ESA Earth Home Missions Data Products Resources Applications
    24-Jul-2014
EO Data Access
How to Apply
How to Access
Index
  Acknowledgements and contact
List of figures and tables
List of tables
 Additional information on the GOMOS measurements
Periods of data unavailabilities
Evolution of the IPF
Glossary
List of abbreviations and acronyms
Product types and structure
Other sources of information
GOMOS-related ESA web pages
Publications (peer-reviewed publications and proceedings of conferences)
Secondary products
Scintillation and turbulence
Aerosols and PSC
Product validation
NO2 and NO3 measurements
O3 measurements
CAL/VAL activities
Retrieval and processing issues
Assimilation of GOMOS products
Mesosphere
Specific events
GOMOS-related theses
Other technical reports by members of GOMOS SAG, ESL and QWG
Level2 processing
How to
FAQ
Access tools
GOMOS products toolbox
EnviView
Data selection
PCD summary in the Level2 products
Obliquity
Star properties
Data availability
Presentation of the GOMOS products
Description of the products
Product content
Product structure
Time availability
Data size
Presentation of the instrument and the measurements
Scientific achievements
Validation results
Measurement characteristics
Accuracy
Occultation obliquity
Geographical and time coverage
Star characteristics
Mission planning
Modified mission scenario since August 2005
Instrument description and measurement principle
Calibration phase and monitoring activities
Measurement technique
GOMOS Product Handbook
Content of the products
Services
Site Map
Frequently asked questions
Glossary
Credits
Terms of use
Contact us
Search


 
 
 


1.5.4 Accuracy

The special feature of the GOMOS retrieval is that the accuracy of a species product varies from one occultation to another. The key factors are the characteristics of the occulted star, and some specificities of the occultation geometry. We discuss here how those key factors may impact the accuracy of the products, and we present the vertical profiles of accuracy for the local density of O3, according to the star type and to the occultation geometry.

In order to select the highest quality data, it is recommended to apply some criteria on the stars and on the occultation geometry. Those criteria are detailed in the Chapter "How to?".

1.5.4.1 Effect of the star characteristics

1.5.4.1.1 Star brightness

Due to the weakness of the stellar fluxes, the star visual magnitude is an important factor to the accuracy; it directly affects the S/N ratio. This is a crucial factor for NO2, NO3 and H2O profiles, for which only the occultations of the brightest stars should be used. It also affects the quality of O3 vertical profiles in the stratosphere, for which occultations of dimmest stars produce noisy values of the profiles in the lowest stratosphere.

The star brightness for each measured occultation is provided in the corresponding products.

1.5.4.1.2 Star temperature

Above 30 km, where ozone retrieval is based on UV-wavelengths, the ozone profile retrieval accuracy strongly depends on the star effective temperature, as cold stars radiate only little at UV-wavelengths. Only occultations of hot stars are able to provide information about ozone at high altitudes. A star with an effective temperature of around 10000K provides about the same accuracy for all altitude levels between 25 km and 60 km. A star with an effective temperature higher than 10000K provides an improved accuracy for the upper stratosphere and lower mesosphere. The best accuracy can be achieved around 60 km. At this level, a 30000K star provides a retrieval accuracy around 0.5% (visual magnitude equal to 0) or 3% (visual magnitude equal to 2).

At altitudes lower than 30 km, where the Chappuis band is used to determine ozone, the accuracy does not depend on the star temperature. The accuracy at 25 km altitude is about 3% (for a star with a visual magnitude equal to 0) to 10% (visual magnitude equal to 2).

The star effective temperature for each measured occultation is provided in the corresponding products.

1.5.4.2 Effect of the occultation geometry

1.5.4.2.1 Occultation obliquity

Small-scale irregularities of air density cause scintillations in the light intensity detected by GOMOS. These variations of light intensity do not occur simultaneously at all wavelengths, thus the spectrum of a star seen through the atmosphere is not only affected by absorption, but is also deformed by scintillations. A scintillation correction is set up from the measurements of scintillations by the two fast-photometers, allowing removal the deformation of individual transmission spectra. The scintillation correction is considered as reliable only for vertical (in the orbital plane) occultations. The incomplete scintillation correction for the oblique occultations between 20 km and 40 km is reflected by corresponding error bars (additional turbulence error added to the overall error budget).

The O3 error estimates of oblique occultations are comparable to the ones of vertical occultations. They are even lower for oblique occultations in some cases and some altitude ranges, especially in the mesosphere. This is more detailed in section 1.5.4.4 .

The occultation obliquity also impacts the HRTP validity altitude range and accuracy.

  • For vertical occultations, the validity range is 18 km-35 km (the upper limit depends on the scintillation strength). For oblique occultations, it is 20 km-30 km.
  • The best accuracy of HRTP products is reached for vertical occultations. It is estimated to be about 1 or 2 K between 18 km and 30 km for vertical and close to vertical occultations.

The obliquity for each measured occultation is provided in the corresponding products.

1.5.4.2.2 Illumination conditions

On the day-side bright limb, the limb brightness is an extended source competing with the stellar signal. Day-side occultations achieve less accurate results than night-side occultations because of a larger noise. Three other illumination conditions than full dark (night-side measurements) and bright limb (day-side measurements) conditions have been characterized: twilight, straylight, and twilight+straylight. For each occultation, the illumination conditions are inferred from the computation of the sun zenith angles at both instrument and tangent point locations.

The illumination condition for each measured occultation is provided in the corresponding products.

1.5.4.3 Altitude range of validity per species

Within the retrieved altitude range, because of some limitations of the processing and of the dependence of the accuracy with the star and the occultation geometry, it is recommended to use the vertical profiles of species in reduced altitude ranges only.

Those validity altitude ranges are detailed in Table 1.3.

Species

Validity/altitude range

Vertical resolution in the validity range (km)

O3

valid at all altitudes for hot stars;

for cold stars, data above 40 km should be considered with caution

2 in the lower stratosphere;

3 in the upper stratosphere and above

NO2

valid between 20 km and 50 km;

data at other altitudes should be considered with caution

4

NO3

valid between 25 km and 45 km, but noisy retrieved values within this altitude range;

data at other altitudes should be considered with caution

4

Aerosols

data below 10 km and above 35 km should be considered with caution

4

H2O

retrieved only up to 50 km; 

in this altitude range, poor results, with large noise

2 in the lower stratosphere;

4 in the upper stratosphere

O2

valid at all altitudes; noise is observed on some profiles

3 in the lower stratosphere;

5 in the upper stratosphere

HRTP

18-35 km for vertical occultations;

20-30 km for oblique occultations

0.2

Table 1.3: Validity assessment and vertical resolution by altitude range and by species.

1.5.4.4 Error estimates

The error estimates provided in the GOMOS products are calculated without taking into account the modelling errors and assuming normal Gaussian error statistics. Those error estimates are not fully validated. The specification of the modelling errors is still an ongoing activity, and the addition of the modelling errors to the error budget is currently under investigation.

We present here vertical profiles of the median of the relative error estimates for the local density of several species, from GOMOS night-time measurements during 2003. Results are presented by category of star brightness, by category of star temperature and for two categories of verticality occultation, in order to highlight the possible impact of the star characteristics and of the occultation geometry on the accuracy of the vertical profiles.

The three categories of stars according to their brightness are defined by range of visual magnitude (Mv):

  • bright stars for Mv lower than 0.8;
  • stars of medium brightness for Mv between 0.8 and 2.0;
  • dim stars for Mv higher than 2.0.

The three categories of stars according to their temperature are defined as:

  • cold stars for T lower than 6000 K;
  • stars of medium temperature with T between 6000 K and 10000 K;
  • hot stars for T higher than 10000 K.

The two categories of occultations according to their obliquity are defined as:

  • occultations very close to the vertical, with a verticality value between -5° and +5° ("V" category in the figure legend or vertical occultations in the text hereafter);
  • oblique occultations, with a verticality between 55° and 65° ("O" category in the figure legend).

1.5.4.4.1 Error estimates of O3 local density

The comparison of the average profiles of O3 error estimates for the oblique occultations and for the vertical occultations shows that generally values for the oblique occultations are comparable to values for the vertical occultations. In some cases and within some altitude ranges, especially in the mesosphere, the error estimates of the oblique occultations are even lower than the error estimates of the vertical occultations.

This is illustrated in Figure 1.8 , plotting the vertical profiles of average O3 error estimates by category of star T and of occultation obliquity, for different categories of star magnitude. The accuracy degradation for cold stars in the upper stratosphere and in the mesosphere is obvious. As expected for O3 at altitudes lower than 30 km, the average value of the error estimates are similar in most cases for all categories of star temperature (for typical and dim stars). At these levels, the average error estimates show lower values for brighter stars than for dimmer stars. For stars of medium or of low magnitude, the minimum relative error estimates up to 25 km-30 km for the oblique occultations are lower than for the vertical ones for most categories of star T. For hot stars of medium or low magnitude, the accuracy of oblique occultations at 90 km is similar to the accuracy around 30km; it is very good between 50 km and 70 km (and similar to the precision of the cross-section). For hot stars of medium and low magnitude, the relative error is lower in the mesosphere for the oblique occultations than for the vertical ones. For hotter and brightest stars, the relative error estimate for the oblique occultations is excellent around 20 km. However, between about 30 km or 35 km and 40 km, the error estimates for the vertical occultations are lower than for the oblique occultations, illustrating the effect of the additional turbulence error on the oblique occultations.

Figure 1.8: Vertical profiles of the median of the relative error estimates for O3 local density, from GOMOS night-time measurements during 2003; each plot corresponds to a category of star brightness (bright stars of visual magnitude lower than 0.8; typical stars of visual magnitude between 0.8 and 2.0; dim stars of visual magnitude higher than 2.0). Results on each plot are given by category of star T (hot stars of T > 10000K; typical stars of T between 6000K and 10000K; cold stars of T lower than 6000K) and by occultation obliquity (V for close to vertical occultations; O for oblique occultations; see text for details). The number of profiles used to calculate the median profile for each category is given in the curve label.

1.5.4.4.2 Error estimates of NO2 local density

Figure 1.9: Vertical profiles of the median of the relative error estimates for NO2 local density, from GOMOS night-time measurements during 2003; each plot corresponds to a category of star T (hot stars of T > 10000K; typical stars of T between 6000K and 10000K; cold stars of T lower than 6000K). Results on each plot are given by category of star brightness (bright stars of visual magnitude lower than 0.8; typical stars of visual magnitude between 0.8 and 2.0; dim stars of visual magnitude higher than 2.0) and by occultation obliquity (V for close to vertical occultations; O for oblique occultations; see text for details).  The number of profiles used to calculate the median profile for each category is given in the curve label.

Figu re 1.10: Vertical profiles of the median of the relative error estimates for NO2 local density, from GOMOS night-time measurements during 2003; each plot corresponds to a category of star brightness (bright stars of visual magnitude lower than 0.8; typical stars of visual magnitude between 0.8 and 2.0; dim stars of visual magnitude higher than 2.0). Results on each plot are given by category of star T (hot stars of T > 10000K; typical stars of T between 6000K and 10000K; cold stars of T lower than 6000K) and by occultation obliquity (V for close to vertical occultations; O for oblique occultations; see text for details).  The number of profiles used to calculate the median profile for each category is given in the curve label.

1.5.4.4.3 Error estimates of NO3 local density

Figure 1.11: Vertical profiles of the median of the relative error estimates for NO3 local density, from GOMOS night-time measurements during 2003; each plot corresponds to a category of star T (hot stars of T > 10000K; typical stars of T between 6000K and 10000K; cold stars of T lower than 6000K). Results on each plot are given by category of star brightness (bright stars of visual magnitude lower than 0.8; typical stars of visual magnitude between 0.8 and 2.0; dim stars of visual magnitude higher than 2.0) and by occultation obliquity (V for close to vertical occultations; O for oblique occultations; see text for details).  The number of profiles used to calculate the median profile for each category is given in the curve label.

Figure 1.12: Vertical profiles of the median of the relative error estimates for NO3 local density, from GOMOS night-time measurements during 2003; each plot corresponds to a category of star brightness (bright stars of visual magnitude lower than 0.8; typical stars of visual magnitude between 0.8 and 2.0; dim stars of visual magnitude higher than 2.0). Results on each plot are given by category of star T (hot stars of T > 10000K; typical stars of T between 6000K and 10000K; cold stars of T lower than 6000K) and by occultation obliquity (V for close to vertical occultations; O for oblique occultations; see text for details).  The number of profiles used to calculate the median profile for each category is given in the curve label.