Ozone profile retrieval from limb scatter measurements in the HARTLEY bands: methodology, algorithm description, sensitivity studies, and validation

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. Ozone profile retrieval from limb scatter measurements in the HARTLEY bands: methodology, algorithm description, sensitivity studies, and validation G. J. Rohen, C. V. Savigny, J. W. Kaiser, E. J. Llewellyn, L. Froidevaux, M. López-Puertas, T. Steck, M. Palm, H. Winkler, J. P. Burrows, et al.


Introduction
Retrievals of mesospheric ozone profiles from satellite limb scatter measurements using the backscatter technique have been done successfully for the first time from Solar Mesosphere Explorer (SME) spectra (Thomas et al., 1980;Rusch et al., 1983).An attempt to retrieve ozone in the Hartley bands from spectra measured by the Shuttle Ozone Limb Sounder Experiment (SOLSE) failed (McPeters et al., 2000) but introduced the retrieval approach using the so-called visible triplet instead.(Flittner et al., 2000).
Mesospheric ozone profiles are of special interest, e.g., for investigating of mesospheric properties or Sun-Earth interactions like solar proton events (Rohen et al., 2005) or solar cycle effects.The may also be required for the modeling of the atmospheric dynamics.At least, they help understand the atmospheric chemistry in the upper stratosphere and lower mesosphere.

EGU
This article presents results from mesospheric ozone profile retrievals from SCIAMACHY limb spectra using the backscatter absorption technique.A special attention is payed to the ozone content at highest possible altitudes by this special retrieval technique if wavelengths in the center of the Hartley bands are used.Also, disturbing emissions and stray light from outside the field of view of the spectrometer, as well as other error sources are discussed which may be a reason for the overestimation of the retrieved profiles above 50 km.

SCIAMACHY limb measurements
The Scanning Imaging Absorption spectroMeter for Atmospheric CHartographY (SCIAMACHY) (Bovensmann et al., 1999) is one of ten instruments on board ESA's Environmental satellite (Envisat).The satellite was launched in March 2002 into a sunsynchronous polar orbit at 799.8 km altitude with an inclination angle of 98.55 • .This implies a fixed equator crossing time for the descending node; the local overflight times vary mostly around 10:00 a.m. during the day, depending on the latitude (see Fig. 1).
The SCIAMACHY spectrometer itself consists of eight channels between 214 and 2384 nm.The first channel -whose measurements are used for the presented profile retrieval -covers the lowest spectral region up to 314 nm with a spectral resolution of 0.21 nm.
SCIAMACHY's instantaneous field of view in limb mode is 110×2.6km (horizontal×vertical) at the tangent point.The horizontal across track coverage is 960 km with a resolution of 240 km at best.During a vertical scan within 60 s, Envisat moves 447 km; during the same time period, the tangent point moves due to the sphericity of the Earth in the opposite direction by about 145 km; both movements of Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
EGU the tangent point result in an along track resolution of approximately 300 km.
Up to now, stratospheric ozone (up to 35 km) (von Savigny et al., 2003a), bromine oxide, nitrogen dioxide and chlorine dioxide (Rozanov et al., 2005) concentration profiles have been inferred successfully from SCIAMACHY limb scatter measurements.

Methodology
The retrieval approach is based on the fitting of measured and modeled limb radiance profiles at discrete wavelengths, i.e., small spectral integration ranges of 2 nm.This technique was first proposed for air-borne limb measurements (e.g., Cunnold et al., 1973;Aruga and Heath, 1982) but was soon applied to SME satellite measurements (Rusch et al., 1983).
Although basically the maximum of ozone absorption cross sections in the Hartley bands (around 255 nm) appears more appropriate for ozone profile retrievals at higher altitudes (Wayne, 1987;Burrows et al., 1999;Bogumil et al., 2003), only wavelengths above 265 nm have been employed by now (Rusch et al., 1983), likely due to a low instrument signal.Higher accuracy is proposed if multiple wavelengths are used (Aruga and Heath, 1982).
A proper fit over a continuous wavelength range is difficult to obtain because emissions contribute to the measurements at large tangent heights.
No justification for the selection of wavelengths have been given for previous ozone profile retrievals.The inversion technique applied to SME measurements, up to now the solely successful ozone retrieval from satellite-borne limb backscatter measurements in the ultraviolet-B, exploits radiances at 265 and 296.4 nm (Rusch et al., 1983).
Ozone profile retrievals at wavelengths in the Hartley bands from SOLSE observations (McPeters et al., 2000) failed due to a bad signal-to-noise ratio, and even in the related theoretical elaboration (Flittner et al., 2000) no reasons for a selection of wave-  lengths at 300, 310, 322, 355, 525, 600, and 675 nm have been given.For their routine profile retrieval, radiances at eleven discrete wavelengths between 300 and 355 nm have been used (McPeters et al., 2000).Contamination of the spectra by emission features may originate from atmospheric trace gases, or also from space debris entering the atmosphere (see, e.g., Carbary et al., 2003;Fritzenwallner and Kopp, 1998;Clemesha, 1995).

EGU
In the ultraviolet, the NO-γ emissions (see, e.g., Herzberg, 1950) impact considerably limb measurements at high altitudes (see Fig. 2) because of the increasing NO concentration with altitude.Their contribution to the total limb radiances, e.g., at 80 km tangent height reaches 20 % at 255 nm; below 220 nm, these can be as strong as the ambient radiation (L ópez-Puertas, 2000).The Vegard-Kaplan bands of N 2 (X 1 Σ + g ←A 3 Σ + u ) overlay the entire ultraviolet and visible spectral region but usually contribute less than 1 % to the ambient radiation.Several N 2 Second Positive (2P) emissions mainly above 295 nm are negligible (L ópez-Puertas, 2000).Three atomic oxygen lines are also expected in the ultraviolet; two are caused by the ( 2 P− 4 S) transition at 247.0 and 247.1 nm and another at 297.2 nm, originating from the ( 1 S− 3 P) transition (Rees, 1989).An averaged SCIAMACHY limb spectrum indicates emissions in the ultraviolet (Fig. 3).The NO-γ electronic transitions dominate the spectrum.Only emissions of magnesium at 280 nm can be identified, other metal emissions are not observable at the wavelength range of interest.
Figure 4 displays the shape of limb radiance profiles in the ultraviolet.In an optically thin atmosphere, limb radiance increases with decreasing altitude due to enhanced scattering.At a certain tangent height, however, the atmosphere becomes optically thick due to absorption (e.g., ozone) and Rayleigh-scattering, and no additional pho-

EGU
tons from the outer line of sight reach the instrument.Below this "knee", radiance decreases or increases again depending on the ratio of Rayleigh scattering and ozone absorption.
In the first step of the inversion, the ultraviolet limb radiances I ik (i denotes the index for wavelengths and k the index for tangent heights) are normalized with respect to the limb radiance I ik ref at a certain empirically chosen reference tangent height above the knee, reducing multiplicative errors.For a better convergence, the base 10 logarithm is used to compose a retrieval vector 5 of 13 wavelengths×21 tangent heights=273 elements.For better convergence the base 10 logarithm is used.
The inversion of the measurement vector is performed by a nonlinear Optimal Estimation iteration scheme (Rodgers, 1976), where x n+1 corresponds to the logarithm of the ozone profile estimate after n+1 iterations.x 0 denotes the logarithm of the a priori profile which is provided by the five-year United Kingdom Universities Global Atmospheric Modeling Programme (UGAMP) climatology (Li and Shine, 1995) based on SME, Solar Backscatter UltraViolet (SBUV) (Fleig et al., 1990), Stratospheric Aerosol and Gas Experiment (SAGE II) (McCormick et al., 1989), and on ozone sonde measurements.y denotes the measurement vector.y n is the modeled measurement after n iteration steps.K in Eq. 2 denotes the respective weighting function matrix and S x 0 the a priori covariance matrix.The off-diagonal elements of the covariance matrix are the covariances

EGU
with respect to the elements x a,i and x a,j taken from the climatology.The following a priori covariance matrix was used, (3) (σ x 0 ) ii is the standard deviation of the climatology and is set to 0.65, z(i ) and z(j ) are the corresponding altitudes, the length scale h is set to 3.3 km.The measurement variance matrix S y is set to a standard deviation of 0.1.The iterative inversion is truncated if (see Rodgers, 1976) is below a predefined value which is found empirically.Here, the summands are and The radiative transfer model SCIARAYS (Kaiser and Burrows, 2003) derives single Rayleigh and aerosol scattering, refractive bending and trace gas absorption in a fully spherical geometry.Aerosol absorption, clouds, and thermal emissions are not considered.GOME FM ozone cross sections (Burrows et al., 1999) are used for the derivation of the ozone absorption.Weighting functions are derived analytically.
Air densities and temperatures are provided by a monthly resoluted climatology derived by a chemical transport model, employing pressure and temperatures (Nagatani and Rosenfield, 1993) and ozone columns (McPeters, 1993) from a compilation of several models and measurements (McLinden et al., 2000).The spatial resolution of this climatology is 5 • both in longitude and latitude.
Although SCIARAYS derives also Rayleigh scattering up to second order, only first order of scattering is used for this inversion since ozone in the Hartley bands absorbs Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU massively and allows almost no multiple scattering.SCIARAYS derives a solution of the radiative transfer equation, in integral form, i.e., radiances on each path through the atmosphere are integrated.The relative contributions of the respective paths as modeled by SCIARAYS are depicted in Fig. 5, demonstrating the dominance of single scattering in the spectral region of interest in this work.
The retrieval methodology provides sensitivity to ozone between about 30 and 60 km, sometimes 65 km (see first results Rohen et al., 2005).Weighting functions (Fig. 6) indicate a continuous coverage at all altitudes.
Figure 7 shows that fits and residuals for all thirteen wavelengths.Averaged residuals between 35 and 65 km are below 3 % (Rohen et al., 2005).

Pointing errors
The accuracy of the tangent height registration can be improved by different techniques, e.g., using the shape of the radiance profiles (Janz et al., 1996;McPeters et al., 2000;Sioris et al., 2003;von Savigny et al., 2003b), or using CO 2 emissions (von Clarmann et al., 2003).For this presented retrieval, Tangent height Retrieval by Ultraviolet-B Exploitation (TRUE) VERSION 1.4 (Kaiser et al., 2004;von Savigny et al., 2005a) was employed which applies the former technique.TRUE is supposed to improve the accuracy to 500 m (von Savigny et al., 2005a).
An example indicates the errors caused by incorrect tangent heights for a sample profile retrieval (Fig. 8); an error of 2.5 km results in 90 % difference in ozone, and a shift of the radiance profile by 0.5 km leads to an error of up to 17 %.Error increases with altitude due to decreasing ozone scale heights.The retrieval is sensitive to the solar zenith angle, in particular at large solar zenith angles whereas the solar azimuth angle has almost no impact on the retrieval (von Savigny, 2002).Angles are registered for the left and for the right side, and for the center of the field of view; the latter is taken for the retrieval.Both the movement of the satellite and the averaging along the swath of 960 km can cause an angle inaccuracy of up to 2 • .The effect of this inaccuracy is shown in Fig. 9, for a small and a large solar zenith angle, respectively.Induced errors reach 13 % whereas almost no impact is observable at small solar zenith angles.

Stray light
External stray light is referred to as the contamination from outside the field of view, e.g., from cities or clouds.Stray light increases strongly where the signal-to-noise ratio is low, e.g., at high altitudes.
External stray light has been investigated, e.g., for the Optical Spectrograph and InfraRed Imaging System (OSIRIS) (Llewellyn et al., 2004); at 400 nm, it reaches 65 % at 70 km tangent height, and even 440 % at larger wavelengths and higher altitudes (Llewellyn and Gattinger, 1998).
The SCIAMACHY stray light exhibits approximately the same magnitude as those of OSIRIS (van Soest, 2005).By subtracting the background signal, channel 1 of SCIAMACHY was found to be not as strongly affected as the channel 2 whose contaminations at high tangent heights can be five to ten times larger than the ambient radiation.
In our sensitivity study, SCIAMACHY stray light was estimated at 310 nm by subtraction of the modeled from the measured radiances.Radiances smaller than 310 nm are less affected (van Soest, 2005), see also Fig. 7.The impact on a sample retrieval is shown in Fig. 10.

EGU
Stray light at 76 km reaches 120 % of the in-field signal whereas the impact on the profile retrieval is small.Induced errors never exceed 1 % between 35 and 65 km; in summary, stray light has a large impact on the measurements only above around 70 km, and hence has almost no effect on the retrieval in the region of interest between 30 and 60 km.This shows that the overstimation of the SCIAMACHY profiles cannot be reasoned only by this error source.

Scattering modes
The small impact of multiple scatter events in the Hartley bands has been demonstrated by model studies (Oikarinen et al., 1999;Loughman et al., 2005).Oikarinen et al. (1999) derived the single scattering contribution below 300 nm to be larger than 99 %.
Figure 5 shows the fraction of double scattered radiation in the atmosphere as derived by SCIARAYS.At 60 km, this fraction is about 5 % at 310 nm and decreases to 1 % below 290 nm.
Figure 11 now shows the impact of double scattered photons on the retrieval.Maximum multiple scattering is expected for SCIAMACHY measurements at about 35 • solar zenith angle, which is approximately the minimum solar zenith angle along SCIAMACHY orbits.Omitting twice scattered photons causes retrieval errors below 3 %, even at 33 km.Thus, in order to save computing time, only single scattering is considered for the retrieval.
Noctilucent clouds (Gadsden and Schr öder, 1989;Kokhanovsky, 2005;von Savigny et al., 2005c) affect the majority of limb measurements at polar latitudes in the summer hemisphere (von Savigny et al., 2005c EGU not usable for retrievals.An upstream cloud detector was implemented to reject measurements with noctilucent clouds signatures. The negligible impact of aerosols in the upper stratosphere has already been shown for OSIRIS as well as for SCIAMACHY (von Savigny, 2002;von Savigny et al., 2005b).Thus, aerosols are not considered for the retrieval.

Ground albedo
The contribution of reflected photons to SCIAMACHY limb radiance measurements below 310 nm is less than 1 % due to massive ozone absorption (Fig. 5).Ground albedo is therefore also expected to have a negligible impact on the retrieval of profiles above 35 km.
Figure 12 shows the induced errors due to omitting surface reflection in the case of a SCIAMACHY measurement at a relatively small solar zenith angle.The contribution of reflected photons at 50 km tangent height is almost zero and increases to 2.2 % at 33 km which agrees with findings for stratospheric ozone profile retrievals with maximum errors between 3.5 and 5 % (55 • solar zenith angle, 40 km altitude von Savigny, 2002).

A priori information
Presentations of the averaging kernels show the independence of the retrieval on a priori information (Fig. 3 of Rohen et al., 2006) for altitudes between 35 and 60 km.Retrieved ozone profiles using different a priori information (50 % difference) exhibit up to 4 % differences (Fig. 13).
At the edge of the retrieval sensitive altitude range, the a priori affects also the effects of the other error sources on the retrieval; the a priori may amplify or lower the effect of the error source depending on the relative position of the a priori and the ozone guess from the fitting.Since the sensitivity studies are based on different a priori profiles from the climatology this may reason why errors are sometimes higher or lower than found Introduction

Conclusions References
Tables Figures Printer-friendly Version Interactive Discussion

EGU
through the sensitivity studies.We will use this later as one explanation for the high differences at the profile comparisons.
4.8 Temperature dependence of ozone cross sections, in the derivation of density, background density Tabulated Global Ozone Monitoring Experiment (GOME) flight model (FM) ozone absorption cross sections (Burrows et al., 1999) are interpolated for usage in the retrieval, available at five temperatures differing by 20 K.
Figure 14 shows the errors induced by applying improper cross sections at 273 K and 293 K.
At lower altitudes, i.e., where ozone absorbs in the Huggins bands, errors of up to 4 % are induced.Absorption at higher altitudes (i.e., in the Hartley bands) is almost independent on temperature.
The temperature also determines the air density through the ideal gas law.The effect caused by lowering the air density by 5 % (which is a rough estimate) is depicted in Fig. 15.
Table 1 gives an overview of the sensitivity studies.

Validation
The validation presents comparisons with ozone profiles from MIPAS on Envisat, the Radiometer for Atmospheric Measurements (RAM) in Spitsbergen, the HALogen Occultation Experiment (HALOE) on the Upper Atmospheric Research Satellite (UARS) and the Microwave Limb Sounder (MLS) on AURA.SCIAMACHY (L0-L1) processor VERSION 5.4 and TRUE VERSION 1.4 have been used.
Figure 16 shows the overall statistics for all instruments which are discussed in the next sections.
The mean difference is below 10 or 20 %, depending on the altitude, with mean 1σ It should be noted at this place that also observations of ozone depletion during the Halloween solar storm -using profiles from the presented profile retrievalagree fairly well with observations of e.g., Global Ozone Monitoring by Occultation of Stars (GOMOS), or the Solar Backscatter Ultraviolet Instrument (SBUV II) (Rohen et al., 2005).Those comparisons may also help to state the quality of the retrieval, even for measurements made under extreme conditions like a during a solar storm.
We will discuss the comparisons as shown in Fig. 16 in more detail, beginning with MIPAS.
MIPAS and POAM III (Lumpe et al., 2003) ozone profiles between 10 km and 60 km show agreement within 10 %, as well as comparisons with several LIDAR and ozone sonde measurements, and with ozone profiles from HALOE (Steck et al., 2007).MIPAS profiles have also been compared to SCIAMACHY stratospheric ozone observations and those of Global Ozone Monitoring by Occultation of Stars (GOMOS) (Bracher et al., 2005).In terms of GOMOS, the agreement of the compared profiles is also within 10 % up to 60 km altitude.
For comparisons, collocated measurements must have been performed at the same solar zenith angle, the maximum difference of the solar zenith angle is set to less than 2 • .The number of coincident SCIAMACHY and MIPAS profiles is large due to the fact that both instruments are on Envisat.Thus, the spatial radius of coincident observations can be chosen small, 100 km or 200 km.
The general agreement is fairly good, but increasing differences with altitude are 12109 Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU observable which are enlarged above about 50 km.The increasing differences with altitude are due to the increasing ozone scale height.The reason for the increased differences above 50 km may be due to the fact that the TRUE retrieval uses ozone climatologies below 50 km what may result in an more inaccurate tangent height retrieval above this altitude.Solar zenith angle as a possible reason is most likely excluded since this error sources affect all altitudes in the same extent.The same is valid for the a priori error source.Stray light was shown to have not such strong effects on the retrieval (see the sensitivity studies).

Possible latitudinal and solar zenith angle dependence
Comparisons with MIPAS profiles in the polar and mid latitude region including the tropics are shown in Fig. 17 which indicate only a slight but characteristic dependence of the profile retrieval on the latitude.The already mentioned overestimation above 50 km can be seen for all latitude bands.Increasing differences at 45 to 50 km -possibly caused by a wavelength gap around 300 nm in the retrieval -can also be seen, but only as a weak feature at midlatitude bands.In contrast to that, the increasing differences due to the tangent height misregistration is clearly observable at midlatitudes but not that clear at the polar regions.This may also be related to the fact that the TRUE tangent height retrieval uses only tropical ozone climatologies.
Almost the same features like in Fig. 17 can be seen at the comparisons for different solar zenith angles (Fig. 18).Again, a clear overestimation between 50 and 60 km altitude can be observed in all figures, but the previously described feature at 45 km appears only at solar zenith angles below 60 • , what shows that this feature is not caused by the methodology solely.It should be noted as this place that only measurements at solar zenith angle below 83 • are used for the profile retrieval.Thus, Fig. 18 has no expressiveness for solar zenith angles above this number of degree.Anyway, the retrieval is almost not affected by the position of the Sun, although, again, the main already described increasing differences with altitude remain.Next, comparisons with profiles from the HALOE instrument will be shown.The current HALOE UARS (Russell et al., 1993) ozone retrieval VERSION 1.9 provides profiles up to 70 km.Comparisons have been done with numerous instruments, including ozone sondes, LIDARs, balloons, rocketsondes, and satellites (Br ühl et al., 1996;Rusch et al., 1997).HALOE profiles up to 50 km are in good agreement with SAGE II measurements.Above 50 km, HALOE observations are supposed to overestimate ozone by up to 20 % due to incorrect photochemical factors (Natarajan et al., 2005;Nazarayan et al., 2005), but this needs to be confirmed by more statistical analysis in the future, since only samples have been taken in order to show the improvements.

Photochemical modeling of solar occultation observations
Solar occultation measurements like made by HALOE are performed at local times which are characterized by minimum ozone concentrations (see Fig. 19).
In order to compare those HALOE solar occultation observations with SCIAMACHY limb observations, collocated HALOE measurements have been modeled photochemically, using a one dimensional version of the Single Layer Isentropic Model of Chemistry And Transport (SLIMCAT) model (Chipperfield, 1999) (Fig. 20).
For instance, differences between ozone at 90 • and 74 • solar zenith angle are of up to 20 % below 60 km and up to several hundred percent above 60 km.The detailed model errors and a summary of uncertainties for SCIAMACHY and HALOE ozone profiles are shown in Rohen (2006) and will therefore not be shown a second time.
The mean differences between the compared profiles indicate a fairly good agreement in terms of the statistical mean (Fig. 16).But the modeled HALOE ozone profiles provide large errors and make it therefore almost impossible to give a reliable statement about the quality of the profiles, at least above 50 km.Below 50 km, the agreement is fairly good, like at the comparisons with MIPAS profiles.However, this may confirm at least the good quality of the retrieved profiles below 50 km.78• N, 11 • E), Norway (Langer, 1999).
RAM is a millimeter-wave radiometer tuned to the frequency of an ozone transition line at 142 GHZ.Retrieved ozone profiles cover 15 to 55 km and exhibit a vertical resolution between 8 km and 20 km (Palm et al., 2005).RAM profiles are validated by ozone sondes between 18 km and 24 km within 10 %, and with LIDAR instruments between 16 km and 34 km within 11 % (Palm et al., 2005).MLS observations have been used for validation between 20 and 55 km which also show an agreement within 10 % (Langer, 1999).
In order to compare SCIAMACHY and RAM ozone, SCIAMACHY ozone profiles were convolved by the averaging kernels of the RAM ozone retrieval.In spite of the poor vertical resolution, collocated ozone profiles indicate an agreement within 10 % (Fig. 16).Only a slight overestimation is observable above 50 km, but this smaller differences above 50 km -as seen in the MIPAS comparisons -may be due to the poor vertical resolution of the RAM profiles.Again, a reliable statement of the quality of the retrieved profiles above 50 km through this comparison is difficult to give.

Comparisons with MLS AURA measurements
Profiles as retrieved from MLS AURA may clear the quality of the retrieved profiles at higher latitudes at last.Two years after launch, the first early validation analyses of VERSION 1.5 MLS on AURA ozone profiles have been published (Froidevaux et al., 2006).Validation has been done up to altitudes of 0.1 mbar (about 48 km) with ozone profiles measured by HALOE, Stratospheric Aerosol and Gas Experiment II (SAGE II), POAM III, and Advanced Composition Explorer (ACE).The MLS ozone profiles agree well within 5 % with the compared profiles in the lower stratosphere, but also an underestimation in

EGU
the upper stratosphere and an overestimation towards the mesosphere is observable.
Although an upper limit of 0.46 mbar (about 54 km altitude) was recommended for the ozone product, good sensitivity in the upper mesosphere is also proposed (Livesey et al., 2005).
A statistics from collocated SCIAMACHY and MLS profiles within 300 km radius (Fig. 16, right, bottom) shows a remarkable good agreement up to 55 km, but again an overestimation of the SCIAMACHY profiles above that height.But, in contrast to the MIPAS comparisons, deviations become considerable only above approximately 56 km, 5 km lower than in the case of MIPAS, and the effects of the tangent height misalignment is less stronger.The increased differences at 45 km are also indicated.

Conclusions
We used SCIAMACHY/Envisat limb scatter measurements in the Hartley bands of ozone to retrieve ozone profiles in the upper stratosphere and lower mesosphere between approximately 35 and 65 km, with a special attention to prove whether this methodology is capable to retrieve information about the content at higher altitudes up to 70 km.At least, sensitivity at higher altitudes (up to 70 km) was reached at large solar zenith angles, but in general, good information content can be archieved up to about 60 or 65 km.Since one purpose of this work was to provide a methodology to retrieve ozone at altitudes as high as possible through the backscatter technique, thirteen discrete wavelengths near the maximum of the ozone absorption cross sections below 255 nm have been chosen for the inversion.
In general, the retrieval provides fairly reliable ozone profiles within 10 % accuracy up to 50 or 55 km.Above this altitudes, the observations overestimate the true ozone content by up to 30 % at the highest altitudes of sensitivity.This is most likely reasoned by the worser tangent height retrieval with increasing altitude whose induced errors are characterized by increasing differences with altitude.Stray light was found to contami-Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU nate the limb radiances above 60 km to a remarkable extent up to 100 % relative to the ambient radiation at 75 km, but this error source was also identified to have no impact on the profile retrieval in the altitude range of interest.
Retrieval errors are also caused by inaccuracies of the solar zenith angle specification, although to a smaller extent and only at measurements at large solar zenith angles.
Comparisons with profiles from MIPAS on Envisat, HALOE, a ground based radiometer in Norway, and MLS on board AURA are mainly within the expected errors as found through sensitivity studies.The comparisons show agreement within 10 % in the upper stratosphere, and indicate an overestimation in the lower mesosphere.HALOE profiles have been modeled to the solar zenith angle condition of the collocated SCIAMACHY measurements.Comparisons of measurements in the tropics are more clearly affected by the tangent height errors than at those in the polar regions.
In summary, the methodology applied to the SCIAMACHY instrument provides reliable ozone profiles in the upper stratosphere, but reaches its limits in the mesosphere.The improvement of the tangent height misregistration will play a key role for a better ozone product from the SCIAMACHY limb scatter measurements.For the most of the scientific questions in the aimed altitude range, the retrieved ozone profiles will be accurate enough, e.g., to investigate the morphology of ozone with respect to solar cycles or atmospheric events like during solar storms if the overestimation above 50 or 55 km is considered.

Table 1 .
Eichmann, K.-U., No ël, S., Rozanov, V., Sinnhuber, B.-M., Weber, M., Burrows, J. P., and Kaiser, J. W.: The ozone hole breakup in September 2002 as seen by SCIAMACHY on Envisat., J.Atmos.Sci., 62, 721-734, 2005b.Introduction Summary of the sensitivity studies.The errors are given in %.Envisat's local overflight times, beginning at 0 • latitude on the night side of the Earth.Although not indicated here, both Earth poles are not overflown in a regular orbit due to an inclination angle unequal 90 Fits of limb radiance profiles for all thirteen wavelengths (sample profile retrieval, orbit 10945, number 1521, state 3).Dots in the respective left panels denote the measurement, the solid lines the model.Radiance profiles are normalized at a certain altitude, respectively.Residuals are depicted in the right panel of the figures.Above 70 km, the measurements are underestimated by the model due to stray light contaminations.Introduction Errors of sample profile induced by virtual tangent height shifts of -2.5 km (left) and -0.5 km (right).Introduction Retrieval errors caused by an inacurracy in the solar zenith angle of 2 • at 82 and 50 • ((orbit 05387, number 1244, state 9) and (orbit 05387, number 1244, state 4)).Introduction