrieval accuracy. The spatial resolution of the AIRS retrieval is approximately 50 km at nadir.

[13] The AIRS V5 retrieval algorithm, although similar to the V4, has substantial improvements. The major updates include improvements to the radiative transfer algorithm (RTA) [Strow et al., 2006; DeSouza-Machado et al., 2007], initial cloud-clearing methodology and the surface emissivity retrieval [Zhou et al., 2008]. Table 1 lists some of the important changes between the V4 and the V5 algorithm. The major change in ozone retrieval is the initial solution (also referred as first guess) used by the PR. The V5 PR uses a priori ozone climatology [McPeters et al., 2007] as the initial solution in lieu of the REG_FG solution used by V4 PR. Figure 1 shows the AIRS instrument sensitivity in the 9.7μm band for a 1% change in ozone in each of the 100 Radiative Transfer Algorithm (RTA) layers for the individual V4 and V5 ozone channels. The V4 PR uses 29 ozone channels (dashed gray scale curves) and the V5_PR uses an expanded set of 41 channels (12 more channels plotted as thin gray scale curves in addition to the V4 PR channels). The average response for the V4 PR (thick dashed curve) and for the V5 PR (thick solid curve) over all channels for tropical, midlatitude, and polar situations is also shown in Figure 1. For each channel and situation, the response is calculated by perturbing the ozone by 1% in each RTA layer and differencing the calculated brightness temperature from a calculation which held the ozone amount fixed. This difference is then divided by the thickness of the RTA layer in kilometers in order to “normalize” the response for small layers as compared to larger RTA layers. In general, both the V4 and V5 channel sensitivity indicates that the AIRS measurements possess approximately 1 piece of information in the vertical. As is evident from the mean response curves, the addition of channels to the V5 algorithm most noticeably increases the sensitivity of ozone in the stratosphere. Also noteworthy is the fact that the peak altitude of the instrument response increases at low latitudes (tropics) as compared to higher latitudes (poles). This is in part due to the fact that the vertical maxima in ozone abundance are at a higher altitude in the tropics than the poles, and also due to increased water vapor absorption at low latitudes. The limited sensitivity of the ozone channels for the lower troposphere coupled with the upward shift of the sensitivity functions (from poles to the tropics) makes the retrieval accuracies degraded for the lower troposphere and especially so over the tropics. Also, the shape of the total column weighting as indicated by the sensitivity functions for the polar, midlatitude, and tropical cases indicates that a tropical total column retrieval with its associated weighting is different from a midlatitude or polar total column retrieval. Some of the other changes listed in Table 1 also affect the V4 and V5 ozone retrievals and are discussed along with the results presented in section 4.

image

Figure 1. The AIRS instrument sensitivity functions for the ozone channels used in V4 and V5 physical retrieval algorithms. The V4_PR uses 29 ozone channels (dashed gray scale curves), and the V5_PR uses an expanded set of 41 channels (12 more channels plotted as thin gray scale curves in addition to the V4_PR channels). The average response for the V4_PR (dashed dark curve) and for the V5 PR (solid dark curve) over all channels for tropical, midlatitude, and polar situations is also shown.

Download figure to PowerPoint

Table 1. Summary of Differences Between the V4 and V5 Algorithms That Affect Ozone Retrievals
Description of ChangesV4V5
Number of ozone channels used in the physical retrieval2941
Ozone first guessregression solution based on ECMWF training [Goldberg et al., 2003]ozone climatology [McPeters et al., 2007]
Damping parameter [Susskind et al., 2003, equation 38]0.751.0
Ad hoc model error term [Susskind et al., 2003, equation 24]ONOFF
Number of functions [Susskind et al., 2003, equation 37]79
HITRAN O3 parametersHITRAN98HITRAN98/HITRAN04
Basis for transmittance tuningECMWF analysis fields collocated with AIRS overpasses [Strow et al., 2006]collocated RS-90 radiosondes coordinated with AIRS overpasses
Basis for radiance tuningECMWF analysis fields collocated with AIRS overpasses [Strow et al., 2006]collocated RS-90 radiosondes coordinated with AIRS overpasses
3.1.2. OMI Total Ozone Retrievals

[14] The OMI measures the solar radiation backscattered by the Earth's atmosphere and surface in the wavelength range 270 to 500 nm with a spectral resolution of 0.5 nm. A detailed description of the OMI instrument characteristics, total column ozone retrieval algorithm and recent updates is provided in the Algorithm Theoretical Basis Document (ATBD, Volume 2) available at http://www.knmi.nl/omi/research/documents/.

[15] The OMI retrievals have a spatial resolution of 13 km along track × 24 km across track at nadir and increase in size as the scan approaches the outer scan positions. Daily OMI overpass retrievals taken at approximately 1345 LT, and matched within 25 km to the BD measurement locations are generated and archived at the Aura validation website (http://avdc.gsfc.nasa.gov/phpgdv2/avdc_tablefile.php?id = 28). The OMI retrievals are matched to the corresponding total ozone measurements from the respective BD stations. The subset of matches that include simultaneous OMI retrievals is from October 2004 to the end of 2005. The entire collocated data set (2002–2005) has a sample size of 4096 matches, and the OMI subset has a sample size of 694 matches.

3.1.3. Aqua-AIRS, Aura-OMI, and NOAA 16 SBUV/2 Global Grids

[16] The AIRS V4 and V5 gridded monthly maps are of 3° latitude × 3° longitude resolution and are generated for the years 2004 and 2005 using the daily total ozone global grid product generated at NOAA/NESDIS [Goldberg et al., 2004]. The Aura-OMI Version 8 gridded monthly averages are available starting from August 2004 to date at 1.0° latitude × 1.25° longitude resolution (ftp://toms.gsfc.nasa.gov/pub/omi/data/monthly_averages/ozone) and are used to generate monthly maps for the years 2004–2005. Also, the NOAA 16 SBUV/2 Version 6 [Bhartia et al., 1996] gridded monthly averages are available at 2.5° latitude × 2.5° longitude (http://www.cpc.ncep.noaa.gov/products/stratosphere/SMOBA/) and are used to generate monthly maps. All these data are remapped to AIRS global grid resolution (3° × 3°) for analysis and intercomparison.

3.2. O3SND and Total Ozone BD Measurement Matches

3.2.1. O3SND Matches

[17] The ozone profile measurements are from the O3SND data reported to the WOUDC (http://es-ee.tor.ec.gc.ca/cgi-bin/ozonesondeflights). The AIRS validation data system [Divakarla et al., 2006] extracts O3SND profile measurements matched to the AIRS retrievals with given collocation criteria; a distance match of 100 km in radius and a time coincidence ±3 h. A sample of 353 AIRS-O3SND matches accepted by the AIRS quality assurance (QA) procedures is obtained from 3 years of data (2002–2005). Because of the limited number of tropical O3SND stations and sparse sampling in the tropics, only 4 matches are found from the tropical stations (WOUDC STN IDs 191, 328, and 443). Except in the boundary layer, and in the lower troposphere, the ozone variations over a single day at the levels reported by O3SNDs are usually small in the tropics. Hence, a relaxed time match criterion (±12 h) is used to obtain a reasonable number of samples for the tropics. The relaxed time match yields a sample size of 838 matches from 31 O3SND stations over the globe, and contains 54 matches from the tropical stations (STN IDs 191, 328, 175, 205, and 443). A list of O3SND station data matches for the 3 h and 12 h collocations is provided as Table S1 in the accompanying auxiliary material. Table S1 also provides data matches from the BD measurement sites discussed in section 3.2.2. Most of the O3SNDs are of either Electrochemical Concentration Cell (ECC) or Brewer Mast (BM) type. Some of the O3SNDs are of Carbon Iodine (CI) type and some others are of unknown type. There may be systematic differences due to O3SND technique, instrument type, sensor solution and data processing [World Meteorological Organization, 1994; Smit et al., 1996; Liu et al., 2006; Fioletov et al., 2006], but no distinction is made in this validation exercise. Figure 2 shows the locations of the AIRS-O3SND matches for 26 stations that have contributed to the match-up samples with black dots for a time collocation of ±3 h and open circles when a time collocation of ±12 h is used. The geographic sampling of the AIRS-O3SND matches inherits the characteristic sampling pattern of Aqua satellite orbit (1330 LT) with a time match criterion of ±3 h and a distance match criterion of 100 km to the ground measurements. Table 2 shows the distribution of matches over different latitude zones for the 3 h and the 12 h collocations. Also shown in Table 2 is the distribution of total ozone BD measurement matches discussed in section 3.2.2. Except for some of the Southern Hemisphere (SH) stations operating around 60–70°S, most of the O3SND matches are from the Northern Hemisphere (NH), and are predominantly from the midlatitude and high-latitude regions. The Arctic station (station ID 089, Ny Alesund, 78.9°N, 11.9°E), and the Antarctic station (STN ID 101, Syowa, 69.0°S, 39.6°E) have a relatively large number of samples followed by stations from the midlatitudes and high latitudes of the NH (see auxiliary material).

image

Figure 2. Locations of O3SND measurements collocated with AIRS observations for ±3 h (black dots) and ±12 h (open circles around black dots) time collocations. Many factors, the Aqua satellite orbit, the time and distance collocation criteria, number of O3SND stations, and the frequency of O3SND launches affect the number of matches.

Download figure to PowerPoint

Table 2. Latitudinal Distributions of AIRS-O3SND Matches for 3 h and 12 h Collocations and Brewer/Dobson (BD) Total Ozone Measurement Matches
Latitude RangeO3SND Profile Measurement Matches (N)BD Total Ozone Measurement Matches (N): Daily
±3 h±12 h
90–80°N000
80–70°N65175167
70–60°N653446
60–50°N47313
50–40°N191267940
40–30°N911364
30–20°N212498
20–10°N00414
10–10°S454212
10–20°S000
20–30°S612119
30–40°S000
40–50°S01117
50–60°S000
60–70°S66237457
70–80°S00115
80–90°S0034
Total samples3548394096
3.2.2. Total Ozone BD Measurement Matches

[18] The total ozone measurements are from the WOUDC BD station measurements archived as a digital video disk (DVD) by the Environment Canada in association with the World Meteorological Organization–Global Atmosphere Watch (WMO-GAW) program (WOUDC, http://www.woudc.org/ ODW DVD #1, containing data reported by the individual monitoring stations around the world). The DVD contains total ozone measurements from Brewer/Dobson/Filter stations for the years 1926–2005 submitted up to February 2006. The Brewer/Dobson ozone spectrophotometers employed at BD stations measure total column ozone in the atmosphere using differential absorption spectroscopy techniques [Dobson, 1957]. A small sample of BD stations has filter measurements, and some others may contain both the direct sun and zenith sky measurements, but no distinction is made in using the data in this validation exercise. The AIRS validation system [Divakarla et al., 2006] matches AIRS retrieval locations with the BD station locations and extracts BD total ozone measurement matches. The BD measurements usually match within ±3 h with the Aqua orbit time (1330 LT). Thus, the availability of total ozone measurement from a BD station is considered a match if a daytime AIRS retrieval is located within 100 km radius of the BD measurement site. Figure 3 shows the geographic locations of 56 BD stations (open squares) that have contributed a total of 4,096 accepted matches (see auxiliary material for details on the type of measurement and the number of matches obtained from each BD station). Column 4 of Table 2 shows the distribution of these samples over different latitude bands. About 37% of the samples are from the polar region (90–60°N; 90–60°S), 56% are from midlatitudes (60–23°N; 60–23°S) and the remaining 7% are from the tropics. In addition, about 36 stations (shown as black dots inside open squares in Figure 3) have matched OMI total ozone retrievals.

image

Figure 3. Locations of the AIRS-BD total ozone measurement station matches (open squares). Also shown are BD station locations that have simultaneous AIRS and OMI retrieval matches (black dots inside open squares).

Download figure to PowerPoint

3.3. ECMWF Ozone

[19] The ECMWF model uses a tracer transport equation with parameterizations for photochemical sources and sinks [Cariolle and Deque, 1986] in the determination of ozone fields. An outline of the ECMWF ozone model and the quality of the resulting ozone profiles and the total ozone fields are discussed by Dethof and Holm [2004]. The ECMWF ozone profile data pertaining to O3PRF_DSET and the ECMWF total ozone pertaining to TO3_DSET data set are generated using the procedures described by Divakarla et al. [2006]. Bias and root-mean-squared (RMS) differences are computed for the ECMWF ozone profiles and the total ozone using O3SNDs and BD measurement matches, respectively. It may be noted that the REG_FG used in the AIRS V4 algorithm uses the ECMWF data for training [Goldberg et al., 2003], and an evaluation of the ECMWF ozone provides a way to assess the quality of the ECMWF data used in training the REG_FG.

3.4. Statistical Metrics

[20] The O3SNDs data contain high-resolution measurements of temperature and ozone partial pressures (mPa) from the surface to10 hPa. The ozone partial pressure data are first transformed into level-to-space column densities, and then interpolated to the AIRS 100 level pressure levels for comparison with the AIRS ozone profile retrievals. The measurements, though merged into AIRS 100 level pressures, are still too high a resolution compared to the AIRS retrievals that have very few vertical degrees of freedom [Susskind et al., 2003]. Hence, statistics are computed for 9 layers for the O3PRF_DSET. The pressure boundaries for layers 1–9 are 1100–260 hPa, 260–126 hPa, 126–66 hPa, 66–32 hPa, 32–16 hPa, 16–8 hPa, 8–4 hPa, 4–2 hPa, and 2–0 hPa respectively. Layers 8 and 9 are above the O3SND balloon burst pressure and contain only the extrapolated ozone amounts. These two layers are excluded in the validation. Some of the O3SNDs may reach high enough altitudes (beyond typical balloon burst pressure of 10 hPa) and may have measurements contributing to layer 7 (8–4 hPa), but the sample size is quite small. The O3SND balloon burst pressure is taken into consideration to avoid having extrapolated data entering into statistics. Bias and RMS difference statistics are computed for column densities converted to layer ozone amounts taking into account the number of samples pooled into each layer. The computations are consistent with the conventions used by Susskind et al. [2003]. The percent error for each layer is computed by weighting the RMS difference with the reference ozone amount in the layer. The ozone bias is computed as a percentage of the reference ozone amount in the layer (100 × (AIRS – O3SND)/O3SND). The procedure is repeated for both the ±3 h and ±12 h time coincident matches. Similarly, bias and RMS differences are computed between the AIRS derived total ozone and BD station measurements, and for other correlative data sets.

4. Results and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[21] Validation results for the AIRS ozone profile retrievals are discussed in section 4.1. Results on the validation of total ozone are discussed in section 4.2. An analysis performed on the AIRS V4 and V5 global monthly maps in conjunction with OMI and SBUV/2 maps is presented in section 4.3. Section 4.4 presents a summary on the performance of the AIRS V4 and V5 algorithms based on the results discussed in sections 4.14.3. The captions used for the AIRS V4 version are V4_FG for the fast regression first guess initial solution and V4_PR for the final physical retrieval solution. Similar notations are used to represent V5 a priori initial solution (V5_FG), and the physical retrieval (V5_PR).

4.1. V4 and V5 Ozone Profile Validations

[22] Section 4.1.1 presents the AIRS retrieval statistics for the global ensemble of O3SND matches. Statistics for the tropical region and for other stations from the SH and the NH that have special significance to ozone events are discussed in section 4.1.2.

4.1.1. Global Statistics

[23] The sample size of AIRS-O3SND 3-h matches is not sufficient to subgroup the data into different regions. Hence, statistics are computed with the global ensemble for the AIRS initial solutions (V4_FG and V5_FG), for the physical retrieval solutions (V4_PR and V5_PR), and for the ECMWF ozone profiles with O3SND matches as the truth. The data set consists of 353 matches from all over the globe. Layer ozone scatterplots (see auxiliary material) for V4_FG, V4_PR, V5_FG, V5_PR, and for the ECMWF are analyzed with reference to the ozone amounts from the O3SND measurements. Results from the analysis indicate that in general, the V4_FG underestimates larger amounts of ozone and overestimates smaller amounts of ozone. The ECMWF in general overestimates the ozone amount for most of the layers. The physical retrievals (V4_PR and V5_PR) corrects the first guess and tend to agree better with the O3SND measurements. Figures 4a and 4b show the physical retrieval solution for the V4 and V5 for layers 1–7. The V4_PR exhibits larger scatter than V5_PR, and the degree of agreement brought in by the V5_PR is better than that achieved by V4_PR. The V4_PR overestimates ozone amount in the lowest layer (1000–260 hPa, layer 1) compared to V5_PR. Another difficulty seen with the V4_PR is its inability to reach extreme values which is consistent with the smaller damping parameter 0.75 (more damping) used in V4_PR. In general, high ozone measurements that are underestimated in the V4 retrieval are increased by the V5 retrieval; and low ozone measurements that are overestimated in the V4 retrieval are decreased by the V5 retrieval. However, while improving these extreme values, the V5_PR, in general, shows a slight overestimation. This overestimation is prominent for layers 3 and 4 (Figure 4b, V5_PR_L03, violet asterisks; V5_PR_L04, red crosses). For the lower troposphere (layer 1) where the ozone amounts are considerably smaller, the V5_PR algorithm improves the retrieval, reduces the scatter and shows much better correlation with the layer ozone amounts from O3SND measurements. Thus, the differences in the V4_PR and V5_PR performance are somewhat dependent on the damping parameter that sets a threshold for the propagation of noise in the solution. Table 3 lists the Pearson correlation (R2) computed for these layers for the V4 and V5 retrievals. The V5_PR shows much higher R2 values than V4_PR for all the layers, and also reflects improvement over V5_FG for all the layers. For the lowest layer (1000–260 hPa) the improvement seen with the V5_PR compared to V5_FG is marginal (R2 values 75% versus 71%). The R2 values for the V4_FG and for the V5_FG are comparable except for the lowest layer where the V5_FG shows a larger correlation (33% versus 71%). Thus, the improvement seen in the V5_PR for the lowest layer is partially achieved through a better initial solution provided by a priori climatology. The R2 values for the ECMWF layer ozone amounts are better than V4_PR, but the V5_PR retrieval shows much better correlation for the lowest layer.

image

Figure 4. Scatterplots of AIRS retrieved layer ozone amounts for layers 1 to 7: 1100–260 hPa, 260–126 hPa, 126–66 hPa, 66–32 hPa, 32–16 hPa, 16–8 hPa, and 8–4 hPa. (a) V4 physical retrieval (V4_PR). (b) V5 physical retrieval (V5_PR).

Download figure to PowerPoint

Table 3. Layer Correlations (R2 * 100) for AIRS V4 and V5 Retrievals and ECMWF Profiles With the O3SND Data
Layer Pressure Boundaries (hPa)Layer NumberSamples (N)AIRS V4 Versus O3SNDsAIRS V5 Versus O3SNDsECMWF Versus O3SNDs
First Guess V4_FGPhysical Retrieval V4_PRFirst Guess V5_FGPhysical Retrieval V5_PR
4–87466563587372
8–1662547977768385
16–3253226260687979
32–6643416259758488
66–12633508189789190
126–26023508188749294
260–110013503339717546

[24] Figures 5a and 5b show the bias and RMS difference statistics computed for V4_PR with reference to O3SNDs (solid squares). Similar statistics for V4_FG (solid circles), and for the ECMWF (solid triangles) are also shown in Figures 5a and 5b. The analysis presented here is for layers 1–6 that span from the surface to 10 hPa (typical sonde burst pressure). An examination of Figures 5a and 5b reveals the following points.

image

Figure 5. Ozone profile retrieval statistics for all the samples accepted by the AIRS V4 first guess (V4_FG, solid circles), physical retrieval (V4_PR, solid squares) and the ECMWF (ECMWF, solid triangles) with reference to WOUDC global O3SND ascents. (a) Percent bias. (b) Percent RMS difference.

Download figure to PowerPoint

[25] 1. The V4_PR shows a bias less than 5% from 10 hPa to 200 hPa, and increases to about 22% for the lowest layer (1000–260 hPa). The V4_FG shows a slightly negative bias for the stratosphere and upper troposphere, and shows a positive bias in the lower troposphere region. These results are consistent with the data shown as scatterplots for V4 and V5 retrievals (see auxiliary material) and support the contention that the AIRS first guess solution has difficulties with the extreme values of ozone.

[26] 2. The V4_PR shows better skill from 50 to 400 hPa compared to V4_FG. The RMS difference is about 15% for layer 6 (8–16 hPa), about 20% for most of the stratosphere and upper troposphere, and degrades to about 37% near 700 hPa. The V4_PR makes no improvement for the lowest layer (1000–260 hPa). This is expected because the ozone channels have a very limited sensitivity to the lower troposphere (Figure 1), and the V4_PR uses V4_FG as a fallback option. The V4_FG also shows larger RMS difference for the lowest layer. Thus, neither the V4_FG nor the ozone channels used in the V4_PR helps the retrieval for the lowest layer.

[27] 3. The ECMWF shows positive bias for all the layers from the surface to10 hPa, consistent with the layer ozone scatterplots (see auxiliary material). The RMS difference for the ECMWF ozone is slightly better than V4_PR. It may be noted that while the radiosonde temperature and water vapor are assimilated into the ECMWF analysis, ozone data from the O3SNDs are not assimilated. In our earlier analysis [Divakarla et al., 2006], the ECMWF temperature and water vapor shows very good agreement (both bias and RMS difference) with the radiosonde temperature and moisture, partly because of the fact that the model forecasts heavily utilize radiosonde information in the analysis. Another interesting fact is that although the V4_FG solution uses the ECMWF for training, the V4_FG and the ECMWF show significant differences. In the V4 algorithm, the initial solution offered by the fast regression first guess is much harder to characterize because the solution is a convoluted function of the radiances, statistical correlations between temperature profile T(p) and O3 and between other interacting trace gases such as CO and O3, etc. Moreover, the training ensemble used in generating regression coefficients might be better behaved because of selection processes [Goldberg et al., 2003] than the retrieval scenarios considered here. The V4_PR is thus confounded with a poor initial solution, and as a consequence, does little to improve the solution.

[28] Figures 6a and 6b show the bias and RMS difference statistics computed for the AIRS V5 retrieval (V5_PR, solid squares; V5_FG, solid circles). The ECMWF statistics are the same as shown in Figures 5a and 5b, and hence are not shown. A comparison of V4 and V5 statistics from Figures 5a and 5b and Figures 6a and 6b reveals the following.

image

Figure 6. Ozone profile retrieval statistics for all the samples accepted by the AIRS V5 first guess (V5_FG, solid circles) and the physical retrieval (V5_PR, solid squares) with reference to WOUDC global O3SND ascents. (a) Percent bias. (b) Percent RMS difference.

Download figure to PowerPoint

[29] 1. The V5_PR shows slightly larger positive biases for the stratospheric layers compared to V4_PR. For the lowest layer (1000–260 hPa), the V5_PR exhibits a very small positive bias (∼3%) in comparison with V4_PR (∼22%, Figure 5a). These results are consistent with the scatterplots for the layer ozone amounts (Figures 4a and 4b).

[30] 2. The V5_PR in general shows smaller RMS differences for most of the layers. The RMS difference for the stratospheric layers is about 15%, an improvement compared to the 20% RMS difference (layer 4, 66–32hPa) seen with V4_PR. The V5_PR shows dramatic improvement for the lowest two layers (1000–260 hPa and 260–126 hPa) with an RMS difference of approximately 20%. With regards to initial solutions, the V5_FG has smaller RMS differences than V4_FG for most of the layers. The improvement seen with the V5_FG for the lowest layer (also larger R2 value of 71% in Table 3) signifies that the improvement in V5_PR is mostly due to the improved initial solution (V5_FG), and only partly due to the additional ozone channels used in the V5_PR. In contrast to the regression solution used by the V4_PR, the initial solution (V5_FG) derived from the ozone climatology offers flexibility for the V5_PR to improve upon the initial solution. The V5_PR thus addresses the shortcomings of the regression approach and provides a solution from channel observations sensitive to the ozone distribution. The profile retrieval skill offered by the V5_PR is also analyzed using a Taylor diagram [Taylor, 2001] for the coarse layers used in this paper and for the total column ozone (see auxiliary material).

4.1.2. Statistics for Tropics, NH, and SH Polar Stations

[31] To evaluate V4 and V5 retrievals for different regions, subsets of data samples from the tropical stations (±12 h matches from STN IDs 191, 328, 175, 205, and 443), from the SH polar station (±3 h matches from STN ID 101, Syowa, 69.0°S, 39.6°E), and from the NH polar station (±3 h matches from STN ID 089, Ny Alesund, 78.9°N 11.9°E) that have special significance to ozone events are selected. The Antarctica station 101 and the Arctic station 089 have a reasonable number of matches from ±3 h collocations and have samples from all the seasons and experience the ozone variability expected in the Antarctic and the Arctic regions, respectively.

[32] Figures 7a and 7b show the bias and RMS difference plots for the V4_PR for the subset of matches obtained from (1) the Arctic station 089 (large-small dashes with solid circles) and (2) the Antarctica station 101 (small dashes with solid diamonds). Also shown in Figures 7a and 7b are the statistics for the tropics (large dashes with solid triangles). Global statistics from ±3 h matches (reproduced from Figures 5a and 5b, solid line with solid squares), and global statistics computed for ±12 h matches (large-small-small dashes with open squares) are also plotted. Table 4 provides the sample size (N), O3SND layer ozone amounts and the standard deviation in Dobson Units (DU) for these data sets. The total ozone amount shown in Table 4 is a computed quantity from the O3SNDs with estimated ozone values above the balloon burst pressure. The NH station 089 has the highest mean ozone amount (356.6 DU) and has a standard deviation of 56.9 DU. The SH station 101 experiences polar vortex during the southern winter and exhibits large variations during the southern spring due to the ozone hole events. The station also launches a relatively large number of O3SNDs to intensely observe ozone hole events, and as a consequence, shows a smaller mean ozone amount and a large variability. The station data also shows large standard deviation for the layers most affected by ozone hole events. Data for the tropical stations show minimum total ozone and least variability. The tropical data set also shows larger ozone amounts at higher altitudes (source region). The tropospheric layers contribute approximately 13% to the total ozone amount for the tropical data set.

image

Figure 7. Ozone profile retrieval statistics for the AIRS V4 physical retrieval algorithm for the subset of samples from the station 089 (STN ID 089, Ny Alesund, 78.9°N, 11.9°E, V4_PR_STN089, solid circles), station 101 (STN ID 101, Syowa, 69.0°S, 39.6°E, V4_PR_STN101, solid diamonds), from a set of tropical stations (STN IDs 191, 328, 175, 205, and 443, V4_PR_TRP, solid triangles), for all the global samples with 3 h matches (V4_PR, solid squares), and for all the global samples with 12 h matches (V4_PR_12H, open squares). (a) Percent bias. (b) Percent RMS difference.

Download figure to PowerPoint

Table 4. Mean Ozone Amount and SD From O3SNDs for Individual Station Locations (STN IDs 089 and 101), Over the Tropics, and for 3 h and 12 h Global Collocationsa
Layer Pressure Boundaries (hPa)Layer NumberNH Station 089 Ny Alesund 78.9°N, 11.9°ESH Station 101 Syowa 69.0°S, 39.6°ETropical Stations 191, 328, 175, 205, and 443Global ±3 h MatchesGlobal ±12 h Matches
NMean (DU)SD (DU)NMean (DU)SD (DU)NMean (DU)SD (DU)NMean (DU)SD (DU)NMean (DU)SD (DU)
  • a

    SD, standard deviation. Unit is Dobson units (DU).

4–87413.64.8820.52.31228.83.34619.03.412019.54.7
8–1663628.06.33531.95.24356.75.825437.18.857536.910.3
16–3254745.510.34748.011.35066.75.632256.512.375755.312.7
32–6645077.812.05062.426.35238.56.934172.018.381368.921.0
66–12635369.314.35045.820.6546.91.835050.121.283448.825.1
126–26025349.916.45028.55.4544.62.535034.017.583433.620.0
260–100015334.68.15018.62.95417.95.635029.98.983427.79.8
Total O3 53356.656.950276.460.054223.924.6350324.565.9834315.475.8

[33] An examination of the bias and RMS differences (Figures 7a and 7b) for the station 101, station 089 and the tropical stations reveals the effect due to the upward movement of the ozone kernel functions and the associated sensitivity changes (Figure 1) from higher latitudes to the tropics. In general, retrievals over the tropics exhibit smaller RMS differences for the upper stratosphere layers and retrievals over high latitudes exhibit smaller RMS differences for the lower tropospheric layers. In addition, deficiencies in the first guess solution given to the physical retrieval also affect the bias and RMS differences. For the tropics, the ECMWF, and the V4_FG initial solution have large bias and RMS differences (not shown here). Since the ozone channels also have very limited sensitivity to the lower troposphere, and especially so over the tropics (Figure 1), the V4_PR shows large bias and RMS differences for the lowest layers. With regards to station 101 (Syowa, 69.0°S, 39.6°E), many of the retrieved profiles from the ECMWF, V4_FG, and V4_PR fail to reproduce ozone hole characteristics (not shown here) and show large bias and RMS differences. The inability of V4_FG to reproduce ozone hole events could be partly due to the deficiencies in the ECMWF training data used in the generation of ozone regression coefficients [Goldberg et al., 2003]. The training data may not have enough cases to represent ozone hole events from the SH springtime, and the V4_PR is confounded to an incorrect but accepted retrieval because of a flawed first guess solution (V4_FG). With regards to station 089 (Ny Alesund, 78.9°N, 11.9°E), the improvement seen in the RMS difference for the lowest layers is probably due to the increased sensitivity of the ozone channels for the lower troposphere at polar latitudes.

[34] Figures 8a and 8b show similar plots for the AIRS V5 retrievals. The V5_PR shows relatively smaller RMS difference for the tropics (large dashes with solid triangles), and for the Antarctica station 101 (small dashes with solid diamonds). The use of a priori climatology as the initial solution in the V5_PR helps both the tropical cases as well as the SH station, and provides more versatility to the V5_PR to improve upon the initial solution. The RMS difference for the tropics is close to 10% for the uppermost layers, and grows to about 30% for layer 3 (66–126 hPa), 50% for layer 2 (126–260 hPa), and reduces to about 40% for the lowest layer (260–1100 hPa). Both the V4_PR and V5_PR show negative bias for the uppermost layers, and a positive bias for the layers covering the troposphere. Since most of the ozone resides in the midstratosphere source region in the tropics, the small percentage of negative bias for a large amount of ozone nullifies the large positive bias seen for the tropospheric layers 1–2 that contribute only about 13% to the total ozone (Table 4). This, leads to an underestimation of the total ozone in the tropical region. The negative bias for the upper stratospheric layers coupled with reduced positive bias for the tropospheric layers from V5_PR makes the total ozone estimates a little lower compared to the estimates from the V4_PR. The statistics presented for the tropics are found to be consistent with the statistics presented by other investigators using ±3 h AIRS retrieval matches with Southern Hemisphere Additional Ozonesondes [Thompson et al., 2004] and other validation experiments (F. Irion, NASA, JPL, unpublished data, 2006). With regards to station 101, the initial solution from V5_FG helps V5_PR in retrieving the ozone hole characteristics reasonably. With regards to station 089, the V5_PR tends to overestimate ozone for the stratospheric layers, and underestimate ozone for the tropospheric layers. Statistics are also computed grouping station matches covering the Arctic Circle (stations IDs 18, 315, 439, 460) and the Antarctic Circle (station IDs 323, 450). It is found that the statistics computed for the stations 089 and 101 very closely represent the Arctic Circle and the Antarctic Circle respectively. Evaluation of V4_PR statistics for the ±3 h and ±12 h matches reveals that the 3-h matches show approximately 5% improvement in the RMS difference (Figure 7b, solid squares for ±3 h matches, open squares for ±12 h matches), and a minor improvement of few percentage points in bias for the lowest layers 1, 2 and 3 (1000–260 hPa, 260–126 hPa, and 126–66 hPa). The V5_PR shows an improvement of about 2–3% in RMS difference for the lowest layers for the 3-h matches (Figure 8b, solid squares for ±3 h matches, open squares for ±12 h matches). No appreciable change in the bias is observed. The initial solution (V4_FG) used by the V4_PR for a 3 h match and for a 12 h match AIRS observation might be changing on the basis of the fast regression first guess solution. The V5_FG solution from a priori climatology is static for the same day and for the same location for the 3 and 12 h matches of AIRS observations. Consequently, the percent change in RMS difference for the V5_PR is smaller than that of V4_PR. Fioletov et al. [2006] have evaluated the ozone variability and instrument uncertainties using same day matches of ozonesonde data, SBUV/2 and other instrument retrievals. They found that as horizontal matchup distance exceeds 200 km, the standard deviation of differences between SBUV/2 lowest layer (1000–63.9 hPa) retrieval and the ozone measurement has increased markedly. However, differences in the time coincidences within a day had little effect on the standard deviation of ozone differences for that layer. Since the AIRS ozone information comes from a smaller FOV (∼50 km) compared to the SBUV/2 retrieval (∼200 km), and appears to be sensitive to changes in the ozone in upper troposphere and lower stratosphere, we thought that the RMS difference for the two lowest layers with the 12 h collocations would be much larger than 3%. The small change in RMS difference with a larger time collocation window warrants a more detailed study of the spatial and temporal variability in ozone using the measured O3SND profiles.

image

Figure 8. Same as in Figures 7a and 7b but for the AIRS V5 physical retrieval algorithm. (a) Percent bias. (b) Percent RMS difference.

Download figure to PowerPoint

4.2. Total Ozone Validation

[35] The AIRS-Brewer/Dobson (BD) station matches have total ozone data from 50 stations (Figure 3) with a reasonably sufficient number of samples from each station. Statistics are computed for all individual stations that have at least five matches. In addition, matches falling into tropical (23°S–23°N), midlatitude (60–23°N; 60–23°S), and polar regions (90–60°N; 90–60°S) are grouped to compute statistics for each region. Data from all the matches are used to generate global statistics. Sections 4.2.1 and 4.2.2 discuss the results of total ozone validation for individual stations, and for different regions respectively. Section 4.2.3 presents an assessment of the total ozone retrieved from the Aqua-AIRS and Aura-OMI instruments. This assessment has helped to validate the use of OMI maps as a transfer standard for ground truth in evaluating AIRS global grids (discussed in section 4.3).

4.2.1. Individual Stations

[36] Figure 9 shows the scatterplot of total ozone measurements from BD stations plotted against AIRS V4 total ozone from the physical retrieval (V4_PR, blue solid squares), from the first guess solution (V4_FG, red solid circles), and from the ECMWF total ozone (ECMWF, cyan solid triangles). The data consist of 4,096 accepted matches from all the stations from all over the globe. The characteristics seen for the total ozone computed from V4_PR, V4_FG and ECMWF are very similar to the observations made from layer ozone scatterplots (see auxiliary material). The ECMWF (solid triangles) overestimates total ozone with reference to the BD measurements. The V4_FG overestimates smaller amounts of total ozone and underestimates larger amounts of total ozone (solid circles). The V4_PR (solid squares) tries to correct the trend for the extreme values, yet shows difficulty in reaching extreme values. The R2 correlation with respect to BD measurements for V4_PR, ECMWF, and V4_FG are 72%, 79%, and 67% respectively. Figure 10 shows a similar scatterplot for V5_PR and V5_FG. The ECMWF ozone data matches are the same as plotted in Figure 9, and hence are not repeated. The R2 correlation with BD measurements for V5_PR and V5_FG is 85% and 58% respectively. Examination of Figures 9 and 10 reveals that V5_PR reduces the scatter seen with V4_PR, and the improvement in correlation from V5_FG to V5_PR is much larger than the improvement seen from V4_FG to V4_PR, thus illustrating superiority of the V5_PR in retrieving the total ozone. Unlike the V4_FG solution from REG_FG, the V5_FG from a priori climatology used as the initial solution for the V5_PR is static and significantly improves the V5_PR in terms of its ability to characterize averaging kernels. However, the V5_PR, while correcting the V5_FG and reaching the extreme values, still shows a tendency for slight overestimation for ozone values above 250 DU, and slight underestimation for low ozone values. Most of the larger ozone values are from the higher latitudes, and ozone values in the range 220–250 DU are found to be mostly from the tropical stations.

image

Figure 9. Scatter diagram showing global total ozone retrievals from the AIRS V4 first guess (V4_FG, red solid circles), from the physical retrieval (V4_PR, blue solid squares), and from the ECMWF ozone (ECMWF, cyan solid triangles) with reference to Brewer/Dobson station measurements.

Download figure to PowerPoint

image

Figure 10. Same as in Figure 9 but for the AIRS V5 first guess (V5_FG, red solid circles) and for the physical retrieval (V5_PR, blue solid squares).

Download figure to PowerPoint

[37] Figures 11 and 12 show the statistics plotted for V4 and V5 retrievals for each BD station. The percentage bias for V4_PR (solid line with solid squares), V4_FG (small dashes with solid circles) and for the ECMWF (large dashes with solid triangles) are shown in Figure 11a. The RMS differences are shown in Figure 11b (symbols used are same as in Figure 11a). The BD stations are sorted from the SH to the NH on the x axis. The caption for each station has the station number followed by a letter representing the region (P for polar, M for midlatitudes, and T for tropics), and 3 characters representing the latitude and the hemisphere. For example, the polar station 111 at 90S is designated on the x axis as 111_P_90S. Similar plots of bias and RMS differences for the V5_PR and V5_FG are shown as Figures 12a and 12b. The ECMWF statistics are the same as shown in Figures 11a and 11b and hence are not repeated. Total ozone measurements from station 376 (376_M_31N, MRSA Matrough, 31.3°N, 31.4°E) appear to have problems since all the retrievals and the ECMWF show similar signs of biases of the same order of magnitude. An examination of Figures 11 and 12 reveals that except for the station over the Saharan desert (Station ID 002_T_23N, Tamanrasset, 22.8°N, 5.5°E), and for the station at the tip of the South Pole (Station ID 111_P_90S, Amundsen-Scott, 89.9°S, 24.8° W), the V4_FG (solid circles) underestimates total ozone for most of the stations. The V4_PR (solid squares) slightly increases the positive bias seen at these two stations due to a flawed initial solution (V4_FG). For all other stations, the V4_PR shows an overall improvement over V4_FG with a bias in the range –5% to 5% and an RMS difference that is less than 10%. The ECMWF ozone (solid triangles) shows positive bias for most of the stations and is consistent with the results seen for ECMWF profiles (Figure 5a). The V5_PR shows relatively smaller bias and RMS differences (Figures 12a and 12b, solid squares) than V4_PR (Figures 11a and 11b, solid squares) for the stations 111 and 002. The V5_PR also show a better degree of RMS difference with the measurements over the Syowa station (101_P_69S, Syowa, 69.0°S latitude, 39.6°E longitude) that experiences SH springtime ozone hole events. For the tropical stations other than station 002, the V5_PR slightly underestimates total ozone (stations 191, 207, 214). Finally, the V5_PR tends to show a consistently positive bias for most of the midlatitude and high-latitude stations. The characteristics seen for the V4_PR and the V5_PR for the total ozone are mostly in conformity with the ozone profile validations discussed in section 4.1.1 and 4.1.2.

image

Figure 11. Total ozone statistics for the V4 first guess (V4_FG, solid circles), physical retrieval (V4_PR, solid squares), and the ECMWF (solid triangles) data over individual Brewer/Dobson measurement stations. (a) Percent bias. (b) Percent RMS difference.

Download figure to PowerPoint

image

Figure 12. Same as in Figure 11 but for the AIRS V5 climatology first guess (V5_FG, solid circles) and the physical retrieval (V5_PR, solid squares). (a) Percent bias. (b) Percent RMS difference.

Download figure to PowerPoint

[38] The deficiencies observed with V4_PR and the improvement seen in the V5_PR over the stations 002 and 111 can be pointed to the initial emissivity retrieval solution offered by V4 and V5 algorithms. As shown in Figure 3, station 002 is in the Saharan desert region, and the station 111 is at a high elevation on the tip of the South Pole, and both have unique emissivity characteristics. The AIRS V4 emissivity retrieval uses the NOAA regression emissivity product [Goldberg et al., 2003] as a first guess over land. The NOAA approach is based on clear radiances simulated from the ECMWF forecast and a surface emissivity training data set. The training data set used for the AIRS V4 algorithm had a limited number of soil, ice, and snow types and very little emissivity variability in the training ensemble. In the AIRS V5 version, the regression coefficient set has been upgraded using a number of published emissivity spectra (12 spectra for ice/snow, 14 for land) blended randomly for land and ice [Zhou et al., 2008]. These improvements generate a better emissivity first guess for use with the V5_PR, and improve retrievals over the desert regions. The improvement in V5 emissivity also seems to help retrievals over the South Pole station 111 and regions around the Antarctica with similar emissivity characteristics. Nevertheless, the V5_PR bias and RMS difference seen over the Saharan desert station 002 suggest that the V5 algorithm, although better thanV4, may require further refinements for better surface emissivity retrievals. A part of the improvement in V5_PR is also due to the better initial solution (V5_FG) provided by the a priori climatology. With regards to the shortcomings of V5_PR, one of the possible reasons for the consistent positive bias seen for the midlatitude and high-latitude regions could be due to systematic errors in the bias corrected difference between AIRS observations and the calculated radiances. In addition to updates to the fast model fast transmittance coefficients for the 9.6 μm O3 band [Strow et al., 2006; DeSouza-Machado et al., 2007], the V5 retrieval algorithm incorporates a static radiance bias correction for all retrieval channels (L. Strow, unpublished data, 2007). These static tuning coefficients are derived using the difference in AIRS observations relative to calculations using a high-quality set of tropical and midlatitude RS-90 radiosondes coordinated with AIRS overpasses. These calculations use tropospheric temperature and moisture profiles from the radiosondes and retrievals of surface skin temperature, ozone profiles, stratospheric temperatures and moisture derived from AIRS. Therefore, the lack of high-latitude radiosonde profiles in the tuning ensemble may explain the consistent bias at these higher latitudes. In addition, uncertainties related to the retrieved spectral emissivity, errors in the cloud clearing also affect the retrieval error statistics. Further, measurement ambiguities due to interinstrument differences (Brewer/Dobson/Filter) and the type of measurement (direct sun and zenith sky) also affect the statistics. With the availability of increased number of matches in future, the match-up data set could alleviate the effects from all these uncertainties. Overall, the improvements and deficiencies seen with the V5_PR over individual stations can be extended to other regions of similar surface and atmospheric characterization, and should be applicable in the evaluation of global grids. The V5_PR algorithm may require a little more optimization. The major advantage with V5_PR is that the reasons for these deficiencies are understandable, and by performing experiments with matched data sets possible remedies can be implemented in upcoming algorithm updates.

4.2.2. Statistics for Tropics, Midlatitudes, Polar, and Global Regions

[39] Figure 13 shows the bias and RMS differences for the V4_PR and the V5_PR when all the station data matches (except for the stations 002, 111 and 376) are binned into tropics (23°N–23°S), midlatitudes (60–23°N; 60–23°S), and polar region (90–60°N; 90–60°S). Also shown in Figure 13 are global statistics (90°N–90°S) for the whole ensemble. Data from stations 002, 111 and 376 are excluded because the V4_PR has known deficiencies due to surface emissivity for stations 002 and 111, and station 376 has spurious total ozone measurements. For the polar regions, the V5_PR shows an overestimation of about 5% with an RMS difference of about 8%. Results for the midlatitude regions also show a similar pattern with a slight reduction in the bias and RMS differences for both the V4 and V5 physical retrievals. For the tropical regions, the V4_PR shows a very slight underestimation (0.5%), and the V5_PR underestimates total ozone by about 4%. Global statistics from the whole ensemble shows V4_PR with a slightly smaller bias than V5_PR. The V5_PR shows an overestimation by about 4% and is well within the expected precision of about 8%. Thus, in terms of validation statistics, the differences between V4 and V5 are relatively small. However, the V5 algorithm offers a better characterization of the ozone product in terms of its ability to capture trends and seasonality patterns as discussed in the subsequent section.

image

Figure 13. Total ozone statistics for AIRS V4_PR and AIRS_V5_PR with matches from all the stations pooled into different latitude zones: polar (60–90°N; 60–90°S), midlatitudes (60–23°N; 60–23°S), tropics (23°N–23°S), and global (90°N–90°S latitude). V4_PR_Bias (bars with horizontal hatches), V4_PR_RMS (bars with vertical hatches), V5_PR_Bias (bars with upward diagonal hatches), and V5_PR_RMS (bars with downward diagonal hatches).

Download figure to PowerPoint

4.2.3. Assessment of Aqua-AIRS and Aura-OMI Retrievals With BD Matches

[40] As stated earlier in section 3.1.2 simultaneous total ozone matches from Aqua-AIRS, Aura-OMI, and BD measurements are analyzed for a relative performance assessment, and to ascertain that OMI global grid data could be used as a transfer standard for truth to evaluate AIRS global products. Figure 14 shows the scatterplot of AIRS V5_PR (solid squares) and OMI retrievals (open circles) with simultaneous BD measurement matches. The data set has 694 matches spanned over the period October 2004 to December 2005 over 36 BD station locations (Figure 3, black dots inside open squares). Collocated data from station 376 are excluded because of questionable total ozone measurements over the station. The OMI retrievals are overpass matches collocated within 25 km to the BD station locations and have a finer spatial resolution (15 km). The AIRS retrievals are matched within100 km radius, and have a spatial resolution of 50 km. The OMI data shows very good agreement with the BD measurement matches with a correlation of 97% (R2). The AIRS V5_PR retrievals show a little larger scatter and have a correlation of 73% with BD measurements. Some of the scatter seen in V5_PR retrievals could be attributed to temporal and spatial match-up differences, the AIRS field-of-view differences, and the resolution of the AIRS retrieval. Figure 15 shows the bias and RMS differences for the OMI retrievals (dashed line with open circles) and for the V5_PR retrievals (solid line with solid squares) with BD station measurements. The OMI retrieval bias is of the order 3% for most of the stations, and the AIRS retrieval shows an overestimation on the order of 6% for many high-latitude stations. It may be noted that many other investigators have performed a detailed validation of the OMI retrievals with BD measurements using many years of data [Ahn et al., 2004; McPeters et al., 2008] and the analysis presented here with simultaneous OMI and AIRS retrievals is mainly intended to demonstrate confidence in the OMI retrievals so that the OMI derived gridded monthly averages can be used to analyze AIRS V4 and V5 global products.

image

Figure 14. Scatterplot of simultaneous Aqua-AIRS (V5_PR, solid squares) and Aura-OMI (OMI, open circles) total ozone retrievals with Brewer/Dobson station measurement matches.

Download figure to PowerPoint

image

Figure 15. Total ozone statistics for AIRS physical retrieval (V5_PR, solid squares) and for OMI total ozone retrieval (OMI, open circles) for different BD station locations. (a) Percent bias. (b) RMS difference.

Download figure to PowerPoint

4.3. Analysis of Total Ozone Monthly Maps

[41] To analyze the trends, patterns, and annual cycles depicted by the AIRS retrievals, global monthly maps of total ozone are generated for the year 2005 for V4_FG, V4_PR, V5_FG and V5_PR. These maps are evaluated in conjunction with the Aura-OMI and NOAA 16 SBUV/2 maps. The OMI maps are taken as the truth in studying the characteristics of the AIRS retrievals. Figures 16 and 17 show the total ozone gridded monthly maps for February and September of 2005, respectively. Figures 16 and 17 contain monthly averages derived for both the AIRS versions (V4_FG, V4_PR, V5_FG and V5_PR), for the OMI, and for the NOAA 16 SBUV/2 retrievals. The SBUV/2 maps shown in Figures 16 and 17 are appended with National Center for Environmental Prediction (NCEP) objective analysis fields beyond poleward of 80° latitude and for the polar night regions. The OMI maps contain retrievals for the sunlit areas. The AIRS instrument produces retrievals for all conditions, including polar night scenes except when clouds are spatially uniform over the retrieved foot print. All the retrieval maps shown in Figures 16 and 17 reveal in general, the total ozone characteristics expected during NH winter (February 2005) and the SH spring (September 2005). The February maps (Figure 16) show a larger gradient in the NH, and a relatively smoother variation in the SH. The September maps (Figure 17) show larger gradients in the SH, and the depletion of stratospheric ozone that occurs annually over the Antarctic region. As seen with the AIRS-BD matches (Figure 11a), the V4_FG shows an underestimation in both the tropics and the high latitudes (dark blue and yellow in Figures 16f and 17f) in comparison to OMI and SBUV/2 maps (light blue and bright red in Figures 16c and 16d; 17c and 17d). The V5_FG (Figures 16a and 17a) reveals the basic climatology given as the initial solution to the V5_PR (Figures 16b and 17b). Coming to the physical retrieval solutions, the V4_PR (Figures 16e and 17e) has difficulty depicting extreme values such as large ozone values in the polar regions, and small ozone values in the ozone hole region. The trends and patterns observed in the V4_PR are also ambiguous. The V4_FG initial solution (derived from the ECWMF trained regression) is difficult to interpret and thus confounds the ability to interpret the V4_PR solution. In contrast, the V5_FG initial solution (derived from ozone climatology) offered to the V5_PR enables a better separation of the a priori and instrument signals (i.e., the trends and patterns not found in the latitudinal and seasonal dependent a priori are solely derived from the AIRS instrument). The V5_PR (Figures 16b and 17b) shows trends and patterns to a reasonable degree of agreement with the OMI and SBUV/2 (Figures 16c, 16d, 17c, and 17d). The V5_PR derived total ozone is lower than the OMI in the tropics (darker blue tone) and higher in the high-latitude regions (bright red). Again, the assessment from these maps is in conformity with the results seen for tropical and high-latitude station validations (Figures 12 and 15). Difference maps produced (not shown here) from the AIRS and OMI monthly maps complement the validation results seen with the BD measurements. The NOAA 16 SBUV/2 and OMI maps compare fairly well with a bias less than 5% for most of the globe. Even with a slightly larger bias than the V4 retrieval, the V5 total ozone retrieval is well within the expected product goal accuracy, and depicts seasonal trends and patterns in concurrence with OMI and SBUV/2 depictions. It may be noted that the monthly maps analyzed for the whole year 2005 are available as a part of the auxiliary material. Also analysis performed using 2 years of AIRS ozone profile retrievals (2004–2005) to verify Brewer/Dobson circulations features is available as a movie loop (see auxiliary material).

image

Figure 16. Total ozone monthly maps for February 2005. (a) V5_FG, (b) V5_PR, (c) OMI, (d) SBUV/2, (e) V4_PR, and (f) V4_FG retrievals. The SBUV/2 map is appended with NCEP analysis fields beyond poleward of 80° latitude and for the polar night regions. The OMI map contains retrievals for the sunlit areas. The AIRS instrument retrievals contain data for all the regions including polar night regions.

Download figure to PowerPoint

image

Figure 17. Same as in Figure 16 but for the month of September 2005. (a) V5_FG, (b)V5_PR, (c) OMI, (d) SBUV/2, (e) V4_PR, and (f) V4_FG retrievals.

Download figure to PowerPoint

[42] Figure 18 shows the annual cycle depicted by the AIRS V4 and V5 ozone algorithm. Also shown in Figure 18 are the annual cycles derived from the gridded monthly averages of OMI and SBUV/2. Qualitative assessment of these maps reveals that the AIRS V5_PR (Figure 18b) is able to reproduce the annual cycle of ozone and agrees well with OMI and SBUV/2 maps (Figures 18c and 18d). The annual cycle from AIRS V4_PR (Figure 18e) has minor deficiencies in polar regions, and in depicting the ozone hole. Using the AIRS global products generated for the years 2004–2007 we are currently investigating further to quantify the findings. The results presented in this section clearly indicate that the AIRS V5 algorithm with its improved first guess climatology provides a better physical retrieval solution and shows a reasonable degree of agreement with OMI and SBUV/2.

image

Figure 18. Total ozone annual cycle for the year 2005 as derived from the gridded monthly averages: (a) V5_FG, (b) V5_PR, (c) OMI, (d) SBUV/2, (e) V4_PR, and (f) V4_FG retrievals. The SBUV/2 map is appended with NCEP analysis fields beyond poleward of 80° latitude and for the polar night regions. The OMI map contains retrievals for the sunlit areas. The AIRS instrument retrievals contain data for all the regions including polar night regions.

Download figure to PowerPoint

4.4. Summary on AIRS V4 and V5 Assessment

[43] The analysis performed to assess AIRS V4 and V5 retrievals using WOUDC-O3SNDS, BD Measurements, SBUV/2 and OMI global grids reveals the following:

[44] 1. The AIRS retrieved ozone profiles from both the V4 and V5 versions match well with the global O3SNDs. The V5 algorithm significantly improves the retrieval bias and RMS differences for the lower troposphere and especially over the tropical regions, where the V4 retrievals show significant discrepancies with the O3SND measurements. The V5 retrieval biases with global O3SNDs are less than 5% for both the stratosphere and the troposphere. The RMS differences are less than 20% for the upper stratosphere and are close to 20% for the lower stratosphere and the troposphere. A part of the improvement in the retrieval capability should also be attributed to the better initial solution provided by the V5 first guess. The improvement seen for the tropospheric layer (1000–260 hPa) is mainly due to the first guess used in the V5 physical retrieval. The improvement seen in the ozone profile retrievals over the SH regions affected by the ozone hole events is also partly due to the better initial solution provided by the V5 first guess.

[45] 2. Total ozone amounts from both versions agree well with the global BD station measurements with a bias of less than 4% and an RMS difference of about 8%. The surface emissivity retrieval improvements implemented in the AIRS V5 algorithm significantly improves ozone retrievals over desert regions where V4 has inherent difficulties. The V5 physical retrieval improves the total ozone derivations for the ozone hole events that occur annually over the Antarctic region.

[46] 3. Analysis of the V4 and V5 total ozone monthly maps in conjunction with the OMI and SBUV/2 maps reveals the superiority of AIRS V5 retrievals over V4 in depicting seasonal trends and patterns. The V4 fast regression first guess operator for the ozone retrieval seems to provide a solution that is hard to interpret and confounds the physical retrieval solution thus producing ambiguous trends and patterns. The use of an improved climatology as the first guess in the V5 ozone retrieval seems to help the physical retrieval to improve on the initial solution. The V5 retrievals show trends and patterns in considerable agreement with OMI and SBUV/2 monthly maps.

[47] 4. Statistics computed for the tropics, midlatitudes and high latitudes reveal that the V5 algorithm, although better in many ways than V4, slightly underestimates total ozone in the tropics and shows a slight overestimation for the midlatitude and high-latitude regions. Analysis performed on the gridded monthly averages in conjunction with OMI and SBUV/2 maps also confirms the disparity. This could be partly due to the tuning procedures adapted in the V5 retrieval algorithm, and partly due to the damping factor used in V5_PR retrieval. This has to be studied for a possible upgrade and implementation in the upcoming retrieval version. Although the V5_PR algorithm improves retrievals over the desert regions in comparison with the V4_PR retrievals, the improvements seen are moderate and there is room for further enhancements to the surface emissivity retrievals. This also requires a little more exploration and possible upgrades in the upcoming retrieval version.

5. Future Development

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[48] It should be noted that the results presented in this paper represent the current state of the AIRS ozone retrieval. We will continue to conduct experiments with future versions of the AIRS retrieval system. The match-up system currently operating at NOAA/NESDIS could be implemented for other satellite systems that have close orbital times (NASA-Aqua, NASA-Aura, NOAA afternoon satellite series, and other A-train satellites) and have instruments to measure ozone and other geophysical parameters. The in situ ozone data set and the RAOB collocations [Divakarla et al., 2006] can be used as a common validation data set and the retrieved products from these instruments can be intercompared and effectively used to produce merged data products, and in the generation of climate quality products.

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[49] The main advantage offered by the V5 algorithm is its ability to capture trends and seasonality patterns. Unlike the regression solution, the ozone climatology used as the initial solution for the V5 physical retrieval is not influenced by the temperature and moisture profiles, and other trace gas correlations. Thus, the ozone retrieval in V5_PR is solely due to the channel radiances sensitive to ozone distribution and provides a clear picture of the ability of the V5_PR in the retrieval of ozone profile and the total ozone determination. The V5 physical retrieval thus addresses the shortcomings of the regression approach and enables better characterization of the results. Some of the shortcomings observed with the V5 algorithm are the slight underestimation of the total ozone in the tropics, and an overestimation in the midlatitude and high-latitude regions. We are currently focusing on developing updates to the static tuning coefficients used in the current V5 retrieval and performing an information content analysis of the AIRS measurements to fully characterize the dependence of the physical retrieval on the climatology used as an initial solution.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[50] The authors wish to thank the AIRS Science Team members for their comments and suggestions during many presentations at the AIRS Science Team meetings. Thanks are due to Shuntai Zhou for many discussions and for providing SBUV/2 data used in this paper. The OMI overpass retrievals used in this paper have been downloaded from the TOMS Web site. Murty Divakarla thanks Vitali Fioletov of Canadian Meteorological Services for his initial help with the WOUDC ozonesondes and for many clarifications and e-mail exchanges. Suggestions and comments from Frank Tilley have helped to improve the style and organization of the manuscript. Thanks are also due to Eric Beach for discussions on the WOUDC total ozone DVD and other data sets. Finally, the authors would like to thank the personnel and agencies responsible for operating the ozonesonde, Dobson, and Brewer stations without whom this validation study would have been impossible. The manuscript contents are solely the opinions of the authors and do not constitute a statement of policy, decision, or position on behalf of the NOAA, NASA, or the U.S. Government.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Validation Data
  5. 3. Data Set Preparation
  6. 4. Results and Discussion
  7. 5. Future Development
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

Auxiliary material for this article contains four figures, two animations, and one table.

Auxiliary material files may require downloading to a local drive depending on platform, browser, configuration, and size. To open auxiliary materials in a browser, click on the label. To download, Right-click and select “Save Target As…” (PC) or CTRL-click and select “Download Link to Disk” (Mac).

See Plugins for a list of applications and supported file formats.

Additional file information is provided in the readme.txt.

FilenameFormatSizeDescription
jgrd14361-sup-0001-readme.txtplain text document8Kreadme.txt
jgrd14361-sup-0002-fs01.epsPS document546KFigure S1. Scatterplots of layer ozone values in Dobson Units (DU) for layers 1–4 covering the pressure slabs 1000–260 hPa, 260–126 hPa, 126–66 Pa, and 66–32 hPa, respectively.
jgrd14361-sup-0003-fs02.epsPS document435KFigure S2. Similar to Figure S1 but for the AIRS V5 retrievals.
jgrd14361-sup-0004-fs03.epsPS document25KFigure S3. “Profile skill” of the retrieval using a Taylor Diagram for the AIRS V5 retrievals for 5 coarse layers spread throughout the atmosphere, and for the total column ozone.
jgrd14361-sup-0005-fs04.epsPS document12KFigure S4. Taylor diagram for the layers used in the manuscript.
jgrd14361-sup-0006-ms01.gifGIF image12556KAnimation S1. Animation of version 5 retrieved ozone profiles for the period October 2004 to December 2005.
jgrd14361-sup-0007-ms02.gifGIF image1517KAnimation S2. Gridded monthly average maps of total ozone (in Dobson Units, DU) computed for both the AIRS versions (V5_FG, V5_PR, V4_FG, V4_PR), for the OMI, and for the NOAA 16 SBUV/2 retrievals.
jgrd14361-sup-0008-ts01.txtplain text document3KTable S1. List of ozonesonde (O3SND) stations and the BD total ozone measurement stations used in the evaluation.
jgrd14361-sup-0009-t01.txtplain text document1KTab-delimited Table 1.
jgrd14361-sup-0010-t02.txtplain text document1KTab-delimited Table 2.
jgrd14361-sup-0011-t03.txtplain text document1KTab-delimited Table 3.
jgrd14361-sup-0012-t04.txtplain text document1KTab-delimited Table 4.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.